a scenario calculator for effects of grazing land management on carbon stocks in australian...

18
Environmental Modelling & Software 18 (2003) 627–644 www.elsevier.com/locate/envsoft A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands M.J. Hill a,, R. Braaten a11 , G.M. McKeon b a Bureau of Rural Sciences and CRC for Greenhouse Accounting, Edmund Barton Building, Barton, PO Box 858, Canberra, ACT 2600, Australia b Queensland Centre for Climate Applications and CRC for Greenhouse Accounting, Gate 4, 80 Meiers Road, Indooroopilly, Queensland, Australia Received 2 August 2002; received in revised form 7 January 2003; accepted 12 February 2003 Abstract Management of grazing lands has been included in land-based mechanisms for reduction of greenhouse gas emissions. This paper describes a spatial system for scenario analysis of the effect of changes in grazing management on rangeland carbon balances. The system is based on identification of alternative biophysical carbon states and incorporates the effects of management changes and socio-economic and cultural barriers to changes. The management factors include grazing pressure, fire management, control or spread of woody weeds and introduction of browse shrubs. The impact of these factors may be influenced by frequency of good and poor growth years, frequency of droughts and be discounted due to social and economic barriers to adoption. The system appeared to plausibly represent the rangeland responses to management when tested for responses to climate variability and changes in stocking rate, and the impact of prescribed burning in the Tropical Woodlands. Responses are highly sensitive to the knowledge- based estimates of proportion of area in zones for each carbon state and the value of the relative carbon index for each state. A complete and ecologically sound representation of this simple model of carbon state dynamics and climate/vegetation interactions is needed to ensure that scenario analysis is useful and valid for scoping studies. Crown copyright 2003 Published by Elsevier Science Ltd. All rights reserved. Keywords: Carbon sequestration; Rangelands; Spatial tool; State and transition model Software availability Name of software: Range-ASSESS Developers: Michael Hill, Robert Braaten and Greg McKeon Contact address: Bureau of Rural Sciences, PO Box 858, Canberra, ACT 2600, Australia Telephone: +61-2-6272-5317 Fax: +61-2-6272-5161 Email: [email protected] Availability: expected to be implemented on the website for the CRC for Greenhouse Accounting http://:www.greenhouse.crc.org.au/ within 6 months. Unlikely to be available for distribution Corresponding author. Tel.: +61-2-6272-3933; fax: +61-2-6272- 5161. E-mail address: [email protected] (M.J. Hill). 1 Current address: Department of Land and Water Conservation, GPO Box 39, Sydney, NSW 2001, Australia. 1364-8152/03/$ - see front matter Crown copyright 2003 Published by Elsevier Science Ltd. All rights reserved. doi:10.1016/S1364-8152(03)00050-1 due to the sensitive nature of Greenhouse car- bon issues. 1. Introduction The analysis of the impacts of human activities on carbon stocks and fluxes has become an important topic for research given the latest projections for the impacts of greenhouse gas emissions on the global climate and terrestrial ecosystems (IPCC, 2001). Models are a pri- mary tool in this analysis and include every level of sophistication from Global Climate Models to simple spreadsheet-based calculators. Carbon accounting is at various levels of development across the world (e.g., AGO, 2002; Brack and Richards, 2002; Maclaren, 1999), and uncertainty about the global carbon budget (Francois et al., 2002) and the extent of climate change still exists (Pittock, 2002; Schneider, 2002). Under cur-

Upload: mj-hill

Post on 05-Jul-2016

224 views

Category:

Documents


6 download

TRANSCRIPT

Page 1: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

Environmental Modelling amp Software 18 (2003) 627ndash644wwwelseviercomlocateenvsoft

A scenario calculator for effects of grazing land management oncarbon stocks in Australian rangelands

MJ Hill alowast R Braatena 1 1 GM McKeonb

a Bureau of Rural Sciences and CRC for Greenhouse Accounting Edmund Barton Building Barton PO Box 858 Canberra ACT 2600Australia

b Queensland Centre for Climate Applications and CRC for Greenhouse Accounting Gate 4 80 Meiers Road Indooroopilly QueenslandAustralia

Received 2 August 2002 received in revised form 7 January 2003 accepted 12 February 2003

Abstract

Management of grazing lands has been included in land-based mechanisms for reduction of greenhouse gas emissions This paperdescribes a spatial system for scenario analysis of the effect of changes in grazing management on rangeland carbon balances Thesystem is based on identification of alternative biophysical carbon states and incorporates the effects of management changes andsocio-economic and cultural barriers to changes The management factors include grazing pressure fire management control orspread of woody weeds and introduction of browse shrubs The impact of these factors may be influenced by frequency of goodand poor growth years frequency of droughts and be discounted due to social and economic barriers to adoption The systemappeared to plausibly represent the rangeland responses to management when tested for responses to climate variability and changesin stocking rate and the impact of prescribed burning in the Tropical Woodlands Responses are highly sensitive to the knowledge-based estimates of proportion of area in zones for each carbon state and the value of the relative carbon index for each state Acomplete and ecologically sound representation of this simple model of carbon state dynamics and climatevegetation interactionsis needed to ensure that scenario analysis is useful and valid for scoping studiesCrown copyright 2003 Published by Elsevier Science Ltd All rights reserved

Keywords Carbon sequestration Rangelands Spatial tool State and transition model

Software availabilityName of software Range-ASSESSDevelopers Michael Hill Robert Braaten and Greg

McKeonContact address Bureau of Rural Sciences PO Box

858 Canberra ACT 2600 AustraliaTelephone+61-2-6272-5317Fax +61-2-6272-5161Email michaelhillbrsgovauAvailability expected to be implemented on the website

for the CRC for Greenhouse Accountinghttpwwwgreenhousecrcorgauwithin 6months Unlikely to be available for distribution

lowast Corresponding author Tel+61-2-6272-3933 fax+61-2-6272-5161

E-mail address michaelhillbrsgovau (MJ Hill)1 Current address Department of Land and Water Conservation

GPO Box 39 Sydney NSW 2001 Australia

1364-815203$ - see front matter Crown copyright 2003 Published by Elsevier Science Ltd All rights reserveddoi101016S1364-8152(03)00050-1

due to the sensitive nature of Greenhouse car-bon issues

1 Introduction

The analysis of the impacts of human activities oncarbon stocks and fluxes has become an important topicfor research given the latest projections for the impactsof greenhouse gas emissions on the global climate andterrestrial ecosystems (IPCC 2001) Models are a pri-mary tool in this analysis and include every level ofsophistication from Global Climate Models to simplespreadsheet-based calculators Carbon accounting is atvarious levels of development across the world (egAGO 2002 Brack and Richards 2002 Maclaren1999) and uncertainty about the global carbon budget(Francois et al 2002) and the extent of climate changestill exists (Pittock 2002 Schneider 2002) Under cur-

628 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

rent circumstances analytical systems that can take theavailable data within a defined level of confidence anderror and provide potential scenario analysis for com-plex human-biophysical interactions involved in globalcarbon management can be a valuable intermediate toolwhile knowledge and mechanistic representation of sys-tems dynamics is improved

Spatial decision support tools have been developed formany applications With complex human-biophysicalsystems problems often require the integration of manydifferent types of data from many different sourcesMulti-criteria assessment and evaluation (MCAE) seeksto represent the significance of many different data formswithin standardised ratings or indices and often uses sim-ple additive or multiplicative models with and withoutweighting to develop suitability capability diversity orhealth assessments for natural systems (eg Veitch1997 Jones et al 1997) Such multi-criteria systems cantake diverse data knowledge and factors and integratethese with quantitative measurements of the state of ter-restrial bio-geophysical systems for the purpose of sim-ple scenario calculation

There is a current need to evaluate options for range-land management based on best available estimatesbecause of the international importance of accounting forcarbon stock changes and greenhouse gas emissions Asscientists gain more accurate estimates of rangeland car-bon storage and sink potential there is a need to developsystems frameworks within which to apply these resultsas they become available This paper describesimplementation and some examples of responses for aspatial system for analysis of the impact of changes inmanagement on rangeland carbon balances The systemis implemented using ASSESS (A System for SelectingSuitable Sites) which represents an implementationMCAE using the ArcInfo GIS to provide spatial displayand menu panels for site selection and land use decisionfunctions (Veitch and Bowyer 1996) The approach tothe problem is based on a conceptual model developedby Stafford-Smith et al (1997) that describes a processfor assessment of biophysical socio-economic and cul-tural factors affecting changes in rangeland managementthat may result in increased carbon storage

2 ASSESSmdashA System for Selecting Suitable Sites

ASSESS was developed by the Natural ResourceInformation Centre (NRIC) of the Bureau of RuralSciences (Veitch and Bowyer 1996) ASSESS is basedon an approach pioneered in the GEM system by Daviset al (1988) and developed into ARX (Whigham andDavis 1989) one of the only explicitly spatial expertsystem designs to become operational The system isconstructed in Arc Macro Language the programminglanguage built into the ArcInfo GIS ASSESS is a user-

friendly interface to the full functionality of the ArcInfoGrid module for manipulating raster data Using theASSESS framework any series of Grid commands formanipulating and displaying rasters can be assigned toobjects such as menus buttons checkboxes and sliderbars

Applications with ASSESS have principally involvedMCAE for natural resource management problems suchas selecting a suitable disposal site for Australiarsquos low-level radioactive waste (Veitch and Bowyer 1996Veitch 1997) analysis of soil suitability in the MurrayDarling Basin and assessment of catchment condition forthe intensive land use zone of Australia (Braaten et al2001 Walker et al 2002) It has also been used to createa scenario calculator to quantify the effects of introduc-tion of trees into agricultural landscapes in order toreduce subsurface water flow and mitigate against dry-land salinity It is in a mixture of calculator and MCEAroles that the ASSESS system has been used to createRange-ASSESS a scenario analysis system for evalu-ation of management effects on carbon sequestration inAustralian rangelands The concept of this rangelandsscenario analysis system is realisable within any pro-gramming environment with suitable interface and spa-tial operational capability Therefore whilst ASSESSwas a convenient platform for this work the emphasishere is on the systems analysis of how to formally calcu-late the impact of rangeland management optionsASSESS has provided the computing platform to dem-onstrate ldquoproof-of-conceptrdquo

3 Rangelands and carbon sequestration

Grazing land management has been included as oneof the options under Article 34 for Annex 1 parties toaccount for anthropogenic greenhouse gas emissions bysources and removals by sinks under recent agreementson the Kyoto Protocol (UNFCC 2001) In Australiagrazing of rangelands is the most extensive land use andchanges in rangelands management could have a sig-nificant impact on the countryrsquos carbon balance (Bakeret al 2000) Implementation of practices with positivecarbon storage outcomes may have a major bearing onthe economics of livestock production and may be lim-ited by social and cultural factors Carbon gains couldbe obtained from increases or reduction in losses ofplant biomass and from accumulation or reduction inlosses of soil carbon An estimate for the USA suggeststhat improved management of rangelands could providegains of 01ndash03 t ha1 year1 and avoid loses of 43Mtyear (Schuman et al 2002) However there is con-siderable uncertainty surrounding the magnitude andreliability of management-induced sinks in arid andsemi-arid rangelands These environments are highlysensitive to climatic cycles that may create periods of

629MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

nil gain or drought and degradation episodes that couldresult in significant carbon losses (Fensham 1998McKeon and Hall 2001)

The Australian rangelands have been subjected to avariety of impacts through pastoral development in thelast 120 years (Griffin and Friedel 1985) and these mayprovide a number of opportunities for management inter-vention

31 Reserves and rehabilitation

The exclusion of livestock and feral animals and theuse of an appropriate fire regime may enable rangelandproductivity and carbon stocks to increase Benefits fromconversion of cropland to grassland under the Conser-vation Reserve Program (CRP) in the USA may be ashigh as 03 t ha1 year1 but rates may be much slowerand vary with management and locations (Burke et al1995) Effects may be less beneficial where the land-scape has already been modified by ingress of exotic andindigenous woody weeds forming a vegetation stateoffering little opportunity for further carbon gains

32 Grazing management

Heavy grazing can damage or kill palatable shrubs(Graetz and Wilson 1984) and trees and degrade theunderstorey grasslands (Morrissey 1984 McIvor et al1995) Since soil carbon content is dependent uponinputs of organic matter from senescent vegetation androot turnover destruction of this vegetation reduces orremoves inputs and leads to exposure of surface soil toerosion and increased oxidation of carbon The limiteddata on the effects of grazing on tussock density(measured as grass basal cover) and soil carbon in trop-ical woodlands suggest that soil carbon is closely asso-ciated with grass tussocks and that reductions in tussockdensity lead to expanded areas of low soil carbon in adegraded system (Northup and Brown 1999ab) Heavygrazing pressure may cause major changes in carbon dis-tribution after about 4 years representing a relativelyrapid loss (Northup and Brown 1999b) Whilst grassbasal cover can recover quickly with favourable climaticconditions and reduced grazing pressure soil carbonrecovery is slower Thus soil carbon stocks in range-lands are sensitive to the management of total grazingpressure and sustainable grazing management isimportant for avoidance of soil carbon losses Based onquantitative assessment of the condition of native pas-tures in Northern Australia it has been suggested thatreducing stocking rates to improve perennial grass basalcover could sequester 315 M t of carbon in the top 10cm of soil over a 30 year period (Ash et al 1995)

33 Fire

Fire is a key factor in Australiarsquos rangelands andoften an annual event in the Tropical and Sub-tropicalWoodlands of northern Australia Fire has a complexhistorical record of influence in the Australian landscape(Craig 1999) through Aboriginal burning practices(Kimber 1983) the influence on species persistence(Morrison et al 1995) and changes resulting from Euro-pean occupation and disruption to Aboriginal life(Griffin and Friedel 1985) The role of fire in determin-ing pre-settlement vegetation cover is controversial asboth indigenous management and lightning have beenmajor sources of ignition (Pyne 1991) Some authors(eg Hodgkinson et al 1984) state that the success ofthe early pastoral industry was largely dependent uponthe grass-dominant communities created by these count-less fires Fire management provides opportunities tocontrol fire susceptible woody vegetation reduce highfuel loads and stimulate grass growth In northern Aus-tralia wildfire frequency and intensity may be so severeas to cause tree death and reduced recruitment (Williamset al 1999) In a simulation study Howden et al (1999)found that there were large differences in carbon seques-tration between an ungrazed-never burnt scenario (highcarbon store) and a grazing-annual burning scenario inmulga grasslands of south-west Queensland (low carbonstore) due to the negative effects of burning on thewoody biomass Hence management (suppression) ofwildfires by prescribed burning to reduce fuel loadsmight have benefits to carbon storage However interac-tion between grazing fuel loads woody weeds fire fre-quency and livestock production systems are complexand detailed understanding of the dynamics across spec-ies associations is lacking

34 Woody weeds and thickening

Woody weed ingress and thickening of woody plantsmay occur in a variety of rangeland ecosystems (egmulga Harrington et al 1984ab tropical areas Griceand Brown 1999 sub-tropical eucalypt woodlands Bur-rows et al 2002) reducing pasture production (eg Bur-rows et al 1990) Woody weeds show a variety of adap-tation mechanisms and sensitivities to fire andmechanical intervention (Fensham 1998 Grice andBrown 1999 Burrows et al 2002 Henry et al 2003)Whether the result of invasion or just regrowth thicken-ing poses challenges to carbon accounting since categ-orisation as a natural or anthropogenic effect is problem-atical (Gifford and Howden 2001) Once land isincluded in Kyoto-based accounting then all changesmust be accounted for indefinitely and understanding thedynamics of exotic and native woody weeds becomesimportant

630 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

35 Exotic woody browse shrubs

Establishment of exotic and native woody browseplants and fodder trees in grazing systems might result insubstantial increases in above-ground biomass (Sampsonand Scholes 2000) The tactical use of browse to reducegrazing pressure on associated pasture may increase car-bon stocks through increased perennial plant density(eg grass basal cover) over the whole grazing property(Lauder 2000ab) However the use of fodder trees suchas mulga (Acacia aneura) to feed animals in extendeddroughts can result in substantial decline in tree biomassFurthermore the retention of these stock on pasture atthe break of drought reduces the regeneration of grassesand shrubs potentially leading to a decline in soil carbonand an overall decline in carbon stocks

The above description of rangeland managementoptions highlights the wide variation in potential changesof carbon stocks and greenhouse gas emissions and thepotential utility of a scenario analysis system to allowrapid evaluation of likely outcomes and the implicationsof their inclusion in global carbon accounting

4 Framework for analysis

Stafford-Smith et al (1997) described a frameworkfor the assessment of opportunities to improve carbonstorage in rangelands The approach sought strategies toavoid carbon loss and encourage carbon gains by (a)identifying biophysical options (b) identifying socio-economic opportunities (c) identifying cultural con-straints and (d) combining these to seek options withthe highest net realistic benefit (Stafford-Smith et al1997) We used this framework as the basis for thedevelopment of the scenario analysis system Thebiophysical component is based on the state and tran-sition (SampT) model of ecosystem change (Westoby etal 1989) The SampT model describes rangeland dynam-ics in terms of a set of discrete vegetation states occur-ring in one geographical location at a time (eg Jonesand Burrows 1994) The transitions between states aretriggered by management actions natural events andtheir interactions (Westoby et al 1989) In the contextof Range-ASSESS states are defined by classes of veg-etation condition and composition that correspond to sig-nificant differences in storage of carbon in biomassandor soil Implementation of this approach requiressome GIS-based analysis to define the spatial extent ofrangeland vegetation associations with different currentcarbon stocks It also requires the collection of data andknowledge about rangeland ecosystems in order todefine carbon states and potential transitions Further itrequires an assessment of relationships between benefitsin terms of carbon storage economic benefit to land useand socialcultural feasibility In our analysis most

emphasis is placed on the biophysical system Whilstsocio-economic and cultural factors are important at thisstage they have only been addressed in a rudimentaryway because of their complexity and inherent uncer-tainty

5 Rangeland regionalisation

In order to obtain a relatively simple but workablesub-division of Australiarsquos rangelands for assessment ofcarbon sequestration we used the Atlas of AustralianVegetation (AUSLIG 1990) to create eight communityzones approximating those defined by Harrington et al(1984a Table 1 Fig 1) In some cases such as the Trop-ical and Sub-tropical Woodlands the derived zone doesnot exactly match that mapped in Mott and Tothill(1984) as there were some difficulties in making a gen-eralised translation from the Atlas classification to matchthe Harrington zones The overall rangeland zone wasdefined conservatively The area therefore excludesmuch of the native grassland in central Queensland andnorthern NSW where rangeland is intermingled withmore intensive land uses This simplified our task inreconstructing the Harrington zones Future versions willprovide a more accurate rendition of the rangelands asdefined by the extent of native grasslands For examplethe operational spatial model of rangeland productionAussie GRASS (Carter et al 2000) is based on 185 pas-ture communities identified by individual successionalvegetation states However at this lsquoproof-of-conceptrsquostage the lumping of pasture communities was neces-sary

6 Populating a carbon state and transitionstructure for the rangeland zones

The SampT models required the following inputs

1 current and possible carbon states and the proportionof each zone in each state

2 carbon indices describing the magnitude of stocks ineach state relative to an original pre-settlementstate and

3 possible transitions between states and the drivers ofthese changes

Rangeland experts were gathered for a two-day work-shop in Canberra (Hill et al 2002) The experts wereasked to construct zone-specific SampT models define areaproportion for carbon states within each zone populatethe models with indices of carbon storage relative to theoriginal state identify the transitions and list the driversof change for the eight rangeland communities (Table 1Fig 1) In some cases these areas had to be split into

631MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 1Rangeland zones used in Range-ASSESS based on Harrington et al (1984a) and constructed using the Atlas of Australian Resources (AUSLIG1990) Carnahan classification of pre-settlement Australian vegetation

Zone Description Reference

Mitchell Grasslands Astrebla spp grasslands on cracking clays soils (Orr and Holmes 1984)Mallee Mallee eucalypts with diverse understoreys (Noble 1984)Tropical and Subtropical Woodlands Eucalyptus spp overstorey four different grassland associations (Mott and Tothill 1984)Arid Mulga Acacia aneura woodlands with four different understorey communities (Morrissey 1984)Hummock Grasslands Spinifex grasses Triodia spp and Plectrachne spp (Griffin 1984)Saltbush and Bluebush Chenopod shrublands (Graetz and Wilson 1984)Semi-Arid Woodlands Made up of a number of communities poplar boxmdashCallitris shrub (Harrington et al 1984b)

woodlands mulga low woodlands Acacia shrub thickets poplar box-mulga shrub woodlands and rosewood-belah shrub woodlands

Central Arid Woodlands Mosaic of communities connected to landform its effect on water (Foran 1984)distribution and soil type

Fig 1 Range-ASSESS interface showing the rangeland regionalisation into eight zones based on Harrington et al (1984ab) and the menu forapplying management changes

subsets to cover major within-zone differences The ran-geland zonation based on Harrington et al (1984a)proved to be readily recognisable by the range expertsand was a key to the success of the process Fig 2 showsan example of the SampT models for Mitchell Grasslandsand Arid Mulga Examples of the indices and driversderived from the workshop process are shown in Tables2 and 3

7 Current carbon stocks

Current carbon stocks were derived by applying theknowledge-based carbon indices for each carbon state in

each zone to simulated steady-state pre-settlement car-bon stock data These estimated pre-settlement carbonstocks were derived from the VAST 10 (Vegetation AndSoil carbon Transfer Barrett 2001 Barrett et al 2001)model which predicts the magnitude and uncertainty ofsteady state net primary productivity biomass lit-termass soil-C stocks and mean residence time of car-bon for the continental terrestrial biosphere of AustraliaVAST 10 consists of a set of statistical models cali-brated by a high quality dataset of observations from thepublished literature including soil bulk density and soilcarbon depth profiles and incorporating informationfrom continental rasterized data sets of climate soil andvegetation The version of VAST used here is the

632 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 2 State and transition models for the Mitchell Grasslands andthe Arid Mulga

empirical steady state version in which the rate of changeof C stocks is assumed to be zero VAST 10 depicts thespatial distribution of stocks under conditions of minimaldisturbance A full description of the VAST 10 modelis given in Hill et al (2002) and Barrett (2001) Barrettet al 2001 Soil properties are not taken into account

Table 2States and relative carbon indices for the Mitchell Grasslands and Arid Mulga The carbon indices represent an index of carbon levels relative topre-European carbon stocks

State Type Area NT-WA Area Qld Soil C Index Biomass C Index

Mitchell GrasslandsG1 Perennial grassland 08 08 10 10G2 Annual grassland 02 015 10 02G3 Annual grassland with woody invasion (Acacia nilotica) 0 005 08 100

Arid Mulga WA areaM1 Mulga with low shrubs and grasses 04ndash05 na 10 10M2 Mulga without shrubs and grasses 03 na 085 07M3 Sheet eroded 01 na 06 0

Table 3Transitions and drivers of change between Mitchell Grassland statesand Arid Mulga states

Transitions Drivers

Mitchell GrasslandsG1 to G2 Heavy grazing and droughtG2 to G1 Reduced grazing and rainG2 to G3 Seed introduction with grazing and no fireG3 to G2 No occurrenceG1 to G3 Seed introduction with grazing and no fireG3 to G1 No occurrence

Arid MulgaM1 to M2 Grazing and droughtM2 to M1 Reduced grazing and rainM2 to M3 Heavy grazing and droughtM3 to M1 Mechanical intervention and rain

in determination of soil carbon change in Range-ASSESS as this would necessitate having many moresub-zones to correspond to soil classes within veg-etation zones

8 Drivers of changemdashmanagement factors

We gathered together spatial data from a wide rangeof sources to attempt to capture quantitative or qualitat-ive description of the major drivers of rangeland changeidentified at the workshop (Table 4)

81 Grazing pressurestocking rate

Grazing pressure is represented by the sum of thestocking rates of all grazing animalsmdashlivestock feralgrazers rabbits and kangaroos Data layers for thesegroups of grazing animals were constructed from a var-iety of categorical and quantitative data sourcesmdashsheepand cattle from agricultural census data as dry sheepequivalents at SLA level converted to 5 km grid celllayers and feral and native animals all as 5 km grid celldata layers based upon previously published mapsadjusted and augmented with other published data (Table

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 2: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

628 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

rent circumstances analytical systems that can take theavailable data within a defined level of confidence anderror and provide potential scenario analysis for com-plex human-biophysical interactions involved in globalcarbon management can be a valuable intermediate toolwhile knowledge and mechanistic representation of sys-tems dynamics is improved

Spatial decision support tools have been developed formany applications With complex human-biophysicalsystems problems often require the integration of manydifferent types of data from many different sourcesMulti-criteria assessment and evaluation (MCAE) seeksto represent the significance of many different data formswithin standardised ratings or indices and often uses sim-ple additive or multiplicative models with and withoutweighting to develop suitability capability diversity orhealth assessments for natural systems (eg Veitch1997 Jones et al 1997) Such multi-criteria systems cantake diverse data knowledge and factors and integratethese with quantitative measurements of the state of ter-restrial bio-geophysical systems for the purpose of sim-ple scenario calculation

There is a current need to evaluate options for range-land management based on best available estimatesbecause of the international importance of accounting forcarbon stock changes and greenhouse gas emissions Asscientists gain more accurate estimates of rangeland car-bon storage and sink potential there is a need to developsystems frameworks within which to apply these resultsas they become available This paper describesimplementation and some examples of responses for aspatial system for analysis of the impact of changes inmanagement on rangeland carbon balances The systemis implemented using ASSESS (A System for SelectingSuitable Sites) which represents an implementationMCAE using the ArcInfo GIS to provide spatial displayand menu panels for site selection and land use decisionfunctions (Veitch and Bowyer 1996) The approach tothe problem is based on a conceptual model developedby Stafford-Smith et al (1997) that describes a processfor assessment of biophysical socio-economic and cul-tural factors affecting changes in rangeland managementthat may result in increased carbon storage

2 ASSESSmdashA System for Selecting Suitable Sites

ASSESS was developed by the Natural ResourceInformation Centre (NRIC) of the Bureau of RuralSciences (Veitch and Bowyer 1996) ASSESS is basedon an approach pioneered in the GEM system by Daviset al (1988) and developed into ARX (Whigham andDavis 1989) one of the only explicitly spatial expertsystem designs to become operational The system isconstructed in Arc Macro Language the programminglanguage built into the ArcInfo GIS ASSESS is a user-

friendly interface to the full functionality of the ArcInfoGrid module for manipulating raster data Using theASSESS framework any series of Grid commands formanipulating and displaying rasters can be assigned toobjects such as menus buttons checkboxes and sliderbars

Applications with ASSESS have principally involvedMCAE for natural resource management problems suchas selecting a suitable disposal site for Australiarsquos low-level radioactive waste (Veitch and Bowyer 1996Veitch 1997) analysis of soil suitability in the MurrayDarling Basin and assessment of catchment condition forthe intensive land use zone of Australia (Braaten et al2001 Walker et al 2002) It has also been used to createa scenario calculator to quantify the effects of introduc-tion of trees into agricultural landscapes in order toreduce subsurface water flow and mitigate against dry-land salinity It is in a mixture of calculator and MCEAroles that the ASSESS system has been used to createRange-ASSESS a scenario analysis system for evalu-ation of management effects on carbon sequestration inAustralian rangelands The concept of this rangelandsscenario analysis system is realisable within any pro-gramming environment with suitable interface and spa-tial operational capability Therefore whilst ASSESSwas a convenient platform for this work the emphasishere is on the systems analysis of how to formally calcu-late the impact of rangeland management optionsASSESS has provided the computing platform to dem-onstrate ldquoproof-of-conceptrdquo

3 Rangelands and carbon sequestration

Grazing land management has been included as oneof the options under Article 34 for Annex 1 parties toaccount for anthropogenic greenhouse gas emissions bysources and removals by sinks under recent agreementson the Kyoto Protocol (UNFCC 2001) In Australiagrazing of rangelands is the most extensive land use andchanges in rangelands management could have a sig-nificant impact on the countryrsquos carbon balance (Bakeret al 2000) Implementation of practices with positivecarbon storage outcomes may have a major bearing onthe economics of livestock production and may be lim-ited by social and cultural factors Carbon gains couldbe obtained from increases or reduction in losses ofplant biomass and from accumulation or reduction inlosses of soil carbon An estimate for the USA suggeststhat improved management of rangelands could providegains of 01ndash03 t ha1 year1 and avoid loses of 43Mtyear (Schuman et al 2002) However there is con-siderable uncertainty surrounding the magnitude andreliability of management-induced sinks in arid andsemi-arid rangelands These environments are highlysensitive to climatic cycles that may create periods of

629MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

nil gain or drought and degradation episodes that couldresult in significant carbon losses (Fensham 1998McKeon and Hall 2001)

The Australian rangelands have been subjected to avariety of impacts through pastoral development in thelast 120 years (Griffin and Friedel 1985) and these mayprovide a number of opportunities for management inter-vention

31 Reserves and rehabilitation

The exclusion of livestock and feral animals and theuse of an appropriate fire regime may enable rangelandproductivity and carbon stocks to increase Benefits fromconversion of cropland to grassland under the Conser-vation Reserve Program (CRP) in the USA may be ashigh as 03 t ha1 year1 but rates may be much slowerand vary with management and locations (Burke et al1995) Effects may be less beneficial where the land-scape has already been modified by ingress of exotic andindigenous woody weeds forming a vegetation stateoffering little opportunity for further carbon gains

32 Grazing management

Heavy grazing can damage or kill palatable shrubs(Graetz and Wilson 1984) and trees and degrade theunderstorey grasslands (Morrissey 1984 McIvor et al1995) Since soil carbon content is dependent uponinputs of organic matter from senescent vegetation androot turnover destruction of this vegetation reduces orremoves inputs and leads to exposure of surface soil toerosion and increased oxidation of carbon The limiteddata on the effects of grazing on tussock density(measured as grass basal cover) and soil carbon in trop-ical woodlands suggest that soil carbon is closely asso-ciated with grass tussocks and that reductions in tussockdensity lead to expanded areas of low soil carbon in adegraded system (Northup and Brown 1999ab) Heavygrazing pressure may cause major changes in carbon dis-tribution after about 4 years representing a relativelyrapid loss (Northup and Brown 1999b) Whilst grassbasal cover can recover quickly with favourable climaticconditions and reduced grazing pressure soil carbonrecovery is slower Thus soil carbon stocks in range-lands are sensitive to the management of total grazingpressure and sustainable grazing management isimportant for avoidance of soil carbon losses Based onquantitative assessment of the condition of native pas-tures in Northern Australia it has been suggested thatreducing stocking rates to improve perennial grass basalcover could sequester 315 M t of carbon in the top 10cm of soil over a 30 year period (Ash et al 1995)

33 Fire

Fire is a key factor in Australiarsquos rangelands andoften an annual event in the Tropical and Sub-tropicalWoodlands of northern Australia Fire has a complexhistorical record of influence in the Australian landscape(Craig 1999) through Aboriginal burning practices(Kimber 1983) the influence on species persistence(Morrison et al 1995) and changes resulting from Euro-pean occupation and disruption to Aboriginal life(Griffin and Friedel 1985) The role of fire in determin-ing pre-settlement vegetation cover is controversial asboth indigenous management and lightning have beenmajor sources of ignition (Pyne 1991) Some authors(eg Hodgkinson et al 1984) state that the success ofthe early pastoral industry was largely dependent uponthe grass-dominant communities created by these count-less fires Fire management provides opportunities tocontrol fire susceptible woody vegetation reduce highfuel loads and stimulate grass growth In northern Aus-tralia wildfire frequency and intensity may be so severeas to cause tree death and reduced recruitment (Williamset al 1999) In a simulation study Howden et al (1999)found that there were large differences in carbon seques-tration between an ungrazed-never burnt scenario (highcarbon store) and a grazing-annual burning scenario inmulga grasslands of south-west Queensland (low carbonstore) due to the negative effects of burning on thewoody biomass Hence management (suppression) ofwildfires by prescribed burning to reduce fuel loadsmight have benefits to carbon storage However interac-tion between grazing fuel loads woody weeds fire fre-quency and livestock production systems are complexand detailed understanding of the dynamics across spec-ies associations is lacking

34 Woody weeds and thickening

Woody weed ingress and thickening of woody plantsmay occur in a variety of rangeland ecosystems (egmulga Harrington et al 1984ab tropical areas Griceand Brown 1999 sub-tropical eucalypt woodlands Bur-rows et al 2002) reducing pasture production (eg Bur-rows et al 1990) Woody weeds show a variety of adap-tation mechanisms and sensitivities to fire andmechanical intervention (Fensham 1998 Grice andBrown 1999 Burrows et al 2002 Henry et al 2003)Whether the result of invasion or just regrowth thicken-ing poses challenges to carbon accounting since categ-orisation as a natural or anthropogenic effect is problem-atical (Gifford and Howden 2001) Once land isincluded in Kyoto-based accounting then all changesmust be accounted for indefinitely and understanding thedynamics of exotic and native woody weeds becomesimportant

630 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

35 Exotic woody browse shrubs

Establishment of exotic and native woody browseplants and fodder trees in grazing systems might result insubstantial increases in above-ground biomass (Sampsonand Scholes 2000) The tactical use of browse to reducegrazing pressure on associated pasture may increase car-bon stocks through increased perennial plant density(eg grass basal cover) over the whole grazing property(Lauder 2000ab) However the use of fodder trees suchas mulga (Acacia aneura) to feed animals in extendeddroughts can result in substantial decline in tree biomassFurthermore the retention of these stock on pasture atthe break of drought reduces the regeneration of grassesand shrubs potentially leading to a decline in soil carbonand an overall decline in carbon stocks

The above description of rangeland managementoptions highlights the wide variation in potential changesof carbon stocks and greenhouse gas emissions and thepotential utility of a scenario analysis system to allowrapid evaluation of likely outcomes and the implicationsof their inclusion in global carbon accounting

4 Framework for analysis

Stafford-Smith et al (1997) described a frameworkfor the assessment of opportunities to improve carbonstorage in rangelands The approach sought strategies toavoid carbon loss and encourage carbon gains by (a)identifying biophysical options (b) identifying socio-economic opportunities (c) identifying cultural con-straints and (d) combining these to seek options withthe highest net realistic benefit (Stafford-Smith et al1997) We used this framework as the basis for thedevelopment of the scenario analysis system Thebiophysical component is based on the state and tran-sition (SampT) model of ecosystem change (Westoby etal 1989) The SampT model describes rangeland dynam-ics in terms of a set of discrete vegetation states occur-ring in one geographical location at a time (eg Jonesand Burrows 1994) The transitions between states aretriggered by management actions natural events andtheir interactions (Westoby et al 1989) In the contextof Range-ASSESS states are defined by classes of veg-etation condition and composition that correspond to sig-nificant differences in storage of carbon in biomassandor soil Implementation of this approach requiressome GIS-based analysis to define the spatial extent ofrangeland vegetation associations with different currentcarbon stocks It also requires the collection of data andknowledge about rangeland ecosystems in order todefine carbon states and potential transitions Further itrequires an assessment of relationships between benefitsin terms of carbon storage economic benefit to land useand socialcultural feasibility In our analysis most

emphasis is placed on the biophysical system Whilstsocio-economic and cultural factors are important at thisstage they have only been addressed in a rudimentaryway because of their complexity and inherent uncer-tainty

5 Rangeland regionalisation

In order to obtain a relatively simple but workablesub-division of Australiarsquos rangelands for assessment ofcarbon sequestration we used the Atlas of AustralianVegetation (AUSLIG 1990) to create eight communityzones approximating those defined by Harrington et al(1984a Table 1 Fig 1) In some cases such as the Trop-ical and Sub-tropical Woodlands the derived zone doesnot exactly match that mapped in Mott and Tothill(1984) as there were some difficulties in making a gen-eralised translation from the Atlas classification to matchthe Harrington zones The overall rangeland zone wasdefined conservatively The area therefore excludesmuch of the native grassland in central Queensland andnorthern NSW where rangeland is intermingled withmore intensive land uses This simplified our task inreconstructing the Harrington zones Future versions willprovide a more accurate rendition of the rangelands asdefined by the extent of native grasslands For examplethe operational spatial model of rangeland productionAussie GRASS (Carter et al 2000) is based on 185 pas-ture communities identified by individual successionalvegetation states However at this lsquoproof-of-conceptrsquostage the lumping of pasture communities was neces-sary

6 Populating a carbon state and transitionstructure for the rangeland zones

The SampT models required the following inputs

1 current and possible carbon states and the proportionof each zone in each state

2 carbon indices describing the magnitude of stocks ineach state relative to an original pre-settlementstate and

3 possible transitions between states and the drivers ofthese changes

Rangeland experts were gathered for a two-day work-shop in Canberra (Hill et al 2002) The experts wereasked to construct zone-specific SampT models define areaproportion for carbon states within each zone populatethe models with indices of carbon storage relative to theoriginal state identify the transitions and list the driversof change for the eight rangeland communities (Table 1Fig 1) In some cases these areas had to be split into

631MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 1Rangeland zones used in Range-ASSESS based on Harrington et al (1984a) and constructed using the Atlas of Australian Resources (AUSLIG1990) Carnahan classification of pre-settlement Australian vegetation

Zone Description Reference

Mitchell Grasslands Astrebla spp grasslands on cracking clays soils (Orr and Holmes 1984)Mallee Mallee eucalypts with diverse understoreys (Noble 1984)Tropical and Subtropical Woodlands Eucalyptus spp overstorey four different grassland associations (Mott and Tothill 1984)Arid Mulga Acacia aneura woodlands with four different understorey communities (Morrissey 1984)Hummock Grasslands Spinifex grasses Triodia spp and Plectrachne spp (Griffin 1984)Saltbush and Bluebush Chenopod shrublands (Graetz and Wilson 1984)Semi-Arid Woodlands Made up of a number of communities poplar boxmdashCallitris shrub (Harrington et al 1984b)

woodlands mulga low woodlands Acacia shrub thickets poplar box-mulga shrub woodlands and rosewood-belah shrub woodlands

Central Arid Woodlands Mosaic of communities connected to landform its effect on water (Foran 1984)distribution and soil type

Fig 1 Range-ASSESS interface showing the rangeland regionalisation into eight zones based on Harrington et al (1984ab) and the menu forapplying management changes

subsets to cover major within-zone differences The ran-geland zonation based on Harrington et al (1984a)proved to be readily recognisable by the range expertsand was a key to the success of the process Fig 2 showsan example of the SampT models for Mitchell Grasslandsand Arid Mulga Examples of the indices and driversderived from the workshop process are shown in Tables2 and 3

7 Current carbon stocks

Current carbon stocks were derived by applying theknowledge-based carbon indices for each carbon state in

each zone to simulated steady-state pre-settlement car-bon stock data These estimated pre-settlement carbonstocks were derived from the VAST 10 (Vegetation AndSoil carbon Transfer Barrett 2001 Barrett et al 2001)model which predicts the magnitude and uncertainty ofsteady state net primary productivity biomass lit-termass soil-C stocks and mean residence time of car-bon for the continental terrestrial biosphere of AustraliaVAST 10 consists of a set of statistical models cali-brated by a high quality dataset of observations from thepublished literature including soil bulk density and soilcarbon depth profiles and incorporating informationfrom continental rasterized data sets of climate soil andvegetation The version of VAST used here is the

632 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 2 State and transition models for the Mitchell Grasslands andthe Arid Mulga

empirical steady state version in which the rate of changeof C stocks is assumed to be zero VAST 10 depicts thespatial distribution of stocks under conditions of minimaldisturbance A full description of the VAST 10 modelis given in Hill et al (2002) and Barrett (2001) Barrettet al 2001 Soil properties are not taken into account

Table 2States and relative carbon indices for the Mitchell Grasslands and Arid Mulga The carbon indices represent an index of carbon levels relative topre-European carbon stocks

State Type Area NT-WA Area Qld Soil C Index Biomass C Index

Mitchell GrasslandsG1 Perennial grassland 08 08 10 10G2 Annual grassland 02 015 10 02G3 Annual grassland with woody invasion (Acacia nilotica) 0 005 08 100

Arid Mulga WA areaM1 Mulga with low shrubs and grasses 04ndash05 na 10 10M2 Mulga without shrubs and grasses 03 na 085 07M3 Sheet eroded 01 na 06 0

Table 3Transitions and drivers of change between Mitchell Grassland statesand Arid Mulga states

Transitions Drivers

Mitchell GrasslandsG1 to G2 Heavy grazing and droughtG2 to G1 Reduced grazing and rainG2 to G3 Seed introduction with grazing and no fireG3 to G2 No occurrenceG1 to G3 Seed introduction with grazing and no fireG3 to G1 No occurrence

Arid MulgaM1 to M2 Grazing and droughtM2 to M1 Reduced grazing and rainM2 to M3 Heavy grazing and droughtM3 to M1 Mechanical intervention and rain

in determination of soil carbon change in Range-ASSESS as this would necessitate having many moresub-zones to correspond to soil classes within veg-etation zones

8 Drivers of changemdashmanagement factors

We gathered together spatial data from a wide rangeof sources to attempt to capture quantitative or qualitat-ive description of the major drivers of rangeland changeidentified at the workshop (Table 4)

81 Grazing pressurestocking rate

Grazing pressure is represented by the sum of thestocking rates of all grazing animalsmdashlivestock feralgrazers rabbits and kangaroos Data layers for thesegroups of grazing animals were constructed from a var-iety of categorical and quantitative data sourcesmdashsheepand cattle from agricultural census data as dry sheepequivalents at SLA level converted to 5 km grid celllayers and feral and native animals all as 5 km grid celldata layers based upon previously published mapsadjusted and augmented with other published data (Table

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 3: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

629MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

nil gain or drought and degradation episodes that couldresult in significant carbon losses (Fensham 1998McKeon and Hall 2001)

The Australian rangelands have been subjected to avariety of impacts through pastoral development in thelast 120 years (Griffin and Friedel 1985) and these mayprovide a number of opportunities for management inter-vention

31 Reserves and rehabilitation

The exclusion of livestock and feral animals and theuse of an appropriate fire regime may enable rangelandproductivity and carbon stocks to increase Benefits fromconversion of cropland to grassland under the Conser-vation Reserve Program (CRP) in the USA may be ashigh as 03 t ha1 year1 but rates may be much slowerand vary with management and locations (Burke et al1995) Effects may be less beneficial where the land-scape has already been modified by ingress of exotic andindigenous woody weeds forming a vegetation stateoffering little opportunity for further carbon gains

32 Grazing management

Heavy grazing can damage or kill palatable shrubs(Graetz and Wilson 1984) and trees and degrade theunderstorey grasslands (Morrissey 1984 McIvor et al1995) Since soil carbon content is dependent uponinputs of organic matter from senescent vegetation androot turnover destruction of this vegetation reduces orremoves inputs and leads to exposure of surface soil toerosion and increased oxidation of carbon The limiteddata on the effects of grazing on tussock density(measured as grass basal cover) and soil carbon in trop-ical woodlands suggest that soil carbon is closely asso-ciated with grass tussocks and that reductions in tussockdensity lead to expanded areas of low soil carbon in adegraded system (Northup and Brown 1999ab) Heavygrazing pressure may cause major changes in carbon dis-tribution after about 4 years representing a relativelyrapid loss (Northup and Brown 1999b) Whilst grassbasal cover can recover quickly with favourable climaticconditions and reduced grazing pressure soil carbonrecovery is slower Thus soil carbon stocks in range-lands are sensitive to the management of total grazingpressure and sustainable grazing management isimportant for avoidance of soil carbon losses Based onquantitative assessment of the condition of native pas-tures in Northern Australia it has been suggested thatreducing stocking rates to improve perennial grass basalcover could sequester 315 M t of carbon in the top 10cm of soil over a 30 year period (Ash et al 1995)

33 Fire

Fire is a key factor in Australiarsquos rangelands andoften an annual event in the Tropical and Sub-tropicalWoodlands of northern Australia Fire has a complexhistorical record of influence in the Australian landscape(Craig 1999) through Aboriginal burning practices(Kimber 1983) the influence on species persistence(Morrison et al 1995) and changes resulting from Euro-pean occupation and disruption to Aboriginal life(Griffin and Friedel 1985) The role of fire in determin-ing pre-settlement vegetation cover is controversial asboth indigenous management and lightning have beenmajor sources of ignition (Pyne 1991) Some authors(eg Hodgkinson et al 1984) state that the success ofthe early pastoral industry was largely dependent uponthe grass-dominant communities created by these count-less fires Fire management provides opportunities tocontrol fire susceptible woody vegetation reduce highfuel loads and stimulate grass growth In northern Aus-tralia wildfire frequency and intensity may be so severeas to cause tree death and reduced recruitment (Williamset al 1999) In a simulation study Howden et al (1999)found that there were large differences in carbon seques-tration between an ungrazed-never burnt scenario (highcarbon store) and a grazing-annual burning scenario inmulga grasslands of south-west Queensland (low carbonstore) due to the negative effects of burning on thewoody biomass Hence management (suppression) ofwildfires by prescribed burning to reduce fuel loadsmight have benefits to carbon storage However interac-tion between grazing fuel loads woody weeds fire fre-quency and livestock production systems are complexand detailed understanding of the dynamics across spec-ies associations is lacking

34 Woody weeds and thickening

Woody weed ingress and thickening of woody plantsmay occur in a variety of rangeland ecosystems (egmulga Harrington et al 1984ab tropical areas Griceand Brown 1999 sub-tropical eucalypt woodlands Bur-rows et al 2002) reducing pasture production (eg Bur-rows et al 1990) Woody weeds show a variety of adap-tation mechanisms and sensitivities to fire andmechanical intervention (Fensham 1998 Grice andBrown 1999 Burrows et al 2002 Henry et al 2003)Whether the result of invasion or just regrowth thicken-ing poses challenges to carbon accounting since categ-orisation as a natural or anthropogenic effect is problem-atical (Gifford and Howden 2001) Once land isincluded in Kyoto-based accounting then all changesmust be accounted for indefinitely and understanding thedynamics of exotic and native woody weeds becomesimportant

630 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

35 Exotic woody browse shrubs

Establishment of exotic and native woody browseplants and fodder trees in grazing systems might result insubstantial increases in above-ground biomass (Sampsonand Scholes 2000) The tactical use of browse to reducegrazing pressure on associated pasture may increase car-bon stocks through increased perennial plant density(eg grass basal cover) over the whole grazing property(Lauder 2000ab) However the use of fodder trees suchas mulga (Acacia aneura) to feed animals in extendeddroughts can result in substantial decline in tree biomassFurthermore the retention of these stock on pasture atthe break of drought reduces the regeneration of grassesand shrubs potentially leading to a decline in soil carbonand an overall decline in carbon stocks

The above description of rangeland managementoptions highlights the wide variation in potential changesof carbon stocks and greenhouse gas emissions and thepotential utility of a scenario analysis system to allowrapid evaluation of likely outcomes and the implicationsof their inclusion in global carbon accounting

4 Framework for analysis

Stafford-Smith et al (1997) described a frameworkfor the assessment of opportunities to improve carbonstorage in rangelands The approach sought strategies toavoid carbon loss and encourage carbon gains by (a)identifying biophysical options (b) identifying socio-economic opportunities (c) identifying cultural con-straints and (d) combining these to seek options withthe highest net realistic benefit (Stafford-Smith et al1997) We used this framework as the basis for thedevelopment of the scenario analysis system Thebiophysical component is based on the state and tran-sition (SampT) model of ecosystem change (Westoby etal 1989) The SampT model describes rangeland dynam-ics in terms of a set of discrete vegetation states occur-ring in one geographical location at a time (eg Jonesand Burrows 1994) The transitions between states aretriggered by management actions natural events andtheir interactions (Westoby et al 1989) In the contextof Range-ASSESS states are defined by classes of veg-etation condition and composition that correspond to sig-nificant differences in storage of carbon in biomassandor soil Implementation of this approach requiressome GIS-based analysis to define the spatial extent ofrangeland vegetation associations with different currentcarbon stocks It also requires the collection of data andknowledge about rangeland ecosystems in order todefine carbon states and potential transitions Further itrequires an assessment of relationships between benefitsin terms of carbon storage economic benefit to land useand socialcultural feasibility In our analysis most

emphasis is placed on the biophysical system Whilstsocio-economic and cultural factors are important at thisstage they have only been addressed in a rudimentaryway because of their complexity and inherent uncer-tainty

5 Rangeland regionalisation

In order to obtain a relatively simple but workablesub-division of Australiarsquos rangelands for assessment ofcarbon sequestration we used the Atlas of AustralianVegetation (AUSLIG 1990) to create eight communityzones approximating those defined by Harrington et al(1984a Table 1 Fig 1) In some cases such as the Trop-ical and Sub-tropical Woodlands the derived zone doesnot exactly match that mapped in Mott and Tothill(1984) as there were some difficulties in making a gen-eralised translation from the Atlas classification to matchthe Harrington zones The overall rangeland zone wasdefined conservatively The area therefore excludesmuch of the native grassland in central Queensland andnorthern NSW where rangeland is intermingled withmore intensive land uses This simplified our task inreconstructing the Harrington zones Future versions willprovide a more accurate rendition of the rangelands asdefined by the extent of native grasslands For examplethe operational spatial model of rangeland productionAussie GRASS (Carter et al 2000) is based on 185 pas-ture communities identified by individual successionalvegetation states However at this lsquoproof-of-conceptrsquostage the lumping of pasture communities was neces-sary

6 Populating a carbon state and transitionstructure for the rangeland zones

The SampT models required the following inputs

1 current and possible carbon states and the proportionof each zone in each state

2 carbon indices describing the magnitude of stocks ineach state relative to an original pre-settlementstate and

3 possible transitions between states and the drivers ofthese changes

Rangeland experts were gathered for a two-day work-shop in Canberra (Hill et al 2002) The experts wereasked to construct zone-specific SampT models define areaproportion for carbon states within each zone populatethe models with indices of carbon storage relative to theoriginal state identify the transitions and list the driversof change for the eight rangeland communities (Table 1Fig 1) In some cases these areas had to be split into

631MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 1Rangeland zones used in Range-ASSESS based on Harrington et al (1984a) and constructed using the Atlas of Australian Resources (AUSLIG1990) Carnahan classification of pre-settlement Australian vegetation

Zone Description Reference

Mitchell Grasslands Astrebla spp grasslands on cracking clays soils (Orr and Holmes 1984)Mallee Mallee eucalypts with diverse understoreys (Noble 1984)Tropical and Subtropical Woodlands Eucalyptus spp overstorey four different grassland associations (Mott and Tothill 1984)Arid Mulga Acacia aneura woodlands with four different understorey communities (Morrissey 1984)Hummock Grasslands Spinifex grasses Triodia spp and Plectrachne spp (Griffin 1984)Saltbush and Bluebush Chenopod shrublands (Graetz and Wilson 1984)Semi-Arid Woodlands Made up of a number of communities poplar boxmdashCallitris shrub (Harrington et al 1984b)

woodlands mulga low woodlands Acacia shrub thickets poplar box-mulga shrub woodlands and rosewood-belah shrub woodlands

Central Arid Woodlands Mosaic of communities connected to landform its effect on water (Foran 1984)distribution and soil type

Fig 1 Range-ASSESS interface showing the rangeland regionalisation into eight zones based on Harrington et al (1984ab) and the menu forapplying management changes

subsets to cover major within-zone differences The ran-geland zonation based on Harrington et al (1984a)proved to be readily recognisable by the range expertsand was a key to the success of the process Fig 2 showsan example of the SampT models for Mitchell Grasslandsand Arid Mulga Examples of the indices and driversderived from the workshop process are shown in Tables2 and 3

7 Current carbon stocks

Current carbon stocks were derived by applying theknowledge-based carbon indices for each carbon state in

each zone to simulated steady-state pre-settlement car-bon stock data These estimated pre-settlement carbonstocks were derived from the VAST 10 (Vegetation AndSoil carbon Transfer Barrett 2001 Barrett et al 2001)model which predicts the magnitude and uncertainty ofsteady state net primary productivity biomass lit-termass soil-C stocks and mean residence time of car-bon for the continental terrestrial biosphere of AustraliaVAST 10 consists of a set of statistical models cali-brated by a high quality dataset of observations from thepublished literature including soil bulk density and soilcarbon depth profiles and incorporating informationfrom continental rasterized data sets of climate soil andvegetation The version of VAST used here is the

632 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 2 State and transition models for the Mitchell Grasslands andthe Arid Mulga

empirical steady state version in which the rate of changeof C stocks is assumed to be zero VAST 10 depicts thespatial distribution of stocks under conditions of minimaldisturbance A full description of the VAST 10 modelis given in Hill et al (2002) and Barrett (2001) Barrettet al 2001 Soil properties are not taken into account

Table 2States and relative carbon indices for the Mitchell Grasslands and Arid Mulga The carbon indices represent an index of carbon levels relative topre-European carbon stocks

State Type Area NT-WA Area Qld Soil C Index Biomass C Index

Mitchell GrasslandsG1 Perennial grassland 08 08 10 10G2 Annual grassland 02 015 10 02G3 Annual grassland with woody invasion (Acacia nilotica) 0 005 08 100

Arid Mulga WA areaM1 Mulga with low shrubs and grasses 04ndash05 na 10 10M2 Mulga without shrubs and grasses 03 na 085 07M3 Sheet eroded 01 na 06 0

Table 3Transitions and drivers of change between Mitchell Grassland statesand Arid Mulga states

Transitions Drivers

Mitchell GrasslandsG1 to G2 Heavy grazing and droughtG2 to G1 Reduced grazing and rainG2 to G3 Seed introduction with grazing and no fireG3 to G2 No occurrenceG1 to G3 Seed introduction with grazing and no fireG3 to G1 No occurrence

Arid MulgaM1 to M2 Grazing and droughtM2 to M1 Reduced grazing and rainM2 to M3 Heavy grazing and droughtM3 to M1 Mechanical intervention and rain

in determination of soil carbon change in Range-ASSESS as this would necessitate having many moresub-zones to correspond to soil classes within veg-etation zones

8 Drivers of changemdashmanagement factors

We gathered together spatial data from a wide rangeof sources to attempt to capture quantitative or qualitat-ive description of the major drivers of rangeland changeidentified at the workshop (Table 4)

81 Grazing pressurestocking rate

Grazing pressure is represented by the sum of thestocking rates of all grazing animalsmdashlivestock feralgrazers rabbits and kangaroos Data layers for thesegroups of grazing animals were constructed from a var-iety of categorical and quantitative data sourcesmdashsheepand cattle from agricultural census data as dry sheepequivalents at SLA level converted to 5 km grid celllayers and feral and native animals all as 5 km grid celldata layers based upon previously published mapsadjusted and augmented with other published data (Table

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 4: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

630 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

35 Exotic woody browse shrubs

Establishment of exotic and native woody browseplants and fodder trees in grazing systems might result insubstantial increases in above-ground biomass (Sampsonand Scholes 2000) The tactical use of browse to reducegrazing pressure on associated pasture may increase car-bon stocks through increased perennial plant density(eg grass basal cover) over the whole grazing property(Lauder 2000ab) However the use of fodder trees suchas mulga (Acacia aneura) to feed animals in extendeddroughts can result in substantial decline in tree biomassFurthermore the retention of these stock on pasture atthe break of drought reduces the regeneration of grassesand shrubs potentially leading to a decline in soil carbonand an overall decline in carbon stocks

The above description of rangeland managementoptions highlights the wide variation in potential changesof carbon stocks and greenhouse gas emissions and thepotential utility of a scenario analysis system to allowrapid evaluation of likely outcomes and the implicationsof their inclusion in global carbon accounting

4 Framework for analysis

Stafford-Smith et al (1997) described a frameworkfor the assessment of opportunities to improve carbonstorage in rangelands The approach sought strategies toavoid carbon loss and encourage carbon gains by (a)identifying biophysical options (b) identifying socio-economic opportunities (c) identifying cultural con-straints and (d) combining these to seek options withthe highest net realistic benefit (Stafford-Smith et al1997) We used this framework as the basis for thedevelopment of the scenario analysis system Thebiophysical component is based on the state and tran-sition (SampT) model of ecosystem change (Westoby etal 1989) The SampT model describes rangeland dynam-ics in terms of a set of discrete vegetation states occur-ring in one geographical location at a time (eg Jonesand Burrows 1994) The transitions between states aretriggered by management actions natural events andtheir interactions (Westoby et al 1989) In the contextof Range-ASSESS states are defined by classes of veg-etation condition and composition that correspond to sig-nificant differences in storage of carbon in biomassandor soil Implementation of this approach requiressome GIS-based analysis to define the spatial extent ofrangeland vegetation associations with different currentcarbon stocks It also requires the collection of data andknowledge about rangeland ecosystems in order todefine carbon states and potential transitions Further itrequires an assessment of relationships between benefitsin terms of carbon storage economic benefit to land useand socialcultural feasibility In our analysis most

emphasis is placed on the biophysical system Whilstsocio-economic and cultural factors are important at thisstage they have only been addressed in a rudimentaryway because of their complexity and inherent uncer-tainty

5 Rangeland regionalisation

In order to obtain a relatively simple but workablesub-division of Australiarsquos rangelands for assessment ofcarbon sequestration we used the Atlas of AustralianVegetation (AUSLIG 1990) to create eight communityzones approximating those defined by Harrington et al(1984a Table 1 Fig 1) In some cases such as the Trop-ical and Sub-tropical Woodlands the derived zone doesnot exactly match that mapped in Mott and Tothill(1984) as there were some difficulties in making a gen-eralised translation from the Atlas classification to matchthe Harrington zones The overall rangeland zone wasdefined conservatively The area therefore excludesmuch of the native grassland in central Queensland andnorthern NSW where rangeland is intermingled withmore intensive land uses This simplified our task inreconstructing the Harrington zones Future versions willprovide a more accurate rendition of the rangelands asdefined by the extent of native grasslands For examplethe operational spatial model of rangeland productionAussie GRASS (Carter et al 2000) is based on 185 pas-ture communities identified by individual successionalvegetation states However at this lsquoproof-of-conceptrsquostage the lumping of pasture communities was neces-sary

6 Populating a carbon state and transitionstructure for the rangeland zones

The SampT models required the following inputs

1 current and possible carbon states and the proportionof each zone in each state

2 carbon indices describing the magnitude of stocks ineach state relative to an original pre-settlementstate and

3 possible transitions between states and the drivers ofthese changes

Rangeland experts were gathered for a two-day work-shop in Canberra (Hill et al 2002) The experts wereasked to construct zone-specific SampT models define areaproportion for carbon states within each zone populatethe models with indices of carbon storage relative to theoriginal state identify the transitions and list the driversof change for the eight rangeland communities (Table 1Fig 1) In some cases these areas had to be split into

631MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 1Rangeland zones used in Range-ASSESS based on Harrington et al (1984a) and constructed using the Atlas of Australian Resources (AUSLIG1990) Carnahan classification of pre-settlement Australian vegetation

Zone Description Reference

Mitchell Grasslands Astrebla spp grasslands on cracking clays soils (Orr and Holmes 1984)Mallee Mallee eucalypts with diverse understoreys (Noble 1984)Tropical and Subtropical Woodlands Eucalyptus spp overstorey four different grassland associations (Mott and Tothill 1984)Arid Mulga Acacia aneura woodlands with four different understorey communities (Morrissey 1984)Hummock Grasslands Spinifex grasses Triodia spp and Plectrachne spp (Griffin 1984)Saltbush and Bluebush Chenopod shrublands (Graetz and Wilson 1984)Semi-Arid Woodlands Made up of a number of communities poplar boxmdashCallitris shrub (Harrington et al 1984b)

woodlands mulga low woodlands Acacia shrub thickets poplar box-mulga shrub woodlands and rosewood-belah shrub woodlands

Central Arid Woodlands Mosaic of communities connected to landform its effect on water (Foran 1984)distribution and soil type

Fig 1 Range-ASSESS interface showing the rangeland regionalisation into eight zones based on Harrington et al (1984ab) and the menu forapplying management changes

subsets to cover major within-zone differences The ran-geland zonation based on Harrington et al (1984a)proved to be readily recognisable by the range expertsand was a key to the success of the process Fig 2 showsan example of the SampT models for Mitchell Grasslandsand Arid Mulga Examples of the indices and driversderived from the workshop process are shown in Tables2 and 3

7 Current carbon stocks

Current carbon stocks were derived by applying theknowledge-based carbon indices for each carbon state in

each zone to simulated steady-state pre-settlement car-bon stock data These estimated pre-settlement carbonstocks were derived from the VAST 10 (Vegetation AndSoil carbon Transfer Barrett 2001 Barrett et al 2001)model which predicts the magnitude and uncertainty ofsteady state net primary productivity biomass lit-termass soil-C stocks and mean residence time of car-bon for the continental terrestrial biosphere of AustraliaVAST 10 consists of a set of statistical models cali-brated by a high quality dataset of observations from thepublished literature including soil bulk density and soilcarbon depth profiles and incorporating informationfrom continental rasterized data sets of climate soil andvegetation The version of VAST used here is the

632 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 2 State and transition models for the Mitchell Grasslands andthe Arid Mulga

empirical steady state version in which the rate of changeof C stocks is assumed to be zero VAST 10 depicts thespatial distribution of stocks under conditions of minimaldisturbance A full description of the VAST 10 modelis given in Hill et al (2002) and Barrett (2001) Barrettet al 2001 Soil properties are not taken into account

Table 2States and relative carbon indices for the Mitchell Grasslands and Arid Mulga The carbon indices represent an index of carbon levels relative topre-European carbon stocks

State Type Area NT-WA Area Qld Soil C Index Biomass C Index

Mitchell GrasslandsG1 Perennial grassland 08 08 10 10G2 Annual grassland 02 015 10 02G3 Annual grassland with woody invasion (Acacia nilotica) 0 005 08 100

Arid Mulga WA areaM1 Mulga with low shrubs and grasses 04ndash05 na 10 10M2 Mulga without shrubs and grasses 03 na 085 07M3 Sheet eroded 01 na 06 0

Table 3Transitions and drivers of change between Mitchell Grassland statesand Arid Mulga states

Transitions Drivers

Mitchell GrasslandsG1 to G2 Heavy grazing and droughtG2 to G1 Reduced grazing and rainG2 to G3 Seed introduction with grazing and no fireG3 to G2 No occurrenceG1 to G3 Seed introduction with grazing and no fireG3 to G1 No occurrence

Arid MulgaM1 to M2 Grazing and droughtM2 to M1 Reduced grazing and rainM2 to M3 Heavy grazing and droughtM3 to M1 Mechanical intervention and rain

in determination of soil carbon change in Range-ASSESS as this would necessitate having many moresub-zones to correspond to soil classes within veg-etation zones

8 Drivers of changemdashmanagement factors

We gathered together spatial data from a wide rangeof sources to attempt to capture quantitative or qualitat-ive description of the major drivers of rangeland changeidentified at the workshop (Table 4)

81 Grazing pressurestocking rate

Grazing pressure is represented by the sum of thestocking rates of all grazing animalsmdashlivestock feralgrazers rabbits and kangaroos Data layers for thesegroups of grazing animals were constructed from a var-iety of categorical and quantitative data sourcesmdashsheepand cattle from agricultural census data as dry sheepequivalents at SLA level converted to 5 km grid celllayers and feral and native animals all as 5 km grid celldata layers based upon previously published mapsadjusted and augmented with other published data (Table

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 5: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

631MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 1Rangeland zones used in Range-ASSESS based on Harrington et al (1984a) and constructed using the Atlas of Australian Resources (AUSLIG1990) Carnahan classification of pre-settlement Australian vegetation

Zone Description Reference

Mitchell Grasslands Astrebla spp grasslands on cracking clays soils (Orr and Holmes 1984)Mallee Mallee eucalypts with diverse understoreys (Noble 1984)Tropical and Subtropical Woodlands Eucalyptus spp overstorey four different grassland associations (Mott and Tothill 1984)Arid Mulga Acacia aneura woodlands with four different understorey communities (Morrissey 1984)Hummock Grasslands Spinifex grasses Triodia spp and Plectrachne spp (Griffin 1984)Saltbush and Bluebush Chenopod shrublands (Graetz and Wilson 1984)Semi-Arid Woodlands Made up of a number of communities poplar boxmdashCallitris shrub (Harrington et al 1984b)

woodlands mulga low woodlands Acacia shrub thickets poplar box-mulga shrub woodlands and rosewood-belah shrub woodlands

Central Arid Woodlands Mosaic of communities connected to landform its effect on water (Foran 1984)distribution and soil type

Fig 1 Range-ASSESS interface showing the rangeland regionalisation into eight zones based on Harrington et al (1984ab) and the menu forapplying management changes

subsets to cover major within-zone differences The ran-geland zonation based on Harrington et al (1984a)proved to be readily recognisable by the range expertsand was a key to the success of the process Fig 2 showsan example of the SampT models for Mitchell Grasslandsand Arid Mulga Examples of the indices and driversderived from the workshop process are shown in Tables2 and 3

7 Current carbon stocks

Current carbon stocks were derived by applying theknowledge-based carbon indices for each carbon state in

each zone to simulated steady-state pre-settlement car-bon stock data These estimated pre-settlement carbonstocks were derived from the VAST 10 (Vegetation AndSoil carbon Transfer Barrett 2001 Barrett et al 2001)model which predicts the magnitude and uncertainty ofsteady state net primary productivity biomass lit-termass soil-C stocks and mean residence time of car-bon for the continental terrestrial biosphere of AustraliaVAST 10 consists of a set of statistical models cali-brated by a high quality dataset of observations from thepublished literature including soil bulk density and soilcarbon depth profiles and incorporating informationfrom continental rasterized data sets of climate soil andvegetation The version of VAST used here is the

632 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 2 State and transition models for the Mitchell Grasslands andthe Arid Mulga

empirical steady state version in which the rate of changeof C stocks is assumed to be zero VAST 10 depicts thespatial distribution of stocks under conditions of minimaldisturbance A full description of the VAST 10 modelis given in Hill et al (2002) and Barrett (2001) Barrettet al 2001 Soil properties are not taken into account

Table 2States and relative carbon indices for the Mitchell Grasslands and Arid Mulga The carbon indices represent an index of carbon levels relative topre-European carbon stocks

State Type Area NT-WA Area Qld Soil C Index Biomass C Index

Mitchell GrasslandsG1 Perennial grassland 08 08 10 10G2 Annual grassland 02 015 10 02G3 Annual grassland with woody invasion (Acacia nilotica) 0 005 08 100

Arid Mulga WA areaM1 Mulga with low shrubs and grasses 04ndash05 na 10 10M2 Mulga without shrubs and grasses 03 na 085 07M3 Sheet eroded 01 na 06 0

Table 3Transitions and drivers of change between Mitchell Grassland statesand Arid Mulga states

Transitions Drivers

Mitchell GrasslandsG1 to G2 Heavy grazing and droughtG2 to G1 Reduced grazing and rainG2 to G3 Seed introduction with grazing and no fireG3 to G2 No occurrenceG1 to G3 Seed introduction with grazing and no fireG3 to G1 No occurrence

Arid MulgaM1 to M2 Grazing and droughtM2 to M1 Reduced grazing and rainM2 to M3 Heavy grazing and droughtM3 to M1 Mechanical intervention and rain

in determination of soil carbon change in Range-ASSESS as this would necessitate having many moresub-zones to correspond to soil classes within veg-etation zones

8 Drivers of changemdashmanagement factors

We gathered together spatial data from a wide rangeof sources to attempt to capture quantitative or qualitat-ive description of the major drivers of rangeland changeidentified at the workshop (Table 4)

81 Grazing pressurestocking rate

Grazing pressure is represented by the sum of thestocking rates of all grazing animalsmdashlivestock feralgrazers rabbits and kangaroos Data layers for thesegroups of grazing animals were constructed from a var-iety of categorical and quantitative data sourcesmdashsheepand cattle from agricultural census data as dry sheepequivalents at SLA level converted to 5 km grid celllayers and feral and native animals all as 5 km grid celldata layers based upon previously published mapsadjusted and augmented with other published data (Table

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 6: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

632 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 2 State and transition models for the Mitchell Grasslands andthe Arid Mulga

empirical steady state version in which the rate of changeof C stocks is assumed to be zero VAST 10 depicts thespatial distribution of stocks under conditions of minimaldisturbance A full description of the VAST 10 modelis given in Hill et al (2002) and Barrett (2001) Barrettet al 2001 Soil properties are not taken into account

Table 2States and relative carbon indices for the Mitchell Grasslands and Arid Mulga The carbon indices represent an index of carbon levels relative topre-European carbon stocks

State Type Area NT-WA Area Qld Soil C Index Biomass C Index

Mitchell GrasslandsG1 Perennial grassland 08 08 10 10G2 Annual grassland 02 015 10 02G3 Annual grassland with woody invasion (Acacia nilotica) 0 005 08 100

Arid Mulga WA areaM1 Mulga with low shrubs and grasses 04ndash05 na 10 10M2 Mulga without shrubs and grasses 03 na 085 07M3 Sheet eroded 01 na 06 0

Table 3Transitions and drivers of change between Mitchell Grassland statesand Arid Mulga states

Transitions Drivers

Mitchell GrasslandsG1 to G2 Heavy grazing and droughtG2 to G1 Reduced grazing and rainG2 to G3 Seed introduction with grazing and no fireG3 to G2 No occurrenceG1 to G3 Seed introduction with grazing and no fireG3 to G1 No occurrence

Arid MulgaM1 to M2 Grazing and droughtM2 to M1 Reduced grazing and rainM2 to M3 Heavy grazing and droughtM3 to M1 Mechanical intervention and rain

in determination of soil carbon change in Range-ASSESS as this would necessitate having many moresub-zones to correspond to soil classes within veg-etation zones

8 Drivers of changemdashmanagement factors

We gathered together spatial data from a wide rangeof sources to attempt to capture quantitative or qualitat-ive description of the major drivers of rangeland changeidentified at the workshop (Table 4)

81 Grazing pressurestocking rate

Grazing pressure is represented by the sum of thestocking rates of all grazing animalsmdashlivestock feralgrazers rabbits and kangaroos Data layers for thesegroups of grazing animals were constructed from a var-iety of categorical and quantitative data sourcesmdashsheepand cattle from agricultural census data as dry sheepequivalents at SLA level converted to 5 km grid celllayers and feral and native animals all as 5 km grid celldata layers based upon previously published mapsadjusted and augmented with other published data (Table

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 7: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

633MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 4Data layers and structure of Range-ASSESS system

Process stage Data layers and non-spatial scalars Sources

1 Select zonemdashone or all zones Rangeland zones (Harrington et al 1984ab)Modify area by State boundaries Pastoral zone (ABARE 1999)land tenure etc

State boundaries land tenure map (AUSLIG 1990)2 Change managementSheep stocking rate Sheep density in dseha Grazing pressure (Guppy unpublished)Cattle stocking rate Cattle density in dseha (Guppy unpublished)Feral grazing animal control Feral grazing animal (Wilson et al 1992 plus many others)

densityRabbit control Rabbit density in dseha (Wilson et al 1992 Neave 1999)

by 5 km gridKangaroo harvest Kangaroo density in (Pople and Grigg 1999 Caughley 1987 Short

dseha 1985)Carrying capacity in dsehamdashused with grazing (Wilson and Harrington 1984)pressure to give GRAZINDEX

Control woody weeds Density of fire susceptible and fire resistant woody (Thorp and Lynch 2000)weeds

Introduce browse shrubs Adaptation zones for Leucaena spp Tagasaste spp Estimated from published data using climate and soiland saltbush species surfaces

Introduce prescribed burning Frequency of fires (annual) From NOAA AVHRR fire scar data for 1997ndash2000(Tropical and sub-tropical (Marsden et al 2001)woodlands only)3 Assess risks Non-spatial scalars forClimate SOIIPO year types (Table 7) for growth and (McKeon and Hall 2001)

drought (DRYNESS)4 Social User specified index between 0 and 1 of social and

economic barriers to adoptionEconomicRun the scenarioNew stocks Continental 1 km data for biomass litter and soil (Barrett 1999 2001)

carbon stocks modified by scenario changesChanges from old stocks

4) Details of methods used for development of theselayers are given in Hill et al (2002) available as a pdffile at httpwwwgreenhousecrcorg

The effect of grazing pressure on carbon sequestrationmust be evaluated in relation to the carrying capacity ofthe rangeland vegetation types A quantitative basis foranalysis of this relationship between grazing pressureand carrying capacity is critical to plausible carbonsequestration estimation for rangeland managementscenarios At this proof-of-concept stage carryingcapacity is described by simple polynomial equationsfitted to the data in Wilson and Harrington (1984) com-bined with an annual rainfall layer classified into winterdominant rainfall (W) and summer dominant rainfall (S)to create a carrying capacity (C) map for Australia(Fig 3)

Winter rainfall CW 020950002423W (1)

1027e 005W2

Summer rainfall CS 014440001763S (2)

662e 006S2

However we recognise that these simplisticregression relationships over-estimate carrying capacityin the following areas

1 dry monsoonal regions where rainfall occurs in arestricted period (3ndash5 months) and nutrients also limitpasture growth (Mott et al 1985)

2 regions where woody plant density is naturally rela-tively high and competes with pasture growth foravailable water and nutrients (eg areas of semi-aridand tropical woodlands)

3 landscape units including elements of meso-scaletopographically and edaphically determined veg-etation mosaics where low soil fertility restricts pas-ture growth and reduces resilience of perennialgrasses to heavy defoliation (eg semi-arid and cen-tral arid woodlands)

Conversely carrying capacity is likely to be underesti-mated in pasture communities with high fertility and lowtree density (eg Mitchell grasslands)

Alternative estimates of safe carrying capacity include

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 8: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

634 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 3 Example layers describing stocking density and carryingcapacity that may be used in analysis of management change scenariosRainfall-based carrying capacity shown here overestimates carryingcapacity in Northern Australia

regional recommendations (eg Wilcox and McKinnon1972) formal regional calculation procedures (Condon1968 Johnston et al 1996ab) and more general model-ling approaches (Hall et al 1998) based on simulatedpasture growth and safe utilisation rates In the nextstage of development simulated pasture growth fromAussie GRASS will be used Carrying capacity will becalculated from estimates of safe utilisation rates basedon indices of soil fertility

82 Climate

The Inter-decadal Pacific Oscillation (IPO) hasrecently been described as a longer term 10ndash20 yearlsquocyclersquo underlying the shorter term influence of theSouthern Oscillation Index (SOI) on Australian rainfall

(Power et al 1999) McKeon and Hall (2001) classifiedhistorical years into six combinations of IPO and SOI(Table 5) These year types are associated with decreasesor increases in simulated growth of pasture and changesin frequency of droughts The analysis was based onaverage rainfall and simulated pasture growth (McKeonet al 1982) for nine locations associated with historicaldegradation episodes (McKeon and Hall 2001) Sevenlocations were in eastern Australia thus the averages foreach year type (Table 5) reflect the strong effects of IPOand SOI on eastern Australia

The year types are used to provide drought frequencyand relative potential for carbon accumulation or lossAt this time these data are used in a non-spatial mannerHowever values for Australia at 5 km resolution havebeen calculated with the Aussie GRASS model(McKeon and Hall 2001) and will be included in thenext stage of development Current research is estimat-ing underlying secular trends in rainfall in each year typeallowing climate change trends to also be included at alater stage

83 Fire

An annual fire frequency map was created fromNOAA AVHRR fire scar map for the period 1997ndash2000(Marsden et al 2001) Monthly fire scar data weremerged to form annual fire scar maps These annualmaps were then converted to grids with 1 km resolutionand added together to give a single layer with rastervalues corresponding to the number of fires occurring ina four year period for WA and NT and a three yearperiod for Queensland

84 Exotic woody weed encroachment

Exotic woody weed distribution was constructed frommapping of current and potential distribution of a largenumber of weeds of national significance (Thorp andLynch 2000) These data provide current distributionmaps for six exotic woody weeds Chinee apple (Pyrusprunifolia) mesquite (Prosopis spp) mimosa (Mimosapigra) Parkinsonia (Parkinsonia aculeata) prickly aca-cia (Acacia nilotica spp indica) and rubber vine(Cryptostegia grandiflora) These weeds could be sig-nificant for analysis of carbon balances if associated landis brought into Kyoto-based accounting by grazing landmanagement activities (as discussed in Section 34)Maps provide presenceabsence for Western Australiaand low medium and high density for presence in theother States on a 05 025 or 0125 square degree basis(Thorp and Lynch 2000) Data were combined to formtwo layers that indicated low medium or high densityof fire susceptible and fire resistant exotic woody weeds(Table 5)

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 9: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

635MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 5Classification of years by phase of SOI and IPO used to develop future climate scenario impacts on carbon sequestration (after Hill et al 2002)

Year Type No of years Rain deviation Growth deviation No years rainfall decile 1 Dryness index

SOI-4 IPO0 1 16 18 18 2 5SOI 4 IPO0 2 17 33 44 0 1SOI4 IPO0 3 17 14 25 3 5SOI4 IPO0 4 11 5 10 05 1SOI-4 amp 4 IPO0 5 17 12 11 1 3SOI4 amp 4 IPO0 6 30 10 11 4 5

85 Introduction of browse shrubs

The potential adaptation zones for three woodybrowse plantsmdashleucaena (Leucaena leucocephala)tagasaste (Chaemicytisus spp) and saltbush (Atriplexspp)mdashthat could be introduced into the rangelands werecreated as single category raster layers on the basis ofpublished information on climate and soil constraintsand spatial data layers for soil and climate Leucaenasuitability zone was defined by 600 mm annual rain-fall no frost soil pH60 and no risk of waterlogging(Chamberlain et al 1999) Tagasaste suitability wasdefined by the 350ndash550 mm rainfall zone on deep well-drained sandy soils (Wiley 2000) The saltbush suit-ability zone was defined by soils at risk of salinity withless than 350 mm annual rainfall and winter dominanceThe zones reported here are quite approximate and needto be refined using more precise criteria to define suit-ability in subsequent versions of Range-ASSESS

9 Transition implementation

The drivers of transition between states are each rep-resented by a rating scale from 1 to 5 relating to thedegree of stress they exert on the vegetation (Table 6)Rules are then used to control the effect of drivers onstate transitionsmdashexamples are shown for MitchellGrassland and Arid Mulga (Table 7) The mechanismsfor the five drivers currently implemented (to varyingextents) are described below

91 Grazing pressurecarrying capacity

The total grazing pressure is provided by the sum ofthe grazing animal data layers Changes in any of thegrazing factors result in a recalculation of this total graz-ing pressure The relationship between the grazing press-ure and carrying capacity a simple ratio GRAZEIN-DEX drives the transition between states in conjunctionwith the effect of climate (Table 7)

92 Drought

The drought frequency expected for each year typewas used to create a drought index DRYNESS (Table6) and enables changes in the proportions of year typesin the target period to affect drought likelihood andhence transition to a lower carbon state if stocking ratesexceeded a threshold (Table 7)

93 Woody weed spread or control

Woody weeds may be increased or decreased This isinitiated by the user by increasing the WEEDINDEXwhich changes the values in the WEEDINDEX layer andacts on the threshold in the Mitchell Grass SampT model(Table 6) At present Weedindex only operates for theMitchell Grasslands the effect of woody understorey istaken into account in the relative carbon index for rel-evant states in a number of other zones but is notdynamic in the modelling

94 Fire

In this framework we restricted our consideration offire to the influence it has on the transition betweengrassland and invasion by woody shrubs and to theeffects of control of wildfires by prescribed burning onthinningthickening of woody vegetation in the Tropicaland Sub-Tropical Woodlands zone only A high fre-quency of wild fires is assumed to cause damage towoodland biomass (Williams et al 1985) Introductionof prescribed burning is assumed to control shrub estab-lishment and diminish the risk of wildfire therebyallowing tree growth and recruitment The threshold ofwildfire frequency (FIREINDEX Table 6) used for tran-sitions between the state in which tree damage and thin-ning may occur and the state where woody biomass isstable or potentially thickening is difficult to define inthe absence of long term regional data on vegetationchange and fire frequency We have initially used rela-tively generous values in terms of potential damage orthinning of between 1 in 2 and 1 in 25 years for this

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 10: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

636 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Table 6Conversion of drivers to indices

Index Relationship Conversion

GRAZEINDEX Ratio of grazing pressure (dse) 105Carrying capacity(dse) 2=05ndash08

3 08ndash104 10ndash155 15

DRYNESS Droughts per decade 1052=05ndash083=08ndash104=10ndash15515

FIREINDEX Fires per year 102 low3=02ndash04 moderate504 high

WEEDINDEX Density of fire susceptible or fire resistant woody weeds 1=none2=low3=moderate4=high5=very high

BROWSEINDEX Biomass carbon accumulation (tha) over 10 years by browse 3=3 saltbushshrubs

5=5 Leucaena and TagasasteSOCIALINDEX User supplied 0ndash1 multiplierECONINDEX User supplied 0ndash1 multiplier

Table 7Rules for transition between carbon states for Mitchell Grasslands and Arid Mulga

Starting state Rule DescriptionMitchell Grasslands

G1 If (grazeindex ge 4 and dryness ge 4) state 1=state 2 High grazing pressure drought and low weediness pushes state 1 tostate 2

Else if (weediness ge 3) state 1=state 3 High weediness leads to state 3Else state 1=state 1

G2 If (grazeindex le 3 and dryness le 3) state 2=state 1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1

Else if (weedinex ge 3) state 2=state 3 High weediness leads to state 3Else state 2=state 2

G3 If (grazeindex ge 1) state 3 No recovery from woody weed dominance except mechanicalArid Mulga

M1 If (grazeindex ge 4 and dryness ge 4) state1=2 High grazing pressure and drought push state 1 to state 2Else state1=1

M2 If (grazeindex le 3 and dryness le 3) state2=1 Low grazing pressure and good rainfall allows state 2 to recover tostate 1 (but some understorey such as chenopod shrubs do notrecover)

Else if (grazeindex ge 4 and dryness ge 4) state 2=3 High grazing pressure and drought pushes state 2 to state 3 (Mulgais grazed and damaged by sheep goats and camels)

Else state 2=2M3 If (grazindex 1) state 3 No recovery from sheet erosion

transition for the purposes of illustrative scenarios(Table 6)

95 Introduction of browse plants

Browse plants can only be introduced within theirzone of adaptation They are assumed to then accumulate

carbon for 10 years to a maximum level(BROWSEINDEX Table 6) Agronomic evaluation ofwhere browse can be established is still in progress andsuitability zones currently used are quite approximate

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 11: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

637MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

10 Constraints on management effects

101 Climate

The frequency of occurrence of IPOSOI year typesand the percentage change in simulated pasture growth(Table 5) was used to create a weighted-average multi-plier for sequestration rate over 50 years or any shorterperiod in question The proportions of years 1ndash6 (Table5) can be changed thus influencing the weighted averagemultiplier and hence the rate of carbon storage or lossSince the year types are not random in distribution dueto their association with underlying cycles for periodsof analysis shorter than 50 years it is assumed that thenumber of possible year types diminishes in proportionto the length of the period Hence for a 20 year analysisthe average length for a single IPO phase three of thesix potential year types could occur We plan to incor-porate maps describing spatial patterns of influence forthese IPOSOI year types in the near future (McKeonand Hall 2001)

102 Socio-economic barriers to adoption andimplementation

In order to include socio-economic limits to adoptionin the scenario framework in a simple form an index foreach of socialcultural (SOCIALINDEX Table 6) andeconomic (ECONINDEX Table 6) likelihood of man-agement changes actually occurring was included Theindices allow the user to mentally incorporate all thecomplexity of adoption constraints and issues into a sin-gle scalar The two indices may be set to a value rangingfrom 0 to 1 (Table 6) Initial values are set to 10assuming that management change is able to be fullyadopted and implemented The index of economic riskof management changes occurring is intended to capturesuch risks as negative changes in the terms of trade forsheep and cattle or increases in costs of acquiring capi-tal The index of social and cultural risk provides forsocial impediments to management changes like lack oftechnical skills as well as cultural resistance to changesfor instance the opposition of some pastoralists to adop-tion of conservative stocking practices because of short-term financial pressures (Young et al 1984 Morrissey1984) These two indices are used to adjust downwardthe modelled changes in carbon status associated withmodifications to management

Improvements in this very simplified approach areneeded The indices could be explicity broken down insub-indices for each of the items listed above Howeverit would be preferable to base the sub-indices on somerelationships or correlations between economic andsocial conditions and adoption obtained from appropri-ate literature

11 Range-ASSESSmdashstructure operation andinterface

The overall structure of the Range-ASSESS system isshown in Fig 4 Carbon status is modelled using thefollowing process

1 The fundamental inputs are the vegetation zones (A)carbon layers (B) and the SampT tables (C) These dataare used to calculate present carbon stocks

2 The modelling operates on continental carbon stocklayers adjusted to current conditions using the areaproportions and relative carbon index levels providedin the state and transition models (eg Table 2)

3 The distribution of the drivers is given by the mapdata (D) These data and the IPOSOI table providevalues for the indices used to drive transitionsbetween states (E) (eg Table 7)

4 The final vegetation states are modelled on a gridcellbasis over the entire zone for each of the three poss-ible starting states

5 Changes in carbon stocks in response to managementare then calculated from changes in the area pro-portions in each carbon state using Eq (3)

Current stock (P1 PC1 hellip Pi PCi) C1 (3)

where P1ndashi are the proportions of the zone in eachstate PC1ndashi are estimates of the current carbon stockfor each state as a proportion of the undisturbed car-bon stock and C1 is the undisturbed carbon stock (orthe carbon stock for state 1 the undisturbed state)from VAST 10 carbon layers

6 A area-weighted average of the modelled carbon lay-ers one corresponding to each possible starting stateis then computed based on the pre-specified pro-portions of the area in each starting state (F)

7 Losses of soil and biomass carbon are assumed tooccur linearly over 2 years whilst gains are assumedto occur linearly over 50 years Thinning in TropicalWoodlands is assumed to occur over 100 yearsAnalysis for shorter periods results in proportionaldiscounting of the carbon change This simplified rulefor gains and losses was adopted to fit with the abruptsteps from one carbon state to another It wasassumed that degradation could occur quickly throughover-grazing and drought but recovery would beslower and dependent upon significant regenerativerainfall events which may have a frequency ofbetween 10 and 20 years in most of the rangelandsIn reality recovery is highly dependent upon the nat-ure of the degraded state the regenerative capacity ofthe dominant species and the type of conditionsrequired for regeneration Vegetation associations inthe different zones vary in their resilience under stresssuch that some systems make take much longer todegrade permanently and others may both degrade

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 12: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

638 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 4 A schema describing the structure of the Range-ASSESS system

and recover quickly with changes in stress At thistime it is not possible to provide zone specific valuesfor loss and gain periods It is possible for the userto change the degradation period for individual zonesto suit different circumstances and information

8 The change in carbon is then modified if appropri-ate by

(a) the fixed value increment due to browse introduction(b) growth adjustment from IPOSOI year typesmdashposi-

tive carbon change resulting from transitions tostates with higher biomass carbon or higher soil car-bon is adjusted based on the percentage change in

growth predicted by the weighted average ofIPOSOI year types (positive adjustments are restric-ted to periods less than 50 yearsmdashif the value of theadjustment multiplied by the time period exceeds 50then the adjustment is constrained to a value of 10as it is assumed that stocks reach an equilibrium orclimax state at 50 years and therefore cannotincrease significantly in response to favourable con-ditions

(c) an adjustment for social and economic barriers toadoption is applied to the change values (G)

The Range-ASSESS interface (Fig 1) allows users to

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 13: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

639MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

change the management of different rangeland zonesand model the effects of these management changes onthe soil and biomass carbon states The system operateson a 5 km grid cell Each cell within a zone is assignedthe same state composition Within a zone the spatialdistribution of different carbon states is not representedthis is a substantial simplification that can be refined withfurther reference to surveys and expert opinion and bet-ter data layers The system consists of four steps(Table 4)

1 Select an area for analysis (Table 4) This area caninclude all or some of the zones and be constrained byAustralian State boundaries andor land tenure type

2 View and modify management factors in the area(Table 4)

3 View or modify of the frequency of occurrence ofIPOSOI year types (Table 6) and set thesocialcultural and economic constraints to adoptionand implementation (Table 4)

4 Model the effects of management changes on carbonstatus and view the mapped and tabular summaries ofmodelled carbon stocks (Table 4)

12 A limited response analysis

The basic functioning of the system is addressed inthis analysis Firstly the system was tested to examinethe responses to changes in climate and grazing pressurekey drivers of management outcomes and to test thesystem representation of changes to fire frequency Sec-ondly the results from any scenario analysis are criti-cally dependent upon the relative proportions of zonesin different carbon states and the relative magnitude ofthe carbon index for each state derived from the knowl-edge of experts

All scenarios were run over a period of 50 yearsGrazing pressure and climate were the only drivers usedFor transitions that depended upon additional drivers(eg fire) the requirements were assumed to be satis-fied A change of state dependent on grazing pressurewas assumed to occur if grazing pressure was greaterthan carrying capacity The critical threshold for achange of state dependent upon drought was set to occurwhen DRYNESS=4 equivalent to a drought frequencygreater than one per decade

1 Climatemdashthe first set of sensitivity analyses involvedvarying the proportion of climate year types fromequal proportions of only the three driest year types(1 3 6) to equal proportions of only the wettest yeartypes (2 4 5) Each pass consisted of a 5 changein the proportions real climate sequences are non-random but for the purposes of this simulation realyear frequencies were not used (Table 6) This also

results in a change to the DRYNESS index such thatthat once dry year types dominate and averagedroughts per decade exceeds 10 DRYNESS exceedsthe threshold for triggering changes in state

2 Livestock stocking densitymdashthe second set of sensi-tivity analyses involved varying stocking rate from 0to 200 of present value in increments of 20 forthree climatic scenarios average based on historicaloccurrence of the year types dry based on equal pro-portions of only the three driest year types (1 3 6)and wet based on equal proportions of only the threewettest year types (2 4 5)

3 Use of prescribed burning for wildfire controlmdashtheoperation of the fire option for the tropical and sub-tropical woodlands was tested by applying prescribedburning to no areas to crown land only or to all landfor an average dry and wet climate

4 Relative carbon index and proportions of area in eachstarting statemdashthe sensitivity of the system to theknowledge-based inputs was tested for the Mitchellgrasslands and arid mulga The carbon indices for soiland biomass for states 2 and 3 were varied from 50to 200 of their original values in increments of 20while all other indices were held constant The pro-portions of the vegetation zone area starting in eachof the three possible states were varied from 0 to 1in intervals of 01 While one state was being variedthe other two were adjusted proportionately to pro-duce a sum of 10

13 Results and discussion

Climate simulations show significant declines in car-bon stocks under prolonged dry conditions at currentstocking rates (Fig 5) As the dry year type dominatesstate transitions are triggered by the DRYNESS indexmoving systems to different soil carbon and biomass car-bon states The balance between stocking rate and carry-ing capacity layers also influences the response Underwet conditions transitions to higher carbon states are nottriggered because these transitions also require a stock-ing rate reduction and wet years do not increase growthsince the adjustment factor can never exceed 10 whenthe model is run for the full 50 years If the stocks actu-ally changed at time increments rather than in a singlestep between the start and end of the time period thena prolonged wet period would enable the maximumstock to be attained earlier However within-perioddynamics are not represented in this system

The brief rise in carbon stocks in the Tropical andSub-tropical Woodlands as conditions get drier is causedby a shift from open woodland with annual grassland towoodland with a woody understorey when DRYNESSreaches a value of 3 This effect is then gradually erodedby transition from open woodland with perennial grass-

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 14: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

640 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 5 Response of total simulated carbon storage to changes in live-stock stocking rates under wet average and dry climatic conditions

land and thinned woodland with a woody understorey toopen woodland with annual grassland when DRYNESSreaches a value of 4 as well as a discounting effect ofthe dry end value of the IPOSOI growth factor on thestate transitions (Table 7) This example illustrates thepotential for unexpected responses and the need torefine transition rules and index thresholds

Stocking rate simulations show a decline in soil car-bon stocks with increasing stocking rate with a steeperdecline under dry conditions (Fig 5) The effect acceler-ates at stocking rates above 80ndash100 of the present lev-els (Fig 5) Simulated carbon storage under a dry scen-ario is substantially lower than for average or wetconditions since transition to a lower carbon state is trig-gered by the DRYNESS index As stocking rate isincreased it exceeds carrying capacity in more areas andtriggers the state transition resulting in gradual declineGiven that carrying capacity is overestimated in the drymonsoonal areas this response could be more significantwith a more accurate carrying capacity layer There islittle difference between the average and wet scenarioseither because the areas available for regeneration underwet conditions are small or the change is impossiblewithout human intervention Allowing for the simplisticrepresentation of the system the behaviour appears tobe plausible and may be indicative of the potential fordegradation and loss of soil carbon under increasedstocking rates

With no prescribed burning (Fig 6) a significant pro-portion of state 1 (open woodland with perennialgrassland) shifts to state 4 (thinned woodland withwoody understorey) in areas where fire frequency is highresulting in lower biomass carbon It is assumed thatseed for regeneration of woody understorey has beenimported by livestock but current grazing pressure isinsufficient to reduce fuel loads With prescribed burnsthis transition is halted allowing the woodland torecover resulting in an increased carbon store Inaddition with no prescribed burning and an average orwet climate a large proportion of state 2 (open woodlandwith annual grassland) goes to state 1 because the cli-mate allows regeneration of the perennial grasses How-ever under a dry climate this transition does not occurresulting in a lower carbon store A more completeunderstanding of fire-vegetation dynamics in grazedwoodlands is needed to enable incorporation of the fireeffects on grazing resources

Figs 7 and 8 demonstrate the sensitivity of the modelto the underlying assumptions In Fig 7 the effect oncarbon storage of varying the proportions of a zone ineach carbon state is shown The responses depend on therelative carbon content of each state For the Mitchellgrasslands changes in the proportion of the area instate1 perennial grassland and state 2 annual grasslandresult in small changes in total carbon as the two statescontain similar amounts of carbon Increases in state 3woody weed infestation result in a significant increasein carbon stocks since biomass carbon is high in state 3

For the arid mulga zone while changes in state 1(mulga with low shrubs and grasses) and 2 (mulga withno understorey) have relatively small effects on carbonstorage an increase in the area of state 3 (sheet eroded)results in a large decrease These results suggest werequire accurate estimates of the initial proportion of thevegetation zone in states that involve large changes incarbon stock and accurate estimates of the magnitudeand likelihood of increases or decreases in the area of

Fig 6 Effect of prescribed burning to reduce wildfires on carbonstocks in the tropical and subtropical woodlands under wet averageand dry climatic conditions

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 15: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

641MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Fig 7 Response of simulated carbon storage to changes in the pro-portion of a rangeland zone in any carbon state for Mitchell Grasslandand Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

these states Therefore as might be expected scenariooutcomes are highly sensitive to the proportion of a zonein any carbon state and therefore also highly sensitiveto zone definition The zones used here are approxi-mations and their boundaries are inexact and highly gen-eralised since they represent an approximate aggre-gation of many Atlas of Australian Vegetation classesThe sensitivity results emphasise that this scenario calcu-lator approach is useful for identifying direction andmagnitude of changes but is unsuitable for accountingfor carbon stocks and greenhouse gas emissions

The effect of changes in relative carbon index onoverall carbon storage (Fig 8) depends on the impor-tance of the biomass or soil component of a given statein determining the overall carbon store for that zone Forexample for the Mitchell grasslands varying the soilcarbon index for state 2 (annual grassland) has a largeeffect on overall carbon as soil carbon makes up themajority of carbon stores in that zone For the aridmulga the large effect of the soil carbon index for state2 mulga with no understorey relates to the importanceof soil carbon in that state and zone Varying the biomasscarbon coefficient for state 3 (sheet eroded) has no effectas all biomass is lost in that state In the SampT modelsstates involving woody weeds are variably rated relativeto state 1 in Mitchell Grasslands with Acacia nilotica

Fig 8 Response of simulated carbon storage to changes in the rela-tive carbon index for two rangelands zones Mitchell Grasslands andthe Arid Mulga Refer to Fig 2 for SampT models for these veg-etation zones

the factor is 10 in tropical woodlands with woodyunderstorey the factor is 2 in semi-arid woodlands withEremophila spp shrubs 12 in central arid woodlandswith thickening or woody understorey the factor isbetween 12 and 2 for sub-zones Scenario outcomes cantherefore be heavily influenced by these factors if thewoody state is expanded over significant areas

For the most part the changes induced by smallchanges in areas and relative indices are modest sug-gesting that the system responds in a relatively robustfashion to errors in expert estimation of the relative car-bon indices and to real changes of carbon state

14 Conclusions

The knowledge-based approach to definition of areasand relative carbon stocks for each state proved to berelatively robust but scenario outcomes are under-standably sensitive to errors in zone definition and inrelative carbon indices However the simplicity of therepresentation of a large and complex system provides

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 16: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

642 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

a powerful means to visualise the scope of responses andassess the magnitude of possible gains and losses fromdiverse scenarios Our representation of the systembehaviour in terms of thresholds and drivers of statetransitions is highly approximate and in some casesmay be highly contestable Our purpose here is to dem-onstrate the concept Future work will correct problemswith representation of the system and provide a quanti-tative analysis of scenario outcomes within the boundsof the uncertainty and errors inherent in the system

The balance between grazing pressure and carryingcapacity and the influence of climate variability are keyfactors that determine the outcomes of scenario simula-tions with Range-ASSESS and of real outcomes in theAustralian rangelands The representation of the systemseeks to combine event-driven and continuous change(Watson et al 1996) but the representation of theseinfluences in the system is still highly imperfect Theanalysis can be improved substantially by

1 Development of a more spatially explicit carryingcapacity layer This will be provided by AussieGRASS in the near future (Carter et al 2000)

2 Assessment and revision of the operation of indicesand scalars Particularly incorporation of climateimpacts and social and economic effects earlier in theprocess rather than as simple modifiers of the finalnumber

3 A modest improvement in the definition of the range-land zones such that conversion approximations fromvegetation classifications are rationalised zones arerefined to explicitly represent understorey differenceswhilst keeping the increase in number of zones to aminimum

4 Replacement of the knowledge-based area pro-portions and relative carbon indices with measureddata with explicit definition of uncertainty and varia-bility at an appropriate moderate spatial scale

5 Realisation of a relatively simple but ecologicallysound and accepted representation of system dynam-ics including fire-grazing-vegetation-climate interac-tion within the analytical framework

With the combination of points (3) and (4) a diversityof temporal behaviours can be represented instead of therudimentary and overly simplistic representation used inthe current version

The GIS-based interface and structure is not wellsuited to iterative Monte-Carlo style modelling Theconcept presented here may be rendered in any suitableprogramming language with spatial operators and inter-face capability The concept has now been ported to theDelphi programming environment to enable refinementand detailed scenario analysis to be undertaken

Acknowledgements

This research was funded by the CooperativeResearch Centre for Greenhouse Accounting We aregrateful to Ron Hacker John Carter Rodd Dyer Marga-ret Friedel and Alexandra Van Vreeswyk for the devel-opment of state and transition models used in the scen-ario system We also thank Damian Barrett for access tocontinental carbon surfaces

References

ABARE 1999 Economic land use zones for Australia Digital dataAustralian Bureau of Agricultural and Resource Economics Can-berra

AGO 2002 National Carbon Accounting System Greenhouse GasEmissions from Land Use Change in Australia An IntegratedApplication of the National Carbon Accounting System AustralianGreenhouse Office Canberra 124 pp

Ash AJ Howden SM McIvor JG West NE 1995 Improvedrangelands management and its implications for carbon seques-tration In West NE (Ed) Proceedings of the Fifth InternationalRangeland Congress Salt Lake City Utah 23ndash28 July 1995 Vol1 Society for Range Management Denver pp 19ndash20

AUSLIG 1990 Atlas of Australian Resources Vegetation 3rd seriesvol 6 Australian Surveying and Land Information Group Depart-ment of Administrative Services Canberra

Baker B Barnett G Howden M 2000 Carbon sequestration inAustraliarsquos rangelands In Keenan R Bugg AL Ainslie H(Eds) Management Options for Carbon Sequestration in ForestAgricultural and Rangeland Ecosystems Cooperative ResearchCentre for Greenhouse Accounting Canberra pp 73ndash82

Barrett DJ 2001 VAST Model NPP dataset Australia Availablefrom httpdaacornlgovnpqnonmdashguidesvastmdashdeshtml

Barrett DJ Galbally IE Graetz RD 2001 Quantifying uncer-tainty in estimates of C emissions from above ground biomass dueto historic land-use change to cropping in Australia Global ChangeBiology 7 883ndash902

Braaten R Dowling T Walker J Veitch S 2001 A frameworkfor using indicators and environmental software systems for analy-sis of catchment condition Proceedings of the 4th InternationalSymposium on Environmental Software Systems May 22ndash25Banff Canada Environmental Software Systems 4 13ndash27

Brack CL Richards GP 2002 Carbon accounting model for forestsin Australia Environmental Pollution 116 187ndash194

Burke IC Lauenroth WK Coffin DP 1995 Soil organic matterrecovery in semiarid grasslands implications for the ConservationReserve Program Ecological Applications 5 793ndash801

Burrows WH Carter JO Scanlan JC Anderson ER 1990Management of savannas for livestock production in northeast Aus-tralia contrasts across the tree grass continuum Journal of Bioge-ography 17 503ndash512

Burrows WH Henry BK Back PV Hoffman MB Tait LJAnderson ER Menke N Danaher T Carter JO and McKeonGM 2002 Growth and carbon stock change in eucalypt wood-lands in north-east Australia ecological and greenhouse sink impli-cations Global Change Biology 8 769ndash784

Carter JO Hall WB Brook KD McKeon GM Day KA PaulCJ 2000 Aussie GRASS Australian grassland and rangelandassessment by spatial simulation In Hammer G Nicholls NMitchell C (Eds) Applications of Seasonal Climate Forecastingin Agricultural and Natural Ecosystemsmdashthe Australian Experi-ence the Netherlands Kluwer Academic Press pp 329ndash350

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 17: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

643MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Caughey G 1997 Introduction to the sheep rangelands In CaugheyG Shepherd N Short J (Eds) Kangaroos Their Ecology andManagement in the Sheep Rangelands of Australia CambridgeUniversity Press Cambridge pp 1ndash13

Chamberlain J Lambert G Middleton C 1999 Leucaena in Cen-tral Queensland DPI Note Department of Primary IndustriesQueensland Available fromhttpwwwdpiqldgovaudpinotespasturesbi99123html

Condon RW 1968 Estimation of grazing capacity on arid grazinglands In Stewart GA (Ed) Land Evaluation Papers of a CSIROSymposium organised in cooperation with UNESCO CSIRODivision of Land Research Canberra pp 112ndash124

Craig AB 1999 Fire management of rangelands in the Kimberleylow-rainfall zone a review Rangeland Journal 21 39ndash70

Davis JR Whigman P Grant IW 1988 Representing and apply-ing knowledge about spatial processes in environmental manage-ment AI Applications 2 17ndash25

Fensham RJ 1998 The influence of cattle grazing on tree mortalityafter drought in savanna woodland in north Queensland AustralianJournal of Ecology 23 405ndash407

Foran BD 1984 Central Arid Woodlands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 299ndash316

Francois L Faure H Probst J-L 2002 The global carbon cycleand its change over glacial-interglacial cycles Global and PlanetaryChange 33 viindashviii

Gifford RM Howden M 2001 Vegetation thickening in an ecologi-cal perspective significance to national greenhouse gas inventoriesEnvironmental Science amp Policy 4 59ndash72

Graetz RD Wilson AD 1984 Saltbush and Bluebush In Harring-ton GN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 209ndash222

Grice AC Brown JR 1999 Invasive shrubs in Australian tropicalrangelands past present and future In Eldridge D Freuden-berger D (Eds) People and Rangelands Building the Future Pro-ceedings of the VIth International Rangeland Congress VI Inter-national Rangeland Congress Inc Aitkenville Queensland pp603ndash604

Griffin GF 1984 Hummock Grasslands In Harrington GN Wil-son AD Young MD (Eds) Management of Australiarsquos Rangel-ands CSIRO Australia pp 271ndash284

Griffin GF Friedel MH 1985 Discontinuous change in centralAustralia some implications of major ecological events for landmanagement Journal of Arid Environments 9 63ndash80

Hall WB McKeon GM Carter JO Day KA Howden SMScanlan JC Johnston PW Burrows WH 1998 Climatechange and Queenslandrsquos grazing lands II An assessment of theimpact on animal production from native pastures Rangeland Jour-nal 20 174ndash202

Harrington GN Wilson AD Young MD (Eds) 1984a Manage-ment of Australiarsquos Rangeland CSIRO Australia 354 pp

Harrington GN Mills DMD Pressland AJ Hodgkinson KC1984b Semi-arid woodlands In Harrington GN Wilson ADYoung MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 189ndash208

Henry BK Danaher T McKeon GM Burrows WB 2003 Areview of the potential role of greenhouse gas abatement in nativevegetation management in Queenslandrsquos rangelands The Range-land Journal (in press)

Hill MJ Braaten R McKeon G Barrett D Dyer R et al 2002Range-ASSESS a spatial framework for analysis of potential forcarbon sequestration in rangelands Technical Publication No 1CRC for Greenhouse Accounting ANU Canberra 43 pp

Hodgkinson KC Harrington GN Griffin GF Noble JC YoungMD 1984 Management of vegetation with fire In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 141ndash156

Howden SM Moore JL McKeon GM Reyenga PJ CarterJO Scanlan JC 1999 Impacts of fire dynamics in the mulgawoodlands of south-west Queensland In Howden SM McKeonGM Reyenga PJ(compilers) Global Change Impacts on Aus-tralian Rangelands Report to the Australian Greenhouse OfficeCSIRO Wildlife and Ecology Working Paper Series 9909 pp23ndash32

IPCC 2001 Climate Change 2001 The Scientific BasismdashSummaryfor Policymakers and Technical Summary of the Working GroupI Report Intergovernmental Panel on Climate Change CambridgeUniversity Press

Johnston PW McKeon GM Day KA 1996a Objective lsquosafersquograzing capacities for south-west Queensland Australia develop-ment of a model for individual properties Rangeland Journal 18224ndash258

Johnston PW Tannock PR Beale IF 1996b Objective lsquosafersquograzing capacities for south-west Queensland Australia modelapplication and evaluation Rangeland Journal 18 259ndash269

Jones P Burrows WH 1994 State and transition models for rangel-ands 13 A state and transition model for the mulga zone of south-west Queensland Tropical Grasslands 28 279ndash283

Jones BK Ritters KH Wickham JD Tankersley Jr RD OrsquoNe-ill RV Chaloud DJ Smith ER Neale AC 1997 An Eco-logical Assessment of the United States Mid-Atlantic RegionUnited States Environmental Protection Agency Office ofResearch and Development Washington DC EPA600R-97130

Kimber R 1983 Black lightning Aborigines and fire in Central Aus-tralia and the Western Desert Archaeology in Oceania 18 38ndash45

Lauder A 2000a Good pastures are the key to profit In HeywoodJ Hodgkinson K Marsden S Pahl L (Eds) Graziersrsquo Experi-ences in Managing Mulga Country Department of Primary Indus-tries Brisbane pp 52ndash62

Lauder A 2000b Who does Drought Visit and When A Manual onEnjoying More Good Seasons Report for the Drought RegionalInitiative Project 2000

Maclaren P 1999 Carbon accounting and forestry a review of thesubsequent papers Environmental Science and Policy 2 93ndash99

Marsden AJ Smith RGC Craig RI Heath BC Raisbeck-Brown NA Steber MT Adams J Church KF 2001 Conti-nental mapping of fire risk across Australia using the NOAA-AVHRR In IEEE 2001 International Geoscience and RemoteSensing Proceedings University of New South Wales SydneyAustralia 9ndash13 July IEEE Piscataway NJ

McIvor JG Ash AJ Cook GD 1995 Land condition in the trop-ical tallgrass pasture lands 1 Effects on herbage production Rang-eland Journal 17 69ndash85

McKeon GM Hall WB 2001 Can seasonal climate forecastingprevent degradation of Australiarsquos grazing lands Final report forthe Climate Variability in Agriculture Program Land and WaterAustralia Canberra

McKeon GM Rickert KG Ash AJ Cooksley DG ScattiniWJ 1982 Pasture production model Proceedings of the Aus-tralian Society of Animal Production 14 201ndash204

Morrissey JG 1984 Arid Mulga Woodlands In Harrington GNWilson AD Young MD (Eds) Management of AustraliarsquosRangelands CSIRO Australia pp 285ndash298

Morrison DA Cary GJ Pengelly SM Ross DG Mullins BJThomas CR Anderson TS 1995 Effects of fire frequency onplant species composition in the Sydney region inter-fire intervaland time-since-fire Australian Journal of Ecology 20 239ndash247

Mott JJ Tothill JC 1984 Tropical and subtropical woodlands InHarrington GN Wilson AD Young MD (Eds) Managementof Australiarsquos Rangelands CSIRO Australia pp 255ndash270

Mott JJ Williams J Andrew MH Gillison AN 1985 Aus-tralian savanna ecosystems In Tothill JC Mott JJ (Eds) Ecol-ogy and Management of the Worlds Savannas Canberra Aus-tralian Academy of Science pp 56ndash82

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References
Page 18: A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands

644 MJ Hill et al Environmental Modelling amp Software 18 (2003) 627ndash644

Neave H M 1999 Rabbit Calicivirus Disease Program Report Ioverview of effects on Australian wild rabbit populations andimplications for agriculture and biodiversity A report of researchconducted by participants of the Rabbit Calicivirus Disease Moni-toring and Surveillance Program and Epidemiology Research Pro-gram Prepared for the RCD Management Group Bureau of RuralSciences Canberra 118 pp

Noble JC 1984 Mallee In Harrington GN Wilson AD YoungMD (Eds) Management of Australiarsquos Rangelands CSIRO Aus-tralia pp 223ndash240

Northup BK Brown JR 1999a Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia tussock and inter-tussock scales In Eldridge D Freudenberger D (Eds) VIthInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp120ndash121

Northup BK Brown JR 1999b Spatial distribution of soil carbonin grazed woodlands of dry tropical Australia meso-patch to com-munity scales In Eldridge D Freudenberger D (Eds) VithInternational Rangelands Congress Proceedings VIth InternationalRangelands Congress Inc Aitkenville Queensland Vol 1 pp121ndash122

Orr DM Holmes WE 1984 Mitchell grasslands In HarringtonGN Wilson AD Young MD (Eds) Management of Aus-traliarsquos Rangelands CSIRO Australia pp 241ndash254

Pittock AB 2002 What we know and donrsquo t know about climaticchange reflections on the IPCC TAR Climatic Change 53 393ndash411

Pople A Grigg G 1999 Overview of Background Information forKangaroo Management Commercial Harvesting of Kangaroos inAustralia Department of Zoology The University of QueenslandAustralia Report for Environment Australia Biodiversity GroupCanberra Available from httpwwwbiodiversityenvironmentgovauplantswildliferooroobghtm

Power SB Casey T Folland C Coleman A Mehta V 1999Inter-decadal modulation of the impact of ENSO on Australia Cli-mate Dynamics 15 319ndash423

Pyne SJ 1991 Burning Bush A Fire History of Australia HenryHolt and Co New York

Sampson RN Scholes RJ 2000 Additional human-induced activi-tiesmdasharticle 34 In Watson RT Noble IR Bolin B Ravind-ranath NH Verardo DJ Dokken DJ (Eds) Land Use Land-Use Change and Forestry A Special Report of the Intergovern-mental Panel on Climate Change Cambridge University PressCambridge pp 181ndash282

Schneider SH 2002 Can we estimate the likelihood of climaticchanges at 2100 Climatic Chang 52 441ndash451

Schuman GE Janzen HH Herrick JE 2002 Soil carbon dynam-ics and potential carbon sequestration by rangelands Environmen-tal Pollution 116 391ndash396

Short J 1985 The functional response of kangaroos sheep and rab-bits in an arid grazing system Journal of Applied Ecology 22435ndash447

Stafford-Smith M Ojima D Carter J 1997 Integrated approachesto assessing sequestration opportunities for carbon in rangelands

In Squires VR Glenn EP Ayoub AT (Eds) Combating Glo-bal Climate Change by Combating Land Degradation Proceedingsof a Workshop Nairobi Kenya 4ndash8 September 1995 UnitedNations Environment Programme Nairobi pp 305ndash326

Thorp JR Lynch R 2000 The Determination of Weeds of NationalSignificance National Weeds Strategy Executive CommitteeLaunceston 234 pp

UNFCCC 2001 Preparations for the first session of the conference ofthe parties serving as the meeting of the parties to the Kyoto Proto-col (Decision 8CP4) Matters relating to land-use land-use changeand forestry Draft decision -CP6 Land use land-use change andforestry FCCCCP2001L11Rev1 Available fromhttpwwwunfcccdecop6mdash2indexhtml

Veitch SM 1997 Land use decisions and site selection a GIS-basedapproach In Denzer R Swayne DA Schimak G (Eds)Environmental Software Systems Vol 2 Chapman amp Hall Lon-don xx

Veitch SM Bowyer JK 1996 ASSESS a GIS-based system forselecting suitable sites In Morain S Lopez Baros S (Eds) Ras-ter imagery in Geographic Information Systems Onword PressSanta Fe pp 182ndash191

Walker J Veitch S Braaten R Dowling T Guppy L 2002Catchment Condition Project Report National Land and WaterResources Audit Canberra

Watson IW Burnside DG Holm AMcR 1996 Event-driven orcontinuous which is the better model for managers RangelandJournal 18 351ndash369

Westoby M Walker BH Noy-Meir I 1989 Opportunistic man-agement for rangelands not at equilibrium Journal of Range Man-agement 42 266ndash274

Whigham PA Davis JR 1989 Modelling with an integratedGISExpert System Ninth Annual ESRI Users Conference PalmSprings CA May 1989

Wilcox DG McKinnon EA 1972 A report on the condition ofthe Gascoyne catchment Department of Agriculture Western Aus-tralia and Department of Lands and Surveys Western AustraliaPerth

Williams RJ Cook GD Gill AM Moore PHR 1999 Fireregimes fire intensity and tree survival in a tropical savanna innorthern Australia Australian Journal of Ecology 24 50ndash59

Wilson AD Harrington GN 1984 Grazing ecology and animalproduction In Harrington GN Wilson AD Young MD(Eds) Management of Australiarsquos Rangelands CSIRO Australiapp 63ndash77

Wilson G Dexter N OrsquoBrien P Bomford M 1992 Pest Animalsin Australia A Survey of Introduced Wild Animals 64 pp Bureauof Rural Resources and Kangaroo Press Canberra

Wiley T 2000 Designing the paddock layout and suitable site for atagasaste plantation Farmnote 532000 Agriculture Western Aus-tralia Available from httpwwwagricwagovauagencyPubnsfarmnote2000f05300htm

Young MD Walker PA Cocks KD 1984 Distribution of influ-ences on rangeland management In Harrington GN WilsonAD Young MD (Eds) Management of Australiarsquos RangelandsCSIRO Australia pp 333ndash346

  • A scenario calculator for effects of grazing land management on carbon stocks in Australian rangelands
    • Introduction
    • ASSESS-A System for Selecting Suitable Sites
    • Rangelands and carbon sequestration
      • Reserves and rehabilitation
      • Grazing management
      • Fire
      • Woody weeds and thickening
      • Exotic woody browse shrubs
        • Framework for analysis
        • Rangeland regionalisation
        • Populating a carbon state and transition structure for the rangeland zones
        • Current carbon stocks
        • Drivers of change-management factors
          • Grazing pressurestocking rate
          • Climate
          • Fire
          • Exotic woody weed encroachment
          • Introduction of browse shrubs
            • Transition implementation
              • Grazing pressurecarrying capacity
              • Drought
              • Woody weed spread or control
              • Fire
              • Introduction of browse plants
                • Constraints on management effects
                  • Climate
                  • Socio-economic barriers to adoption and implementation
                    • Range-ASSESS-structure operation and interface
                    • A limited response analysis
                    • Results and discussion
                    • Conclusions
                    • Acknowledgements
                      • References