assessing the landscape context and conversion risk of

13
Assessing the landscape context and conversion risk of protected areas using satellite data products Leona K. Svancara a, , J. Michael Scott b , Thomas R. Loveland c , Anna B. Pidgorna d a University of Idaho, Idaho Cooperative Fish and Wildlife Research Unit, and Idaho Department of Fish and Game,121 Sweet Avenue, Moscow, ID, USA b U.S. Geological Survey and University of Idaho, Idaho Cooperative Fish and Wildlife Research Unit, 975 W. 6th Street, Moscow, ID, USA c U.S. Geological Survey, EROS Data Center, 47914 252nd Street, Sioux Falls, SD, USA d University of Idaho, Environmental Science Program, Moscow, ID, USA abstract article info Article history: Received 30 November 2007 Received in revised 24 November 2008 Accepted 26 November 2008 Keywords: Conversion Landscape context Land cover Parks Refuges Satellite data products Risk National Park Service National Wildlife Refuge System U.S. Fish and Wildlife Service Since the establishment of the rst national park (Yellowstone National Park in 1872) and the rst wildlife refuge (Pelican Island in 1903), dramatic changes have occurred in both ecological and cultural landscapes across the U.S. The ability of these protected areas to maintain current levels of biodiversity depend, at least in part, on the integrity of the surrounding landscape. Our objective was to quantify and compare the extent and pattern of natural land cover, risk of conversion, and relationships with demographic and economic variables in counties near National Park Service units and U.S. Fish and Wildlife Service refuges with those counties distant from either type of protected area in the coterminous United States. Our results indicate that landscapes in counties within 10 km of both parks and refuges and those within 10 km of just parks were more natural, more intact, and more protected than those in counties within 10 km of just refuges and counties greater than 10 km from either protected area system. However, they also had greater human population density and change in population, indicating potential conversion risk since the percent of landscape protected averaged b 5% in both groups and human population dynamics are primary drivers of change in many landscapes. Conversion outweighed protection by at least two times (Conservation Risk Index N 2) in 76% of counties near both parks and refuges, 81% of counties near just parks, 91% of counties near just refuges, and 93% of distant counties. Thirteen percent of counties in the coterminous U.S. had moderate to high amounts of natural land cover (N 60%), low protection (b 20%), and the greatest change in population (N 20%). Although these areas are not the most critically endangered, they represent the greatest conservation opportunity, need, and urgency. Our approach is based on national level metrics that are simple, general, informative, and can be understood by broad audiences and by policy makers and managers to assess the health of lands surrounding parks and refuges. Regular monitoring of these metrics with satellite data products in counties surrounding protected areas provides a consistent, national level assessment of management opportunities and potentially adverse changes on adjacent lands. © 2009 Elsevier Inc. All rights reserved. 1. Introduction Forty plus years ago they were emphasized as islands, isolated from surrounding natural and cultural landscapes (Udall, 1962). Today, they form the core of our country's natural resources portfolio and play an integral part in the social, economic, and cultural sectors of surrounding communities. They are protected areas, increasingly recognized and embraced as part of the regions in which they occur (Zube, 1995), and increasingly isolated ecologically. Established in 1916, the National Park Service (NPS) was directed to conserve natural and historical resources within the park system and manage so to “…leave them unimpaired for the enjoyment of future generations.At the time, the majority of parks were in the western U.S. on essentially uninhabited lands and little thought was given to compatible management among parks and surrounding lands. Today, however, this need is recognized as crucial to the effectiveness of parks as conservation areas and the ability of the NPS to manage for the unimpairedmission. Concerns over potential external inuences date as far back as 1933 (Wright et al., 1933), and management of adjacent lands has been identied as one of, if not the most, serious challenge facing park managers over the last 25 years (Shands, 1979; NPCA, 1979; NPS, 1980; Buechner et al., 1992). In 1963, the National Academy of Sciences Advisory Committee recommended that specic attention should be given to assessing changes in land use, resource use, and economic activities on areas adjacent to national parks that likely affect those parks (Robbins et al., 1963). In 1993, the National Park System Advisory Board recom- mended that resource management should be addressed in broader contextand specically recognized the impact of activities outside Remote Sensing of Environment 113 (2009) 13571369 Corresponding author. E-mail address: [email protected] (L.K. Svancara). 0034-4257/$ see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.rse.2008.11.015 Contents lists available at ScienceDirect Remote Sensing of Environment journal homepage: www.elsevier.com/locate/rse

Upload: others

Post on 14-Apr-2022

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Assessing the landscape context and conversion risk of

Remote Sensing of Environment 113 (2009) 1357–1369

Contents lists available at ScienceDirect

Remote Sensing of Environment

j ourna l homepage: www.e lsev ie r.com/ locate / rse

Assessing the landscape context and conversion risk of protected areasusing satellite data products

Leona K. Svancara a,⁎, J. Michael Scott b, Thomas R. Loveland c, Anna B. Pidgorna d

a University of Idaho, Idaho Cooperative Fish and Wildlife Research Unit, and Idaho Department of Fish and Game, 121 Sweet Avenue, Moscow, ID, USAb U.S. Geological Survey and University of Idaho, Idaho Cooperative Fish and Wildlife Research Unit, 975 W. 6th Street, Moscow, ID, USAc U.S. Geological Survey, EROS Data Center, 47914 252nd Street, Sioux Falls, SD, USAd University of Idaho, Environmental Science Program, Moscow, ID, USA

⁎ Corresponding author.E-mail address: [email protected] (L.K. Svancara).

0034-4257/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.rse.2008.11.015

a b s t r a c t

a r t i c l e i n f o

Article history:Received 30 November 2007Received in revised 24 November 2008Accepted 26 November 2008

Keywords:ConversionLandscape contextLand coverParksRefugesSatellite data productsRiskNational Park ServiceNational Wildlife Refuge SystemU.S. Fish and Wildlife Service

Since the establishment of the first national park (Yellowstone National Park in 1872) and the first wildliferefuge (Pelican Island in 1903), dramatic changes have occurred in both ecological and cultural landscapesacross the U.S. The ability of these protected areas to maintain current levels of biodiversity depend, at leastin part, on the integrity of the surrounding landscape. Our objective was to quantify and compare the extentand pattern of natural land cover, risk of conversion, and relationships with demographic and economicvariables in counties near National Park Service units and U.S. Fish and Wildlife Service refuges with thosecounties distant from either type of protected area in the coterminous United States. Our results indicate thatlandscapes in counties within 10 km of both parks and refuges and those within 10 km of just parks weremore natural, more intact, and more protected than those in counties within 10 km of just refuges andcounties greater than 10 km from either protected area system. However, they also had greater humanpopulation density and change in population, indicating potential conversion risk since the percent oflandscape protected averaged b5% in both groups and human population dynamics are primary drivers ofchange in many landscapes. Conversion outweighed protection by at least two times (Conservation RiskIndex N2) in 76% of counties near both parks and refuges, 81% of counties near just parks, 91% of countiesnear just refuges, and 93% of distant counties. Thirteen percent of counties in the coterminous U.S. hadmoderate to high amounts of natural land cover (N60%), low protection (b20%), and the greatest change inpopulation (N20%). Although these areas are not the most critically endangered, they represent the greatestconservation opportunity, need, and urgency. Our approach is based on national level metrics that are simple,general, informative, and can be understood by broad audiences and by policy makers and managers toassess the health of lands surrounding parks and refuges. Regular monitoring of these metrics with satellitedata products in counties surrounding protected areas provides a consistent, national level assessment ofmanagement opportunities and potentially adverse changes on adjacent lands.

© 2009 Elsevier Inc. All rights reserved.

1. Introduction

Forty plus years ago they were emphasized as islands, isolatedfrom surrounding natural and cultural landscapes (Udall, 1962).Today, they form the core of our country's natural resources portfolioand play an integral part in the social, economic, and cultural sectorsof surrounding communities. They are protected areas, increasinglyrecognized and embraced as part of the regions in which they occur(Zube, 1995), and increasingly isolated ecologically.

Established in 1916, the National Park Service (NPS) was directedto conserve natural and historical resources within the park systemand manage so to “…leave them unimpaired for the enjoyment offuture generations.” At the time, the majority of parks were in the

ll rights reserved.

western U.S. on essentially uninhabited lands and little thought wasgiven to compatible management among parks and surroundinglands. Today, however, this need is recognized as crucial to theeffectiveness of parks as conservation areas and the ability of the NPSto manage for the “unimpaired” mission. Concerns over potentialexternal influences date as far back as 1933 (Wright et al., 1933), andmanagement of adjacent lands has been identified as one of, if not themost, serious challenge facing park managers over the last 25 years(Shands, 1979; NPCA, 1979; NPS, 1980; Buechner et al., 1992).

In 1963, the National Academy of Sciences Advisory Committeerecommended that specific attention should be given to assessingchanges in land use, resource use, and economic activities on areasadjacent to national parks that likely affect those parks (Robbins et al.,1963). In 1993, the National Park System Advisory Board recom-mended that “resource management should be addressed in broadercontext” and specifically recognized the impact of activities outside

Page 2: Assessing the landscape context and conversion risk of

1358 L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

park boundaries (NPS, 1993). Again, in 2001, NPS managementpolicies addressed the importance of external threats (NPS, 2001)and the NPS Advisory Board indicated the need for broad-scaleresearch and management when they suggested restoring landscape-,regional-, and continental-scale habitat corridors and establishingnew parks or modifying existing park boundaries (NPSAB, 2001).

Such concerns are not unique to the National Park Service. The1997 National Wildlife Refuge System Improvement Act establishedthe mission of the refuge system “to administer a national network oflands and waters for the conservation, management, and whereappropriate, restoration of the fish, wildlife, and plant resources andtheir habitats within the United States for the benefit of present andfuture generations of Americans” (16 U.S.C. §§ 668dd, 668ee). The Actalso requires the U.S. Fish andWildlife Service (USFWS) tomanage therefuge system in a manner consistent with the preservation of itsbiological integrity, diversity, and environmental health (Public Law105-57, 1997, Section 4 (4) (B)). While the Act makes no mention ofsurrounding landscapes, nor provides any indication as to how thoselandscapes should be managed, adjacent land use is just one of manythreats that have raised concerns as to the ability of the system tomaintain viable populations of species (Curtin,1993; Scott et al., 2004;Czech, 2005).

Attributes of surrounding landscapes contribute to the abiotic andbiotic dynamics of remnant areas (Saunders et al., 1991; Meffe andCarroll, 1997) and are major determinants of both short-term andlong-term protection effectiveness (Schonewald-Cox, 1988). In parti-cular, changes in land use (i.e., development) can have numerousimpacts on ecological processes. In many landscapes, humans are theprimary drivers of landscape change (Vitousek et al., 1997; Sala et al.,2000) and anthropogenic disturbances, land ownership, and land useinteract with natural disturbances, landforms, and land cover to definelandscapes (Forman and Godron, 1986). In 1980, over 50% of threatsreported across the NPS were from external sources; development onadjacent lands, air pollution, urban encroachment, and roads andrailroads were most frequently cited (NPS, 1980). More recently, landuse change (Hansen & Rotella, 2002), fragmentation (Ambrose &Bratton, 1990), and human population density (Newmark et al., 1994;Parks and Harcourt, 2002), have been documented as threats toindividual parks. Exurban development, or what Dale et al. (2005)term “rural life”, results in four primary ecological changes: speciesdistribution and composition, land cover extent and juxtaposition,disturbance regimes, and biogeochemical cycles (Dale et al., 2005). Inturn, these changes can have profound impacts on resources inprotected areas. It has been hypothesized that only protected areaswith adequate expanses of surrounding habitat and linkages to otherprotected areas will be able to support current levels of biodiversityinto the future (Hansen et al., 2001).

Given that human land uses tend to expand over time (Wade et al.,2003) and habitat loss is non-random (Seabloom et al., 2002),quantifying and monitoring the extent and pattern of natural landcover around protected areas is essential. Analyzing and monitoringloss and fragmentation of natural land cover across broad extents (e.g.,regional, national, global) in a cost effective manner has only becomepossible in the last decade as continuous and consistent remotelysensed data and derived land use/land cover products have becomeavailable (Kupfer, 2006). In addition, only recently has a consistentnational level land cover change product (Fry et al., 2009) becomeavailable. These data provide the basis for measures of landscapepattern often sought by policy makers and land managers to aid indecision-making processes (Noss, 1999; Lindenmayer et al., 2002;Wiens et al., 2009-this issue).

Loss or conversion of natural land cover results in three distinctchanges in landscape patterns: reduced natural area, creation ofedges, and increased isolation of resulting fragments (Ewers &Didham, 2006; Kupfer, 2006). Assessing these changes in land covercondition through the application of landscape pattern metrics to

satellite data products has become common practice (e.g., Colomboet al., 2004; Griffith et al., 2003), particularly at local levels. Four broadtypes of landscapes, each associated with particular levels of loss andconnectivity, have been described (McIntyre and Hobbs, 1999, 2000;Hobbs, 2005). These landscapes cover a gradient from intact (N90%habitat remaining) to variegated (60–90% remaining), fragmented(10–60% remaining), and relictual (b10% remaining). Divisionsbetween these four categories can be seen as critical thresholds(McIntyre & Hobbs, 1999) where small changes in the spatialpatterning of resources can produce abrupt ecological responses(Turner & Gardner, 1991). The distinction between variegated andfragmented landscapes is supported by percolation theory where,assuming a random distribution, a critical threshold exists such that ifthe land cover type of interest comprises ≥59.28% of the landscape,any individual capable of using the habitat should be able to passthrough the landscape (Stauffer, 1985; Gardner et al., 1987; With &Crist, 1995). Dropping below this threshold results in smaller, isolatedpatches of habitat and potentially clumped, disjunct populations.

While these categories will likely require further subcategories inany given situation, they provide a first step in guiding where on thelandscape to allocate efforts toward different management actions(Hobbs, 2005; McIntyre & Hobbs, 2000). For example, monitoring theproportion of natural land cover in surrounding areas can be a generalindicator of the degree of fragmentation and may be helpful for landmanagers wanting to reduce impacts of land conversion by maximiz-ing landscape connectivity. When the proportion of natural land coveris ≥60%, protective measures might be taken, while values between40 and 60% might indicate a need for restoration (Wade et al., 2003).Critical thresholds supported by percolation theory have beensuggested as a landscape monitoring approach at the state level(O'Neill et al., 1997) and, at the national level, such assessments maysupply information useful from both a management and policyperspective (Kupfer, 2006), though these approaches supplementand provide context for, but should not replace, local level monitoring.

Our objective was to quantify and compare the extent and patternof natural land cover, relationships among demographic and economicvariables, and risk of conversion in counties surrounding NPS unitsand USFWS refuges with those counties distant from either type ofprotected area in the coterminous United States. Thus providing aspatially explicit mapping of threats and opportunities for thoseinterested in the future of our parks and refuges.

2. Data sets and methods

The NPS Natural Resource Challenge identified approximately 270units as having “significant natural resources” (NPS, 1999), of which243 occur in the coterminous U.S. We selected boundaries for these243 units from the 12/07/2006 version of the NPS boundariesdatabase (available at: http://science.nature.nps.gov/nrdata/). TheNPS Lands division is currently undergoing an update and approvalprocess for each boundary in this database. In the version we used, 65units had been approved and 1 (the Appalachian Scenic Trail) wasmissing, which we obtained from the NPS Northeast TemperateNetwork. While this selection of units with significant naturalresources represents many different management designations (e.g.,reserves, preserves, national parks, historical sites, etc.), we refer tothem all as “parks”. Boundaries for USFWS refuges in the coterminousU.S. (n=495) were obtained from Pidgorna (2007).

We defined “context” based on county boundaries and separatedcounties into four groups: 1) counties within 10 km of both parks andrefuges, 2) counties within 10 km of just parks, 3) counties within10 km of just refuges, and 4) counties N10 km from either a park orrefuge. Variables describing the extent and pattern of land cover, riskof conversion, and demographic and economic characteristics of eachcounty were assessed using a geographic information system (ESRI,2008). Group differences over all variables simultaneously were

Page 3: Assessing the landscape context and conversion risk of

1359L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

analyzed using a single factor MANOVA. We used subsequentunivariate ANOVAs and Tukey's HSD multiple comparisons todetermine significant differences (α=0.05) in group means for eachvariable. All analyses of variance were done using JMP version 7.0.2statistical package (JMP, 2007). While counties may not be theappropriate sampling unit for assessing land cover composition orpatterns (Riitters et al., 2006), demographic and economic informa-tion are most readily available at this level. In addition, land usedecisions are often made at the county level and those, often non-ecologically based, decisions influence the composition and config-uration of local landscapes.

To assess the extent and pattern of natural and converted landcover adjacent to protected areas, we required consistent andcomparable data across the U.S. (see also Wiens et al., 2009-thisissue).We used the 2001 National Land Cover Database (NLCD, Homeret al., 2004), resampled to 120 m×120 m cell resolution based onnearest neighbor for computational reasons. Accuracy assessments ofthe 1992 NLCD (Stehman et al., 2003;Wickham et al., 2004) suggestedmeasurements of composition and pattern should be based on eightgeneralized categories (Riitters et al., 2006). Given that similarassessments for the 2001 NLCD are not yet completed, we aggregatedthe original 21 classes to 3 generalized categories: urban, agriculture,and natural (Table 1). The NLCD is strictly land cover and does notisolate disturbances. For example, an area with a great deal of forestclearing (e.g., logging) may be classified as forest, grass, or shrubs byNLCD, all of which would be considered ‘natural’ under ouraggregation even though it is highly disturbed. Similarly, weconsidered two classes (barren land and grassland/herbaceous)natural even though they may contain areas subject to anthropogenicimpacts such as strip mines, gravel pits, or grazing. Aggregating allnon-converted lands (forest, grasslands, shrublands) as ‘natural’focuses on the effects of agriculture and urban development onlandscape patterns surrounding protected areas. In addition, it may bemore appropriate for capturing the natural heterogeneity of somecommunities (e.g., forests in the southwestern U.S.) (Heinz Center,2006).

The total change in natural land cover is perhaps the simplestindicator of biotic integrity (O'Neill et al., 1997) and may be betterthan fragmentation metrics in distinguishing patterns in human-modified landscapes (Theobald, 2004). We described the extent ofland cover by quantifying the percent of natural, urban, andagriculture land cover outside of parks and refuges in each county.We calculated the proportion of converted (agriculture and urban)land, also known as the U-index (O'Neill et al., 1988), to measuregeneral land use pressure by humans. Following McIntyre and Hobbs(1999, 2000) we described each county as intact, variegated,fragmented, or relict based on the percent of natural land coveroutside of parks and refuges. Using the NLCD change product (Fryet al., 2009), we calculated the percent of natural land cover outside ofparks and refuges converted to urban or agriculture between circa1992 and 2001. The NLCD change product was developed to overcomethe challenges inherent in directly comparing the NLCD 1992 andNLCD 2001 data and offers a current, consistent, and seamless landcover change product for the U.S. It is currently available in provisional

Table 1Aggregation of original 2001 National Land Cover Database land cover classes to generalurban, agriculture, and natural categories.

Generalcategory

NLCD classes

Urban Low intensity developed, medium intensity developed, high intensitydeveloped, open space developed

Agriculture Pasture/hay, cultivated cropsNatural Grassland/herbaceous, shrub/scrub, mixed forest, evergreen forest,

deciduous forest, barren land, perennial ice/snow, woody wetlands,emergent herbaceous wetlands, open water

status until a formal accuracy assessment can be completed (Fry et al.,2009).

Several studies have assessed the relationships among land coverpattern metrics and the reliability of metrics across different spatialresolutions and extents (Turner et al., 1989; Benson & MacKenzie,1995; Riitters et al., 1995; Wickham & Riitters, 1995; Wu et al., 2000,2002; Saura, 2004;Wu, 2004; Li et al., 2005; Frohn & Hao, 2006; Saura& Castro, 2007). We selected three simple and common patternmetrics to address landscape pattern: 1) mean patch size of naturalland cover, 2) density of natural/converted edge, and 3) isolation ofremaining natural patches asmeasured by the Landscape Shape Index.These metrics have been recommended in the literature for theirsensitivity to changes in pattern parameters, high predictability, andreliability across scales (Frohn & Hao, 2006; Saura, 2004; Wu et al.,2002; Wu, 2004; Li et al., 2005; Saura & Castro, 2007).

At the landscape level, mean patch size (MPS) represents theaverage size of all natural land cover fragments within an area ofinterest and can serve as a habitat fragmentation index, with a lowerMPS indicating greater fragmentation. Because the NLCD 2001 datawere developed via a pixel-based classification of satellite imageryand we further aggregated thematic classes, we broadly defined apatch as a block of contiguous pixels of natural land cover (seeTable 1). Edge density (ED) is the amount of natural/converted landcover edge relative to the landscape area (in this case, the area of thecounty outside of parks and/or refuges). Interpretation of ED requiresknowing the extent of natural land cover because values of ED can besimilar at both high and low amounts of cover. Though dependent on(and thus redundant with) other patch shape and size metrics, ED is avaluable metric given that edge is a habitat element identified to beimportant for several species (Griffith et al., 2000). The LandscapeShape Index (LSI) is a measure of patch type aggregation across thelandscape such that values equal to 1 represent a landscape consistingof a whole patch and values increase N1 as the patch type becomesmore disaggregated (McGarigal & Marks, 1995). Like ED,interpretation of LSI requires knowing the extent of natural landcover. Landscape pattern metrics were calculated using FRAGSTATS(McGarigal & Marks, 1995) and Patch Analyst spatial statistics byregions (Rempel, 2006).

In each county, we calculated the percent of protected land(defined as National Gap Analysis Program Status 1 or 2 lands, Scottet al., 1993) and multiple-use public land (Gap Status 3) outside ofparks and refuges. These data came from the Protected AreasDatabase, Version 4 (PAD, Conservation Biology Institute, 2006)modified to reflect the NPS and USFWS refuge boundary databasesmentioned earlier.

Demographic and economic characteristics drive urban develop-ment (Alig et al., 2004) and can have significant effects on resources inprotected areas (Parks & Harcourt, 2002). We compared humanpopulation density, per capita income, housing density, and popula-tion change (1990–2000) in each of the four county groups. Weobtained data on population, housing, and population change fromthe U.S. Census Database, 2000 (U.S. Census Bureau, 2000) andcalculated population density and housing density based on theamount of land in each county outside protected areas (GAP Status 1and 2). Because population data collected through the U.S. CensusBureau are tied to primary residences, we included housing density totake into account recreational and second homes and to moreprecisely indicate influences of land use (Brown et al., 2005; Liuet al., 2003). Per capita income data were obtained from the U. S.Bureau of Economic Analysis (2003).

Following Hoekstra et al. (2005) we calculated the ConservationRisk Index (CRI) for each county as the ratio of percent area convertedto percent area protected. The CRI represents a comprehensivemeasure of whether particular land units (e.g., counties) protecttheir natural environments on the same scale as those converted. Theratio is easily interpreted as for every hectare converted, “x” are

Page 4: Assessing the landscape context and conversion risk of

1360 L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

protected, with CRI values N1 indicating the number of timesconversion outweighs protection. For example, in Edmonson County,KY the area outside of Mammoth Cave NP is 35% converted with only0.8% protected resulting in a CRI of 44. Thus, conversion is 44 timesgreater than protection and for every hectare that has been convertedoutside the park, only 0.02 ha have been protected. In contrast, landsoutside of Craters of the Moon National Monument and Preserve inBlaine County, ID are 5% converted and 21% protected resulting in a CRIof 0.2. In this county, for every hectare converted, 4.2 have beenprotected.

Calculating the CRI requires the assumption that every samplingunit has at least a portion of its area under protection. This was not thecase in 16% of counties near both parks and refuges, 20% of countiesnear just parks, 25% of counties near just refuges, and 36% of distantcounties. To avoid dividing by zero, we assigned an extremely lowvalue for the percent protected (0.0001%) for those areas lackingprotection.

Lastly, we identified potential opportunities and risk of conversionfor both the NPS and USFWS based on the percent natural land cover,percent protected, and human population change. Potential risk ofconversion followed Hoekstra et al. (2005), such that counties withN20% conversion and CRI N2 were classified as ‘vulnerable,’ those withN40% conversion and CRI N10 as ‘endangered’ and those with N50%conversion and CRI N24 (including counties lacking any protection) as‘critically endangered.’ We classified counties with b20% converted orCRI b2 as ‘less risk’. We selected arbitrary values of ≤20% protected,≥20% population change, and the percolation threshold of ≥60%natural land cover to help identify counties with the greatestconservation need, urgency, and opportunity. By definition, countiesfitting these criteriawere considered ‘vulnerable’ or not classified (i.e.,‘less risk’) under Hoekstra et al. (2005).

Fig. 1. Counties within 10 km of National Park Service units (NPS, n=464), U.S. Fish andWilcounties N10 km from either system (DIST, n=1763).

3. Results

Together, parks and refuges represent 2.16% of the coterminous U.S.Of all counties in the coterminous U.S., 7% (n=220) occur within10 km of both parks and refuges, 15% (n=464) occur within 10 km ofjust parks, 21% (n=660) occur within 10 km of just refuges, and 57%(n=1763) are N10 km from either a park or refuge (Fig. 1).

Based on our aggregation of the 2001 NLCD cover types, 28% of thecoterminous U.S. has been converted to urban or agriculture (Fig. 2).Given our aggregation approach and the spatial characteristics ofLandsat data, this is a conservative estimate. Counties near both parksand refuges as well as counties near just parks were more likely tohave intact landscapes (N90% natural land cover) (Fig. 3). Conversely,counties near refuges and counties distant from either parks orrefuges were more likely to be fragmented (10–60% natural landcover).While 7% of distant counties could be described as having relictlandscapes, less than 2% of counties near just parks or just refuges andonly 1 county near both parks and refuges (San Francisco County, CA)could be described as such. Counties classified as intact occurredprimarily in the western U.S., Maine, and Great Lakes region whilecounties with relictual landscapes occurred primarily in the Mis-sissippi River Valley (Fig. 4). Overall, 70% of counties near both parksand refuges and 70% of those near just parks had natural land covergreater than the critical threshold of 60% predicted by percolationtheory. In contrast, ≤50% of counties near just refuges or countiesN10 km from either parks or refuges met this threshold.

Few lands surrounding parks and refuges were protected for thelong-termmaintenance of biodiversity and even fewer were protectedin distant counties. Sixteen percent of counties near both parks andrefuges, 20% of counties near just parks, and 25% of those near justrefuges did not have any protected lands outside of the parks and

dlife Service refuges (NWR, n=660), both protected area systems (BOTH, n=220), and

Page 5: Assessing the landscape context and conversion risk of

Fig. 2. Spatial relationship of natural and converted land cover types relative to National Park Service units (n=243) and U.S. Fish and Wildlife Service refuges (n=495). Convertedland cover included only urban and agricultural classes, all others (including water) were considered natural. Protected area boundaries enhanced for viewing.

1361L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

refuges themselves. Thirty-six percent of distant counties lacked anyprotection. These counties occurred primarily in the central portionsof the U.S. The vast majority of counties (67% near both parks andrefuges, 67% near just parks, 68% near just refuges, and 60% distant)have 0.01–10% protected. Just 7% of counties near both parks andrefuges have N20% of surrounding lands protected, as do just 4%, 2%,and 2% of park, refuge, and distant counties, respectively. Countieswith the greatest proportion of protected land surrounding parks andrefuges occurred in the western states with the exception of MonroeCounty, FL, which represented a unique situation in that virtually all ofthe 271,500 ha land areawas considered protected. Everglades and BigCypress National Parks encompassed over 80% of the county as well as

Fig. 3. Percent of counties within 10 km of National Park Service units (NPS, n=464),U.S. Fish and Wildlife Service refuges (NWR, n=660), both protected area systems(BOTH, n=220), and counties N10 km from either system (DIST, n=1763) in each offour landscape types defined by McIntyre and Hobbs (1999, 2000).

substantial portions of neighboring counties. Of the remaining area,93% was encompassed by at least 1 protected area (e.g., Florida KeysNational Marine Sanctuary) and 4% was considered multiple-use land(GAP Status 3, Key West Naval Air Station). Because it was such anoutlier, we excluded Monroe County from analyses.

The MANOVA analysis revealed a significant difference in countygroups for the 10 assessed variables (Wilks' lambda=0.8698,F=14.7354, pb0.0001). We excluded housing density from theseanalyses because it was significantly correlated with populationdensity (r=0.9881). Results of the ANOVAs and pair-wise compar-isons indicate that counties near both parks and refuges and thosenear just parks have significantly more natural land cover than thosecounties near refuges or distant counties (Fig. 5). Counties near parkshad significantly larger MPS of natural land cover than either countiesnear just refuges or those distant, but not counties near to both parksand refuges. Counties near both parks and refuges have significantlylower edge density than counties near just refuges and those distant.Although counties near just parks appear to have significantly higheredge density as well, these counties have similar amounts of naturalland cover as those counties near both parks and refuges. Given thepotential ambiguity of edge density, this may not be a biologicallysignificant difference. Similarly, given that both distant counties andcounties near just refuges have low amounts of natural land cover, thestatistically different values of LSI may not be biologically meaningful.Landscapes in counties near just refuges are significantly moredisaggregated (as measured by LSI) than either counties near justparks or those near both parks and refuges. Based on the NLCD 1992/2001 change product, a significantly greater percent of natural landcover has been lost in counties near just parks than in counties nearjust refuges. No significant difference was seen between counties nearboth parks and refuges or distant counties.

Page 6: Assessing the landscape context and conversion risk of

Fig. 4. Percent natural land cover outside of National Park Service units and U.S. Fish and Wildlife Service refuges in counties of the coterminous U.S. Categories reflect the fourlandscape types defined by McIntyre and Hobbs (1999, 2000). Protected area boundaries enhanced for viewing.

1362 L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

Analyses of demographic variables revealed that counties nearboth parks and refuges as well as counties near just parks hadsignificantly higher human population density on surrounding land-scapes and they experienced a greater percent change in populationbetween 1990 and 2000 (Fig. 5). In addition, the per capita income incounties near both parks and refugeswas significantly higher than anyother groups. Counties near just parks also had significantly higher percapita income than distant counties. Counties with the greatestincrease in population occurred across the Southwest and Great Basinas well as in select areas of southeastern states while counties in theMidwest and New England states experienced population declines(Fig. 6).

Conversion outweighed protection by at least two times (CRI N2)in 76% of counties near both parks and refuges, as well as 81%, 91%, and93% of those near just parks, just refuges, and distant counties,respectively, predominantly in the Midwest and eastern regions of theU.S. (Fig. 7). Landscapes in counties near both parks and refuges andthose counties near just parks had significantly more area protectedbut overall means still were b5% (Fig. 5). Conservation risk (asmeasured by the CRI) was significantly higher in counties distant fromboth protected area systems. While conservation risk in counties nearjust refuges was not significantly different from those near just parks,it was significantly higher than those counties near both parks andrefuges. Modifying the CRI to specifically address urban development,we found that 63% of counties near both parks and refuges, 68% ofcounties near just parks, 79% of counties near just refuges, and 85% ofdistant counties had more than 2 times the level of urban areas thanthey did protection. Looking strictly at agriculture, 63% of countiesnear parks and refuges, 73% of counties near just parks, 87% ofcounties near just refuges, and 90% of distant counties had more than2 times the level of agriculture as protection.

Counties near refuges and those distant from both protected areasystems were more likely to be classified as critically endangeredbased on the CRI and percent of converted land as defined by Hoekstraet al. (2005) (Fig. 8). Conversely, counties near both parks and refugesor just parks were less likely to be at risk. Thirty percent of countiesnear both parks and refuges were classified as vulnerable toconversion, as were 32% of counties near just parks, 26% of countiesnear just refuges, and 26% of distant counties. Fewer counties wereclassified as endangered (9%, 10%, 12%, and 10%, respectively).

Based on percent land protected (b20%), population change(N20%) and natural land cover (N60%), 39 counties near both parksand refuges, 92 counties near just parks, and 79 counties near justrefuges represent the greatest conservation need, urgency, andopportunity (Fig. 9). An additional 200 counties distant from eitherparks or refuges also fit these criteria. These counties are scatteredacross the U.S. with regional groupings in the Great Basin, NorthernRockies, South Texas, northern Florida, northernMichigan, and centralMinnesota (Fig. 10).

4. Discussion

A majority of protected areas are dependent on adjacent landssimply because their boundaries fail to encompass habitats andprocesses (e.g., migratory species, fire regimes, hydrologic regimes,succession) necessary to maintain complete species communities(Myers, 1972; Western, 1982; Curry-Lindahl, 1972; Garratt, 1984). Lossof natural land cover types on surrounding lands can impact protectedareas, even if they are maintained within the protected area boundaryitself (DeFries et al., 2005). For example, studies in the GreaterYellowstone Ecosystem have shown that some species cannot persistin Yellowstone National Park without access to low elevation, riparian,

Page 7: Assessing the landscape context and conversion risk of

Fig. 5. Mean and standard error of variables used to assess differences in counties within 10 km of National Park Service units (NPS, n=464), U.S. Fish and Wildlife Service refuges(NWR, n=660), both protected area systems (BOTH, n=220), and counties N10 km from either system (DIST, n=1763). Groups not connected by the same letter are significantlydifferent at α=0.05.

1363L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

or grassland habitats on adjacent lands (Hansen & Rotella, 2002). Ourresults indicate that landscapes in counties near both parks andrefuges and those near just parks were more natural, more intact, andmore protected than those in counties near just refuges and countiesdistant from either protected area system. However, they also hadgreater human population density and change in population, indicat-ing potential conversion risk since the percent of landscape protectedaveraged b5% in both groups and human population dynamics areprimary drivers of change in many landscapes (Vitousek et al., 1997;Sala et al., 2000).

Across all landscapes, maintaining landscape connectivity is vital,particularly when coupled with the projected effects of climatechange (e.g., shifting distributions, shifting phenology). Each of thethree changes in land cover pattern associated with loss or conversionof natural land cover (reduced natural area, creation of edges, andincreased isolation of resulting remnants) influence landscapeconnectivity and may affect biodiversity in various ways dependingon the interaction between species dispersal behavior, mode and scaleof movement, requirements for suitable habitat, arrangement of thehabitat, and type of landscape modification (O'Neill et al., 1988;Pearson et al., 1996; With & Crist, 1995; McIntyre & Hobbs, 1999;Kupfer, 2006; Fischer & Lindenmayer, 2007). Connectivity is thought

to be threshold-based, disrupting abruptly and possibly initiatingother ecological thresholds in a “threshold cascade” (With, 2005).While a single threshold value cannot adequately describe responsesof all species to changes in landscape pattern or extent (Andrén, 1994;With & Crist, 1995; Svancara et al., 2005), certain levels of natural landcover may act like “red-flags” for some species (Hansen & Urban,1992). Our results indicate that more counties near refuges are belowthe critical threshold of 60% predicted by percolation theory underrandom landscapes than counties near just parks or both parks andrefuges. Associated with the loss of natural land cover, counties nearjust refuges showed decreased mean patch size, increased isolation ofthe remaining natural patches, and increased amount of natural/converted edge. This suggests that, on average, species in refuges mayhave a more difficult time dispersing across the landscape in the faceof threats (such as climate change), unless a park is nearby. However,if those species have a large dispersal range, high fecundity, and highsurvivorship, they may be able to persist even when suitable habitatcovers only 25–50% of the landscape (Lande, 1987). For species withlow demographic potential that may not persist even with 80% of thelandscape in suitable habitat (Lande,1987), our results indicate that inonly select areas (14% of all counties) will either protected area systemoffer long-term support.

Page 8: Assessing the landscape context and conversion risk of

Fig. 6. Population change from 1990–2000, expressed as percent of 1990 population, in counties of the coterminous U.S. Boundaries of National Park Service units and U.S. Fish andWildlife Service refuges enhanced for viewing.

1364 L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

Though we chose to focus on the 60% threshold based onpercolation theory, other thresholds could just as easily have beenchosen. In real landscapes, this threshold is likely to be much lowerand several studies have suggested that below 30% natural land cover,connectivity loss is particularly severe resulting in increased loss ofspecies dependent on native vegetation (Andrén, 1994; With & Crist,1995; Fahrig, 2003; Radford et al., 2005). Using the 30% threshold, 8%of counties near parks and refuges,10% of counties near just parks, and21% of counties near just refuges have experienced severe loss inconnectivity and associated species. Obviously, these ‘rules’ do notapply to all species or ecosystems (Andrén, 1994; Bascompte & Solé,1996; Parker & Mac Nally, 2002; Lindenmayer et al., 2005) but doprovide a general indication as to the overall landscape conditionsurrounding parks and refuges.

The land cover categories we used (natural versus converted)imply a simple dichotomy. In reality, of course, this is not the case.Both categories represent gradients of condition that likely changepositively and negatively with time. The transition areas betweennatural and converted are also gradients variable with temporal andspatial scale. Aggregating to these categories facilitated the focus oneffects of conversion on natural land cover patterns. Because of thisaggregation, however, we are unable to differentiate the specificnatural land cover types associated with the assessed landscape,demographic, and economic variables. For example, Riitters et al.(2002) and Wade et al. (2003) determined that forest fragmentationwas similar across the U.S., mainly associated with shrubland andgrassland types in the West and mostly agricultural, nonforestwetland, and urban land cover in the East. In our study, landscapeswould not appear as fragmented.

Habitat loss may contribute more to species losses than humanpopulation size (Wiersma et al., 2004) yet, cultural, political, and

socioeconomic factors all contribute to land use decisions (Naveh,1995; Nassauer, 2005; With 2005) and are widely used indicators oflandscape quality or threats to biodiversity (Nassauer, 2005; Cincottaet al., 2000). Our results indicate that counties near both parks andrefuges and those near just parks had higher population densities andexperienced a greater change in population between 1990 and 2000.Even with more intact surrounding landscapes, species in thesecounties may be at greater risk given that high human populationdensity has been shown to adversely affect the persistence of habitatsand species (Kerr & Currie, 1995; Woodroffe, 2000; Parks & Harcourt,2002; Luck, 2007). The NLCD 2001 urban classes, which we used todetermine the extent of conversion, likely missed low-density,exurban areas with less than one housing unit per hectare (Wardet al., 2000). Therefore our measurements of the extent of conversionmay be underestimates and, from a conservation perspective, growthin low-density, exurban areas is worrisome. Expansion of exurbandevelopment (1 home/0.4–16.2 ha, Brown et al., 2005) can havenumerous biological impacts (Hansen et al., 2002; Hansen et al.,2005) and is increasingly recognized as a primary driver of ecologicalprocesses and biodiversity (McKinney, 2002; Miller & Hobbs, 2002).In the Greater Yellowstone Ecosystem, exurban development ispredicted to result in up to 40% habitat conversion by 2020 (Gudeet al., 2007). If trends identified by Brown et al. (2005) continue,population density will increase most in Florida, the northeast coast,Great Lakes region, Northwest coast, and California, most of whichcurrently have greater proportions of natural land cover and relativelylow proportions of protected area.

Although our analysis was not designed to assess cause and effect,the relationship we found with per capita income is interesting tonote. Economic activity has been shown to impact biodiversity(Naidoo & Adamowicz, 2001; McKinney, 2002) but there is current

Page 9: Assessing the landscape context and conversion risk of

Fig. 7. Conservation Risk Index in counties of the coterminous U.S. following Hoekstra et al. (2005) with greater values indicating more lands are converted than are protected.Boundaries of National Park Service units and U.S. Fish and Wildlife Service refuges enhanced for viewing.

1365L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

debate whether increasing per capita income fuels land conversion(Sisk et al., 1994) or is necessary to solve environmental problems(Beckerman,1992). Some postulate that as per capita income rises, thedemand for environmental quality and resource availability also rises(Beckerman, 1992; World Bank, 1992) and percent of area protectedincreases (McKinney, 2002). We found that counties near both parksand refuges had significantly higher per capita income than othercounty groups and counties near just parks or just refuges hadsignificantly higher incomes than distant counties, lending support tothis hypothesis. On the other hand, economic growth has also beenrelated to resource exploitation, increased development, and loss ofboth cropland and forest (Sisk et al., 1994; Lo & Yang, 2002; Alig et al.,

Fig. 8. Percent of counties within 10 km of National Park Service units (NPS, n=464), U.S.Fish and Wildlife Service refuges (NWR, n=660), both protected area systems (BOTH,n=220), and counties N10 km from either system (DIST, n=1763) at risk of conversionbasedon theConservationRisk Index andpercentof converted landasdefinedbyHoekstraet al. (2005).

2004) suggesting counties near both parks and refuges, or just parksmay be at greater risk of landscape fragmentation.

Creation or expansion of protected areas is the traditional humanresponse to habitat loss. Hoekstra et al. (2005) argue that at very highand very low levels of conversion, habitat protection declines. Basedon their study, biomes with intermediate levels of habitat conversionhave greater levels of protection. At the county level, however, thereappears to be little relationship between conversion and protection.This may be due, in part, to the low levels of protection overall.Protection outweighed conversion in only 17%, 15%, and 5% of countiesnear parks and refuges, just parks, and just refuges, respectively. Incountieswith large areas of public, multiple-use lands, protected areasmay be “buffered” enough that their effective size is greater than theactual boundary. However, resources on multiple-use lands are notpermanently protected and, in some cases, neither is the land itself(e.g., ski areas on national forests).

Given that species respond differently to different habitat featuresat different scales, there is no simple and direct way to transformlandscape features into an index for conservation potential (Opdamet al., 2003). However, effects of habitat conversion on biodiversity arewell documented. At the level of biomes and ecoregions, Hoekstraet al. (2005) proposed the CRI as a useful measure for identifyingthose most at risk of conversion and potential opportunities forconservation. At the level of counties, CRI provides a useful first step indescribing the relationship between landscape condition and protec-tion, however interpreting this measure in terms of opportunities maybemisleading. Hoekstra et al. (2005) suggest that conservation effortsshould be directed toward those areas classified as vulnerable,endangered, or critically endangered. From a biological perspective,however, those areas identified as endangered or critically endan-gered may offer little conservation opportunity because the few

Page 10: Assessing the landscape context and conversion risk of

Fig. 9. Conservation need, urgency, and opportunity based on percent protected, percent population change (1990–2000), and percent natural land cover for counties within 10 km ofNational Park Service units and U.S. Fish and Wildlife Service refuges. Counties meeting the criteria of b20% protected, N20% population change, and N60% natural land cover arehighlighted in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

1366 L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

remaining habitats (including the protected areas themselves) aretruly relicts, too small and disjunct to support ecologically viablecommunities (see also Czech, 2005). While their status as remainingremnants is worthy and they can be important complements to largerareas (Fischer & Lindenmayer, 2002; Tscharntke et al., 2002) or to

Fig. 10. Conservation need, urgency, and opportunity in counties of the lower 48 states basedand percent in natural land cover (N60%). Categories reflect the number of criteria that each crefuges enhanced for viewing.

maintain relictual populations of narrow endemics, there is little leftof the surrounding natural landscape to protect. In addition, theseareas are generally experiencing slow growing, or even decliningpopulations (Fig. 6). Conversely, many of the areas identified asvulnerable, or not classified at all according to Hoekstra et al. (2005),

on percent of land protected (b20%), percent population change (1990–2000) (N20%),ountymeets. Boundaries of National Park Service units and U.S. Fish andWildlife Service

Page 11: Assessing the landscape context and conversion risk of

1367L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

are experiencing often extremely high population growth. These areaswithmoderate to high amounts of natural habitat, low protection, andgreatest change in demographics represent the greatest conservationopportunity, need, and urgency (Figs. 9 and 10).

In the end, the success of protected areas depends upon managingthe entire landscape, including adjacent land use activities (Wiens,1996; Jongman, 2005). Increasing ecological isolation, as is apparentwith refuges, further increases the value and need of reintegratingprotected areas into the surrounding matrix. Current county-levelland use planning approaches, often implemented in isolation ofsurrounding counties, may not effectively protect resources. Even inthose counties where a large portion of land outside of protected areasis considered multiple-use, the low elevation, high soil productiveareas are often private, unprotected, and most susceptible todevelopment (Scott et al., 2001). For example, Gude et al. (2007)suggest that existing county growth management policies in theGreater Yellowstone Ecosystem will provide minimal protection, yetaggressive, regionally coordinated, growth management can protecthabitats most at risk while not limiting housing development. Aproactive, broad-scale planning approach is often most effective,particularly when it incorporates an ecological perspective (Shafer,1999; Theobald et al., 2005; Dale et al., 2005).

Many have sought measures of landscape pattern that can beincorporated into monitoring plans and aid in decision-making (Noss,1999; Herzog et al., 2001; Lindenmayer et al., 2002; McAlpine & Eyre,2002; others this issue). As Li et al. (2005) point out, no landscapepattern metric is a “magic bullet” and representative and simpleindices should be chosen according to the purpose and context of thestudy. We used national level metrics that are simple, general,informative, understood by broad audiences, and provide a methodfor monitoring regional changes. Our approach is intended forcharacterizing and monitoring changes in the structural compositionand configuration of natural land cover, without accounting for theinherent complexity and differences among species, ecologicalprocesses, and ecosystems (Kupfer, 2006; Fischer & Lindenmayer,2007). In almost all cases, species-specific information and finer scalelandscape variables will further refine our results and efforts at locallevels should validate these measurements, ideally linking changes inthese patterns directly to specific ecological consequences (Kupfer,2006; Tischendorf, 2001). We provide a spatially explicit context foradditional detailed studies addressing this complexity as well asstructure, primary productivity, sub-canopy composition, and dis-turbance indicators using remote-sensing derived products.

Both parks and refuges have been established for a myriad ofpurposes, on an “opportunistic” rather than strategic basis (Leopoldet al., 1968 cited in Fischman, 2003; p.23), and have diversemanagement standards (Fischman, 2003; Shafer, 1999). This rangeof establishment andmanagement objectives obviously contributes tothe results of our study. Priorities for continued management andconservation will likely vary on a local level according to the type oflandscape. For instance, intact and variegated landscapes prioritiesmay focus on maintaining the least modified habitats, whilerestoration and buffering of the few remaining fragments are likelypriorities for fragmented and relictual landscapes (Wade et al., 2003;Hobbs, 2005). Through the use of satellite data products and broad,simplistic landscape pattern metrics, regular monitoring of the extentand configuration of natural land cover, protected lands, andpopulation information in counties surrounding parks and refugesprovides a consistent assessment of adverse changes on adjacentlands. Use of these products in answering natural resource policy andmanagement questions has a long history. However, recent advancesin remote sensing science, product availability, and computationalabilities suggest that it will be increasingly important as managersbegin to assess continental and global threats to the integrity, diversityand health of the conservation landscape. Given that management ofprotected area systems occurs at multiple scales (i.e., park/refuge,

regional, and national), the ability to reliably and consistently applymetrics across scales is important. The metrics used in our study havebeen shown to have consistent and robust scaling relations (Wu,2004; Saura, 2004; Saura & Castro, 2007) supporting the possibility ofincorporating finer scale data and applying these methods to levelsperhaps more applicable to the local protected area manager. Nestingour approach in a hierarchical monitoring framework that spansbiomes, ecoregions, political entities (states), and local planning areas(counties) provides context and addresses the protected areas systemas a whole. Such a strategy explicitly acknowledges that any oneagency or organization cannot do it alone, and allows multiplestakeholders to focus regional and local efforts in areas at greatest risk.

Acknowledgements

We thank Gary Davis, Jean McKendry, Ray Sauvajot, John Dennis,Mike Soukup, and John Gross for informative discussions. We alsothank Charles Van Riper, William Halvorson, Gina Wilson, and 3anonymous reviewers for comments on earlier drafts. Fundingsupport for L. K. Svancara was provided by the National Park Service.

References

Alig, R. J., Kline, J. D., & Lichtenstein, M. (2004). Urbanization on the U.S. landscape:Looking ahead in the 21st century. Landscape and Urban Planning, 69, 219−234.

Ambrose, J. P., & Bratton, S. P. (1990). Trends in landscape heterogeneity along theborders of Great Smoky Mountains National Park. Conservation Biology, 4, 35−143.

Andrén, H. (1994). Effects of habitat fragmentation on birds and mammals in landscapeswith different proportions of suitable habitat: A review. Oikos, 71, 355−366.

Bascompte, J., & Solé, R. V. (1996). Habitat fragmentation and extinction thresholds inspatially explicit models. Journal of Animal Ecology, 65, 465−473.

Beckerman,W. (1992). Economic growth and the environment:Whose growth?Whoseenvironment? World Development, 20, 481−496.

Benson, B. J., & MacKenzie, M. D. (1995). Effects of sensor spatial resolution onlandscape structure parameters. Landscape Ecology, 10, 113−120.

Brown,D.G., Johnson,K.M., Loveland, T. R., & Theobald, D.M. (2005). Rural land-use trendsin the conterminousUnited States,1950–2000. Ecological Applications, 15,1851−1863.

Buechner, M., Schonewald-Cox, C., Sauvajot, R., & Wilcox, B. A. (1992). Cross-boundaryissues for national parks: What works “on the ground.”. Environmental Manage-ment, 16, 799−809.

Cincotta, R. P.,Wisnewski, J., & Engelman, R. (2000). Humanpopulation in the biodiversityhotspots. Nature, 404, 990−992.

Colombo, S., Chica-Olmo, M., Abarca, R., & Eva, H. (2004). Variographic analysis of tropicalforest cover from multi-scale remotely sensed imagery. Journal of Photogrammetry &Remote Sensing, 58, 330−341.

Conservation Biology Institute (2006). Protected areas database, version 4. Oregon:Corvallis.

Curry-Lindahl, K. (1972). Ecological research and managemen. In R. Osten (Ed.), WorldNational Parks: Progress and opportunities (pp. 197−213). Brussels: IUCN.

Curtin, C. G. (1993). The U. S. National Wildlife Refuge System. Conservation Biology, 7,30−38.

Czech, B. (2005). The capacity of the National Wildlife Refuge System to conservethreatened and endangered animal species in the United States. ConservationBiology, 19, 1246−1253.

Dale, V., Archer, S., Chang, M., & Ojima, D. (2005). Ecological impacts and mitigationstrategies for rural land management. Ecological Applications, 15, 1879−1892.

DeFries, R., Hansen, A., Newton, A.C., &Hansen,M. C. (2005). Increasing isolationof protectedareas in tropical forests over the past twenty years. Ecological Applications, 15, 19−26.

ESRI (2008). ArcMap 9.3. Redlands, CA: Environmental Systems Research Institute.Ewers, R. M., & Didham, R. K. (2006). Confounding factors in the detection of species

responses to habitat fragmentation. Biological Review, 81, 117−142.Fahrig, L. (2003). Effects of habitat fragmentation on biodiversity. Annual Review of

Ecology Evolution and Systematics, 34, 487−515.Fischer, J., & Lindenmayer, D. B. (2002). Small patches can be valuable for biodiversity

conservation: Two case studies on birds in southeastern Australia. BiologicalConservation, 106, 129−136.

Fischer, J., & Lindenmayer, D. B. (2007). Landscape modification and habitatfragmentation: A synthesis. Global Ecology and Biogeography, 16, 265−280.

Fischman, R. L. (2003). The National Wildlife Refuges: Coordinating a conservation systemthrough law. Washington, DC: Island Press.

Forman, R. T. T., & Godron, M. (1986). Landscape ecology. New York: JohnWiley and Sons.Frohn, R. C., & Hao, Y. (2006). Landscape metric performance in analyzing two decades

of deforestation in the Amazon Basin of Rondonia, Brazil. Remote Sensing ofEnvironment, 100, 237−251.

Fry, J. A., Coan, M. J., Homer, C. G., Meyer, D. K., & Wickham, J. D. (2009). Completion ofthe National Land Cover Database (NLCD) 1992-2001 Land Cover Change Retrofitproduct. U.S. Geological Survey Open-File Report 2008-1379 (18 pp).

Gardner, R. H., Milne, B. T., Turner, M. G., & O'Neill, R. V. (1987). Neutral models for theanalysis of broad-scale landscape pattern. Landscape Ecology, 1, 19−28.

Page 12: Assessing the landscape context and conversion risk of

1368 L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

Garratt, K. (1984). The relationship between adjacent lands and protected areas: Issuesof concern for the protected area manager. In J. A. McNeely, & K. R. Miller (Eds.),National Parks, conservation and development: The role of protected areas insustaining society (pp. 65−71). Washington, DC: Smithsonian Institution Press.

Griffith, J. A., Martinko, E. A., & Price, K. P. (2000). Landscape structure analysis of Kansasat three scales. Landscape and Urban Planning, 52, 45−61.

Griffith, J. A., Stehman, S. V., Sohl, T. L., & Loveland, T. R. (2003). Detecting trends inlandscape pattern metrics over a 20-year period using a sampling-basedmonitoring programme. International Journal of Remote Sensing, 24, 175−181.

Gude, P. H., Hansen, A. J., & Jones, D. A. (2007). Biodiversity consequences of alternativefuture land scenarios in Greater Yellowstone. Ecological Applications, 17, 1004−1018.

Hansen, A. J., Knight, R. L., Marzluff, J., Powell, S., Brown, K., Gude, P. H., et al. (2005).Effects of exurban development on biodiversity: Patterns, mechanisms, andresearch needs. Ecological Applications, 15, 1893−1905.

Hansen, A. J., Neilson, R. P., Dale, V. H., Flather, C. H., Iverson, L. R., Currie, D. J., et al.(2001). Global change in forests: Responses of species, communities, and biomes.BioScience, 51, 765−779.

Hansen, A. J., Rasker, R., Maxwell, B., Rotella, J. J., Johnson, J., Wright Parmenter, A., et al.(2002). Ecological causes and consequences of demographic change in the NewWest. BioScience, 52, 151−168.

Hansen, A. J., & Rotella, J. J. (2002). Biophysical factors, land use, and species viability inand around nature reserves. Conservation Biology, 16, 1112−1122.

Hansen, A. J., & Urban, D. L. (1992). Avian response to landscape pattern: The role ofspecies' life histories. Landscape Ecology, 7, 163−180.

Heinz Center (The H. JohnHeinz III Center for Sceince, Economics and the Environment)(2006). Landscape pattern indicators for the nation: A report from the Heinz Center'sLandscape Pattern Task Group. Washington, DC: The H. John Heinz III Center forScience, Economics and the Environment.

Herzog, F., Lausch, A., Muller, E., Thulke, H. H., Steinhardt, U., & Lehmann, S. (2001).Landscape metrics for assessment of landscape destruction and rehabilitation.Environmental Management, 27, 91−107.

Hobbs, R. J. (2005). Restoration ecology and landscape ecology. In J. A.Wiens &M. R.Moss(Eds.), Issues and perspectives in landscape ecology (pp. 217−229). CambridgeUniversity Press.

Hoekstra, J. M., Boucher, T. M., Ricketts, T. H., & Roberts, C. (2005). Confronting a biomecrisis: Global disparities of habitat loss and protection. Ecology Letters, 8, 23−29.

Homer, C., Huang, C., Yang, L., Wylie, B., & Coan, M. (2004). Development of a 2001National Landcover Database for the United States. Photogrammetric Engineeringand Remote Sensing, 70, 829−840.

JMP (2007). JMP, version 7.0.2. Cary, North Carolina: SAS Institute Inc.Jongman, R. H. G. (2005). Landscape ecology in land-use planning. In J. A. Wiens & M. R.

Moss (Eds.), Issues and perspectives in landscape ecology (pp. 316−328). CambridgeUniversity Press.

Kerr, J. T., & Currie, D. J. (1995). Effects of human activity on global extinction risk.Conservation Biology, 9, 1528−1538.

Kupfer, J. A. (2006). National assessments of forest fragmentation in the U.S. GlobalEnvironmental Change, 16, 73−82.

Lande, R. (1987). Extinction thresholds in demographic models of territorial popula-tions. American Naturalist, 130, 624−635.

Leopold, A. S., Cottam, C., Cowan, I., Gabrielson, I. N., & Kimball, T. L. (1968). The NationalWildlife Refuge System, Report of the Advisory Committee on Wildlife Manage-ment. Final Environmental Statement, Operation of the National Wildlife RefugeSystem Washington, D.C.: U.S. Department of the Interior, U.S. Fish and WildlifeService W-1.

Li, X., He, H. S., Bu, R., Wen, Q., Chang, Y., Hu, Y., et al. (2005). The adequacy of differentlandscapemetrics for various landscape patterns. Pattern Recognition, 38, 2626−2638.

Lindenmayer, D. B., Cunningham, R. B., Donnelly, C. F., & Lesslie, R. (2002). On the use oflandscape surrogates as ecological indicators in fragmented forests. Forest Ecologyand Management, 159, 203−216.

Lindenmayer, D. B., Cunningham, R. B., & Fischer, J. (2005). Vegetation cover thresholdsand species responses. Biological Conservation, 124, 311−316.

Liu, J., Daily, G. C., Ehrlich, P. R., & Luck, G. W. (2003). Effects of household dynamics onresource consumption and biodiversity. Nature, 421, 530−533.

Lo, C. P., & Yang, X. (2002). Drivers of land-use/land-cover changes and dynamicmodeling for the Atlanta, Georgia metropolitan area. Photogrammetric Engineering& Remote Sensing, 68, 1073−1082.

Luck, G. W. (2007). A review of the relationships between human population densityand biodiversity. Biological Review, 82, 607−645.

McAlpine, C. A., & Eyre, T. J. (2002). Testing landscape metrics as indicators of habitatloss and fragmentation in continuous eucalypt forests (Queensland, Australia).Landscape Ecology, 17, 711−728.

McGarigal, K., & Marks, B. J. (1995). FRAGSTATS: Spatial pattern analysis program forquantifying landscape structure. General technical report PNW-GTR-351 Portland,OR: USDA Forest Service, Pacific Northwest Research Station.

McIntyre, S., & Hobbs, R. J. (2000). Human impacts on landscapes: Matrix condition andmanagement priorities. In J. Craig, D. A. Saunders, & N. Mitchell (Eds.), Natureconservation 5: Nature conservation in production environments (pp. 301−307).Chipping Norton, NSW: Surrey Beatty.

McIntyre, S., & Hobbs, R. J. (1999). A framework for conceptualizing human impacts onlandscapes and its relevance to management and research. Conservation Biology, 13,1282−1292.

McKinney, M. L. (2002). Urbanization, biodiversity, and conservation. BioScience, 52,883−890.

Meffe, G. K., & Carroll, C. R. (1997). Conservation reserves in heterogeneous landscapes. InG. K.Meffe, & C. R. Carroll (Eds.), Principles of conservation biology (pp. 305−346)., 2ndEdition Sunderland, Massachusetts: Sinauer Associates.

Miller, J. R., & Hobbs, R. J. (2002). Conservation where people live and work. Conser-vation Biology, 16, 330−337.

Myers, N. (1972). National parks in savannah Africa. Science, 178, 1255−1263.Naidoo, R., & Adamowicz, W. L. (2001). Effects of economic prosperity on numbers of

threatened species. Conservation Biology, 15, 1021−1029.Nassauer, J. I. (2005). Using cultural knowledge to make new landscape patterns. In J. A.

Wiens &M. R. Moss (Eds.), Issues and perspectives in landscape ecology (pp. 274−280).Cambridge University Press.

National Park Service (NPS) (1980). State of the Parks — 1980: A report to the congress.Washington, D.C.: National Park Service, U.S. Department of Interior.

National Park Service (NPS) (1993). Science and the National Parks II: Adapting to change.Washington, DC: Government Printing Office.

National Park Service (NPS) (1999).Natural Resource Challenge: The National Park Service'sAction Plan for Preserving Natural Resources.Washington, D.C.: U.S. Department ofInterior Available online at http://www.nature.nps.gov/challengedoc

National Park Service (NPS) (2001). Management policies 2001. Washington, D.C.: U.S.Department of Interior.

National Park Service Advisory Board (NPSAB) (2001). Rethinking the National Parks forthe 21st century.Washington, D.C.: National Park Service, U. S. Department ofInterior Available online http://www.nps.gov/policy/futurereport.htm

National Parks and Conservation Association (NPCA) (1979). NPCA adjacent landssurvey: No park is an island. National Parks and Conservation Magazine, 53, 4−9.

Naveh, Z. (1995). Interactions of landscapes and cultures. Landscape and UrbanPlanning, 32, 43−54.

Newmark, W. D., Manyanza, D. N., Gamassa, D. M., & Sariko, H. I. (1994). The conflictbetween wildlife and local people living adjacent to protected areas in Tanzania:Human density as a predictor. Conservation Biology, 8, 249−255.

Noss, R. F. (1999). Assessing andmonitoring forest biodiversity: A suggested frameworkand indicators. Forest Ecology and Management, 115, 135−146.

O'Neill, R. V., Hunsaker, C. T., Jones, K. B., Riitters, K. H.,Wickham, J. D., Schwartz, S.M., et al.(1997). Monitoring environmental quality at the landscape scale. BioScience, 47,513−519.

O'Neill, R. V., Krummel, J. R., Gardner, R. H., Sugihara, G., Jackson, B., DeAngelis, D. L., et al.(1988). Indices of landscape pattern. Landscape Ecology, 1, 153−162.

Opdam, P., Verboom, J., & Pouwels, R. (2003). Landscape cohesion: An index for theconservation potential of landscapes for biodiversity. Landscape Ecology, 18, 113−126.

Parker, M., &Mac Nally, R. (2002). Habitat loss and the habitat fragmentation threshold:An experimental evaluation of impacts on richness and total abundances usinggrassland invertebrates. Biological Conservation, 105, 217−229.

Parks, S. A., & Harcourt, A. H. (2002). Reserve size, local human density, andmammalianextinctions in U.S. protected areas. Conservation Biology, 16, 800−808.

Pearson, S. M., Turner, M. G., Gardner, R. H., & O'Neill, R. V. (1996). An organism-basedperspective of habitat fragmentation. In R. C. Szaro & D. W. Johnston (Eds.), Biodi-versity in managed landscapes: Theory and practice (pp. 77−95). New York: OxfordUniversity Press.

Pidgorna, A., (2007). Representation, Redundancy, and Resilience: Waterfowl and theNational Refuge System. PhD Dissertation, University of Idaho, Moscow, Idaho.

Radford, J. Q., Bennett, A. F., & Cheers, G. J. (2005). Landscape-level thresholds of habitatcover for woodland-dependent birds. Biological Conservation, 124, 317−337.

Rempel, R. (2006). Patch analyst, version 3. Thunder Bay, Ontario, Canada: LakeheadUniversity.

Riitters, K. H., O'Neill, R. V., Hunsaker, C. T., Wickham, J. D., Yankee, D. H., Timmins, S. P.,et al. (1995). A factor analysis of landscape pattern and structuremetrics. LandscapeEcology, 10, 23−39.

Riitters, K. H., Wickham, J. D., O'Neill, R. V., Jones, K. B., Smith, E. R., Coulston, J. W., et al.(2002). Fragmentation of continental United States forests. Ecosystems, 5, 815−822.

Riitters, K. H., Wickham, J. D., & Wade, T. G. (2006). Evaluating ecoregions for samplingand mapping land-cover patterns. Photogrammetric Engineering and RemoteSensing, 72, 781−788.

Robbins, W. J., Ackerman, E. A., Bates, M., Cain, S. A., Darling, F. D., Fogg, J. M., Jr., et al.(1963). National Academy of Sciences Advisory Committee on Research in theNational Parks: The Robbins Report Available online http://www.cr.nps.gov/history/online_books/robbins/robbins.htm

Sala, O. E., Chapin, F. S., III, Arnesto, J. J., Berlow, E., Bloomfield, J., Dirzo, R., et al. (2000).Global biodiversity scenarios for the year 2100. Science, 287, 1770−1774.

Saunders, D. A., Hobbs, R. J., & Margules, C. R. (1991). Biological consequences ofecosystem fragmentation: A review. Conservation Biology, 5, 18−32.

Saura, S. (2004). Effects of remote sensor spatial resolution and data aggregation onselected fragmentation indices. Landscape Ecology, 19, 197−209.

Saura, S., & Castro, S. (2007). Scaling functions for landscape pattern metrics derivedfrom remotely sensed data: Are their subpixel estimates really accurate? Journal ofPhotogrammetry & Remote Sensing, 62, 201−216.

Schonewald-Cox, C.M. (1988). Boundaries in the protectionof nature reserves: Translatingmultidisciplinary knowledge into practical conservation. BioScience, 38, 480−486.

Scott, J. M., Davis, F., Csuti, B., Noss, R. F., Butterfield, B., Groves, C., et al. (1993). Gapanalysis: A geographic approach to protection of biological diversity. WildlifeMonographs No. 123.

Scott, J. M., Davis, F. W., Gavin McGhie, R., Wright, R. G., Groves, C., & Estes, J. (2001).Nature reserves: do they capture the full range of America's biological diversity?Ecological Applications, 11, 999−1007.

Scott, J. M., Loveland, T. R., Gergely, K., Strittholt, J., & Staus, N. (2004). National WildlifeRefuge System: Ecological context and integrity. Natural Resources Journal, 44,1041−1066.

Seabloom, E. W., Dobson, A., & Stoms, D. B. (2002). Extinction rates under nonrandompatterns of habitat loss. Proceedings of the National Academy of Sciences of the UnitedStates of America, 99, 11229−11234.

Page 13: Assessing the landscape context and conversion risk of

1369L.K. Svancara et al. / Remote Sensing of Environment 113 (2009) 1357–1369

Shafer, C. L. (1999). National park and reserve planning to protect biological diversity:Some basic elements. Landscape and Urban Planning, 44, 123−153.

Shands, W. E. (1979). Federal resource lands and their neighbors. Washington, DC:Conservation Foundation.

Sisk, T. D., Launer, A. E., Switky, K. R., & Ehrlich, P. R. (1994). Identifying extinctionthreats: Global analyses of the distribution of biodiversity and the expansion of thehuman enterprise. BioScience, 44, 592−604.

Stauffer, D. (1985). Introduction to percolation theory. Philadelphia, Pennsylvania, USA:Taylor and Francis.

Stehman, S. V., Wickham, J. D., Smith, J. H., & Yang, L. (2003). Thematic accuracy of the1992 National Land Cover Data (NLCD) for the eastern United States: Statisticalmethodology and regional results. Remote Sensing of Environment, 86, 500−516.

Svancara, L. K., Brannon, R., Scott, J. M., Groves, C. R., Noss, R. F., & Pressey, R. L. (2005).Policy-driven versus evidence-based conservation: A review of political targets andbiological needs. BioScience, 55, 989−995.

Theobald, D. M. (2004). Placing exurban land-use change in a human modificationframework. Frontiers in Ecology and the Environment, 2, 139−144.

Theobald, D.M., Spies, T., Kline, J., Maxwell, B., Hobbs, N. T., & Dale, V. H. (2005). Ecologicalsupport for rural land-use planning. Ecological Applications, 15, 1906−1914.

Tischendorf, L. (2001). Can landscape indices predict ecological processes consistently?Landscape Ecology, 16, 235−254.

Tscharntke, T., Steffan-Dewenter, I., Kruess, A., & Thies, C. (2002). Contribution of smallhabitat fragments to conservation of insect communities of grassland–croplandlandscapes. Ecological Applications, 12, 354−363.

Turner, M. G., & Gardner, R. H. (1991). Quantitative methods in landscape ecology: Anintroduction. In M. G. Turner, & R. H. Gardner (Eds.), Quantitative methods inlandscape ecology (pp. 3−14). New York: Springer-Verlag.

Turner, M. G., O'Neill, R. V., Gardner, R. H., & Milne, B. T. (1989). Effects of changingspatial scale on the analysis of landscape pattern. Landscape Ecology, 3, 153−162.

U.S. Bureau of Economic Analysis (2003). Regional Economic Information System.Washington, DC: U. S. Bureau of Economic Analysis Online: http://www.bea.gov

U.S. Census Bureau (2000). U. S. Census Database, 2000.Washington, DC: U. S. CensusBureau Online: http://www.census.gov

Udall, S. L. (1962). Nature islands for theworld. In A. B. Adams (Ed.), FirstWorld Conferenceon National Parks (pp. 1−10). Washington, DC: U.S. Department of Interior, NationalPark Service.

Vitousek, P. M., Mooney, H. A., Lubchenco, J., & Melillo, J. M. (1997). Human dominationof Earth's ecosystems. Science, 277, 494−499.

Wade, T. G., Riitters, K. H., Wickham, J. D., & Jones, K. B. (2003). Distribution and causesof global forest fragmentation.Conservation Ecology, 7, 7 Online at http://www.consecol.org/vol7/iss2/art7

Ward, D., Phinn, S. R., & Murray, A. T. (2000). Monitoring growth in rapidly urbanizingareas using remotely sensed data. Professional Geographer, 52, 371−386.

Western, D. (1982). Amboseli National Park: Enlisting landowners to conservemigratory wildlife. Ambio, 11, 302−308.

Wickham, J. D., & Riitters, K. H. (1995). Sensitivity of landscape metrics to pixel size.International Journal of Remote Sensing, 16, 3585−3594.

Wickham, J. D., Stehman, S. V., Smith, J. H., & Yang, L. (2004). Thematic accuracy of the1992 National Land Cover Data for the western United States. Remote Sensing ofEnvironment, 91, 452−468.

Wiens, J. A. (1996). Wildlife in patchy environments: Metapopulations, mosaics, andmanagement. InD. R.McCullough (Ed.),Metapopulations and conservation (pp. 53−84).Washington, DC: Island Press.

Wiens, J. A., Sutter, R., Anderson, M., Blanchard, J., Barnett, A., Aguilar-Amuchastegui, N.,et al. (2009). Selecting and conserving lands for biodiversity: the role of remotesensing. Remote Sensing of Environment, 113, 1370−1381 (this issue).

Wiersma, Y. F., Nudds, T. D., & Rivard, D. H. (2004). Models to distinguish effects oflandscape patterns and human population pressures associated with species loss inCanadian national parks. Landscape Ecology, 19, 773−786.

With, K. A. (2005). Landscape conservation: A new paradigm for the conservation ofbiodiversity. In J. A. Wiens, & M. R. Moss (Eds.), Issues and perspectives in landscapeecology (pp. 238−247). : Cambridge University Press.

With, K. A., & Crist, T. O. (1995). Critical thresholds in species' responses to landscapestructure. Ecology, 76, 2446−2459.

Woodroffe, R. (2000). Predators and people: Using human densities to interpretdeclines of large carnivores. Animal Conservation, 3, 165−173.

World Bank (1992). World Development Report 1992: Development and the environment.New York: Oxford University Press.

Wright,G.M.,Dixon, J. S., & Thompson, B.H. (1933). Faunaof thenational parks:Apreliminarysurvey of faunal relations in national parks. Fauna series no. 1 Washington, DC: U.S.Government Printing Office Online: http://www.cr.nps.gov/history/online_books/fauna1/fauna.htm

Wu, J. (2004). Effects of changing scale on landscape pattern analysis: Scaling relations.Landscape Ecology, 19, 125−138.

Wu, J., Jelinski, D. E., Luck, M., & Tueller, P. T. (2000). Multiscale analysis of landscapeheterogeneity: Scale variance and pattern metrics. Geographic Information Sciences,6, 6−19.

Wu, J., Shen, W., Sun, W., & Tueller, P. T. (2002). Empirical patterns of the effects ofchanging scale on landscape metrics. Landscape Ecology, 17, 761−782.

Zube, E. H. (1995). No park is an island. In J. A. McNeely (Ed.), Expanding partnerships inconservation (pp. 169−177). Washington, DC: Island Press.