research article ecosystem services in agricultural landscapes: a spatially explicit...

Research ArticleEcosystem Services in Agricultural Landscapes A SpatiallyExplicit Approach to Support Sustainable Soil Management

Mohsen Forouzangohar1 Neville D Crossman2 Richard J MacEwan3

D Dugal Wallace4 and Lauren T Bennett1

1 Department of Forest and Ecosystem Science The University of Melbourne 4 Water Street Creswick VIC 3363 Australia2 Commonwealth Scientific and Industrial Research Organisation (CSIRO) Ecosystem Sciences PMB 2 Urrbrae SA 5064 Australia3 Future Farming Systems Research Division Department of Environment amp Primary Industries PO Box 3100Bendigo Delivery Centre VIC 3554 Australia

4Agriculture Productivity Group Department of Environment amp Primary Industries 32 Lincoln Square North CarltonVIC 3053 Australia

Correspondence should be addressed to Mohsen Forouzangohar mohsenfunimelbeduau

Received 13 November 2013 Accepted 11 December 2013 Published 30 January 2014

Academic Editors E de Blas and H G Rosatto

Copyright copy 2014 Mohsen Forouzangohar et alThis is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

Soil degradation has been associated with a lack of adequate consideration of soil ecosystem services We demonstrate a broadlyapplicable method for mapping changes in the supply of two priority soil ecosystem services to support decisions about sustainableland-use configurations We used a landscape-scale study area of 302 km2 in northern Victoria south-eastern Australia which hasbeen cleared for intensive agriculture Indicators representing priority soil services (soil carbon sequestration and soil water storage)were quantified and mapped under both a current and a future 25-year land-use scenario (the latter including a greater diversity ofland uses and increased perennial crops and irrigation) We combined diverse methods including soil analysis using mid-infraredspectroscopy soil biophysical modelling and geostatistical interpolation Our analysis suggests that the future land-use scenariowould increase the landscape-level supply of both services over 25 years Soil organic carbon content and water storage to 30 cmdepth were predicted to increase by about 11 and 22 respectively Our service maps revealed the locations of hotspots as wellas potential trade-offs in service supply under new land-use configurations The study highlights the need to consider diverse landuses in sustainable management of soil services in changing agricultural landscapes

1 Introduction

That soils are fundamental to a wide range of ecosystemservices needs to be acknowledged to avoid further soil degra-dation and to identify sustainable land-use change Severalrecent conceptual works have used an ecosystem serviceapproach to highlight the importance of soils to the sustainedprosperity and welfare of humankind [1ndash4] Nonethelessdecreases in the supply of soil ecosystem services like waterquality regulation and soil structure stabilization are symp-tomatic both of ongoing loss of soil natural capital andof ongoing disregard of the full range of soil services inproduction systems [3 4]

Better integration of ecosystem service knowledge in soilmanagement systems requires holistic yet straightforward

approaches to the quantification and mapping of soil ecosys-tem service supply [5] Rutgers et al [6] and van Wijnen etal [7] are recent examples although these studies did notprovide approaches for quantifying changes in multiple soilecosystem services in response to changing managementmdashacritical knowledge gap recognised by Haines-Young et al [8]

Full consideration of all soil ecosystem services at land-scape to regional scales will rarely be possible necessitatingfirstly the identification of priority services and secondly thenomination of soil properties to represent these servicesBennett et al [9] identified 11 ldquofinalrdquo soil ecosystem servicesthat are directly utilised to benefit humans (versus supportingldquointermediaterdquo services) Priority services are identified asthose that offer the greatest benefit within that landscapecontext but that also act as likely ldquosurrogatesrdquo for the provision

Hindawi Publishing Corporatione Scientific World JournalVolume 2014 Article ID 483298 13 pageshttpdxdoiorg1011552014483298

2 The Scientific World Journal

of other services [9] All soil services are essentially aggregatesof soil processes where processes are inputs losses transfersand transformations ofmaterial and energy [2 10] For exam-ple the intermediate soil service ldquoorganic matter cyclingrdquo(nomenclature of [9]) is the result of various processes (littercomminution decomposition and humification) which areexpressed as rates Full temporal quantification of multipleprocess rates at landscape scales is not feasible Instead soilproperties (eg soil organic carbon content) can be used torepresent the end-point of processesservices (eg carbonsequestration) and changes in soil properties used as anindication of change in potential service supply [11]

Accurate prediction of changes in soil servicespropertieswill rely both on the direct measurement of properties andon the use of soil-specific biophysical models In this contextchanges in soil properties are simulated continuously inresponse to changes in soil processes as a result of interactionsamong soil water vegetation and the atmosphere [12]Thereare many soil simulation models some of which have beenrecommended for soil services assessment [13] For exampleAitkenhead et al [14] introduced theMOSESmodel for quan-titative assessment of selected ecosystem services calibratedfor a range of soils in the UK In Australia AgriculturalProduction System Simulator (APSIM) was developed as amodelling framework consisting of soil plant and manage-ment component modules to simulate diverse productionsystems specific to Australian soils and conditions [15 16]APSIM has considerable potential as a tool for modellingchanges in soil properties [13] because soil properties aresimulated continuously in response to changes in weathermanagement and vegetation [15 16] It has been appliedto the prediction of changes in soil organic carbon (SOC)content as a result of changes in land management [17ndash19] In addition APSIM has been successfully tested in thesimulation of soil water balance [20ndash22]

Soil simulation models are traditionally point-basedwhereas soil ecosystem services are best represented as con-tinuous surfaces for decision-making Interpolation is re-quired and kriging offers an objective and rigorous approachto interpolating between spatially-distributed data pointsand thus tomapping soil services [23 24]The use of kriging[25] to produce soil property maps has been demonstratedfrequently for SOC [26ndash29] Although many studies appliedthese techniques at field scales with high sample densities[26 28] some dealt with sparser sampling schemes at muchbroader scales For example Kumar and Lal [27] estimatedthe spatial distribution of SOC across an area of 117599 km2using 920 sample points (sim one sample per 128 km2) Simi-larly Zhang et al [29] produced amap of SOC for the Repub-lic of Ireland based on kriging interpolations using an averagesample density of two per 100 km2 Clearly in the contextof broad-scale changes in soil properties a key requirementis to choose a sampling intensity that accommodates costand logistical constraints but that is also sufficient forcapturing change trends at appropriate scales within acceptedlevels of error Here while demonstrating the advantages ofkriging from grid samples for regional estimations of soilproperties McBratney and Webster [24] concluded that the

best sampling strategy would be a regular grid scheme withthe highest affordable number of observations

We used a novel combination of methods includingfield sampling biophysical modelling (APSIM) and spatialinterpolations (eg kriging) to produce landscape-scalemapsindicating changes in the supply of soil services from currentto future management scenarios We selected a study area innorthern Victoria south-eastern Australia to illustrate theutility of this approach for considering priority soil servicesin decisions relating to changing configurations of land-usepractices at landscape scales We focused on two priorityservices namely carbon sequestration and water storageWe used a landholder defined scenario of future land-useconfigurations (involving increased complexitydiversity) topredict effects of land-use change on the spatial distributionof two soil properties (hereafter ldquoindicatorsrdquo) namely SOCcontent (indicating carbon sequestration) and volume ofwater stored in soil (indicating water storage) Our aims wereto quantify and map the distribution of service indicators(representing priority services) under both current and futureland-use configurations and to demonstrate the utility of ourapproach in supporting land-use decisions by highlightinglocations where management resources could be targetedto maintain or enhance the supply of priority soil services[11 30ndash32]

2 Methods

21 Approach Overview Our unique approach (Figure 1)included representative sampling of soils across the studyarea according to a geostatistically-valid design analysisof selected soil properties in a subset of reference soilsamples development of mid-infrared calibrationmodels forthe prediction of soil properties in the remaining samplesparameterising and configuring the APSIMmodel to predictindicators (representing priority services) under current andfuture land-use scenarios and applying spatial interpolations(kriging) in a GIS environment to produce maps of currentand of likely future changes in the distribution of theseindicators across the entire study area In addition a hotspotmap of soil services change was produced to help evaluate thelikely trade-offs and opportunities arising from future land-use configurations

22 Study Area This study encompasses a case-study land-scape of 30200 ha (302 km2) in northern Victoria between35∘ 24531015840 and 35∘ 41231015840 S and 143∘ 37721015840 and 143∘ 54221015840 Esouth-eastern Australia (Figure 2) The region has a warmgrassland climate (Australian Bureau of Meteorologymdashhttpwwwbomgovauiwkclimate zones) and typicallyexperiences hot and dry summers and cold winters and anaverage annual rainfall of 370mm falling mostly in winterand spring (June to November) Mean monthly minimumtemperatures range from 33∘C (August) to 184∘C (January)and mean monthly maximum temperatures range from131∘C (July) to 358∘C (January climate data based on 25-yeardata from nearby Swan Hill and Kerang weather stations)

There is no detailed soil survey of the study area Accord-ing to the Australian Soil Classification [33] and the Victorian

The Scientific World Journal 3

Study area

10 20 and 30 cm = 180 samples

Characterised 180 soil samples for selected soil properties

Parameterised APSIM soil component for 60 sampling points

Calibration set 60 soil samples for reference analysis

Prediction set 120 soil samples for MIR-PLSR analysis

Configured APSIM for two current land uses

Configured APSIM for five future land uses

Estimated service indicators atpresent under current land uses

Estimated service indicators in 25 years under future land uses

Mapped service Mapped service Mapped service indicators using kriging

for current land usesindicators using kriging

for future land usesindicator changes and

hotspots

Sampled soil at 60 points and 3 depths of 0ndash

Figure 1 Approach overview summarising the multiple steps leading to the production of indicator maps (representing priority services)under current and future land-use configurations

GMU250 geomorphology database [34] the dominant soilorders are Sodosols (or Solonetz in World Reference Base(WRB) soil classification) and Vertosols (or Vertisols inWRBsoil classification) generally in the western and eastern sidesof the study area respectively The region contains severalwater bodies some of which are recognised as part of theRamsar Convention on Wetlands [35] The Victorian LandUse Information System 2009 dataset [36] lists the principalland uses within the study area as domestic livestock grazingmixed farming and grazing and cereal dry cropping Thelandscape has been extensively cleared for agriculture andthe remaining native vegetation is highly fragmented andorrestricted to riparian zones Historical clearing and intensiveagriculture in this landscape have contributed to degradationof soil capital through erosion compaction and localisedsalinisation [37 38]

Many land parcels within the study landscape have comeunder the ownership and management of a single entity andare part of a 25-year plan to extensively reconfigure landuses for improved agricultural production and for improvedenvironmental stewardship Associated changes in thelocation and types of agricultural and restoration practiceswill directly and indirectly affect the supply of soil services[31] making this an ideal landscape for examining service

changes under likely future scenarios The 25-year plan isspecific to 6441 ha a subset of the 30200 ha study area andincludes proposed land uses on a continuum from less inten-sively managed systems (ldquoecological estaterdquo with and withoutminor grazing and eucalypt plantation) to more intensivelymanaged systems (mainly irrigated cropping limitedperennial pasture and perennial horticulture Figure 2)

23 Soil Sampling and Analysis We used a 2 km grid as thestarting point for the soil sampling design on the basis thatgrid sampling is often recommended when ordinary krigingis to be used for spatial distribution predictions [28] Such asampling scheme minimises the kriging variance that is thestandard error of prediction [39 40] A 2 km sampling gridwas proposed in this study because of its capacity to providean adequate sampling intensity within cost and logisticalconstraints When overlain on the study landscape the 2 kmgrid yielded c 60 strata (segments) that coincided withaccessible land parcels covered by the abovementioned 25-year plan One sampling point was chosen at the centre ofeach of these strata or from the nearest available locationleading to an irregular-grid sampling scheme consisting of 60sampling points (Figure 2)


W09W08W07

W06W05W04

W36W35

W34W33

W32W31

W30

W03

W29

W28W27W26W25

W24W23

W22W21W20

W02

W19W18W17W16

W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13R12

R11

R10

R01

Water bodies

N

Landscape boundaryTwo km grid designSampling points

Land useEcological estateEcological estate with limited grazing

Eucalypt plantationIrrigated croppingPerennial horticulturePerennial pastureRural living

0 5 1025(km)

Figure 2 Study area location the distribution of proposed land uses in the future (25-year) land-use scenario (covering a total area of 6441 ha)and the associated irregular-grid used to sample soils in this study

Each sampling point was located in the field and an areaof 10 by 10mwas defined for samplingWithin each sampling-point area three randomly located subsamples were collectedfrom each of three soil depths (0ndash10 10ndash20 and 20ndash30 cm)using a soil auger of 825 cm internal diameter (AMS IncID USA) Subsamples from each depth were bulked to givea single composite sample per depth per sample-point areaSeparate intact cores (custom-made 75 times 10 cm core) werealso collected for bulk density determination

The 180 soil samples (60 sampling points times 3 depths) wereair-dried and passed through a 2mm sieve in preparationfor mid-infrared (MIR) spectral analysis A subset of 60 soilsamples were analysed for selected chemical and physicalproperties (see Table 1) andMIR spectroscopy was then usedto estimate these properties in the remaining 120 samples ForMIR spectral scanning a sub-sample (approximately 7 g) ofeach soil was finely ground to an approximate particle size ofless than 01mm diameter using a vibrating mixer ball mill


Table 1 Analysed soil chemical and physical properties of 60 sampling points (119899 = 60) at each of three depths from the study area in northernVictoria south-eastern Australia Data were produced through conventional analysis methods as well as mid-infrared spectroscopylowast

Soil properties Min Max Mean SD MedianOrganic carbon ()a

0ndash10 cm 03 45 13 08 1210ndash20 cm 02 24 08 04 0720ndash30 cm 01 16 06 03 05

pHCaCl2b


EC (dSm)c


Clay ()d


Sand ()d


Bulk density (gcm3)0ndash10 cm 10 16 13 01 1310ndash20 cm 12 16 14 01 1420ndash30 cm 12 16 14 01 14

aDry combustion and Walkley-Black methods [41]bpH in 001MCaCl2 1 5 extraction ratio [42]cEC in 1 5 water extraction ratio [43]dHydrometer method [44]lowastConventional analysismethodswere applied to 61 reference samples (ie 61 out of the 180 study samples) andMIR spectroscopywas used to predict propertiesin the remaining samples

(Retsch Haan Germany) set on a vibrational frequency of20 sminus1 for 3min

24 Analysis of Soils Using MIR Spectroscopy and PartialLeast-Squares (PLS) Regression We used a Spectrum-One(PerkinElmer Wellesley MA USA) Fourier transform MIRspectrometer to collect diffuse reflectance MIR spectra of thecomplete set of 180 soils following a methodology describedelsewhere [45 46] Soil MIR spectroscopy provides thepossibility of rapid inexpensive and simultaneous charac-terisation of various chemical properties in large numbers ofsamples [2 45 47] Moreover the repeatability and repro-ducibility of this technique among different laboratories havebeen found to be superior to the performance of conventionalsoil analyses [2] prior to partial least-squares (PLS) regres-sion principal component analysis (PCA) was performedto detect any spectral outliers and to select a calibrationsubset of 60 soil samples for PLS model development PCArevealed one sample as a spectral outlier which alongwith thecalibration subset was included in the reference soil propertycharacterisation (Table 1)

The calibrationmodels for predicting SOC pH electricalconductivity (EC) and percent sand silt and clay weredeveloped between MIR spectra (predictor variables) andreference soil property values (response variables) basedon the method introduced by Haaland and Thomas [48]which is popular in quantitative soil MIR analysis [47] Thesoftwares GRAMSAI GRAMS IQ and IQ Predict whichare part of the GRAMS spectroscopy software suite (ThermoFisher ScientificWalthamMA) were used for PCAandPLS-regression model building and predictions The MIR-PLSprediction models from the calibration subset were appliedto the remaining 119 samples (see Forouzangohar et al [4546]) and these predicted soil properties were used in theremainder of the study

25 APSIM for Predicting Indicators to Represent PriorityServices Two priority soil ecosystem services were identifiedfor consideration namely carbon sequestration and waterstorageThese two serviceswere nominated by the landholderas central to their land-use decision making (eg potentialinvolvement in both carbon and water markets) and were


identified as priority services for the benefit of humans andthe environment by Aitkenhead et al [14] We chose toestimate current and future capacity to supply these servicesusing two indicators namely SOC content and soil waterstorage In this case study the service ldquowater storagerdquo caninclude water added as irrigation As such this service rep-resents capacity of the soil to retain a scarce resource (water)either from natural or artificial sources in the landscape andto reliably provide that resource for plant growth

Key APSIM modules used in this study were SOILWAT[20 49] for simulating soil water dynamics and SOILN [15 19]for soil organic carbon In addition the MANAGER com-ponents were utilised to control land-use configurations (asbelow) and the plant growth modules (Agpasture SoybeanOat Wheat Lucerne and Eucalyptus grandis) were used tosimulate the production systems corresponding to currentand future land uses

Two scenarios of indicator predictions were examined forthe study landscape one based on the current distributionof land uses and another based on a proposed future land-use configuration over the next 25 years (Figure 2) To isolatethe effects of land-use change climate change over the 25-year period was assumed to be negligible although somemodels predict lower rainfall and higher temperatures [50] inthe study area which could affect our indicator predictionsData from a recent 25-year period (1985 to 2009) wereused to represent climate in 25-years-time In addition tofurther avoid effects of interannual climate variability dailyweather data were defined for a ldquotypical yearrdquo comprisedof 12 ldquotypicalrdquo months each based on the average monthlyrainfall and the distribution of rain days from two nearbyweather stations (Swan Hill and Kerang located resp to thenorth and south of the study area) Daily weather data forthese weather stations were sourced from the SILO climatedatabase (httpwwwlongpaddockqldgovausilo) hostedby the Queensland Climate Change Centre of Excellence [51]

To represent the current land-use scenario (time = 0years) APSIM was configured to simulate two land uses

(1) dry cropping system as a continuous winter cerealcropping simulated by the Wheat module withremoval of soil surface plant residues 30ndash45 days afterharvest in preparation for opportunistic sowing in lateautumn (leaving the soil surface bare for 4-5months)

(2) pasturegrassland under intensive grazing simulatedby the Agpasture module with regular fortnightlygrazing to a remaining herbage of 500 kgha

In addition to represent the future land-use scenario (time =25 years) APSIM was configured to simulate five new landuses (Figure 2)

(1) ecological estate (protected for restoration of peren-nial native vegetation predominantly grasslands andshrublands) simulated by the Agpasture modulewith no plant biomass removal

(2) ecological estate with limited grazing simulated bythe Agpasture module with two grazing events peryear to a remaining herbage of 1000 kgha

(3) irrigated cropping system as summer-winter rota-tional cropping systems simulated by the SoybeanOats and Wheat modules of crops assuming no-tillage systems and including no removal of soilsurface plant residues after harvest

(4) irrigated permanent lucerne stand (3 year ley) sim-ulated by the Lucerne module assuming regularharvest at flowering and removal of 95 of harvestedplant material

(5) eucalypt plantation simulated by the Egrandis andAgpasture modules using the Canopy module toaccount for light and water competition betweentrees and intertree herbage (including no removalof plant biomass) note that APSIM currently lacksmodules for tree species other than eucalypts sowe treated limited areas of perennial horticulture aseucalypt plantations for the purposes of this study(ie perennial horticulture is not considered further)

Five soil types of theMallee region in north-western VictoriaAustralia were chosen from the APSIM soil database tounderpin APSIM modelling The soils were chosen fromthose generic sites that had similar characteristics to sampledsoils and were further parameterised for SOC EC pH parti-cle size distribution and bulk density using our measures ofthe 60 reference soils

26 Mapping Service Indicators Ordinary kriging was usedto estimate spatial distribution of service indicators over thestudy area under both current and future land-use scenariosWe used ArcGIS 100 (ESRI Inc CA USA) to producedistribution maps of indicators (SOC content and soil waterstorage) based on kriging between the 60 sample pointsHerethe average distance between neighbouring sample pointswas 1300m so that a lag size of 1300m was used in the semi-variogram models (which depict the spatial autocorrelationof the sample points and define the weights of the krigingfunctions) With 10 lags of this size semivariograms of upto 13000m were fitted for each service indicator under eachscenario The exponential functions were best fitted throughsemivariogram modelling and nuggets were set to zero (ieit was assumed that there was no error for model-derivedvalues at each sample point) Using the trend analysis toolsecond-order polynomial trends were identified and fitted toremove directional trends in SOC content data As requiredlog transformations were used to better approximate normaldistributions of input data Given a 2 km sampling grid arange of 2300mwas identified as appropriatewhenfitting thesemivariogrammodels in all cases In spatial interpolation ofSOC content each kriging average was obtained using a four-sector searching neighbourhood searching for a maximumof 5 and a minimum of 2 points in each sector In mappingsoil water storage any points beyond the 2300m range wasnot involved in predictions

27 Mapping Hotspots of Service Change Changes in the twoservice indicators between current and future scenarios werenormalized to a 0-1 scale through linear scaling [52] This


Table 2 Predicted service indicators under both current and new land uses for a predominant clay soil in the study landscape Values werederived from 25-year simulations based on average properties of the clay soil to 30 cm depth

Soil organiccarbon (tha)

Soil waterstorage (m3ha)

Aboveground biomass tosoil water storage kgm3a

Current land usesDry cropping 30b 760 08Intensive grazing 30b 680 09

Future land usesEcological estate 34 710 39Ecological estate with grazing 33 690 23Eucalypt plantation 34 750 845Irrigated no-till cropping 38 1200c 22Irrigated permanent lucerne 27 1170c 11

aAboveground biomass data is derived from APSIM simulationsbInitial soil organic carbon content in equilibrium (modelling assumption)cSoil water topped up by irrigation

allowed comparison of relative estimated changes in the twopriority services with a combined value of minus2 indicatingmaximum negative change in both and of +2 indicatingmaximum positive change in both These combined valueswere then used to produce a ldquohotspotrdquo map (where a hotspotindicates positive change in both [30]) encompassing the 60sampling points

28 Assessing Uncertainties of Spatial Interpolations Theoverall reliability of the indicator prediction surfaces (maps)over the entire study area was assessed using the mean pre-diction error and the standardized Root-Mean-Square Error(standardized RMSE) A mean prediction error close to zero(ie relative to themagnitude of the average prediction error)and a standardized RMSE close to one indicate relativelyunbiased predictions of service supplychangeThe reliabilityof the kriging model predictions for specific points overthe study area was expressed as the ratio of performancedeviation (RPD) which is the ratio of natural variation inthe sample set (standard deviation SD) to the size of theaverage prediction error (ie the higher the RPD the morereliable the prediction [53]) An RPD value of lt14 indicatespoor performance RPD gt 14 indicates acceptable modelperformance for qualitative to semiquantitative analysis andRPD gt 20 indicates excellent performance [54]

3 Results and Discussion

31 Soil Properties There was significant variation in mea-sured soil properties across the study area (Table 1) Forexample SOC concentration to 10 cm depth ranged from 05to 45 over distances of c 10 km in the northern sectionsGenerally SOC concentrations were greatest in northern andeastern sections (gt2) and least in central sections (lt1)ThepHvalues ranged from slightly acidic (52 at 30 cmdepth)to alkaline (93 at 30 cm depth) and EC ranged from 004 to107 dSm reflecting patterns of localised surface salinizationSimilarly soil texture varied across the study area with

percent clay in the top 30 cm ranging from 6 to 61 and sandfrom 16 to 93 (Table 1)

32 Service Indicators under Alternative Land Uses Simu-lations over a twenty-five-year period for a predominantclay soil in the study area predicted strong potential forincreases in service indicators under the five new landuses For example mean SOC content of 30 tha undercurrent land uses was predicted to increase under four ofthe new land uses the highest being an estimated 38 thaunder irrigated no-till cropping (Table 2) Increases in meansoil water storage were only predicted where new landuses involved irrigation (Table 2) However increases in theratio of aboveground biomass to soil water storage werepredicted from 08 and 09 kgm3 under current land uses to22 kgm3 under new irrigated land uses and to 23 39 and845 kgm3 for nonirrigated new land uses of ecological estatewith grazing ecological estate without grazing and eucalyptplantation respectively (Table 2)

33 Soil Organic Carbon Content Current Status and FutureProjections On average soil to 30 cm depth at the 60sampling points under current management stored 35 thasoil organic carbon (ldquoSOCrdquo range 13ndash102 tha Table 3) Thegreatest current SOC stores (gt55 tha) were at the northernend of the study landscape (sampling points W1 W2 W3W4 andW7 see Figure 2 for point locations)where samplingpoints were adjacent to remnant native eucalypt woodlandsalong the Little Murray River Isolated pockets of low relativecarbon stores (lt20 tha) were scattered throughout the studyarea and were associated with general cropping mixedfarming and grazing (W8 W10 W12 W18 W36 R10)

Predictions of soil carbon change under the future land-use scenario indicated a mean increase over the 25-yearperiod of 3 tha to 30 cm depth at the 60 sampling points(Table 3) Over the 6441 ha covered by the 25-year plan thistranslated to an increase of 17938 t SOC (ie change from165119 to 183057 t Table 3) Simulations indicated a range


Table 3 Estimates of service indicator at 60 sampling points (0ndash30 cm depth) and over the entire study area under current and future (25-year) land-use scenarios Overall changes in each indicator were based on differences between supply on an average day under the currentscenario and on an average day in 25-year-time under the future scenario

Indicator Sampling points Entire study areaaMin Max Mean SD

Soil organic carbon (tha)Current 13 102 35 17 165119 tFuture 18 92 39 14 183057 tChange minus11 11 3 4 17938 t

Soil water storage (m3ha)Current 179 843 587 156 2745342m3

Future 189 1220 719 251 3362834m3

Change 0 463 132 146 617492m3

aStudy area of 6441 ha

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22W21W20

W02

W19W18W17W16 W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10

Sampling pointsLandscape boundaryWater bodies

Soil organic carbon change (tha)

0 5 1025(km)

minus11ndashminus42minus42ndashminus06minus06ndash1313ndash2424ndash33ndash3333ndash3939ndash55ndash77ndash11

N

(a)

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10


0ndash2020ndash3434ndash4242ndash5555ndash7575ndash110110ndash165165ndash275275ndash600

Soil water change (m3ha)

(b)

Figure 3 Predicted changes in service indicators (a) soil organic carbon content and (b) soil water storage to 30 cm depth in the studyarea associated with changes from current (time = 0) to future (time = 25 years) land-use scenarios Each prediction surface representsdifferences in the supply of services between an average day under the current scenario and an average day under the future scenario (a)Mean prediction error 004 standardized RMSE 106 average prediction error 24 RPD 15 (b) mean prediction error minus48 standardizedRMSE 109 average prediction error 127 RPD 12

of potential changes in SOC of up to plusmn11 tha (Figure 3(a)Table 3)

Spatial predictions of SOC changes were acceptably reli-able (RPD gt 14 Figure 3(a)) and highlighted potential risksand opportunities in soil carbon management [25 30] Forexample decreases of up to 11 t Cha were predicted near thenorthern end (Figure 3(a)) corresponding with the greatestcurrent SOC stores (gt55 tha) Reasons for the high relativecurrent SOC here were unclear although they could relate to

the close proximity to remnant perennial vegetation andorto time since clearing (unknown) Nonetheless this studyrsquospredictions indicate that the planned change in land usefrom dry cropping to zero-tillage irrigated cropping systemswill not prevent SOC decreases and that alternative landuses should be examined to conserve this carbon content ifsustaining SOC content is a priority Elsewhere predicteddecreases of c 4 t Cha near sampling point W13 in thenorthern section (Figure 3(a)) were associatedwith a planned


land-use change from regular grazing to an irrigated lucernesystem involving regular removal of plant material fromthe soil surface Given the already low carbon stores at thissampling point (284 tha) alternative land uses that retainrather than remove plant material might be considered

Landscape locations that show high potential increasesin SOC might also be considered for targeted managementFor example locations of maximum predicted increases(eg R7 W21 and W35 Figure 3(a)) were associated withchanges from intensive agriculture to a zero-tillage irrigatedcropping system with no plant residue removal These datathus support strong recommendations for adoption of no-till farming by Lal et al [55] as an option for enhancingSOC storage in cropland soils of relatively low current SOCcontent (lt30 t Cha) Another example of an opportunity toincrease SOC stores was indicated at sampling point W12Here the soil was a loamy sand of very low SOC content(132 tha) and a land-use change from mixed farming withintensive grazing to ecological estate with limited grazingled to predicted increases of 57 t Cha (43) over 25 years(Figure 3(a))

Overall changes from intensive agriculture to ecologicalestate ecological estate with limited grazing and eucalyptplantations were associated with moderate increases in SOCof 13 to 5 t Cha over the 25-year period (Figure 3(a)) This isconsistent with previous findings of measured SOC increaseswith conversion from traditional croplands to perennialgrasslands (eg [56ndash58]) Katterer et al [56] reported anaverage of 04 thayr increase in SOC content over 30 yearswhen a croplandwas converted back to grassland in a Swedishfarm although their study involved a wetter (542mmyr)and cooler (mean annual temperature of 51∘C) climate anddouble the starting SOC content In this studyrsquos landscapelimited rainfall and very hot and dry summers are likelyto be limiting factors for grassland growth as simulatedby the APSIM-Agpasture module Nonetheless in termsof percent gain in SOC through conversion of degradedlands to natural grasslands our results show agreement withpreviously reported limits of 15ndash20 increases [56 57 59]

34 Soil Water Storage Current Status and Future ProjectionsCurrent water storage of soils reflected the predominance ofclay-textured soils in the 6441 ha covered by the 25-year planTotal water stored to 30 cm soil depth on an average day underthe current scenario was estimated at 2745342m3 (Table 3)On average soils stored 587m3ha water without irrigationThis value ranged from 179m3ha for a loamy sand soil underregularly-grazed pasture (sample point W12) to 843m3ha ina clay soil under dryland cereal cropping (sample point R7)

Land-use changes in the 25-year plan were predicted toincrease water storage by 22 (ie to a mean water storageof 719m3ha and a total storage capacity of 3362834m3Table 3) Similarly modelled changes in average daily soilwater storage over the entire study landscape ranged from0 to 463m3ha but were mostly gt110m3ha (Figure 3(b))Modelled soil water storage values under current and futureland-use scenarios were consistent with APSIM-generatedpublished works [21 22] For instance volumetric soil water

in the top 30 cm of selected soil profiles was predictedand observed to be between 01 and 04 water volumetotalvolume (c 300 to 1200m3ha) in the nearby Murray-DarlingBasin [22]

Predicted increases in soil water storage were associatedwith changes to irrigated no-till systems and to a lesserdegree increases in perenniality Irrigated no-till croppingsystems add water from outside the system and can increasewater retention through a mulching effect associated withretained plant residues Mean predicted increases in meanwater storage in soils under this land-use change were 54from 669m3ha to 1029m3ha (ie deepest red areas forpoints R5 R7-R9 R14 R17 R19 W3 W4 W7 W9 W11W21 andW35 in Figure 3(b)) Increases in soil water storageover the 25-year period were also indicated for land usesthat increased perenniality namely eucalypt plantation andecological estate (with or without grazing) In these landuses increases of 8ndash15 in average daily water storage wereapparently due to processes affecting water infiltration andretention and were perhaps tempered by increased lossesthrough evapotranspiration associated with the establish-ment of woody species [60]This is also consistent with stud-ies that have indicated that the reverse land-use change (iefrom woody native vegetation to cropping) was associatedwith decreased soil water-holding capacity [61]

35 Relationships among Soil Ecosystem Services The indi-cator change maps highlight locations both of potentialcomplementarities (synergies) among priority services asindicated by mutual increases with changed managementand of potential trade-offs among services as indicated bymarked increases in one service but decreases in anotherOverall we found mostly positive (or ldquosynergicrdquo [62]) rela-tionships among our priority soil services (Figure 4) Thissuggests that they increased at the same time due either tosimultaneous responses to the same land-use change or topositive interactions [62] In this context hotspots are locatedat points where the supply of multiple soil services can bemutually and markedly improved [30]

This study found several potential hotspots as indicatedby high combined service change scores (gt1 to 2 Figure 4)For example land-use change to irrigated no-till cropping ledto substantial estimated increases in both soil water storageand SOC content at sampling points R7 R8 R9 R14 R17R19 W11 W21 and W35 (Figure 4) However the studyalso highlighted at least six instances of possible trade-offsbetween services as indicated by very low combined changescores (lt02) Four of these cases were located at samplingpoints W1 W3 W4 and W13 where future predictionsflagged likely reductions in SOC stocks However we doubtthis is due to any causal relationships among the servicesbut as discussed in Section 33 this appears to be dueto the relatively high initial SOC contents at these points(except W13) which were not sustained by the new landuses Similarly the low combined change score at W13 wasmainly due to predicted decreases in SOC associated withregular removal of plant material under the planned irrigatedlucerne system Elsewhere the low combined change score


W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11

R10Sampling pointsLandscape boundaryWater bodies

Combined change score

0 5 1025(km)

N

19 (highest)

minus005 (lowest)

Figure 4 ldquoHotspotrdquomap indicating combined relative change in the two service indicators (soil organic carbon content and soil water storage)from current (time = 0) to future (time = 25 years) land-use scenarios at each of 60 sampling points Potential combined scores range fromminus2 (maximum negative change in both indicators) to +2 (maximum positive change in both indicators)

at W16 could be attributed to negligible change in land-use practices and thus in service provision under unchangedclimatic conditions

36 Approach Limitations We acknowledge limitations inour approach Our spatial interpolation method of choicekriging from point data carries some limitations Prediction

statistics were acceptable for SOC (mean prediction error004 standardised RMSE 106 and RPD 15) but were lessconvincing for soil water storage (mean prediction errorminus48standardised RMSE 109 and RPD 12) In particular the lowRPD for soil water storage indicated nonreliable estimationsfor at least some locations perhaps due to greater spatialvariation in soil water distributions than SOC


In addition the APSIM model could not equally repre-sent our diversity of land-use changes For example APSIMplant modules have not been specifically developed fordiverse native plant systems (ie our ldquoecological estaterdquo)and for perennial horticulture systems Moreover it was notfeasible to fully parameterise the APSIM soil components forthe 60 sampling points in this study Instead we relied onthe APSIM soil database to provide default values for someof our modelling components This leads to uncertaintiesthe quantification of which is beyond the scope of thisstudy Future work will examine the implications of variousassumptions used not only in the APSIMmodelling but alsoin the entire approach through detailed sensitivity analyses

4 Conclusion

We demonstrated a broadly applicable approach for quan-tifying and mapping service indicators (representing pri-ority soil ecosystem services) in an agricultural landscapein south-eastern Australia A feasible sampling intensitywas combined with soil MIR spectral analysis biophysicalmodeling and spatial interpolations to provide estimatesof two soil indicators under both current and future land-use scenarios We showed that under the future land-useplan the supply of the two priority soil services (based onindicators) would likely increaseThemodelledmaps providea basis for supporting decisions about alternative land usesby indicating ldquohotspotrdquo locations where there are mutualincreases in the supply of soil ecosystem services or locationswhere there is a risk to the supply of some services Suchknowledge informs the management of multiple services ata range of scales from paddock to landscape and regionboth in response to markets for particular services and torequirements or incentives for providing multiple benefits(see eg the proposed ldquocobenefits indexrdquo in Australiarsquos recentldquoCarbon Credits Billrdquo Commonwealth of Australia 2011)

The proposed approach could be useful in designingstrategies for supporting sustainable soil management inthis landscape Coordinated management of multiple landparcels within a landscape rather than isolated managementof single parcels offers potential to maintain supply ofpriority services at the landscape scale by avoiding risksand harnessing opportunities through matching soils withparticular land-use practices Introducing new land uses inany given landscape not only offers scope for increasinglandscape complexity [63 64] but also offers more flexibilityin decisions relating to landscape-level delivery of multipleservices This type of flexibility and responsiveness will beneeded as additional factors like climate change furtherconfound the challenge to managing soils sustainably inchanging agricultural landscapes

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This study was funded by the (then) Land Health UnitNatural Resources Division of the (now) Victorian Depart-ment of Environment and Primary Industries The authorswould also like to thank Kilter Pty Ltd in particular MaloryWeston andDavidHeislers for providing themwith scenarioinformation and access to properties and for help with soilsampling APSIMdevelopment team for providing themwithAPSIM software and technical advice Rai Kookana MikeMcLaughlin and Les Janik from CSIRO Land and WaterAdelaide for access to the soil MIR spectroscopy facility andNarges Milani Saravanan Jangammanaidu Krishnaraj TarekMurshed Jane Lean Mathew Lee Najib Ahmady and LynWilks from theDepartment of Forest and Ecosystem Sciencethe University of Melbourne for their help and technicalassistance in the collection and analysis of soil samples

References

[1] E Dominati M Patterson and A Mackay ldquoA framework forclassifying and quantifying the natural capital and ecosystemservices of soilsrdquo Ecological Economics vol 69 no 9 pp 1858ndash1868 2010

[2] C Palm P Sanchez S Ahamed and A Awiti ldquoSoils acontemporary perspectiverdquo Annual Review of Environment andResources vol 32 pp 99ndash129 2007

[3] D A Robinson N Hockley E Dominati et al ldquoNatural capitalecosystem services and soil change why soil science mustembrace an ecosystems approachrdquo Vadose Zone Journal vol 11no 1 2012

[4] D A Robinson I Lebron andH Vereecken ldquoOn the definitionof the natural capital of soils a framework for descriptionevaluation and monitoringrdquo Soil Science Society of AmericaJournal vol 73 no 6 pp 1904ndash1911 2009

[5] S Frank C Furst L Koschke and F Makeschin ldquoA contri-bution towards a transfer of the ecosystem service conceptto landscape planning using landscape metricsrdquo EcologicalIndicators vol 21 pp 30ndash38 2012

[6] M Rutgers H J van Wijnen A J Schouten et al ldquoA methodto assess ecosystem services developed from soil attributes withstakeholders and data of four arable farmsrdquo Science of the TotalEnvironment vol 415 pp 39ndash48 2012

[7] H J van Wijnen M Rutgers A J Schouten C Mulder D deZwart and A M Breure ldquoHow to calculate the spatial distri-bution of ecosystem servicesmdashnatural attenuation as examplefrom The Netherlandsrdquo Science of the Total Environment vol415 pp 49ndash55 2012

[8] R Haines-Young M Potschin and F Kienast ldquoIndicatorsof ecosystem service potential at European scales mappingmarginal changes and trade-offsrdquo Ecological Indicators vol 21pp 39ndash53 2012

[9] L T Bennett P M Mele S Annett and S Kasel ldquoExamininglinks between soil management soil health and public benefitsin agricultural landscapes an Australian perspectiverdquo Agricul-ture Ecosystems amp Environment vol 139 no 1-2 pp 1ndash12 2010

[10] R W Simonson ldquoOutline of a generalized theory of soilformationrdquo Soil Science Society of America Proceedings vol 23no 2 pp 152ndash156 1959


[11] B Burkhard F Kroll S Nedkov and F Muller ldquoMappingecosystem service supply demand and budgetsrdquo EcologicalIndicators vol 21 pp 17ndash29 2012

[12] B Minasny and A E Hartemink ldquoPredicting soil properties inthe tropicsrdquo Earth-Science Reviews vol 106 no 1-2 pp 52ndash622011

[13] M U F Kirschbaum J O Carter P R Grace et al ldquoBriefdescription of several models for simulating net ecosystemexchange in Australiardquo in Net Ecosystem Exchange M UF Kirschbaum and R Mueller Eds CRC for GreenhouseAccounting Canberra Australia 2001

[14] M J Aitkenhead F Albanito M B Jones and H I J BlackldquoDevelopment and testing of a process-based model (MOSES)for simulating soil processes functions and ecosystem servicesrdquoEcological Modelling vol 222 no 20ndash22 pp 3795ndash3810 2011

[15] B A Keating P S Carberry G L Hammer et al ldquoAn overviewof APSIM a model designed for farming systems simulationrdquoEuropean Journal of Agronomy vol 18 no 3-4 pp 267ndash2882003

[16] R L McCown G L Hammer J N G Hargreaves D PHolzworth and D M Freebairn ldquoAPSIM a novel softwaresystem for model development model testing and simulationin agricultural systems researchrdquo Agricultural Systems vol 50no 3 pp 255ndash271 1996

[17] R J Farquharson G D Schwenke and J D Mullen ldquoShouldwe manage soil organic carbon in Vertosols in the northerngrains region of Australiardquo Australian Journal of ExperimentalAgriculture vol 43 no 3 pp 261ndash270 2003

[18] Z Luo E Wang O J Sun C J Smith and M E ProbertldquoModeling long-term soil carbon dynamics and sequestrationpotential in semi-arid agro-ecosystemsrdquoAgricultural and ForestMeteorology vol 151 no 12 pp 1529ndash1544 2011

[19] K Ranatunga M J Hill M E Probert and R C DalalldquoComparative application of APSIM RothC and Century topredict soil carbon dynamicsrdquo in Proceedings of the Interna-tional Congress on Modelling and Simulation (MODSIM rsquo01) FGhassemi PWhetton R Little andM Littleboy Eds pp 733ndash738 Modelling and Simulation Society of Australia and NewZealand Canberra Australia 2001

[20] B A Keating D Gaydon N I Huth et al ldquoUse of modellingto explore the water balance of dryland farming systems inthe Murray-Darling Basin Australiardquo European Journal ofAgronomy vol 18 no 1-2 pp 159ndash169 2002

[21] M E Probert J P Dimes B A Keating R C Dalal andW MStrong ldquoAPSIMrsquos water and nitrogen modules and simulationof the dynamics of water and nitrogen in fallow systemsrdquoAgricultural Systems vol 56 no 1 pp 1ndash28 1998

[22] J Wang E Wang and D L Liu ldquoModelling the impacts ofclimate change on wheat yield and field water balance overtheMurray-Darling Basin in AustraliardquoTheoretical and AppliedClimatology vol 104 no 3-4 pp 285ndash300 2011

[23] R O Gilbert and J C Simpson ldquoKriging for estimating spatialpattern of contaminants potential and problemsrdquo Environmen-tal Monitoring and Assessment vol 5 no 2 pp 113ndash135 1985

[24] A B McBratney and R Webster ldquoHow many observations areneeded for regional estimation of soil propertiesrdquo Soil Sciencevol 135 no 3 pp 177ndash183 1983

[25] P Longley Geographic Information Systems amp Science JohnWiley amp Sons Hoboken NJ USA 2011

[26] Z L Frogbrook J Bell R I Bradley et al ldquoQuantifyingterrestrial carbon stocks examining the spatial variation in two

upland areas in the UK and a comparison to mapped estimatesof soil carbonrdquo Soil Use andManagement vol 25 no 3 pp 320ndash332 2009

[27] S Kumar and R Lal ldquoMapping the organic carbon stocksof surface soils using local spatial interpolatorrdquo Journal ofEnvironmental Monitoring vol 13 no 11 pp 3128ndash3135 2011

[28] J W van Groenigen W Siderius and A Stein ldquoConstrainedoptimisation of soil sampling for minimisation of the krigingvariancerdquo Geoderma vol 87 no 3-4 pp 239ndash259 1999

[29] C S Zhang Y Tang X L Xu and G Kiely ldquoTowardsspatial geochemical modelling use of geographically weightedregression for mapping soil organic carbon contents in IrelandrdquoApplied Geochemistry vol 26 no 7 pp 1239ndash1248 2011

[30] N D Crossman and B A Bryan ldquoIdentifying cost-effectivehotspots for restoring natural capital and enhancing landscapemultifunctionalityrdquoEcological Economics vol 68 no 3 pp 654ndash668 2009

[31] P M Haygarth and K Ritz ldquoThe future of soils and land use inthe UK soil systems for the provision of land-based ecosystemservicesrdquo Land Use Policy vol 26 no 1 pp S187ndashS197 2009

[32] M Schroter R P Remme and L Hein ldquoHow andwhere tomapsupply and demand of ecosystem services for policy-relevantoutcomesrdquo Ecological Indicators vol 23 pp 220ndash221 2012

[33] R F IsbellThe Australian Soil Classification CSIRO AustraliaVictoria Australia 2002

[34] DPI ldquoGeomorphology of Victoria (GMU250GMU250)rdquoDepartment of Primary Industries Bendigo VictoriaAustralia 2010

[35] Ramsar Convention Secretariat The Ramsar Convention Man-ual AGuide To theConvention onWetlands (Ramsar Iran 1971)Ramsar Convention Secretariat Gland Switzerland 2011

[36] DPI ldquoVictorian land use information system datasetrdquo Depart-ment of Primary Industries Bendigo Victoria Australia 2009

[37] J Connor ldquoThe economics of time delayed salinity impactmanagement in the River Murrayrdquo Water Resources Researchvol 44 no 3 2008

[38] N D Crossman J D Connor B A Bryan DM Summers andJ Ginnivan ldquoReconfiguring an irrigation landscape to improveprovision of ecosystem servicesrdquo Ecological Economics vol 69no 5 pp 1031ndash1042 2010

[39] G Christakos andR AOlea ldquoSampling design for spatially dis-tributed hydrogeologic and environmental processesrdquoAdvancesin Water Resources vol 15 no 4 pp 219ndash237 1992

[40] A W Warrick and D E Myers ldquoOptimization of samplinglocations for variogram calculationsrdquoWater Resources Researchvol 23 no 3 pp 496ndash500 1987

[41] D W Nelson and L E Sommers ldquoTotal carbon organiccarbon and organicMatterrdquo inMethods of Soil Analysismdashpart 3Chemical Methods D L Sparks Ed pp 961ndash1010 Soil ScienceSociety of America Madison Wis USA 1996

[42] G W Thomas ldquoSoil pH and soil acidityrdquo in Methods of SoilAnalysismdashpart 3 Chemical Methods D L Sparks Ed pp 475ndash490 Soil Science Society of America MadisonWis USA 1996

[43] J D Rhoades ldquoSalinity electrical conductivity and total dis-solved solidsrdquo in Methods of Soil Analysismdashpart 3 ChemicalMethods D L Sparks Ed pp 417ndash435 Soil Science Society ofAmerica Madison Wis USA 1996

[44] G W Gee and J W Bauder ldquoParticle-size analysisrdquo inMethodsof Soil Analysismdashpart 1 Physical and Mineralogical Methods DL Sparks Ed pp 383ndash411 Soil Science Society of AmericaMadison Wis USA 1996


[45] M Forouzangohar D Cozzolino R S Kookana R J SmernikS T Forrester and D J Chittleborough ldquoDirect compari-son between visible near- and mid-infrared spectroscopy fordescribing diuron sorption in soilsrdquo Environmental Science ampTechnology vol 43 no 11 pp 4049ndash4055 2009

[46] M Forouzangohar R S Kookana S T Forrester R J Smernikand D J Chittleborough ldquoMidinfrared spectroscopy andchemometrics to predict diuron sorption coefficients in soilsrdquoEnvironmental Science amp Technology vol 42 no 9 pp 3283ndash3288 2008

[47] L J Janik R H Merry and J O Skjemstad ldquoCan mid infrareddiffuse reflectance analysis replace soil extractionsrdquo AustralianJournal of Experimental Agriculture vol 38 no 7 pp 681ndash6961998

[48] DMHaaland andEVThomas ldquoPartial least-squaresmethodsfor spectral analyses 1 Relation to other quantitative calibrationmethods and the extraction of qualitative informationrdquoAnalyt-ical Chemistry vol 60 no 11 pp 1193ndash1202 1988

[49] K Ranatunga E R Nation and D G Barratt ldquoReview of soilwater models and their applications in Australiardquo Environmen-tal Modelling amp Software vol 23 no 9 pp 1182ndash1206 2008

[50] B Timbal and D A Jones ldquoFuture projections of winterrainfall in southeast Australia using a statistical downscalingtechniquerdquo Climatic Change vol 86 no 1-2 pp 165ndash187 2008

[51] S J Jeffrey J O Carter K BMoodie and A R Beswick ldquoUsingspatial interpolation to construct a comprehensive archive ofAustralian climate datardquo Environmental Modelling amp Softwarevol 16 no 4 pp 309ndash330 2001

[52] K Swingler Applying Neural Networks A Practical GuideAcademic Press San Francisco Calif USA 1996

[53] T Fearn ldquoAssessing calibrations SEP RPD RER and R2rdquo NIRNews vol 13 no 6 pp 12ndash14 2002

[54] C-W Chang D A Laird M J Mausbach and C RHurburgh ldquoNear-infrared reflectance spectroscopymdashprincipalcomponents regression analyses of soil propertiesrdquo Soil ScienceSociety of America Journal vol 65 no 2 pp 480ndash490 2001

[55] R Lal M Griffin J Apt L Lave andM GMorgan ldquoManagingsoil carbonrdquo Science vol 304 no 5669 p 393 2004

[56] T Katterer L Andersson O Andren and J Persson ldquoLong-term impact of chronosequential land use change on soil carbonstocks on a Swedish farmrdquo Nutrient Cycling in Agroecosystemsvol 81 no 2 pp 145ndash155 2008

[57] M P McHenry ldquoFarm soil carbon monitoring developmentsand land use change unearthing relationships between pad-dock carbon stocks monitoring technology and new marketoptions inWesternAustraliardquoMitigation andAdaptation Strate-gies for Global Change vol 14 no 6 pp 497ndash512 2009

[58] L P Qiu X R Wei X C Zhang et al ldquoSoil organic carbonlosses due to land use change in a semiarid grasslandrdquo Plant andSoil vol 355 no 1-2 pp 299ndash309 2012

[59] L B Guo and R M Gifford ldquoSoil carbon stocks and land usechange ameta analysisrdquoGlobal Change Biology vol 8 no 4 pp345ndash360 2002

[60] M D Nosetto E G Jobbagy and J M Paruelo ldquoLand-usechange and water losses the case of grassland afforestationacross a soil textural gradient in central Argentinardquo GlobalChange Biology vol 11 no 7 pp 1101ndash1117 2005

[61] G Mahe J-E Paturel E Servat D Conway and A DezetterldquoThe impact of land use change on soil water holding capacityand river flow modelling in the Nakambe River Burkina-FasordquoJournal of Hydrology vol 300 no 1ndash4 pp 33ndash43 2005

[62] E M Bennett G D Peterson and L J Gordon ldquoUnderstand-ing relationships among multiple ecosystem servicesrdquo EcologyLetters vol 12 no 12 pp 1394ndash1404 2009

[63] T Tscharntke J M Tylianakis T A Rand et al ldquoLand-scape moderation of biodiversity patterns and processesmdasheighthypothesesrdquo Biological Reviews vol 87 no 3 pp 661ndash685 2012

[64] J Vandermeer and I Perfecto ldquoThe agricultural matrix and afuture paradigm for conservationrdquoConservationBiology vol 21no 1 pp 274ndash277 2007

Submit your manuscripts athttpwwwhindawicom

Forestry ResearchInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Environmental and Public Health

Journal of



EcosystemsJournal of


MeteorologyAdvances in

EcologyInternational Journal of


Marine BiologyJournal of


Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Advances in


Environmental Chemistry

Atmospheric SciencesInternational Journal of



Waste ManagementJournal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics


Geological ResearchJournal of

EarthquakesJournal of


BiodiversityInternational Journal of


ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyInternational Journal of


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


ClimatologyJournal of


of other services [9] All soil services are essentially aggregatesof soil processes where processes are inputs losses transfersand transformations ofmaterial and energy [2 10] For exam-ple the intermediate soil service ldquoorganic matter cyclingrdquo(nomenclature of [9]) is the result of various processes (littercomminution decomposition and humification) which areexpressed as rates Full temporal quantification of multipleprocess rates at landscape scales is not feasible Instead soilproperties (eg soil organic carbon content) can be used torepresent the end-point of processesservices (eg carbonsequestration) and changes in soil properties used as anindication of change in potential service supply [11]

Accurate prediction of changes in soil servicespropertieswill rely both on the direct measurement of properties andon the use of soil-specific biophysical models In this contextchanges in soil properties are simulated continuously inresponse to changes in soil processes as a result of interactionsamong soil water vegetation and the atmosphere [12]Thereare many soil simulation models some of which have beenrecommended for soil services assessment [13] For exampleAitkenhead et al [14] introduced theMOSESmodel for quan-titative assessment of selected ecosystem services calibratedfor a range of soils in the UK In Australia AgriculturalProduction System Simulator (APSIM) was developed as amodelling framework consisting of soil plant and manage-ment component modules to simulate diverse productionsystems specific to Australian soils and conditions [15 16]APSIM has considerable potential as a tool for modellingchanges in soil properties [13] because soil properties aresimulated continuously in response to changes in weathermanagement and vegetation [15 16] It has been appliedto the prediction of changes in soil organic carbon (SOC)content as a result of changes in land management [17ndash19] In addition APSIM has been successfully tested in thesimulation of soil water balance [20ndash22]

Soil simulation models are traditionally point-basedwhereas soil ecosystem services are best represented as con-tinuous surfaces for decision-making Interpolation is re-quired and kriging offers an objective and rigorous approachto interpolating between spatially-distributed data pointsand thus tomapping soil services [23 24]The use of kriging[25] to produce soil property maps has been demonstratedfrequently for SOC [26ndash29] Although many studies appliedthese techniques at field scales with high sample densities[26 28] some dealt with sparser sampling schemes at muchbroader scales For example Kumar and Lal [27] estimatedthe spatial distribution of SOC across an area of 117599 km2using 920 sample points (sim one sample per 128 km2) Simi-larly Zhang et al [29] produced amap of SOC for the Repub-lic of Ireland based on kriging interpolations using an averagesample density of two per 100 km2 Clearly in the contextof broad-scale changes in soil properties a key requirementis to choose a sampling intensity that accommodates costand logistical constraints but that is also sufficient forcapturing change trends at appropriate scales within acceptedlevels of error Here while demonstrating the advantages ofkriging from grid samples for regional estimations of soilproperties McBratney and Webster [24] concluded that the

best sampling strategy would be a regular grid scheme withthe highest affordable number of observations

We used a novel combination of methods includingfield sampling biophysical modelling (APSIM) and spatialinterpolations (eg kriging) to produce landscape-scalemapsindicating changes in the supply of soil services from currentto future management scenarios We selected a study area innorthern Victoria south-eastern Australia to illustrate theutility of this approach for considering priority soil servicesin decisions relating to changing configurations of land-usepractices at landscape scales We focused on two priorityservices namely carbon sequestration and water storageWe used a landholder defined scenario of future land-useconfigurations (involving increased complexitydiversity) topredict effects of land-use change on the spatial distributionof two soil properties (hereafter ldquoindicatorsrdquo) namely SOCcontent (indicating carbon sequestration) and volume ofwater stored in soil (indicating water storage) Our aims wereto quantify and map the distribution of service indicators(representing priority services) under both current and futureland-use configurations and to demonstrate the utility of ourapproach in supporting land-use decisions by highlightinglocations where management resources could be targetedto maintain or enhance the supply of priority soil services[11 30ndash32]

2 Methods

21 Approach Overview Our unique approach (Figure 1)included representative sampling of soils across the studyarea according to a geostatistically-valid design analysisof selected soil properties in a subset of reference soilsamples development of mid-infrared calibrationmodels forthe prediction of soil properties in the remaining samplesparameterising and configuring the APSIMmodel to predictindicators (representing priority services) under current andfuture land-use scenarios and applying spatial interpolations(kriging) in a GIS environment to produce maps of currentand of likely future changes in the distribution of theseindicators across the entire study area In addition a hotspotmap of soil services change was produced to help evaluate thelikely trade-offs and opportunities arising from future land-use configurations

22 Study Area This study encompasses a case-study land-scape of 30200 ha (302 km2) in northern Victoria between35∘ 24531015840 and 35∘ 41231015840 S and 143∘ 37721015840 and 143∘ 54221015840 Esouth-eastern Australia (Figure 2) The region has a warmgrassland climate (Australian Bureau of Meteorologymdashhttpwwwbomgovauiwkclimate zones) and typicallyexperiences hot and dry summers and cold winters and anaverage annual rainfall of 370mm falling mostly in winterand spring (June to November) Mean monthly minimumtemperatures range from 33∘C (August) to 184∘C (January)and mean monthly maximum temperatures range from131∘C (July) to 358∘C (January climate data based on 25-yeardata from nearby Swan Hill and Kerang weather stations)

There is no detailed soil survey of the study area Accord-ing to the Australian Soil Classification [33] and the Victorian


Study area













hotspots








W09W08W07

W06W05W04

W36W35

W34W33

W32W31

W30

W03

W29

W28W27W26W25

W24W23

W22W21W20

W02

W19W18W17W16

W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13R12

R11

R10

R01

Water bodies

N




0 5 1025(km)








pHCaCl2b


EC (dSm)c


Clay ()d


Sand ()d













































Future 189 1220 719 251 3362834m3

Change 0 463 132 146 617492m3


W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22W21W20

W02

W19W18W17W16 W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10



0 5 1025(km)


N

(a)

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10




(b)
















W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11



0 5 1025(km)

N

19 (highest)

minus005 (lowest)







4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics















Study area













hotspots








W09W08W07

W06W05W04

W36W35

W34W33

W32W31

W30

W03

W29

W28W27W26W25

W24W23

W22W21W20

W02

W19W18W17W16

W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13R12

R11

R10

R01

Water bodies

N




0 5 1025(km)








pHCaCl2b


EC (dSm)c


Clay ()d


Sand ()d













































Future 189 1220 719 251 3362834m3

Change 0 463 132 146 617492m3


W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22W21W20

W02

W19W18W17W16 W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10



0 5 1025(km)


N

(a)

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10




(b)
















W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11



0 5 1025(km)

N

19 (highest)

minus005 (lowest)







4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics















W09W08W07

W06W05W04

W36W35

W34W33

W32W31

W30

W03

W29

W28W27W26W25

W24W23

W22W21W20

W02

W19W18W17W16

W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13R12

R11

R10

R01

Water bodies

N




0 5 1025(km)








pHCaCl2b


EC (dSm)c


Clay ()d


Sand ()d













































Future 189 1220 719 251 3362834m3

Change 0 463 132 146 617492m3


W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22W21W20

W02

W19W18W17W16 W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10



0 5 1025(km)


N

(a)

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10




(b)
















W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11



0 5 1025(km)

N

19 (highest)

minus005 (lowest)







4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics


















pHCaCl2b


EC (dSm)c


Clay ()d


Sand ()d













































Future 189 1220 719 251 3362834m3

Change 0 463 132 146 617492m3


W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22W21W20

W02

W19W18W17W16 W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10



0 5 1025(km)


N

(a)

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10




(b)
















W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11



0 5 1025(km)

N

19 (highest)

minus005 (lowest)







4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics



















































Future 189 1220 719 251 3362834m3

Change 0 463 132 146 617492m3


W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22W21W20

W02

W19W18W17W16 W15W14

W13W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10



0 5 1025(km)


N

(a)

W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23R22R21R20

R02

R19R18R17

R16R15R14R13R12R11

R10




(b)
















W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11



0 5 1025(km)

N

19 (highest)

minus005 (lowest)







4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics

























W09W08W07

W06W05W04

W36W35

W34W33W32

W31W30

W03

W29W28W27W26W25

W24 W22

W21W20

W02

W19W18W17W16 W15

W14W13

W12

W11W10

W01

R09

R08R07R06

R05R04R03

R24R23

R22R21R20

R02

R19R18R17

R16R15R14R13

R12R11



0 5 1025(km)

N

19 (highest)

minus005 (lowest)







4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics
















4 Conclusion





Acknowledgments


References








































































Journal of












Volume 2014

Advances in









Geophysics











































































Journal of












Volume 2014

Advances in









Geophysics














research article ecosystem services in agricultural landscapes: a spatially explicit...

Documents