Assessing Land-Use Impacts on Natural ResourcesV. H. DALE*A. W. KINGL. K. MANNR. A. WASHINGTON-ALLENR. A. McCORDEnvironmental Sciences DivisionOak Ridge National LaboratoryOak Ridge, Tennessee 37831-6031, USA

ABSTRACT / Much information is available on changes thatoccur in natural resources from both spatially-explicit dataon environmental conditions and models of the interactionsof these conditions and resources with human activities. Thestrategy for assessing land-use impacts on natural re-sources developed in this paper provides a framework forusing relevant data and models to address questions of howmanagement practices can promote both use and protec-tion of resources. This assessment strategy integrates spa-tially explicit environmental data using geographic informa-

tion systems (GIS) with computer models that simulatechanges in land cover in response to land-use impacts. Thecomputer models also simulate susceptibility of species tochanges in habitat suitability and landscape patterns. Theapproach is applied to management of limestone barrens onthe Oak Ridge Reservation in East Tennessee. Potential lime-stone barrens habitats are identified by overlaying appropri-ate soils, geology, slope, and land-use/land-cover condi-tions. Their validity is tested against known sites containingrare species that occur in these habitats. The location ofhabitats at risk in the aftermath of human activities is deter-mined by using an available area model that identifies thesize and proximity of sites that particular types of speciescan no longer use as habitat. The resulting risk map can beused in land management planning. The approach usesreadily available in situ and remotely sensed data and is ap-plicable to a wide range of locations and land-use scenarios.This approach can be refined based on needs identified byland managers and on the sensitivity of the results to theresolution of available resource information.

Appropriate management of environmental re-sources is of great concern to those involved in the useand protection of these resources (Noss 1992, Swansonand Franklin 1992, USDA Forest Service 1992). Re-search needs for ecosystem management have beendiscussed (Lubchenco and others 1991, Christensenand others 1996, Holling 1993). The need to considersocial and policy aspects of resource management hasbeen emphasized (Clark 1992, Moran 1990, Slocombe1993). Based on an extensive review of the conservationand policy literature, Grumbine (1994) surmises thatecosystem management has several goals: maintainingviable populations, diverse ecosystem representation,and ecological processes; protecting evolutionary poten-tial of species and ecosystems; and accommodatinghuman use. Some specific management alternativeshave been debated (e.g., movement corridors—Simber-loff and others 1992), and ecosystem management hasbeen implemented in specific locations (e.g., Bur-roughs and Clark 1995). Nevertheless, a general strat-egy that uses computer tools for assessing land-useimpacts on natural resources has not been published.

Land-use activities can cause significant impacts to

natural resources. Forest cutting, agricultural practices,urban and industrial expansion, road development,military training, and alteration of waterways are amongsome of the common human and societal land-useactivities that can significantly alter land cover. Suchland-use activities may remove vegetation cover, com-pact soils, create erosion, alter local and regionalhydrology, or degrade soil productivity by reducingnutrient and organic matter content. Moreover, thepresence and activity of humans alters the balance ofwildlife because some species cannot coexist with hu-mans, while others (e.g., starlings, raccoons) thrive inthe presence of human activity.

Habitat loss and fragmentation that result fromland-use changes are major factors contributing to thedecline of many biological populations. Many species orpopulations at risk are associated with specific habitats,and the loss of those habitats can directly or indirectlyimpact the species. For example in the southern Appala-chian region, 84% of the 51 federally listed threatenedand endangered plant and animal species are associ-ated with rare community and streamside habitats(SAMAB 1996). As another example, in the UnitedStates at least one third of the threatened or endan-gered species live in wetlands (Murdock 1994). Re-cently, ecologists have recognized that patterns of landuse and fragmentation of the landscape can have assignificant an effect on biological populations as an

KEY WORDS: Land management; Assessment; Habitat characteriza-tion; Limestone barrens; Ecological modeling; Geo-graphic information systems

*Author to whom correspondence should be addressed.

Environmental Management Vol. 22, No. 2, pp. 203–211 r 1998 Springer-Verlag New York Inc.

absolute change in habitat area (Forman and Godron1986, Askins and others 1989, Bierregaard and others1992, Turner 1989).

In this paper we present an assessment strategy fordetermining landscape impacts of human activities.The strategy characterizes the susceptibility of habitatsto specific land management practices in a spatiallyexplicit manner. Once the spatial distributions of habi-tats of special concern are determined, the particularhabitats that are of greatest value and most at risk tochange are identified. The resulting assessment presentsa decision maker with the information necessary tocompare alternative management plans as to theirsuccess in protecting the most critical and vulnerableoccurrences of sensitive habitats while achieving theobjectives of the land manager.

Approach

The assessment strategy we have developed includesfour components (Figure 1): (1) the creation of spa-

tially-explicit data layers within a geographic informa-tion system (GIS), (2) habitat characterization, (3)relating habitat to ecological resources of interest, and(4) the production of biological resource risk maps.After presenting the general features of the strategybelow, we illustrate the approach by using limestonebarrens as the habitat of conservation concern and thepopulations that occupy them as the natural resourceunder consideration.

The GIS

The first step of the assessment approach is tocompile spatially explicit data layers that describe thephysical and biological environment of the area. Theselayers commonly include land use and land cover,geology, soils, topography, hydrology, and administra-tive infrastructure (e.g., roads and buildings). Thisinformation can be obtained from in situ or remotelysensed data. Past disturbances, both natural and anthro-pogenic, are also integrated into a data layer of landuse/land cover. Time-series data are especially useful inanalyzing the historical impact of land managementpractices.

Habitat Characterization

Characterizing a habitat of interest requires knowingthe physical features that define the habitat, having dataat the appropriate temporal and spatial resolution, anddeveloping a GIS algorithm that combines the datalayers to produce a habitat map. The physical featuresdefining a habitat might include soil characteristics,water availability, land use, land cover, geology, andclimate. The appropriate temporal and spatial resolu-tion for the habitat depends on the size and life cycles ofthe organisms of interest and the scale of the manage-ment regimes (e.g., management actions may be revisedquarterly for each site). The process of characterizinghabitat by intersecting ecologically related factors isreferred to as rules of combination (Kiefer 1965,Hopkins 1977, Westman 1985). Most GIS packages havetools for combining data layers, yet ecological knowl-edge of the data and habitat are needed to developalgorithms that appropriately reflect the issues of con-cern. For instance, it may be that soil moisture condi-tions are crucial during the reproductive season of theorganism of interest but are not important at othertimes.

The final step of habitat characterization is perform-ing an assessment of the accuracy of the method inidentifying the habitat. Independently identified habi-tats (e.g., from field studies) can be used as a check ofthe procedure. These independent sites should repre-sent the diversity of features that define the habitat.

Figure 1. Strategy for assessing land-use impacts.

V. H. Dale and others204

Relating Habitats to Ecological Resources

Once a habitat has been defined in terms of GISlayers, it can be related to the spatial distribution ofeither ecological or cultural resources on the landscape.For purposes of illustration, we discuss the relation ofhabitats to populations and focus on species of specialinterest (e.g., rare and endangered). However, thereare two problems in relating habitats to populations.First, the parameters reflecting important features of aspecies’ habitat may not be directly related to a GISlayer. Second, some species are not habitat limited orare impacted by factors only marginally related tohabitat features (e.g., competition).

Habitats can be related to biological populationsusing one or more types of models: habitat associationmodels, spatially explicit population models (Dunningand others 1995), or cell-based landscape models (Gard-ner and others 1989, Turner and others 1989). Habitatassociation models assume a correlation between thepresence of habitat and the presence of certain species.They can be deterministic (e.g., if habitat type h ispresent, then individuals of species s are assumed to bepresent) or probabilistic (e.g., the presence of a speciesis expressed as a probability of occurrence). Habitatquality models, a special kind of habitat associationmodel, characterize each occurrence of the habitat as ameasure, or index, of habitat quality or suitability(Maurer 1986). Examples of habitat quality modelsinclude multivariate statistical models, which relatehabitat variables to population density, and habitatevaluation procedures (HEP) of the US Fish andWildlife Service. HEP models generate a habitat suitabil-ity index (HSI) based on qualitative or simple quantita-tive measures of habitat features (Schamberger andothers 1982). Individuals of the species of concern aremore likely to occur on those habitats with higher HSIscores. Some HSI models predict both the potential fora species to occur and its density.

Spatially-explicit population models describe thespatial distribution of individuals or population densityacross the landscape and how that distribution changeswith time. For example, the density of individuals of aspecies at time t and place s, N(t,s), might be repre-sented by a difference equation formulation of logisticpopulation growth that includes local immigration andemigration:

N(t, s) 5 f [N(t 2 1, s), r(h), k(h), e, I ]

where h is an index of habitat type at place s, r(h) is therate of increase in habitat h, k(h) is the carryingcapacity, e is the emigration rate (which may be afunction of density), and I is the immigration rate ofindividuals from surrounding areas. The greatest den-

sity would occur in areas with habitat most suitable forthe population (e.g., high r and k). In contrast to thehabitat association models and habitat quality models,the spatially explicit models consider the spatial relation-ships among habitat patches. Consequently, they can beused to investigate the impact of changes in land-use/land-cover on the fragmentation and other alterationsof the spatial distribution of habitats. Moreover, spatially-explicit population models relate habitat variables todemographic variables and may avoid some of thelimitations of the density-habitat regressions that under-lie habitat quality models (Maurer 1986). The spatially-explicit models are, however, data intensive and moredifficult to implement.

Cell-based landscape models also consider the spa-tially-explicit relationships among patches of habitat.However, they do not normally represent individuals orpopulation density explicitly. Rather, they describe thestate of discrete cells that define the landscape and thespatial relationship among the cells. Spatial patterns ofpatchiness, connectivity, and the like are related implic-itly (and sometimes explicitly) to populations thatutilize the landscape. For example, individuals may beable to disperse through connected habitat cells belong-ing to a single cluster but not able to disperse acrossintervening cells of nonhabitat. In such cases increasedfragmentation is associated with reduced dispersal. Thecell-based landscape models are, in some sense, acompromise between habitat quality models that lackconsideration of spatial pattern and spatially-explicitpopulation models that require extensive informationof demographic processes. As such, these models arewell suited for spatially-explicit assessment of land-useimpacts. Cell-based models include both spatial pat-terns and some information on how species use thelandscape. We will use a cell-based model, Pearson’savailable area model (Dale and others 1994, Pearsonand others 1996), to illustrate our proposed assessmentstrategy.

Biological Resource Risk Maps

Biological resource risk maps depict habitats orother natural resources at risk under certain landmanagement regimes. Biological resource risk mapscan identify areas that are susceptible to particular landuses and the degree of potential damage. Summarizedas a map or video, this information can be presented tothe land manager to graphically illustrate the resourcesat risk under specified land management practices.Using these risk maps, resource managers can plan thelocation and timing of land management or use in sucha way as to minimize impacts on the natural resource.For example, human activities can be scheduled to avoid

Assessing Land-Use Impacts 205

sensitive times such as during the breeding season or whensoil moisture is high and site degradation is more likely.

An Example: Limestone Barrenson the Oak Ridge Reservation

We illustrate the application of our strategy forland-use assessment using the limestone barrens lo-cated on the Department of Energy’s Oak Ridge Reser-vation (ORR) in Oak Ridge, Tennessee. The 14,266-haORR lies between 84° 268 and 84° 118 W and 35° 528 and36° N in the Ridge-and-Valley Province of East Tennes-see. Major plant communities of the reservation includeoak–hickory forests, native pine–hardwood forests, bot-tomland and riparian hardwood forests, loblolly pineplantations, northern hardwood forests, old fields, grass-lands, and limestone barrens (Mann and others 1996).

Limestone barrens are habitats of sparse woodlandsof mixed hardwoods and eastern red cedar ( Juniperusvirginiana) that contain grassy or herbaceous openings,intermixed with limestone or dolomite rock occurringat or near the surface of the ground. The distribution ofbarrens is determined by local edaphic factors, pastclimate conditions (Delcourt and others 1986), anddisturbance events that may be important in maintain-ing the habitats (DeSelm and Murdock 1993). Theopenings are frequently small (,1 ha), dispersed, andfragmented, yet they are critical for the continuedexistence of the endemic species (mostly plants andinvertebrates) associated with them (Freeman and oth-ers 1979). The barrens flora includes species that arerare or endangered (Cunningham and others 1993).These endemic species may be locally abundant in thebarrens openings, with different openings having ahigh degree of taxonomic similarity (DeSelm and oth-ers 1969). The long-term persistence of these endemicsis aided by, if not dependent upon, dispersal among theindividual barrens and the recolonization of sites thathave experienced local extinction. Barrens are habitatsof conservation concern throughout the southeasternUnited States. The management of barrens involves notonly the management of local conditions (e.g., con-trolled burns to suppress woody invasion) but also themaintenance of a larger-scale spatial distribution ofbarrens that allows dispersal and recolonization.

The strategy for assessing land-use impacts on natu-ral resources provides methods both to describe thedistribution of barrens and to assess impacts of humanactivities on the barrens. The steps for assessmentstrategy are described sequentially below.

The GIS

GIS data layers for ORR were acquired from a varietyof sources. GIS and image processing software, includ-

ing GRASS 4.1 (USA-CERL 1993), ERDAS Imagine 8.2(ERDAS 1994), and ARC/INFO 7.0 (ESRI 1993), wereused to conduct the analyses on workstations using theUNIX operating system. The land-use/land-cover datalayer was generated from spectral reflectance data fromthe Landsat Thematic Mapper (TM) satellite. Fre-quently the spatial distribution of habitats is not avail-able as a primary data layer (land-cover/land-use catego-ries, for example) or is too coarse to represent uniquehabitat types. Consequently, habitat characterizationfrequently requires modeling of the habitat via integra-tion of the attributes in the GIS map layers. Habitatcharacterization thus involves identification of ecologi-cal attributes that uniquely define the habitats and theirspatial distribution. For ORR, the spectral reflectancesignatures were characterized into nine cover classes,adapted from Anderson and others (1976): water,evergreen plantation, barren land, urban, pasture, tran-sitional areas, and evergreen, deciduous, and mixedforest. We included the urban land-use category be-cause its constituents (e.g., roads, trimmed lawns, graveldriveways, greenways, etc.) have spectral signaturessimilar to the exposed rock outcrops or level gravel-likecauseways of some limestone barrens. Digital elevationdata obtained from the USGS were converted to per-cent slope data using ERDAS Imagine’s Topographicutilities. Together, these data provide the necessaryinformation to define limestone barren habitats.

Habitat Characterization:Predicting Limestone Barrens

Limestone barrens occur on the Oak Ridge Reserva-tion in sites with appropriate soils, geology, slope, andland use/land cover (Figure 2). Mann and Kitchings(1982) hypothesized that limestone barrens are charac-terized by soils that are lithic Rendolls, Mollisols (Argiu-dolls) or lithic and typic Hapludalfs, but this hypothesiswas not tested in the field. In Oak Ridge, limestonebarrens occur on a geology of thin or a mixture of thinand medium bedded Chickamauga limestone residuumand colluvium and on slopes less than 25% (Hatcherand others 1992). The land cover/land use on thelimestone barrens of Oak Ridge is either urban, pastureland, barren land, mixed forest land, or transitional.Cells in the intersection of the data layers for soil,geology, slope, and land use/land cover that meet thenecessary criteria for limestone barrens are predicted tobe barrens (Figure 3).

The accuracy of the prediction was assessed byoverlaying the predicted area of limestone barrens witha GIS layer of ORR’s eight ORNL Research Park NaturalAreas (NA), which contain barren species. NAs are siteswhere federally and state of Tennessee listed threatenedand endangered vertebrate animal and plant species of

V. H. Dale and others206

the southern Appalachians are known or are likely tooccur (Pounds and others 1993, King and others 1994).The NAs are administrative units created by ORNL topreserve biological diversity in compliance with theNational Environmental Policy Act and other federaland state regulations (Pounds and others 1993). Of theeight NAs that contain barren species on the ORR, sixwere predicted to be barren habitat by the habitatcharacterization procedure described herein. In termsof area, approximately 166 ha of 189 ha (88%) ofknown barren habitat intersected with 291 ha of pre-dicted barren habitat. The remaining 125 ha (43%) ofpredicted habitat probably consists of sites that remainundiscovered, have changed due to vegetation succes-sion, have disappeared due to land-use change, or were

mispredicted, probably because the urban land-usecategory was included, which contains some roads andother features that may be incorrectly predicted to belimestone barrens.

Relating Habitat to Ecological Resources:An Example Using a Cell-Based Landscape Model

The question of how habitat fragmentation affectsthe spatial distribution of a species on limestone bar-rens is addressed by using Pearson’s available areamodel (Dale and others 1994, Pearson and others,1996). The model assumes that areas of habitat that aresmaller than the minimum area required by an indi-vidual (or group if the organism only occurs in groups)cannot support the species and thus are ‘‘unavailable.’’

Figure 2. GIS layers used to characterize the limestone barren habitat on the Oak Ridge Reservation (ORR). The soil series layer(A) and slope layer (B) are depicted as gray scales of the 122 different soil series and a slope range of 0% (white) to 182% (black),respectively, on the ORR (outlined in yellow). The geology layer (C) has 32 formations, 13 of which are Chickamauga limestonethat support barrens (i.e., Ocy/l, Obe, Obl, Obw, Oca, Och, Ocn, Oh, Olb, Omc, Ops, Ord, and Owi) (Hatcher and others 1992).Layer D is the 1994 land use/land cover of the ORR (outlined in yellow), which was derived from classification of a Landsat TMimage.

Assessing Land-Use Impacts 207

Patches of suitable habitat are defined by the species’gap-crossing ability. By crossing gaps, organisms are ableto utilize otherwise separated occurrences of habitat.Species with wide gap-crossing ability are assumed toview relatively dispersed occurrences of habitat as partof a continuous cluster of available space. Species withnarrow gap-crossing ability are assumed to perceive thesame landscape as more fragmented and as a collectionof smaller isolated habitats. Highly vagile insects orother small but mobile animals might be species thathave low area requirements but that have high gap-crossing ability. Large organisms that prefer to remainconcealed by dense cover might be typical of speciesthat have large area requirements but low gap-crossingability. The total habitat area available to a species is thesum of the areas of patches of habitat large enough tosupport an individual of the species (i.e., those greaterthan the species’ minimum area requirement).

To illustrate the implications of fragmentation, weexamined effects of fragmentation on the cedar barrenslocated in the northeast portion of the ORR. Figure 4shows the predicted barrens area available to specieswith different area requirements and gap-crossing abili-ties. In general, the area of suitable habitat decreases as

gap-crossing ability decreases and as a species arearequirements increase. For species with either smallarea requirements or wide gap-crossing abilities, mostor all of the 102 ha of predicted barrens in the northeastpart of the ORR are available habitat and can be

Figure 3. Areas of predicted limestone barren habitat (shown in white) for the northeast portion of the Oak Ridge Reservation(ORR).

Figure 4. Predicted available habitat for species of differentgap crossing ability and area requirements before habitatremoval.

V. H. Dale and others208

utilized. On the other hand, for those species with alarge area requirement (e.g., 10 ha) and a narrowgap-crossing ability (,25 m), only about 29 ha (28%) ofthe barrens habitat is suitable.

To examine effects of a systematic land change, weassumed a change in land use that removed 22 ha in auniform strip across the ORR barrens, leaving 80 ha inbarrens, a modeled 21% reduction of the total barrenshabitat. The amount of habitat remaining for a speciesdepends on their gap-crossing ability and area require-ment (Figure 5). All species had a slightly higherpercentage of the remaining barrens habitat availableto them as suitable habitat, but species with a large arearequirement and low gap-crossing ability lost morehabitat area than those with large area requirements.For example, the species discussed in the previousparagraph, which has a gap-crossing ability of ,25 mand an area requirement of 10 ha, now has 24 ha of thebarrens habitat as suitable compared to 29 ha with nodisturbance.

Biological Resource Risk Map

A habitat risk map was produced to illustrate the areaof limestone barrens habitat available for a species withgap-crossing ability less than 25 m and area require-ments of 1 ha (Figure 6). Such a species has 67.7 ha ofsuitable habitat before the removal and 54.8 ha of

Figure 5. Predicted suitable habitat for species of differentgap crossing ability and area requirements after 22 ha oflimestone barren was removed in a uniform strip of land.

Figure 6. Map of areas of limestone barrens habitat predicted to be at risk both by direct (blue) and indirect (red) disturbance to22-ha of habitat. The green sites are unimpacted limestone barren habitats.

Assessing Land-Use Impacts 209

suitable habitat after the removal. Although 22 ha oflimestone barrens were removed, not all of the removedarea was available habitat for the species of concern.Furthermore, a portion of remaining habitat was nolonger available to this species type, even though thehabitat was not directly removed by the change in landuse. This secondary removal of habitat occurs as a resultof habitat fragmentation that accompanies the simu-lated disturbance. These indirect habitat losses wouldnot be predicted by an assessment that did not explicitlyconsider the spatial arrangement of habitat.

Conclusions

The ORR limestone barrens example illustrates theapplicability of our proposed strategy for managementof land-use impacts on natural resources. A GIS-basedhabitat characterization provides a map of the spatialdistribution of habitats for the species of concern. Thehabitat characterization map can be used to identifysites at direct risk from land-use change. Sites of indirectrisk can be identified by using spatially explicit ecologi-cal models of populations of concern. These indirecteffects cannot be predicted from knowledge of totalhabitat loss. Presentation of these results as risk mapsprovides for the communication of model results to theland managers. The example we have presented showsthat a habitat characterization, spatially-explicit ecologi-cal models, and risk maps are all essential elements inan integrated analysis of resources at risk from changesin land use.

The approach can be adapted based on the needs ofspecific land managers and the data available. Theintent of the approach is to enhance the ability toachieve the mission for the sites while conservingnatural resources. Although developed primarily foruse on federal lands, this standard approach can helpprioritize the collection and maintenance of naturalresource information for private lands.

Acknowledgments

The assistance of Robin Graham, Pat Scarbrough,and Tom Chatfield in the preparation of GIS data isappreciated. Tom Ashwood, Glen Suter, Craig Allin,and D. S. Slocombe provided insightful reviews of thepaper. Mark Smith provided useful discussions in thedevelopment of this manuscript. The project was fundedby a contract from the Conservation Program of theStrategic Environmental Research and DevelopmentProgram (SERDP) to Oak Ridge National Laboratory(ORNL). ORNL is managed by Lockheed Martin En-ergy Research Corp. for the Department of Energy

under contract DE-AC05-96OR22464. The paper isEnvironmental Sciences Publication number 4661.

Literature CitedAnderson, J. R., E. E. Hardy, J. T. Roach, and R. E. Witmer.

1976. A land use and land cover classification system for usewith remote sensor data. US Geological Survey ProfessionalPaper 964, 28 pp.

Askins, R. A., F. M. Ewert, and R. I. Norton. 1989. Abundanceof wintering migrants in fragmented and continuous forestsin the U.S. Virgin Islands. Pages 197–206 in Proceedings ofManomet Symposium, Smithsonian Institution Press, Wash-ington, D.C.

Bierregaard, R. O., T. E. Lovejoy, V. Kapos, A. A. Dos Santos,and R. W. Hutchings. 1992. The biological dynamics oftropical rainforest fragments. BioScience 42:859–866.

Burroughs, R. H., and T. W. Clark. 1995. Ecosystem manage-ment: A comparison of Greater Yellowstone and GeorgesBank. Environmental Management 19(5):649–663.

Christensen, N. L., A. M., Bartuska, J. H. Brown, S. R.Carpenter, C. D’Antonio, R. Francis, J. F. Franklin, J. A.MacMahon, R. F. Noss, D. J. Parsons, C. H. Peterson, M. G.Turner, and R. G. Woodmansee. 1996. The report of theEcological Society of America committee on the scientificbasis for ecosystem management. Ecological Applications 6(3):665–691.

Clark, T. W. 1992. Practicing natural resource managementwith a policy orientation. Environmental Management 16(4):423–433.

Cunningham, M., L. Pounds, S. Oberholster, P. Parr, L.Edwards, B. Rosensteel, and L. Mann. 1993. Resourcemanagement plan for the Oak Ridge Reservation, volume29: Rare plants on the Oak Ridge Reservation. ORNL/NERP-7, Oak Ridge, Tennessee, 93 pp.

Dale, V. H., S. M. Pearson, H. L. Offerman, and R. V. O’Neill.1994. Relating patterns of land-use change to faunal biodi-versity in the central Amazon. Conservation Biology 8:1027–1036.

Delcourt, P. A., H. R. Delcourt, P. A. Cridlebaugh, and J.Chapman. 1986. Holocene ethnobotanical and paleoecologi-cal record of human impact on vegetation in the LittleTennessee River Valley, Tennessee. Quaternary Research 25:330–349.

DeSelm, H. R., and N. Murdock. 1993. Grass-dominatedcommunities. Pages 87–141 in W. H. Martin, S. G. Boyce,and A. C. Ecternacht (eds.). Biodiversity of the southeasternUnited States. Upland terrestrial communities. John Wiley& Sons, New York.

DeSelm, H. R., P. B. Whitford, and J. S. Olson. 1969. Thebarrens of the Oak Ridge area, Tennessee. American MidlandNaturalist 81:315–330.

Dunning, J. B., Jr., D. J. Stewart, B. J. Danielson, B. R. Noon,T. L. Root, R. H. Lamberson, and E. E. Stevens. 1995.Spatially explicit population models: Current forms andfuture uses. Ecological Applications 5:3–11.

ESRI (Environmental Systems Research Institute). 1993. Un-derstanding GIS: The ARC/INFO method. Longman Scien-tific & Technical, Essex.

V. H. Dale and others210

ERDAS. 1994. ERDAS Field Guide. Erdas, Inc., Atlanta,628 pp.

Forman, R. T. T., and M. Godron. 1986. Landscape ecology.John Wiley & Sons, New York.

Freeman, J. D., A. S. Causey, J. W. Short, and R. R. Haynes.1979. Endangered, threatened, and special concern plantsof Alabama. Journal of the Alabama Academy of Science 50:1–26.

Gardner, R. H., R. V. O’Neill, M. G. Turner, and V. H. Dale.1989. Quantifying scale-dependent effects of animal move-ment with simple percolation models. Landscape Ecology3:217–228.

Grumbine, R. E. 1994. What is ecosystem management?Conservation Biology 8(1):27–38.

Hatcher, R. D., P. J. Lemiszki, R. B. Dreier, R. H. Ketelle, R. R.Lee, D. A. Leitzke, W. M. McMaster, J. L. Foreman, and S. Y.Lee. 1992. Status report on the geology of the Oak RidgeReservation. ORNL/TM-12074, Oak Ridge National Labora-tory, Oak Ridge, Tennessee.

Holling, C. S. 1993. Investing in research for sustainability.Ecological Applications 3:552–555.

Hopkins, L. D. 1977. Methods for generating land suitabilitymaps: A comparative evaluation. Journal of the AmericanInstitute of Planners 43:386–400.

Kiefer, R. W. 1965. Land evaluation for land use planning.Building Sciences 1:105–126.

King, A. L., D. J. Awl, and C. A. Gabrielson. 1994. Environmen-tally sensitive areas surveys program threatened and endan-gered species survey: Progress report. ES/ER/TM-130,Health Sciences Research Division, Oak Ridge NationalLaboratory, 19 pp.

Lubchenco, J. A., A. M. Olson, L. B. Brubaker, S. R. Carpenter,M. M. Holland, S. P. Hubbell, S. A. Levin, J. A. MacMahon,P. A. Matson, J. M. Melillo, H. A. Mooney, C. H. Peterson, H.R. Pulliam, L. A. Real, P. J. Regal, and P. G. Riser. 1991. Thesustainable biosphere initiative: an ecological researchagenda. Ecology 72(2):371–412.

Mann, L. K., and J. T. Kitchings. 1982. Resource data inventoryfor the Oak Ridge area. ORNL/TM-7941.

Mann, L. K., P. D. Parr, L. R. Pounds, and R. L. Graham. 1996.Protection of biota on nonpark public lands: Examples fromthe US Department of Energy Oak Ridge Reservation.Environmental Management 20(2):207–218.

Maurer, B. A. 1986. Predicting habitat quality for grasslandbirds using density–habitat correlations. Journal of WildlifeManagement 50:556–566.

Moran, E. F. 1990. The ecosystem approach to anthropology:From concept to practice. University of Michigan Press, AnnArbor.

Murdock, N. A. 1994. Rare and endangered plants andanimals of the southern Appalachian wetlands. Water, Airand Soil Pollution 77:385–405.

Noss, R. 1992. The wildlands project: Land conservationstrategy. Wild Earth Special Issue No. 1:10–25.

Pearson, S. M., M. G. Turner, R. H. Gardner, and R. V. O’Neill.1996. An organism based perspective of habitat fragmenta-tion. Pages 77–95 in R. C. Szaro and D. W. Johnston, (eds.),Biodiversity in managed landscapes: Theory and practice.Oxford University Press, New York.

Pounds, L. R., P. D. Parr, and M. G. Ryon. 1993. Resourcemanagement plan for the Oak Ridge Reservation, volume30: Oak Ridge National Environmental Research Parknatural areas and reference areas, Oak Ridge Reservationenvironmentally sensitive sites containing special plants,animals, and communities. ORNL/NERP-8, EnvironmentalSciences Division Publication No. 3995.

SAMAB (Southern Appalachian Man and the Biosphere).1996. The southern Appalachian assessment terrestrial tech-nical report. Report 5 of 5. US Department of Agriculture,Forest Service, Southern Region, Atlanta.

Schamberger, M. A., A. H. Farmer, and J. W. Terrell. 1982.Habitat suitability index models: Introduction. US Depart-ment of Interior, Fish and Wildlife Service FWS/OBS-82/10.

Simberloff, D., J. A. Farr, J. Cox, and D. W. Mehlman. 1992.Movement corridors: Conservation bargains or poor invest-ments? Conservation Biology 6(4):493–504.

Slocombe, D. S. 1993. Implementing ecosystem-based manage-ment. BioScience 43:612–622.

Swanson, F. J., and J. F. Franklin. 1992. New forestry principlesfrom ecosystem analysis of Pacific Northwest forests. Ecologi-cal Applications 2:262–274.

Turner, M. G. 1989. Landscape ecology: The effect of patternon process. Annual Review of Ecology and Systematics 20:171–197.

Turner, M. G., R. H. Gardner, V. H. Dale, and R. V. O’Neill.1989. Predicting the spread of disturbance across heteroge-neous landscapes. Oikos 55:121–129.

USA-CERL (United States Army Corps of Engineers Construc-tion Engineering Research Laboratories). 1993. GRASS 4.1User’s reference manual. Open Grass Foundation, Boston,563 pp.

USDA Forest Service. 1992. Ecosystem management of thenational forests and grasslands. Memorandum 1330-1. USDAForest Service, Washington, DC.

Westman, W. E. 1985. Ecology, impact assessment and environ-mental planning. John Wiley & Sons, New York, 532 pp.

Assessing Land-Use Impacts 211