Forest Ecology and Management 304 (2013) 243–253

Contents lists available at SciVerse ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier .com/locate / foreco

Characterizations of anthropogenic disturbance patterns in themixedwood boreal forest of Alberta, Canada

0378-1127/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.foreco.2013.04.031

⇑ Corresponding author. Tel.: +1 604 822 6592; fax: +1 604 822 9106.E-mail address: ppickell@alumni.ubc.ca (P.D. Pickell).

Paul D. Pickell a,⇑, David W. Andison b, Nicholas C. Coops a

a Integrated Remote Sensing Studio, Department of Forest Resources Management, 2424 Main Mall, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z4b Bandaloop Landscape-Ecosystem Services, 1011 Hendecourt Road, North Vancouver, British Columbia, Canada V7K 2X3

a r t i c l e i n f o a b s t r a c t

Article history:Received 1 December 2012Received in revised form 8 April 2013Accepted 15 April 2013Available online 31 May 2013

Keywords:Anthropogenic disturbanceWildfireAggregate harvestDispersed harvestEcosystem-based managementBoreal mixedwood

Ecosystem-based management (EBM) has emerged as a dominant paradigm for the Canadian boreal for-est. One of the principles of EBM is to maintain ecosystem function by means of management activitiesthat approximate the historic patterns or processes responsible for maintaining a range of landscape con-ditions. This ideal has been manifested as planning schemes are shifting away from traditional sustainedyield harvests toward designs based on historic wildfire disturbance patterns. Wildfire disturbance pat-terns represent a coarse-filter management strategy, and are well-suited to the boreal forests of Canada.Forest management professionals in the boreal have been leaders in adopting these strategies over thepast decade. However, two key questions remain unanswered: (1) to what degree have these forest man-agement efforts resulted in disturbance patterns that resemble wildfire burning patterns?; and (2) towhat degree do the other sources of anthropogenic disturbance activities align with historic wildfire pat-terns? In this paper, an existing knowledge of historic range of variability (HRV) of wildfire patterns andthe NEPTUNE (Novel Emulation Pattern Tool for Understanding Natural Events) decision support toolwere used to test both questions.

The results suggest that forest harvest disturbances better approximated historic disturbance pat-terns than did energy extraction disturbances, though in both cases some of the metrics were beyondthe HRV. Significant differences were found between traditional dispersed patterns (e.g., multi-passharvesting) and the more recent aggregate harvest (e.g., single-pass) designs. Aggregate harvests werecharacterized by low proportional area in matrix remnants, moderate levels of combined island andmatrix remnants, and a high proportional area in the single largest disturbed patch (LDP). Dispersedharvests tended to have a higher proportional area in matrix remnants and better approximated theHRV in terms of proportional area in proportional island remnants area and largest island remnant.Aggregate harvest patterns did not perform as well for a few metrics such as proportional area inisland remnants, mean island remnant size and largest island remnant. Overall, the results suggestthat aggregate harvest designs better approximated key HRV patterns such as proportional matrixremnants and LDP, than did dispersed harvest designs on the landscape. The results also suggest thatforest harvesting was significantly more effective at approximating historic disturbance patterns thanthe activities of the energy sector. Energy sector disturbances were smaller and had fewer island rem-nants than the HRV. The composition of surviving remnant trees within anthropogenic disturbanceevents (i.e., matrix and island remnants) remains a critical area of research for approximating HRVpatterns.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Many believe that ecosystem-based management (EBM) ap-proaches to policy and land planning are important for advancingsustainable forestry practices (Seymour and Hunter, 1999). One

of the methodologies hypothesized to achieve the intended out-comes of EBM in the boreal forest is to use disturbance patternsas guides for harvest planning (Hunter, 1993). This principle positsthat anthropogenic disturbances that approximate the historicrange-of-variability (HRV) of ecosystem patterns, processes andconditions are more likely to maintain higher levels of biodiversity(Buddle et al., 2006; Levin, 2000), habitat connectivity (Saunderset al., 1991) and habitat heterogeneity (Turner, 1989) acrossa landscape relative to traditional value-optimization land

244 P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253

management. Range-of-variability1 approaches to forestland man-agement have emerged as a means to characterize the variabilityof landscape pattern that results from disturbance (Bergeron et al.,2007, 2001; Landres et al., 1999). These analysis strategies providestatistical context for evaluating the current configuration of land-scape pattern and, more recently, disturbance patterning (Andison,2012). In order for HRV initiatives to be successful, land managersand forest planners need information and tools that can help them(1) understand the dominant disturbance processes and patternsthat were historically responsible for maintaining ecosystem func-tion and landscape pattern prior to industrialisation, and (2) howthe current disturbance process and patterns differ.

An HRV methodology is theoretically well-suited to the borealforests of Canada. Across most of the boreal, stand-replacing wild-fire is the dominant disturbance agent responsible for forest struc-ture and patterns observed on these landscapes. Although themajority of wildfires are smaller than 1000 ha, the largest 1% oflightning-caused wildfires account for 98% of the total area burned(Cumming, 2001a). Mortality levels within boreal wildfires aregreat enough to replace the previous forest with a new cohort(Johnson, 1992), although recent evidence suggests that borealwildfires leave behind a large range of live vegetation (Andison,2012). At finer spatial scales, there is evidence that forest standcomposition influences burning patterns (Cumming, 2001b), andthe median size of unburned live residuals increases with wildfiresize (Eberhart and Woodard, 1987).

Boreal wildfire characteristics have already influenced harvestregime patterns. Until recently, forest harvesting in Alberta was re-quired to be practiced under a multiple-pass, clear-cut disperseddesign where vegetation is removed from similarly-sized harvestunits (McRae et al., 2001). This harvesting design results in forestedlandscapes often associated with ‘‘checkerboard’’ landscape pat-terns (DeLong, 2002; Franklin and Forman, 1987). After a periodof 10–15 years, the remaining forest adjacent to cut blocks is har-vested. Dispersed harvest designs have been shown to increaseedge density (Franklin and Forman, 1987), reduce interior old for-est area, create significant road networks with associated increasesin ecological and economic costs, reduce snag density and reduceforest patch size (Andison and Marshall, 1999; DeLong et al.,2004), all of which result in negative biological consequences (Bud-dle et al., 2006; Van Wilgenburg and Hobson, 2008).

In response to these ecological and economic concerns, forestmanagement has been shifting away from dispersed harvest de-signs over the last decade in Alberta (Work et al., 2003) and beyond(Andison and Marshall, 1999; Cyr et al., 2009; Hunter, 1993; Non-aka and Spies, 2005) in favor of harvest designs premised on wild-fire disturbance characteristics. These harvest designs aggregateharvest units into single-pass events of a range of sizes, composedof a mosaic of disturbed and undisturbed vegetation of varioussizes, shapes and proportions (Carlson and Kurz, 2007; Dzuset al., 2009) with the explicit aim of reducing the unintended neg-ative outcomes of traditional harvesting pattern on boreal mixed-wood landscapes (Van Wilgenburg and Hobson, 2008).

Consistent with EBM theories, the rationale for an aggregateharvest design is that organisms that have adapted to the spatio-temporal patterns generated by the historic wildfire regime are

1 Landres et al. (1999) define natural variability as the spatiotemporal variabilityobserved of ecological conditions and processes, relatively unaffected by people,within an explicitly-stated extent of time and space appropriate for a given question.However, this classical view is problematic as the term natural may not recognise thesignificant role First Nations may have had on the landscape prior to Europeansettlement of the present-day Americas (Lewis, 2002; Denevan, 1992; Lewis andFerguson, 1988, but also see Vale, 1998 for counter points). Therefore, the presentstudy aligns with Morgan et al. (1994) to adopt the term historic range-of-variability(HRV) to mean the spatiotemporal variability of historic disturbance patterns prior tothe period of wildfire suppression and significant human industrialization.

presumed to be more successful at persisting on the aggregate har-vest landscape due to the higher variability and availability of hab-itat types (Merriam and Wegner, 1992). More specific to the boreal,the increased heterogeneity of landscape structure and pattern as aresult of anthropogenic disturbances modelled after historic wild-fire patterns is also presumed to maintain ecological adaptivecapacity against unintended regime shifts, including climatechange impacts (Drever et al., 2006). Furthermore, the empiricalevidence in the boreal mixedwood supports this hypothesis (e.g.,Van Wilgenburg and Hobson, 2008). However, the efficacy ofaggregate harvest designs largely remains an untested hypothesis.First and foremost, the degree to which aggregate harvest effortsapproximate historic wildfire disturbance patterns is largely un-known. As Andison and Marshall (1999) found, the desire to emu-late wildfire patterns does not necessarily translate so easily intopractice. Perhaps of greater concern is that forest harvesting is onlyone of many stakeholders operating on the boreal landscape. Al-berta is a provincial jurisdiction of Canada endowed with a signif-icant forest and energy resources (e.g., oil, natural gas, and coal).The management of these natural resources over the last severaldecades has generated a complex spatial arrangement of anthropo-genic disturbances across the forested landscapes of Alberta. Theefforts of forest management agencies to emulate historic land-scape patterns may be compromised by the cumulative effects ofother stakeholders (Paine et al., 1998; Schneider et al., 2003;Timoney, 2003; Timoney and Lee, 2001).

To address these questions, disturbance patterns of a landscapein central Alberta that were attributed to forest harvesting and en-ergy extraction were investigated and evaluated for divergencesfrom a regional HRV based on known wildfire patterns. Based onthe works of Carlson and Kurz (2007), DeLong (2002), DeLongand Tanner (1996), and Burton et al. (2008) it was hypothesizedthat current anthropogenic disturbance patterns such as total areain remnants, largest disturbed patch (LDP), number and size of is-land remnants are outside the HRV for the mixedwood boreal for-est of northern Alberta. It was further hypothesized that therelative gap between current and natural patterns varies signifi-cantly by the type of anthropogenic disturbance (e.g., forestry vs.energy sector; and dispersed vs. aggregate harvest). The hypothe-ses were tested on a large forest management area in the borealmixedwood forest of Alberta.

2. Materials and methods

2.1. Study area

The Alberta-Pacific Forest Industries Inc. (Al-Pac) forest man-agement agreement area (FMA) is 5.8 million ha in size and locatedwholly within the Boreal Plains, a bio-geographical ecozone(ESWG, 1996) in north-central Alberta (56�210N, 112�220W;Fig. 1). The Boreal Plains ecozone is a national (Canadian) demarca-tion of a regional ecological system characterized by a continentalinterior climate, mostly flat undulating terrain and is composed ofmixedwood, coniferous and deciduous forests (ESWG, 1996).Approximately half of the Al-Pac FMA is covered by wetlands ofvarious types, many of which are forested (Al-Pac, 1999). In centralAlberta, the mean monthly temperature ranges from �15.5 �C inJanuary to 16.2 �C in July (Strong, 1992). Total annual precipitationis approximately 400 mm of which approximately 60 mm isderived from snowfall (Strong, 1992). The topography exhibits lim-ited variation in landforms with elevation ranging from approxi-mately 300 m to 900 m above sea level. Mesic sites tend to bedominated by Populus tremuloides Michx., Picea glauca Voss. andPopulus balsamifera L. mixedwood stands; xeric sites tend to bedominated by Pinus banksiana Lamb.; and forested wetlands aredominated by Picea mariana Mill. and Larix laricina Du Roi.

Fig. 1. Map of reference wildfires occurring within ecozones of western Canada.

P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253 245

Wildfire is by far the most common and prominent distur-bance agent in the Al-Pac FMA, and thus serves as the foundationfor the HRV estimates. Other endogenous disturbance agents in-clude windthrow and forest tent caterpillar (Malacosoma disstriaHübner), but both create very fine scale disturbances on a far lessfrequent basis and are associated with relatively low levels ofmortality (Roland, 1993). The current range-of-variability (CRV),as used in this study, represents the variability of anthropogenicdisturbance patterns following human industrialization in the re-gion during the 1940s. Since the late 1940s, sources of anthropo-genic disturbance to the Al-Pac FMA include extensive energyexploration near Fort McMurray, the extraction of bitumen fromthe associated bituminous sands via open pit mining and in situoperations, natural gas extraction, timber harvesting, agriculturalland conversion, town settlements, and linear right-of-way dis-turbances such as railways, roadways, seismic lines, transmissionlines and pipelines. Prior to the inception of the FMA in 1993, for-est management was primarily characterized by multi-pass, dis-persed harvesting. Since circa 2000, Al-Pac has shifted theirtimber harvesting planning toward single-pass, aggregate harvestdesigns (Dzus et al., 2009). Anthropogenic disturbances attributedto forest harvesting and energy extraction were both included inthe CRV.

2.2. Data collection

2.2.1. Historic wildfire dataPrevious analysis of 87 aerially-interpreted wildfires with no re-

cord of suppression (Andison and McCleary, in press) provided abaseline for HRV disturbance patterns. All wildfires occurred be-tween 1948 and 2004 with a geographic extent spanning from110�00W to 120�00W and from 54�420N to 60�10N within the BorealShield, Boreal Plains, Taiga Plain and Taiga Shield ecozones of Can-ada, and were mapped to a minimum polygon size of 2 ha (Fig. 1).As detailed in Andison and McCleary (in press), the criteria used toselect unsuppressed wildfires for analysis had an upper size limitof 30,000 ha and therefore does not reflect the historical distribu-tion of wildfire size.

2.2.2. Anthropogenic disturbance dataSampling a FMA presents several challenges. First, the shape is

irregular and, in the case of the Al-Pac FMA, there are several areasof unmanaged forest within the tenure. Additionally, different re-gions of the Al-Pac FMA are managed to different intensities basedon regional stakeholders (e.g., First Nations, energy companies)and timber quality. In terms of the spatial patterns of disturbance,it was assumed that harvest pattern and extent was similar overthe entire FMA. Therefore, data to represent the CRV were obtainedfrom a sub-sample of four areas within the Al-Pac FMA, each 12townships (approximately 112,000 ha) in size (Fig. 2). These foursample areas were chosen to represent the range of CRV sourcesand impacts across the FMA and were qualitatively selected foraggregate and dispersed harvest disturbances by staff at Al-Pac(see Table 1). Further analysis of site conditions for each samplearea found that forested AVI polygons accounted for between77.5% and 85.3% of each total land base. Populus tremuloides and Pi-cea mariana are the dominant forest species across all sampleareas, accounting for between 58.3% and 85.7% of each total for-ested land base.

For each of the four areas, spatial datasets representing the dis-turbance activities of the forestry and energy sectors were createdas follows. First, the most recent (circa 1999) aerially-derived Al-berta vegetation inventory (AVI) data, mapped at a minimum poly-gon size of 2 ha, were obtained (Alberta ESRD, 2005). All majorhighways, well sites, pipelines, permanent water bodies and forestharvest polygons were extracted from the AVI dataset. Second, themost recent spatial data for seismic lines, utility lines, roads, rail-ways and timber harvesting were sourced from GIS databases atAl-Pac. To avoid bias associated with the sample areas, all distur-bance events that partially fell outside of the sample areas wereexcluded from analysis. Linear features, such as roads and seismiclines, were clipped at the sample area borders.

2.3. Disturbance pattern language

A challenge of characterizing disturbance patterns is being con-sistent and unambiguous with the selected pattern terms and

Fig. 2. Map of Al-Pac study area.

Table 1Approximate areal contributions of different sources of disturbance to the land baseof each sample area derived from the AVI dataset.

Disturbance source Study area

A (ha) B C D

Forestry industry 9140.7 8276 2913.6 855.2Energy industry 1343.4 602.9 182.9 426.7Agriculture 0 9.8 0 0Other industry 986.7 44.7 43.3 59.4Wildfire 2728.6 15,630.4 11,827.3 320.7Windthrow 71.5 20.1 58.7 2.4

246 P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253

spatial language (Pickett and White, 1985). Subtle differences inthe delineation of wildfire boundaries can create significant differ-ences in wildfire patterns (Andison, 2012), which can ultimatelycreate regulatory and credibility challenges. In this paper, the dis-turbance pattern language developed by Andison (2012), which ispredicated on the concept of the general area of influence of a wild-fire (sensu Burton et al., 2008), was adopted for analysing the spa-tial layers. This spatial language involves both mapped andgenerated spatial elements, which together delineate spatially-dis-crete and objectively-defined disturbance events (Andison, 2012).The mapped elements derived from aerial interpretation includeboth disturbed patches (i.e., >95% crown mortality) and island rem-nants. A buffering process takes the mapped spatial elements (i.e.,islands and disturbed patches; Fig. 3a) and buffers them by user-defined spatial and temporal parameters (Fig. 3b). The final step

is to apply a negative buffer (or nibble) of the same distance tothe mapped spatial elements (Fig. 3c). These two steps generatea third type of spatial element known as matrix remnants, whichoccur among disturbed patches (Fig. 3d). This process aggregatesdisturbed patches (e.g., harvest areas) that are close in space andtime and delineates a single spatial entity known as the disturbanceevent, which consists of the input spatial layers, the generated ma-trix remnants and the generated disturbance event boundary. Thisthree-level spatial language was adopted because it represents his-torical burning patterns and processes more precisely than the tra-ditional binary system originally developed by MacArthur andWilson (1967) which is predicated on the concept of habitable ‘‘is-lands’’ surrounded by an inhabitable ‘‘matrix’’.

2.4. Data analysis

2.4.1. NEPTUNE decision-support toolDisturbance patterns were generated for historic wildfires and

anthropogenic disturbance data using the web-based NEPTUNE(Novel Emulation Pattern Tool for Understanding Natural Events)decision-support tool (TFC, 2009). NEPTUNE combines the spatiallanguage described in Section 2.3 with the HRV of wildfire patternsas described in Section 2.2.1. NEPTUNE uses a spatial and temporalbuffer to merge disturbed patches into disturbance events (as de-scribed by Andison (2012)). All disturbance polygons were as-signed the same year of disturbance (which allows NEPTUNE toconsider all neighboring polygons). The spatial and temporalparameters of the spatial language have the flexibility to be scaled

Fig. 3. Diagram of the NEPTUNE process for delineating disturbance events.

P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253 247

in NEPTUNE depending on the nature of the disturbances on thelandscape. For the present study, a 200 m buffering thresholdwas selected because Andison (2012) found that, at this distancefor boreal landscapes, most disturbed patches of wildfires werecorrectly aggregated into their respective disturbance events, how-ever, it was not too distant that patches from other disturbanceswere incorrectly aggregated. According to the spatial language,roads are treated as fully disturbed polygons, and, although theyare not used in the creation and delineation of disturbed patches,roads are used to identify island remnants that are wholly sur-rounded by road or that lay between a disturbed patch and a road.As a result, roads, which occur within an anthropogenic distur-bance event, may aid in the creation of island remnants among dis-turbed patches.

2.4.2. Pattern indicesA suite of metrics that characterize disturbance events can be

derived from NEPTUNE including event area, matrix remnantsarea, island remnants area and total remnants area and as a pro-portion of event area, disturbed patch density, and LDP area andas a proportion of event area. Through secondary spatial analysis,island density, largest island area, mean island area, disturbedpatch area and as a proportion of event area, and mean disturbedpatch area were further derived.

2.4.3. Data pre-processingSeismic lines were buffered using ArcMap 10 (ESRI Inc. 2010) by

the width defined within the dataset, which ranged from 1 m to8 m. When a seismic line feature lacked a width identifier, thearithmetic mean width value of all seismic lines (�2 m) was used,which was applied to 147 features representing �5% of total seis-mic line length in the study area. Utility lines and road featureswere buffered by 50 m to either side of the linear feature repre-senting the average width (100 m) of major right-of-way featuresin Alberta. Unimproved roads and trails were excluded from anal-ysis because the right-of-way width could not be reliably deter-mined. Dispersed harvests (e.g., harvesting that occurred beforethe year 2000) were analyzed separately from aggregated harvests(e.g., harvesting that occurred after the year 2000). The AVI dataand fire history records revealed no instances of wildfire in the se-lected harvested areas.

2.4.4. Statistical analysisThe non-parametric Kolmogorov–Smirnov (K–S) test was used

to assay statistical differences between the cumulative distributionfunctions (CDF) of each disturbance source to the empirical cumu-lative distribution function (ECDF) provided by the HRV. TheMann–Whitney–Wilcoxon (MWW) rank sum test was used to as-say significant differences between the location parameters (med-ian) of anthropogenic treatments and HRV. The K–S test statistic Dis a measure of the greatest observed vertical departure betweenany two CDF. Therefore, the K–S test is useful for analysing differ-ences between two distributions that do not meet the assumptionsrequired for more common parametric analysis. The MWW teststatistic U is the sum of all the ranked differences for each observa-tion between any two treatments. The MWW test is useful for indi-cating differences in location between two similarly-shapedsamples with positively skewed data, and, for large sample sizes,the distribution of U approaches a normal distribution. Both testswere performed in R (R Development Core Team, 2011) as two-tailed with significant relationships at a = 0.05.

3. Results

3.1. Historic wildfire patterns

The median area of historic wildfires in island remnants was22%, compared to just 6.4% for matrix remnants (Table 2). All ofthese wildfire disturbances were composed of fewer than 22 totaldisturbed patches and the LDP accounted for 88% of the disturbedarea at the median and 85% on average. The mean size of disturbedpatches was 247 ha at the median and 762 ha on average, whichsuggests the data are positively skewed. Mean island remnant areawas 32 ha on average and 9 ha at the median. The largest single is-land remnant was 459 ha on average and 28 ha at the median. Thenumber of island remnants associated with historic wildfires ran-ged from 0 to 272, where the median and mean were equivalentto 7 and 20, respectively.

3.2. Anthropogenic disturbance patterns

From all of the input spatial layers, 15,431 anthropogenic dis-turbed patches were identified. The NEPTUNE analyses of the

Table 2Summary table of historic range-of-variability of wildfire disturbance patterns.

Metric Range-of-variability

xmin X� xmax X

�(sd)

Event area (ha) 11.3 589.4 28,040 2159 (4732)Disturbed patch area (ha) 9.2 551 27,310 2001 (4538)Island remnants area (ha) 0 77.9 23,774 758.3 (2748)Matrix remnants area (ha) 0 40.5 1384 128.4 (226.4)Total remnants area (ha) 0 132.9 24,450 886.6 (2862)Largest disturbed patch (ha) 9.2 490.7 27,267 1941 (4511)Largest island remnant (ha) 0 28.2 22,634 458.5 (2439)Mean island area (ha) 0 8.7 354.3 32.1 (67.3)Mean disturbed patch area (ha) 9.2 247.4 12,005 761.6 (1563)Number of island remnants 0 7 272 20.4 (37.9)Number of disturbed patches 1 1 22 3 (3.7)Proportional area in disturbed patches (%) 70.2 88.5 99.1 88.5 (6)Proportional area in island remnants (%) 0 21.7 85.1 26 (19.5)Proportional area in matrix remnants (%) 0 6.4 25.5 6.8 (5.5)Proportional area in total remnants (%) 0 30.2 87.6 32.8 (21)Proportional area in largest disturbed patch (%) 44.8 88 99.1 85.1 (11.6)

248 P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253

spatial layers for the study area delineated 1355 anthropogenicdisturbance events. Of those, 126 were forestry-related distur-bance events generated from 1210 forest harvest polygons and1057 were energy-related disturbance events generated from1504 well site polygons. Forest harvest areas were further resam-pled by two harvest designs: 98 disturbance events were generatedfrom 707 dispersed harvest pre-2000 polygons and 74 disturbanceevents were generated from 722 aggregate harvest post-2000 poly-gons. The resampling of forestry-related disturbed features by har-vest designs generated more disturbance events than the forestrysector overall because the spatial arrangement of forestry features(i.e., harvest areas) varied across the landscape by harvest design,resulting in different delineations of disturbance events using thesame set of disturbed features.

3.2.1. Forestry-related disturbancesOverall, patterns attributed to forestry-related disturbances

were characterized by events that were 48 ha in size at the medianand 261 ha on average (Table 3). Approximately 18% of the medianevent area for these disturbances was represented by matrix rem-nants or �12% higher than HRV observations (U = 1,718,p << 0.001; D = 0.58, and p << 0.001) and only 3% was representedby island remnants or �19% lower than HRV observations(U = 6,851, p = 0.002; D = 0.43, and p << 0.001). The median propor-tional area represented by the LDP was 68%, which was �20% low-er than HRV observations (U = 8,006.5, p << 0.001; D = 0.44, andp << 0.001). Mean island area for all forestry-related disturbanceswas 6 ha on average and only 1.1 ha at the median, but withzero-island events omitted the median is adjusted to 5.3 ha.

3.2.1.1. Aggregate harvest designs. Aggregate harvest events werecomposed of few disturbed patches, 1 at the median and 3 on aver-age (Table 3). Proportionally, the LDP accounted for 82% of the dis-turbance event area at the median, which was�6% lower than HRVobservations (U = 4,370, p << 0.001; D = 0.35, and p < 0.001), andranged in size from 3 ha at the minimum to 867 at the maximum.Island and matrix remnants represented 1.6% (U = 4,775,p << 0.001; D = 0.56, and p << 0.001) and 16% (U = 1,073.5,p << 0.001; D = 0.59, and p << 0.001) of the event area at the med-ian, respectively, and were �20% lower and �10% higher than HRVobservations, respectively. Total remnants, that is, matrix and is-land remnants combined, accounted for 25% of event area at themedian and 35% on average, which was �5% lower than HRVobservations and not significantly different from the HRV(U = 3,386, p = 0.572; D = 0.16, and p = 0.281). Mean island area

for aggregate harvests was small, less than 1 ha at the median,but with zero-island events omitted the median is adjusted to5.3 ha, and only 5 ha on average. The number of island remnantswere few, 1 at the median and 5 on average, despite these aggre-gate harvest designs reaching sizes of 45 ha at the median and174 ha on average.

3.2.1.2. Dispersed harvest designs. Proportional area in island rem-nants for dispersed harvest designs was 21% at the median, whichwas �1% lower than HRV observations and not significantly differ-ent from the HRV (U = 4,259, p = 0.99) and 34% on average, but theoverall distribution was significantly different from the HRV(D = 0.29, p = 0.001). Mean island area was quite large, 9 ha at themedian and 18 ha on average. There were few disturbed patchesin these harvest designs, 2 at the median and 3.5 on average. Pro-portionally, the LDP represented 61% of the total disturbance eventarea at the median, which was �27% lower than HRV observationsand significantly different from the HRV (U = 6,313, p << 0.001) and62% on average. There were also few island remnants, 2 at themedian and 5 on average. The largest island remnant in these har-vest designs was 31 ha in size at the median and 51 ha on average.Matrix remnants represented 18% of total event area at the median,which was �12% higher than HRV observations, and 20% on aver-age. Total remnants accounted for 55% of total event area at themedian, which was �25% higher than HRV observations and signif-icantly different from the HRV (U = 2918, p < 0.001).

3.2.2. Energy-related disturbancesA high degree of variability was associated with energy-related

disturbance metrics. For example, the number of island remnantsranged from 0 to 1239, 0 at the median and 1.7 on average witha standard deviation of ±39.7. Although the presence of a denseseismic line network in the largest energy-related event resultedin 1239 individual island remnants, totalling 332 ha in area, the98th percentile of energy-related disturbances contained no islandremnants at all. Energy-related disturbances were significantly dif-ferent from the HRV in terms of all the metrics that were investi-gated. Proportional area in island and matrix remnants was 0%(U = 88,423, p << 0.001; D = 0.92, and p << 0.001) and 0%(U = 77,436.5, p << 0.001; D = 0.7, and p << 0.001) at the median,which was �22% and �6% lower than HRV observations, and0.2% and 2.2% on average, respectively. Total remnants only ac-counted for 0% of total event area at the median, which was�30% lower than HRV observations (U = 86,852, p << 0.001;D = 0.83, and p << 0.001), and 2.4% on average. By contrast, the

Table 3Summary table of comparisons between observed cumulative distribution functions (anthropogenic disturbances) and the empirical distribution function (HRV).

Disturbance Source Metric (x) Range-of-variability MWW testa K–S testa

xmin X� xmax X (sd) U p-Value D p-Value

Forestry (n = 126) Event area (ha) 2.7 47.6 6116 260.8 (817.4) b – b –Disturbed patch area (ha) 2.1 37.1 5631.5 197.7 (663.2) b – b –Island remnants area (ha) 0 2.4 2647 71.1 (283.9) b – b –Matrix remnants area (ha) 0.1 8.2 1758 63.1 (193.5) b – b –Total remnants area (ha) 0.1 18.4 4405 134.2 (456.8) b – b –Largest disturbed patch (ha) 2.1 27.8 3514 131 (424.8) b – b –Largest island remnant (ha) 0 1.8 910.8 24.6 (86) b – b –Mean island area (ha) 0 1.1 54.5 6 (10.2) b – b –Mean disturbed patch area (ha) 2.1 18.8 302.1 33.1 (39.6) b – b –Number of island remnants 0 1 295 9.1 (32.5) b – b –Number of disturbed patches 1 2 50 3.7 (6.6) b – b –Proportional area in disturbed patches (%) 51.1 81.9 99.4 80.1 (11.8) 7808 <<0.001 0.39 <<0.001*

Proportional area in island remnants (%) 0 3.1 98.5 24.3 (33.1) 6851 0.002 0.43 <<0.001*

Proportional area in matrix remnants (%) 0.6 18.2 48.9 19.9 (11.8) 1718 <<0.001 0.58 <<0.001*

Proportional area in total remnants (%) 0.6 33.3 100 44.2 (32.4) 4678 0.07 0.21 0.018*

Proportional area in largest disturbed patch (%) 16.8 68.3 99.4 66.1 (23.6) 8006.5 <<0.001* 0.44 <<0.001*

Aggregate harvest (n = 74) Event area (ha) 3 45.3 2015 173.9 (387.4) b – b –Disturbed patch area (ha) 2.6 34.7 1331.4 125.3 (262.6) b – b –Island remnants area (ha) 0 0.4 279 26.2 (58.1) b – b –Matrix remnants area (ha) 0 5.8 755.5 48.6 (131) b – b –Total remnants area (ha) 0.1 8.6 980.1 74.8 (174.5) b – b –Largest disturbed patch (ha) 2.6 26.4 867.3 83.5 (166.8) b – b –Largest island remnant (ha) 0 0.4 159.4 15 (32.9) b – b –Mean island area (ha) 0 0.4 64.5 5.1 (12.3) b – b –Mean disturbed patch area (ha) 2.6 15.1 176.5 30.8 (36.4) b – b –Number of island remnants 0 1 72 4.9 (12.2) b – b –Number of disturbed patches 1 1 21 3.2 (4) b – b –Proportional area in disturbed patches (%) 54.9 83.6 100 81.9 (10.8) 4432.5 <0.001* 0.35 <0.001*

Proportional area in island remnants (%) 0 1.6 100 16.8 (31) 4775 <<0.001 0.56 <<0.001*

Proportional area in matrix remnants (%) 0 16.5 45.1 18.1 (10.8) 1073.5 <<0.001 0.59 <<0.001*

Proportional area in total remnants (%) 1.5 25.1 100 34.9 (29.1) 3386 0.572 0.16 0.281*

Proportional area in largest disturbed patch (%) 24.1 82.4 100 70.5 (23.1) 4370 <<0.001 0.35 <0.001*

Dispersed harvest (n = 98) Event area (ha) 2.5 48.4 4901.3 158.6 (509.4) b – – –Disturbed patch area (ha) 2.1 43.1 2779.4 105.7 (289.9) b – – –Island remnants area (ha) 0 10 633.3 45.2 (92.3) b – – –Matrix remnants area (ha) 0.1 8.7 2121.8 52.8 (221.1) b – – –Total remnants area (ha) 0.3 25.3 2755.1 98 (300.3) b – – –Largest disturbed patch (ha) 2.1 31 792.6 50.8 (85.4) b – – –Largest island remnant (ha) 0 18.3 173.8 28.4 (34.7) b – – –Mean island area (ha) 0 8.6 127.5 18.1 (24) b – – –Mean disturbed patch area (ha) 1.3 12.5 102.1 19.2 (16.8) b – – –Number of island remnants 0 2 136 5.4 (14.6) b – – –Number of disturbed patches 1 2 54 3.5 (6.2) b – – –Proportional area in disturbed patches (%) 46.9 81.9 99.3 80.2 (12) 6084.5 <<0.001* 0.42 <<0.001*

Proportional area in island remnants (%) 0 20.7 97.4 34.5 (36.1) 4259 0.992 0.29 <<0.001*

Proportional area in matrix remnants (%) 1 18.1 53.1 19.8 (12) 1370 <<0.001* 0.61 <<0.001*

Proportional area in total remnants (%) 2.4 54.7 100 54.3 (35.4) 2918 <0.001 0.38 <<0.001*

Proportional area in largest disturbed patch (%) 7.5 61.1 99.3 61.7 (26.4) 6313 <<0.001 0.49 <<0.001*

Energy (n = 1057) Event area (ha) 0.2 0.9 491.2 2.1 (18.8) b – – –Disturbed patch area (ha) 0.2 0.9 450.8 1.8 (15.6) b – – –Island remnants area (ha) 0 0 331.7 0.5 (10.7) b – – –Matrix remnants area (ha) 0 0 125.5 0.3 (4.3) b – – –Total remnants area (ha) 0 0 372.1 0.8 (13.5) b – – –Largest disturbed patch (ha) 0.2 0.9 409.2 1.6 (13.5) b – – –Largest island remnant (ha) 0 0 8.8 0 (27.9) b – – ––Mean island area (ha) 0 0 0.3 0 (1.8) b – – –Mean disturbed patch area (ha) 0.2 0.9 41 1 (1.3) b – – –Number of island remnants 0 0 1239 1.7 (39.7) b – – –Number of disturbed patches 1 1 55 1.3 (2.1) b – – –Proportional area in disturbed patches (%) 22.7 100 100 97.8 (7.4) 6631.5 <<0.001* 0.83 <<0.001*

Proportional area in island remnants (%) 0 0 67.5 0.2 (3.1) 88,423 <<0.001 0.92 <<0.001*

Proportional area in matrix remnants (%) 0 0 77.3 2.2 (7.4) 77,436.5 <<0.001* 0.70 <<0.001*

Proportional area in total remnants (%) 0 0 77.4 2.4 (8.4) 86,852 <<0.001 0.83 <<0.001*

Proportional area in largest disturbed patch (%) 3 100 100 93.4 (17.1) 10,671.5 <<0.001* 0.81 <<0.001*

a Mann–Whitney–Wilcoxon (MWW) and Kolmogorov–Smirnov (K–S) tests were performed against the HRV (see Table 1).b No tests were performed on absolute, area-based metrics.

* Indicates that ties were present in the data, therefore the calculated p-value is approximate.

P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253 249

250 P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253

LDP represented 100% of total event area at the median, which was12% higher than HRV observations (U = 10,671.5, p << 0.001;D = 0.81, and p << 0.001), and 93% on average.

4. Discussion

The areal pattern metrics are valuable in that they allow com-parisons between different CRV sources, but they are less reliableas indicators of wildfire pattern emulation because they all co-varywith event size. Furthermore, since the 87 wildfires used for theHRV estimate were not chosen randomly, their sizes do not neces-sarily reflect an empirical distribution of wildfire size. Therange-of-variability was recorded for all areal pattern metrics,but comparison tests to the HRV were not performed. However,the results may still be qualitatively compared to other regionalstudies of empirical fire size. For example, Cumming (2001a) esti-mated the 95% confidence interval for maximum fire size in thestudy area to be between 227,000 ha and 1,820,000 ha, but thelargest anthropogenic disturbance recorded in this study was a6116 ha forest harvest. Additionally, the largest 1% of wildfires(>1000 ha) accounted for 98% of the total area burned/disturbed(Cumming, 2001a), but the largest 1% of forestry-related distur-bances accounted for only 37% of the total area disturbed by forestharvesting. In other words, the majority of the landscape disturbedby forest harvesting is derived from many small harvest events.This finding supports other work where landscape patch size isconsidered to be outside an HRV due to forest harvesting and landmanagement (Nonaka and Spies, 2005). The five proportional pat-tern measures are fire size-invariant (Andison, 2012), and thusform the basis for the CRV–HRV comparisons communicated here.

4.1. Disturbance patterns of the forestry and energy sectors

Forestry-related disturbances better approximated the HRVthan did the energy-related disturbances in several importantways. Energy-related disturbances created a large number of extre-mely small, dispersed disturbances with virtually no surviving veg-etation. In sharp contrast, the proportion of total remnants forforestry-related disturbances aligns very well to the HRV. Simi-larly, at the median, the largest disturbed patch for forestry wasalso fairly similar to that of the HRV (83% vs. 88%, respectively),compared to 100% at the median for energy-related disturbances.The high LDP value for energy-related disturbances reflects the factthat most (90%) of the disturbance events by the energy sector in-cluded only a single disturbed patch. Finally, although not statisti-cally comparable to the HRV, it is notable that the mediandisturbance size of forest harvests was 48 ha, compared to lessthan 1 ha for energy-related disturbances. The data support thehypothesis that, relative to forest management, the energy sectorcreates a large number of very small evenly-distributed distur-bance events with little of no residual vegetation.

The most notable trend observed from the analysis was the con-vergence of the proportion of total remnants for forestry-relateddisturbances with the HRV. The fact that the proportional area inisland remnants was under-represented and proportional area inmatrix remnants was over-represented relative to the HRV is ofless concern, but signifies a sampling artifact whereby uncut areaswithin dispersed harvests were coded as matrix remnants. Thephysical distinction between island and matrix remnants is subtle,and in this case most of the islands were partially disturbed stripcuts. However, it is possible that these differences may relate toecological function. Island remnants are physically isolated fromthe nearby matrix of intact forest and thus serve as refugia forsmall-bodied wildlife and seed sources for local flora (Eberhartand Woodard, 1987; Eberhart, 1986; Banks et al., 2011). Matrix

remnants are undisturbed forest residuals of disturbances thatare physically attached to the surrounding matrix of intact forest,and function as corridors and cover for the movement of wildlifebetween forested patches (Renjifo, 2001). The higher representa-tion of proportional matrix remnants in forestry-related distur-bances is an indication of fragmentation and is likely an artifactof dispersed harvesting strategies whereby cut blocks are spacedevenly throughout a landscape, thereby resulting in high levels ofmatrix remnants between cut blocks. However, many of the dis-persed harvests that were sampled had only been through one ofthe two passes. If harvests that had been through two passes weresampled, then matrix remnants would likely be less represented.Although given this limitation, only a handful of the pattern indicesquantified for dispersed harvests would be expected to deviatemarkedly from the current calculations if the second pass had beenincluded. For example, island remnant levels, largest disturbedpatch and mean patch size should not vary greatly from currentquantities because this harvesting strategy reduces variability inpatch size, but matrix remnant levels, total remnant levels, thenumber of disturbed patches and all of the areal metrics would cer-tainly be affected. Furthermore, one challenge associated withsampling anthropogenic disturbances that vary across time, inaddition to space, is assessing when an event is ‘‘complete’’ or dis-crete in time. This creates a special challenge for comparing dis-persed harvests, which can take decades to ‘‘finish’’, withwildfires that may burn over the course of weeks to months.

4.2. Disturbance designs of forest management

The analyses suggest that aggregate harvest patterns in the Al-Pac FMA better approximated the HRV than dispersed harvests interms of event size, mean disturbed patch area, disturbed patchdensity, proportional area in matrix remnants and total remnants,and the size of the LDP (Table 3). However, some dispersed harvestpatterns were noteworthy in relation to wildfire patterns. Forexample, dispersed harvests generally better approximated theHRV in terms of absolute median sizes of spatial elements such asisland, matrix and total remnants area, LDP area, largest island area,mean island area and mean disturbed patch area. Additionally,these disturbances featured more island remnants than aggregateharvests, but were also characterized by a higher proportional arearepresented by matrix remnants. Some of these differences can beexplained by the fact that many of the dispersed harvests that weresampled had only been through one of the two harvesting passes.After only one pass, the area designated for the second pass is ma-trix remnants. Thus what we have captured in the results is a com-parison of the current condition of disturbance patterns. As secondpasses are completed, many of the metrics for the dispersed har-vesting pattern will shift. For example, if two passes are completedover a relatively short period of time, one would expect the LDP toincrease and matrix remnants to decrease. However, extended peri-ods between passes would effectively create two overlappingevents, both with very unnatural patterns. Overall, the effort toemulate wildfire disturbance patterns over the last several yearshas been successful, although further improvement is possible.

Although aggregate harvests did not perform better than dis-persed harvests on absolute area-based metrics (e.g., event area, is-land remnant area, LDP area, etc.), they did converge on historictargets for several key proportional metrics. Proportional metricsof the disturbance event area such as LDP, matrix and island rem-nants area, and disturbed patch area are critical for approximatinghistoric mortality patterns, which are important for enhancingpost-disturbance habitats. The lower proportional area occurringas matrix remnants in aggregate harvests is an indication that thisharvesting strategy is achieving the goal of better approximatingwildfire patterns on the landscape (Fig. 4). The lower proportion

P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253 251

of matrix remnants suggests that more of the disturbance area isallocated to disturbed patches thus resulting in a more compactevent which better approximates historic levels and is more likelyto reduce road density, and thus edge density, in harvest areas.Similarly, the LDP in aggregate harvests tended to account for agreater proportion of the disturbance event area than dispersedharvests, which is an important characteristic of historic fires.The sizes and proportion of island remnants in aggregate harvestsdiverged from historic levels more so than dispersed harvests andforest harvesting overall (Fig. 4). However, most of the islands indispersed harvest designs were partially disturbed strip cuts, manyof which were designed for removal during a second pass. The factthat dispersed designs created larger disturbance events, combinedwith the relatively low area in disturbed patches, demonstratesone of the concerns of these harvesting strategies: vast harvesting‘‘footprints’’ on the landscape. Aggregated harvest designs aremore compact (i.e., lower proportion of matrix remnants) meaningthat they are more likely to leave significant areas of undisturbedforest in which no roads or management activities are requiredto be maintained for extended periods of time. Over time, onewould expect that an aggregate harvesting strategy would resultin fewer and larger forested patches across a given landscape; apattern that is consistent with the historic wildfire regime.

There are some data artifacts worth noting for all disturbancesources. For example, LDP as a proportion of event size is artifi-cially inflated when there is only one disturbed patch per distur-bance event, which was true of 90% of all energy-relateddisturbances. The tendency of dispersed harvests to approximatehistoric levels of island remnant-related metrics may be attributedto the higher likelihood of roads occurring in these events.Although the density of roads was not quantified or analysed inthis study, the effects of roads on the generation of spatial ele-ments like island remnants is related to the spatial language itself.

4.3. Management interpretations

In order for ecosystem-based initiatives to be successful ascoarse-filter approaches to biodiversity conservation, managementstrategies must be implemented that approximate the range-of-variability of historic disturbance patterns. For forest planners,the sizes and proportional area of island and matrix remnants re-mains a critical characteristic of forest disturbances which isimportant for maintaining core and refugia habitats for a varietyof forest mammals and avifauna (Eberhart, 1986; Eberhart andWoodard, 1987; Matthiae and Stearns, 1981; Whitcomb et al.,

Fig. 4. Plot of median island and matrix remnants as a proportion of total eventarea by disturbance class. The whiskers represent the upper bound 75th quartileand the lower bound 25th quartile for both axes.

1981). Although forestry-related disturbances generated less is-land remnants, and more matrix remnants compared to the HRV,combined total remnants were not significantly different fromthe HRV at the median. The physical distinction between islandand matrix remnants is subtle. However, it is possible that thesedifferences may relate to ecological function. Island remnants arephysically isolated from the nearby matrix of intact forest and thusserve as refugia for small-bodied fauna and seed sources for localflora (Eberhart and Woodard, 1987; Eberhart, 1986; Banks et al.,2011). Matrix remnants are undisturbed forest residuals of distur-bances that are physically attached to the surrounding matrix ofintact forest, and functions as corridors and cover for the move-ment of fauna between forested patches (Renjifo, 2001). Historicalremnants patterns may also mitigate concerns about larger anthro-pogenic disturbance events. For example, the inclusion of matrixremnants may allow greater flexibility within larger harvests thatmay be beneficial for larger-bodied mammals, which prefer thedense understory foliage of forests for foraging in the fall (Lewis,2002). Yet, larger harvest areas alone do not necessarily replicatethe heterogeneity of post-burn, structural characteristics of histor-ical forested systems. Thus, disturbance patterns in isolation ofother factors like biotic feedbacks which affect the efficacy of suc-cession-based ecosystem restoration strategies (Song, 2002; Sud-ing et al., 2004). For example, there is no provincial mandate forthe energy sector to re-vegetate seismic lines or well sites in Alber-ta and native species revegetation is costly and voluntary (MacFar-lane, 2003, 1999).

The disturbance event scale of analysis presented here has somedisadvantages for planning at the landscape scale. For example, atthe resolution of the landscape, non-harvested forest matrix be-tween disturbances and matrix remnants between incompletemulti-pass harvests provide cover and function as corridors forfauna, but these elements of landscape composition are not cap-tured by the results. Thus, comparison of HRV and anthropogenicpatterns at larger spatial scales is an area requiring future research.The cumulative nature of anthropogenic disturbances on the Alber-ta landscape was not considered by this study. However, investiga-tions into the sources of and interactions between anthropogenicdisturbances would greatly benefit the cooperative land planninginitiatives between the energy and forestry sectors in Alberta. Fur-thermore, disturbance-based forest research would benefit fromfuture lines of inquiry that examine the cumulative nature ofanthropogenic disturbance activities. Such research should notonly investigate the combined pattern impacts of forestry andthe energy sector, but also the nature and spatial configuration oflinear right-of-ways such as roads, seismic lines, pipelines andtransmission lines. These linear features are ubiquitous in Albertain particular due to rich energy deposits and function as potentialsources of both fragmentation and dispersal networks for mam-mals, forest avifauna and flora (MacFarlane, 2003; Machtans,2006). Ideally, such studies would be augmented with fine filter re-search and effectiveness monitoring required to evaluate if EBMapproaches are actually fulfilling the biotic response assumptionsinherent in their principles.

The historical role of First Nations burning in the central mixed-woods of Alberta is underrepresented in historical ignition data.Similarly, First Nations burning in northern Alberta tended to beseasonal in order to manipulate the distribution of regional faunaand flora (Lewis, 1982). The degree to which historic First Nationsburning altered landscape patterns in the northern Boreal Plainshas been documented because wildfire ignitions by the Plains Creeoccurred well into the 1940s in northern Alberta (Ferguson, 1979).Few studies incorporate these pre-industrial anthropogenic distur-bance activities into account when considering local flora and fau-na, yet the effects of such anthropogenic influences on thespatiotemporal distribution of understorey vegetation remains an

252 P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253

important factor in the consideration of faunal foraging, habitationand movement throughout boreal landscapes (Lewis, 1982).

Finally, the elapsed time between dispersed harvest passes iscrucial for modeling these disturbances. Many of the disturbancesthat are communicated here were incomplete at the planning leveldue to a multi-pass design. In addition, the road network associ-ated with dispersed harvesting can fragment a landscape for sev-eral decades. Completing all harvests passes as near together intime would simulate natural disturbances and reduce the neces-sary road infrastructure.

5. Conclusions

Overall, it was determined that forest management activitiesbetter approximated the HRV relative to the disturbance activitiesof the energy sector. Furthermore, more recent attempts to inte-grate more natural harvest designs via aggregate harvesting pat-terns did in fact create more natural disturbance patterns thanthe traditional two-pass dispersed harvesting system. These find-ings suggest that the ideals of EBM can only in part be achievedthrough the efforts of individual land management agencies. Aconcerted effort to better understand and manage for the cumula-tive effects of anthropogenic disturbance patterns is required if thecurrent delivery of ecosystem services is to be sustained from for-ested landscapes of Alberta.

Role of the funding source

Scientists and technicians at Al-Pac provided data for analysis,were invited to review this manuscript, and were consulted onthe sampling design of their FMA. Primary data analysis, study de-sign, and interpretations were performed by the authors.

Acknowledgements

First and foremost, we would like to thank Alberta-Pacific For-est Industries Inc. (Al-Pac), Hinton Wood Products (West FraserMills), Alberta Newsprint Company (ANC), the Foothills ResearchInstitute, and the National Sciences and Engineering ResearchCouncil of Canada (NSERC) for supporting this project. We alsothank M. Smith (Al-Pac) for curating and providing us with the spa-tial data necessary to carry out our analyses, B. Maier (TFC) for pro-viding technical assistance with NEPTUNE, and E. Dzus (Al-Pac) forsupporting the project and providing valuable feedback during thestudy design and write up. We are grateful for the helpful com-ments of two anonymous reviewers.

References

Alberta Environment and Sustainable Resource Development (ESRD), 2005. AlbertaVegetation Inventory Interpretation Standards. Version 2.1.1. Alberta ResourceInformation Management Branch and Sustainable Resource Development. 99 p.<http://www.srd.alberta.ca/LandsForests/documents/AVI-ABVegetation3-InventoryStan-Mar05.pdf> (accessed 16.08.12.).

Alberta-Pacific Forest Industries, 1999. Detailed Forest Management Plan. Alberta-Pacific Forest Industries, Boyle, 194 p.

Andison, D.W., 2012. The influence of wildfire boundary delineation on ourunderstanding of burning patterns in the Alberta foothills. Can. J. Forest Res.42, 1253–1263.

Andison, D.W., Marshall, P.L., 1999. Simulating the impacts of landscape-levelbiodiversity guidelines: a case study. Forest Chron. 75 (4), 655–665.

Andison, D.W., McCleary, K., in press. Detecting regional wildfire burning patternsin western boreal Canada. Forest Chron.

Banks, S.C., Dujardin, M., McBurney, L., Blair, D., Baker, D., Lindenmayer, D.B., 2011.Starting points for small mammal population recovery after wildfire:recolonisation or residual populations? Oikos 120, 26–37.

Bergeron, Y., Gauthier, S., Kafka, V., Lefort, P., Lesieur, D., 2001. Natural firefrequency for the eastern Canadian boreal forest: consequences for sustainableforestry. Can. J. Forest Res. 31, 384–391.

Bergeron, Y., Drapeau, P., Gauthier, S., Lecomte, N., 2007. Using knowledge ofnatural disturbances to support sustainable forest management in the northernClay Belt. Forest Chron. 83 (3), 326–337.

Buddle, C.M., Langor, D.W., Pohl, G.R., Spence, J.R., 2006. Arthropod responses toharvesting and wildfire: implications for emulation of natural disturbance inforest management. Biol. Conserv. 123, 346–357.

Burton, P.J., Parisien, M.-A., Hicke, J.A., Hall, R.J., Freeburn, J.T., 2008. Large fires asagents of ecological diversity in the North American boreal forest. Int. J.Wildland Fire 17, 754–767.

Carlson, M., Kurz, W.A., 2007. Approximating natural landscape pattern usingaggregated harvest. Can. J. Forest Res. 37, 1846–1853.

Cumming, S.G., 2001a. A parametric model of the fire-size distribution. Can. J. ForestRes. 31, 1297–1303.

Cumming, S.G., 2001b. Forest type and wildfire in the Alberta boreal mixedwood:what do fires burn? Ecol. Appl. 11 (1), 97–110.

Cyr, D., Gauthier, S., Bergeron, Y., Carcaillet, C., 2009. Forest management is drivingthe eastern North American boreal forest outside its natural range of variability.Front Ecol. Environ. 7 (10), 519–524.

DeLong, S.C., 2002. Using nature’s template to best advantage in the Canadianboreal forest. Silva Fenn. 36 (1), 401–408.

DeLong, S.C., Tanner, D., 1996. Managing the pattern of forest harvest: lessons fromwildfire. Biodivers. Conserv. 5, 1191–1205.

DeLong, S.C., Fall, S.A., Sutherland, G.D., 2004. Estimating the impacts of harvestdistribution on road-building and snag abundance. Can. J. Forest Res. 34, 323–331.

Denevan, W.M., 1992. The pristine myth: the landscape of the Americas in 1492.Ann. Assoc. Am. Geogr. 82 (3), 369–385.

Drever, C.R., Peterson, D., Messier, C., Bergeron, Y., Flannigan, M., 2006. Can forestmanagement basted on natural disturbances maintain ecological resilience?Can. J. Forest Res. 36, 2285–2299.

Dzus, E., Grover, S., Dyer, D., Cheyne, D.P., Schieck, J., 2009. Setting, implementingand monitoring targets as a basis for adaptive management: a Canadian forestrycase study. In: Villard, M.-A., Jonsson, B.G. (Eds.), Setting Conservation Targetsfor Managed Forest Landscapes. Cambridge University Press, Cambridge, pp.352–392.

Eberhart, K.E., 1986. Distribution and contribution of residual vegetation associatedwith large fires in Alberta. M.Sc. Thesis, Univ. Alta.

Eberhart, K.E., Woodard, P.M., 1987. Distribution of residual vegetation associatedwith large fires in Alberta. Can. J. Forest Res. 17, 1207–1212.

Environmental Systems Research Institute (ESRI) Inc., 2010. ArcMap GIS software(version 10). ESRI, Redlands.

Ecological Stratification Working Group (ESWG), 1996. A national ecologicalframework for Canada. Agriculture and Agri-Food Canada, Research Branch,Centre for Land and Biological Resources Research and Environment Canada,State of Environment Directorate, Ecozone Analysis Branch, Ottawa, 125 p.

Ferguson, T.A., 1979. Productivity and predictability of resource yield: aboriginalcontrolled burning in the boreal forest. M.A. Thesis, Univ. Alta.

Franklin, J.F., Forman, R.T.T., 1987. Creating landscape patterns by forest cutting:ecological consequences and principles. Landscape Ecol. 1 (1), 5–17.

Hunter, M.L., 1993. Natural fire regimes as spatial models for managing borealforests. Biol. Conserv. 65, 115–120.

Johnson, E.A., 1992. Fire and Vegetation Dynamics: Studies from the NorthAmerican Boreal Forest. Cambridge University Press, Cambridge.

Landres, P.B., Morgan, P., Swanson, F.J., 1999. Overview of the use of naturalvariability concepts in managing ecological systems. Ecol. Appl. 9 (4), 1179–1188.

Levin, S.A., 2000. Multiple scales and the maintenance of biodiversity. Ecosyst 3 (6),498–506.

Lewis, H.T., 1982. A time for burning. Occasional publication no. 17, Boreal Institutefor Northern Studies, Univ. Alta., Edmonton, 62 p.

Lewis, H.T., 2002. An anthropological critique. In: Lewis, H.T., Anderson, M.K.,Stewart, O.C. (Eds.), Forgotten Fires: Native Americans and the TransientWilderness. University of Oklahoma Press, Norman, pp. 17–36.

Lewis, H.T., Ferguson, T.A., 1988. Yards, corridors, and mosaics: how to burn a borealforest. Hum. Ecol. 16 (1), 57–77.

MacArthur, R.H., Wilson, E.O., 1967. The Theory of Island Biogeography. PrincetonUniversity Press, Princeton.

MacFarlane, A.K., 1999. Revegetation of Wellsites and Seismic Lines in the BorealForest. B.Sc. Thesis, Univ. Alta.

MacFarlane, A.K., 2003. Vegetation Response to Seismic Lines: Edge Effects and On-line Succession. M.Sc. Thesis, Univ. Alta.

Machtans, C.S., 2006. Songbird response to seismic lines in the western borealforest: a manipulative experiment. Can. J. Zool. 84, 1421–1430.

Matthiae, P.E., Stearns, F., 1981. Mammals in forest islands in southeasternWisconsin. In: Burgess, R.L., Sharpe, D.M. (Eds.), Forest Island Dynamics inMan-dominated Landscapes. Springer-Verlag, New York, pp. 55–66.

McRae, D.J., Duchesne, L.C., Freedman, B., Lynham, T.J., Woodley, S., 2001.Comparisons between wildfire and forest harvesting and their implications inforest management. Environ. Rev. 9, 223–260.

Merriam, G., Wegner, J., 1992. Local extinctions, habitat fragmentation, andecotones. In: Hansen, A.J., DiCastri, F. (Eds.), Landscape Boundaries:Consequences for Biotic Diversity and Ecological Flows. Springer-Verlag, NewYork, pp. 150–169.

Morgan, P., Aplet, G.H., Haufler, J.B., Humphries, H.C., Moore, M.M., Wilson, W.D.,1994. Historical range of variability: a useful tool for evaluating ecosystemchange. J. Sustain. Forest 2, 87–111.

P.D. Pickell et al. / Forest Ecology and Management 304 (2013) 243–253 253

Nonaka, E., Spies, T.A., 2005. Historical range of variability in landscape structure: asimulation study in Oregon, USA. Ecol. Appl. 15 (5), 1727–1746.

Paine, R.T., Tegner, M.J., Johnson, E.A., 1998. Compounded perturbations yieldecological surprises. Ecosystems 1 (6), 535–545.

Pickett, S.T.A., White, P.S., 1985. Natural disturbance and patch dynamics: anintroduction. In: Pickett, S.T.A., White, P.S. (Eds.), The Ecology of NaturalDisturbances and Patch Dynamics. Academic Press, Orlando, pp. 3–13.

R Development Core Team, 2011. R: a language and environment for statisticalcomputing. R Foundation for Statistical Computing, Vienna, Austria. <http://www.R-project.org/> (ISBN 3-900051-07-0).

Renjifo, M.L., 2001. Effect of natural and anthropogenic landscape matrices on theabundance of subandean bird species. Ecol. Appl. 11 (1), 14–31.

Roland, J., 1993. Large-scale forest fragmentation increases the duration of tentcaterpillar outbreak. Oecologia 93, 25–30.

Saunders, D.A., Hobbs, R.J., Margules, C.R., 1991. Biological consequences ofecosystem fragmentation: a review. Conserv. Biol. 5 (1), 18–32.

Schneider, R.R., Stelfox, J.B., Boutin, S., Wasel, S., 2003. Managing the cumulativeimpacts of land uses in the western Canadian sedimentary basin: a modelingapproach. Conserv. Ecol. 7 (1), 8.

Seymour, R.S., Hunter Jr., M.L., 1999. Principle of ecological forestry. In: Hunter, M.L.(Ed.), Maintaining Biodiversity in Forest Ecosystems. Cambridge UniversityPress, Cambridge, pp. 22–61.

Song, S.J. (Ed.), 2002. Ecological Basis for Stand Management: A Synthesis ofEcological Responses to Wildfire and Harvesting. Alberta Research Council Inc.,Vegreville, 450 p.

Strong, W.L., 1992. Ecoregions and Ecodistricts of Alberta. Energy and NaturalResources Technical Report T/244, Alberta Forestry, Lands and Wildlife,Edmonton, 77 p.

Suding, K.N., Gross, K.L., Houseman, G.R., 2004. Alternative states and positivefeedbacks in restoration ecology. Trends Ecol. Evol. 19 (1), 46–53.

The Forestry Corp (TFC), 2009. NEPTUNE users’ guide. Edmonton, 67 p.Timoney, K.P., 2003. The changing disturbance regime of the boreal forest of the

Canadian prairie provinces. Forest Chron. 79 (3), 502–516.Timoney, K.P., Lee, P., 2001. Environmental management in resource-rich Alberta,

Canada: first world jurisdiction, third world analogue? J. Environ. Manage. 63,387–405.

Turner, M.G., 1989. Landscape ecology: the effect of pattern on process. Annu. Rev.Ecol. Syst. 20, 171–197.

Vale, T.R., 1998. The myth of the humanized landscape: an example from YosemiteNational Park. Nat. Areas J. 18, 231–236.

Van Wilgenburg, S.L., Hobson, K.A., 2008. Landscape-scale disturbance and borealforest birds: can large single-pass harvest approximate fires? Forest Ecol.Manage. 256, 136–146.

Whitcomb, R.F., Lynch, J.F., Klimkiewicz, M.K., Robbins, C.S., Whitcomb, B.L., Bystrak,D., 1981. Effects of forest fragmentation on avifauna of the eastern deciduousforest. In: Burgess, R.L., Sharpe, D.M. (Eds.), Forest Island Dynamics in Man-dominated Landscapes. Springer-Verlag, New York, pp. 125–205.

Work, T.T., Spence, J.R., Volney, W.J.A., Morgantini, L.E., Innes, J.L., 2003. Integratingbiodiversity and forestry practices in western Canada. Forest Chron. 79 (5),906–916.