Landscape Ecology 15: 5���20, 2000. �� 2000 Kluwer Academic Publishers. Printed in the Netherlands. 5 Gap analysis: concepts, methods, and recent results* Michael D. Jennings National Gap Analysis Program, U.S. Geological Survey, 530 S. Asbury St., Suite 1, Moscow, Idaho 83843, USA (e-mail: gap@uidaho.edu) Received 10 March 1999 Accepted 27 July 1999 Key words: biodiversity, conservation, large-area mapping, gap analysis Abstract Rapid progress is being made in the conceptual, technical, and organizational requirements for generating synoptic multi-scale views of the earth���s surface and its biological content. Using the spatially comprehensive data that are now available, researchers, land managers, and land-use planners can, for the first time, quantitatively place landscape units ��� from general categories such as ���Forests��� or ���Cold-Deciduous Shrubland Formation��� to more categories such as ���Picea glauca-Abies balsamea-Populus spp. Forest Alliance��� ��� in their large-area contexts. The National Gap Analysis Program (GAP) has developed the technical and organizational capabilities necessary for the regular production and analysis of such information. This paper provides a brief overview of concepts and methods as well as some recent results from the GAP projects. Clearly, new frameworks for biogeographic information and organizational cooperation are needed if we are to have any hope of documenting the full range of species occurrences and ecological processes in ways meaningful to their management. The GAP experience provides one model for achieving these new frameworks. Introduction As the abundance of humans continues to increase and the current species extinction event extensifies, biogeographic information that is both spatially com- prehensive and of appropriate resolution is becoming more vital for effective management of our biologi- cal resources. Although this tenet may seem obvious to some, its articulation and broad acceptance are recent, responding in part to the perceived conserva- tion imperative and in part to emerging principles and knowledge of landscape ecology. In the USA, some of the most rapid progress in the development of such information has been accomplished through the Na- tional Gap Analysis Program (GAP). Gap analysis is a method for identifying ���gaps��� in the network of conser- vation land and water areas. The conceptual, technical, and organizational bases needed for this work have been developing since the underlying principles of gap analysis were discussed in 1987 (Scott et al. 1987). As *The U.S. Governments��� right to retain a non-exclusive, royalty- free licence in and to any copyright is acknowledged. methods, data, and a massive cooperative experience emerged over the past decade, demand for and ap- plications of GAP information expanded beyond the original intent of ���a quick overview of the distribu- tion and conservation status of several components of biodiversity��� (Scott et al. 1993). While there is debate about the number of species being lost or at risk of extinction (e.g., Lugo 1988 Mann and Plummer 1995 Lawton and May 1995), the number of species at risk has increased since the USA Endangered Species Act was enacted in 1976 (Smith 1996), and it is likely that this trend will con- tinue (Pimm et al. 1995). So far, conservation efforts in the USA have not been effective in slowing the rate of species endangerment. One reason is that most con- servation programs are designed to conserve species already at the verge of extinction and do not address the ultimate problem of continued habitat loss for most species not adapted to human-configured envi- ronments. Until recently no concerted effort was being made to develop spatial information on the actual dis- tributions of ���ordinary��� species (not endangered or

6 threatened with extinction) and their habitats or on the effectiveness of contemporary conservation land and water areas for species conservation. Neither has there been a previous effort to determine, element-by- element, gaps in the current mix of conservation land and water areas as well as other conservation activities. Without spatially explicit data, it is unlikely that the forces causing habitat losses (e.g., invasive species, logging, grazing, mining, infrastructure development, recreation) can be managed effectively or that a net- work of conservation areas can be successfully de- signed. These data must include maps and other spatial information, at resolutions usable by land managers, of (a) species distributions (not to be confused with delineations of general range limits), (b) dominant vegetation cover types (or vegetation alliances FGDC 1997, Grossman et al. 1998), and (c) conservation ar- eas. The lack of this information for states and large regions has, until now, been due partly to a lack of the science and technology needed to create such maps. For example, remotely sensed imagery has not been available, an accepted classification system for veg- etation based on community ecology has only been recently developed, and technical capabilities for cre- ating, assessing the accuracy of, and analysing large spatial data sets has only been available for the past several years. During 1994, the first full set of Land- sat Thematic Mapper (TM) satellite imagery of the 48 contiguous states was assembled for state-by-state mapping of floristically defined vegetation types. Con- temporaneously, a suitable vegetation classification has been established (Grossman et al. 1998 FGDC 1997 Jennings 1993), and computing capabilities have been vastly improved. Of particular importance, the skill pool has been expanded greatly, primarily through support of graduate students associated with GAP projects. In addition to a lack of science and technology, there previously was no organizational or institu- tional catalyst for the development of the information needed. The task is a massive one and impractical for any single organization to achieve. Working as cooperators in GAP, professional biologists, ecolo- gists, computer scientists, geographers, and others have crossed disciplinary and institutional boundaries to address conservation needs (Jennings 1995). Significant concepts Burley (1988) first described a concept for identifica- tion of ���conservation gaps��� as a process to identify and classify the various elements of biodiversity and ex- amine the existing system of protected areas. Then the process was to determine which elements (e.g., vege- tation types, habitat types, species) are not represented or poorly represented in existing conservation areas. Finally, this information was to be used as a way to set priorities for the next steps of conservation actions, such as designing future reserves and planning land acquisitions. Gap analysis is based on Burley���s rather sim- ple concept, yet it requires sophisticated, novel ap- proaches in generating the large amounts of new data that are required. It is a coarse-filter (sensu Noss 1987) information strategy for protecting biodiversity (Scott et al. 1987, 1993) in that it focuses on both community-based units of habitat as well as on each individual species. This approach is intended to work in concert with ���fine-filter��� conservation, which fo- cuses on localized actions for those species in danger of extinction. The method assesses these distribu- tions relative to existing conservation areas and other categories of land and water management, at spatial resolutions useful for understanding and describing the ecological and conservation contexts of a given biodiversity ���element��� (vegetation alliance, habitat, or species) or suite of elements, or of any given land tract of interest. With this information, GAP seeks to identify elements of biodiversity not suffi- ciently represented in conservation areas. These are considered ���conservation gaps��� that may be closed through changes in land or water management prac- tices. (For example, in New Mexico, approximately 93% of Grace���s warbler (Dendroica graciae) habi- tat occurs on land where the species��� habitat needs are not a management consideration (Thompson et al. 1996). Incorporating its habitat needs with existing land-use planning and management may close this conservation gap.) The program also seeks to produce biogeographic information that may be used in regu- lar planning and management of land-based resources. The work is carried out by academic, nongovemment, and agency institutions on a state-by-state basis. GAP is the only USA program attempting to assess the conservation status of all components of the nation���s biodiversity.

7 Proactive rather than reactive management A fundamental assumption GAP makes is that the best time to decrease the probability of a species��� extinction due to human activities is well before its population is diminished to the point of endanger- ment. Waiting until a species is actually endangered or threatened with extinction results in reactive man- agement activities that are expensive, exhibit a low probability of success (Tear et al. 1993), and are often socially divisive. The GAP approach is predicated on the assump- tion that a dual focus on the conservation of habitats and multiple species will be both cheaper and more likely to succeed than conservation programs focused on any single species or population (Scott et al. 1993). At the same time, this approach is intended to provide a biogeography-based stratification for more detailed studies of, for example, composition, structure, and function of individual species, groups of species of interest, and vegetation alliances that are needed for site-level reserve design. The cost of maintaining species in their natural state when they are relatively common and part of self- sustaining ecosystems is less than the cost of intensive management programs needed to save species that are at the brink of extinction (Scott et al. 1987). An effi- cient way to avoid extinction crises is to work with the many different institutions ��� private and public ��� that are involved with land-use management and land-use planning to develop large-area geographic information for overall biodiversity. This information can then be applied to land-use and resource management deci- sions, whether incrementally small everyday decisions such as zoning permits or decisions of broader scope such as state land-use planning. Setting priorities for treating elements of biodiversity Because we cannot practically model all elements of biodiversity in the near tenn, we must set priorities for which elements to treat first (Margules and Austin 1991 Scott et al. 1993 Gap Analysis Program 1998 Csuti and Kiester 1996 Noss and Cooperrider 1994 Jennings et al. 1996). At the same time we must continue to improve assessment capabilities by devel- oping better information about each element and by continually increasing the number of taxa that we de- velop maps of. Initially, gap analysis methods focused on vegetation alliances along with all native species of amphibians, birds, mammals, and reptiles as sur- rogates for biodiversity. We began with this group of vertebrate species because they play a major role in community patterns and processes (Terborgh 1989), and because mapping their distributions at a practical and useful scale was tractable. Vegetation alliances are used because patterns of natural terrestrial land cover are an integrated reflection of the physical and chemical factors that shape the environment of a given land area (Whittaker 1973). They also are determi- nants for overall biological diversity (Franklin 1993 Noss 1990) as their structure and composition sig- nificantly affect species-level interactions. Vegetation alliances are the finest level of biotic assemblages that can be described and mapped over large areas using remotely sensed imagery (though technical limitations to mapping certain alliance types remain). They are constituent parts of landscapes and can be used as a set of equivalence classes in conservation evaluations (Fenner 1974 Austin 1991). In recent years methods have been developed to extend GAP to include ant species (Allen et al. 1998), crayfish, fish, mussels, and snail species (Sowa 1998), and research is under way to develop methods for predicting distributions of plant species (Fertig et al. 1998). These predictive models have broad application for planning, management, and research far beyond GAP conservation assessments, and we anticipate including additional taxa in the future. Clearly, focusing on a limited number of phyla will result in conclusions that are biased toward the mapped elements. Within that limitation, the approach will provide a synoptic spatial framework for linking information which is finer as well as coarser in both thematic description and spatial resolution. For ex- ample, maps of species distributions or habitat types produced for GAP can provide an ecological and ge- ographical context for stand or plot data measuring population or genetic criteria while directly linking these representations to continent-level measurement of biome criteria. Hotspots and reserve selection An early gap analysis hypothesis was that species and alliance maps would allow for identification of biodiversity ���hotspots��� (areas of maximal element co- occurrence, or richness), which might offer efficient conservation opportunities. Work done since that time has tested the concept and changed our understanding of its utility. For example, Prendergast et al. (1993) studied the potential overlap of biodiversity hotspots among birds, butterflies, dragonflies, liverworts, and