Volume No. 10, 2001
Applications
Land use planning for private land is fundamentally important for conserving biodiversity nationwide (Dale et al. 2000). A major opportunity to refine the Gap Analysis methodology is to integrate socioeconomic factors to better assess both levels of protection and risk, particularly on private lands (McKendry and Machlis 1993). Incorporating information about private lands into the GAP methodology is important because private lands contain disproportionately high levels of biodiversity and habitat for rare species (Bean and Wilcove 1997); many of the important causes of habitat loss and habitat fragmentation stem from changes of land use on private lands; and they vary greatly in the degree of human-induced impacts on habitat.
GAP methodology identifies land cover types and species distributions that may be particularly vulnerable given their status in the current array of land ownership and management. However, a main drawback is that the coarse categories (4) of biodiversity management status, based on potential land use activities, may be weakly associated with actual species vulnerability (Stoms 2000). Some types of human activities cover broad expanses of the landscape and result in substantial land cover conversion, such as mono-crop agriculture and urban uses, and these activities typically are well-represented on land cover maps. However, land cover maps miss vast areas under the influence of either broad-extent, low-intensity land uses (e.g., low-density rural residential development) or small-extent, high-intensity activities such as oil and gas wells. Compiling data that more directly relate impacts on biodiversity associated with land uses is challenging (Stoms 2000), but offers a straightforward and reasonable means to identify threats to biodiversity, although actually demonstrating species responses to land use activities is quite challenging in practice (Theobald et al. 1997).
Another opportunity to refine status categories is to move beyond vulnerability and differentiate areas on the landscape (and species habitat) that are currently threatened or likely to be threatened in the future by land use activities associated with human development (e.g., urbanization, intensive agricultural practices, logging, etc.). Without considering these threats to species and habitat, conservation resources overall may not be properly prioritized (Cassidy et al. 2001) to achieve the greatest benefit for the most species (Scott et al. 1993). McKendry and Machlis (1993) described a general framework to extend biodiversity gap analysis by including socioeconomic indicators such as population change, economic trends, government policies, and land use conversion. Although current GAP methodology recognizes this limitation-for example, "We emphasize, however, that GAP only identifies private land as a single homogeneous category and does not differentiate individual private land units or owners…" (Csuti and Crist 2000)-few methods to address these limitations exist.
Recently, Stoms (2000) compared three indicators of development-permitted land use, "roadedness," and human population growth-to stewardship status for two pilot areas in California and found large differences between the more direct indicators and the general proxy of status or protection level. Theobald et al. (1998) developed a preliminary assessment methodology to examine the impacts of private land development on habitat using GAP land cover data, but did not quantify differences between management protection level and other indicators of land use.
Here we present an approach to refine the identification of vulnerable areas to consider what lands are threatened by various human land uses, especially those that have significant impacts and are increasing rapidly, such as urbanization and rural residential development. We utilized data readily available nationwide to develop a methodology to incorporate information about land use on private lands when assessing protection levels on private (and adjacent public) lands, and to forecast future levels of development to identify areas that are most at risk from potential private land development. We illustrate this approach using a case study from Colorado.
Colorado, often referred to as the "bellwether" of the Rocky Mountain West, has seen significant threat to habitat due to development pressures. Indeed, not only is the West's population growing three times as fast as the rest of the US (US Census Bureau 2001; Baron et al. 2000), but demographic and economic trends are changing the pattern and location of development (Riebsame et al. 1997). As a result, more than 60% of the West's counties are experiencing "rural sprawl," where rural areas (outside of city and town limits) are growing at a faster rate than urban areas (US Census Bureau 2001). In Colorado, population growth rates in nearly one-fifth of the counties exceeded 5% from 1990 to 1997, and this growth has caused large expanses of low-density development (Theobald 2000).
We developed two easily mapped measures of development and then used these indicators to assess which land cover types were particularly at risk and to identify where habitat is threatened by development. Our case-study assessment utilized both the land stewardship map and the species distribution maps produced by the Colorado Gap Analysis Project (Schrupp et al. 2001).
We selected two socioeconomic indicators to develop maps for and to test in relation to biodiversity: roads and housing density. The effects of roads on biodiversity and ecological integrity has been well documented (Forman and Alexander 1998). Road and housing density are often thought to be highly correlated, but because mixed results were obtained for a preliminary analysis (Theobald 1997), we chose to model both indicators to further test whether these were highly correlated for statewide areas. Although population density is often used to map human activity patterns, population data is tied to the primary place of residence and so underestimates potential effects on habitat in areas with a high percentage of second and vacation homes (Theobald 2000; Theobald in press).
Moreover, potential impacts to habitat such as removal of native vegetation, alteration of vegetation structure for defensible space for wildfire protection, and introduction of exotic species are more closely related to housing density.
Although road density is typically used as a measure of road effects on biodiversity, we created a "roadedness" map (Figure 1) following the methodology developed in California (Davis et al. 1996; Stoms 2000). Roadedness does not suffer from bias introduced when calculating road density in areas where many roads close together result in very high road densities and better accounts for spatial pattern.
Moreover, an important assumption in creating a map that depicts effects of roads on biodiversity is that larger roads (e.g., highways) typically affect species further from the road than smaller (e.g., local) roads, because larger roads are typically wider and carry more traffic. Therefore, the "roadedness" index estimates the proportion of an area (e.g., watershed, county, status category) that is affected by roads. Roads from US Census Bureau TIGER files were converted to 30 m GRIDs and then were assigned a buffer width according to the schedule in Table 1.
Figure 1. Roaded areas in Colorado.
To map historical and current housing density, we used 1990 US Census Bureau block-groups and blocks, which are subdivisions of the familiar census tract. To account for underestimation of units in previous decades, decennial estimates for 1940-1980 were corrected using a correction factor computed as the ratio of number of units in a county from historical census divided by total housing units summed from current estimates (Theobald 2001b). To map likely future housing density, we developed a model that recognizes and represents land use changes beyond the urban fringe (Figure 2). Although a number of approaches have been developed to forecast future growth patterns, most efforts have focused on urban growth and changes to urban or built-up cover types and are based on land cover types classified from satellite imagery and occasionally from high-altitude aerial photography (e.g., Brown et al. 2000). Recently, Clarke and Gaydos (1998) developed a California-based model to predict urban growth in San Francisco and Baltimore. Stoms (2000) distributed population growth using a rule-based approach that arbitrarily limited growth to 8 km expansion from urban cores.
Figure 2. Housing density in 1990 and 2020.
Rather than rely on urban-centric models of housing growth, we used county-based population projections to derive the number of housing units needed in 2025 and 2050 (Theobald 2001a). We then spread these units throughout the block-groups by assuming that a block-group's density could not exceed the average housing density of its neighbors, for each decadal time step (Theobald et al. 2001).
We then analyzed the threats to habitat by overlaying the roadedness and housing density layers with land cover data.
Over 269,000 kilometers (~167,000 miles) of roads were mapped in Colorado, resulting in 21.7% of Colorado being "roaded." Roaded proportion varies widely by watershed, from a low of 6.1% to a high of 40.9% (mean of 20.7%) (see Figure 3).
Figure 3. Percent roaded by watershed.
Contrary to common belief, there was a poor relationship (R2 = 0.21) between percent roaded and the proportion of public land in each county. Although 10% of Colorado was "protected" (Status 1 and 2), about 13.5% of these protected areas were roaded. Conversely, the majority of Colorado was "unprotected" (Status 4), yet only about one-quarter of this area was roaded. About 5.1% of Colorado was developed in 1990 at densities higher than rural (i.e. urban, suburban, and exurban areas), and an additional 5% of Colorado will be "at risk" from new development forecasted for 2020, located mostly along the foothills of the Front Range and mountain valleys.
In Colorado, 24 of 43 natural land cover types were found to be vulnerable, which we define here as less than 10% protected in Status 1 and 2 (see Table 2). We designated a land cover class as threatened if 20% or more was roaded, or if 15% or more coincided with exurban or greater density development in 1990, was within 2 km of exurban or greater development in 1990, or coincided with areas at risk of development by 2020. Most vulnerable land cover types were also threatened by roads, although ponderosa pine, bristlecone pine, shrub-dominated wetland, and prostrate shrub/tundra were identified as threatened but were not identified as vulnerable. Tallgrass prairie, foothills/mountain grasslands, and bristlecone pine were identified as threatened by future development in 2020. Moreover, a number of land cover types proximal to development were found to be threatened, but were not identified as vulnerable, most notably water, spruce/fir, Douglas fir, ponderosa pine, bristlecone pine, forest-dominated wetland, and most tundra cover types.
Incorporating socioeconomic factors, such as road and housing density, provides an important opportunity to extend the methodology of gap analysis. We found that both road and housing density were useful indicators of potential impacts from activities associated with human land use and could be used to refine analyses of vulnerability to include level of threat (Figure 4). The data to produce these layers were readily available, and methods to convert them into reasonable indicators were straightforward. (Note: The derived maps of housing density are available at http://www.ndis.nrel.colostate.edu/davet/dev_patterns.htm).
In addition to roads and residential land use, there are a number of additional land uses associated with humans that would be useful but are more challenging to incorporate. For example, additional data and methodologies are needed to better incorporate knowledge about the possible effects of grazing, logging, oil and gas wells, and fire suppression in spatially-explicit models of effects.
Figure 4. Patches of land cover ranked by percent "at risk" from development to 2020.
Baron, J.S., D.M. Theobald, and D.B. Fagre. 2000. Management of land use conflicts in the United States Rocky Mountains. Mountain Research and Development 20(1):24-27.
Bean, M.J., and D.S. Wilcove. 1997. The private-land problem. Conservation Biology 11:1-2.
Brown, D.G., B.C. Pijanowski, and J.D. Duh. 2000. Modeling the relationships between land use and land cover on private lands in the Upper Midwest, USA. Journal of Environmental Management 59:247-263.
Cassidy, K.M., C.E. Grue, M.R. Smith, R.E. Johnson, K.M. Dvornich, K.R. McAllister, P.W. Mattocks, Jr., J.E. Cassady, and K.B. Aubry. 2001. Using current protection status to assess conservation priorities. Biological Conservation 97:1-20.
Clarke, K.C., and L.J. Gaydos. 1998. Loose-coupling a cellular automaton model and GIS: Long-term urban growth prediction for San Francisco and Washington/Baltimore. International Journal of Geographical Information Science 12:699-714.
Csuti, B., and P. Crist. 2000.
Mapping and categorizing land stewardship (v. 2.1.0). A handbook for conducting Gap Analysis. Internet WWW page at http://www.gap.uidaho.edu/handbook/Stewardship/default.htm
Dale, V.H., S. Brown, R.A. Haeuber, N.T. Hobbs, N. Huntly, R.J. Naiman, W.E. Riebsame, M.G. Turner, and T.J. Valone. 2000. Ecological principles and guidelines for managing the use of land. Ecological Applications 10:639-670.
Davis, F.W., D.M. Stoms, R.L. Church, W.J. Okin, and K.N. Jonson.
1996. Selecting biodiversity management areas. In Sierra Nevada Ecosystem Project: Final Report to Congress, Vol. II, Assessments and scientific basis for management options.
Forman, R.T.T., and L.E. Alexander. 1998. Roads and their major ecological effects. Annual Review of Ecology and Systematics 29:207-231.
McKendry, J.E., and G.E. Machlis. 1993. The role of geography in extending biodiversity gap analysis. Applied Geography 11:135-152.
Riebsame, W.E., H. Gosnell, and D.M. Theobald. 1997. The Atlas of the New West. Norton Press.
Schrupp, D.L., W.A. Reiners, T.G. Thompson, L.E. O'Brien, J.A. Kindler, M.B. Wunder, J.F. Lowsky, J.C. Buoy, L. Satcowitz, A.L. Cade, J.D. Stark, K.L. Driese, T.W. Owens, S.J. Russo, and F. D'Erchia. 2001. Colorado Gap Analysis Program: A geographic approach to planning for biological diversity. Final report. USGS/BRD Gap Analysis Program and Colorado Division of Wildlife, Denver, Colorado.
Scott, J.M., F. Davis, B. Csuti, R. Noss, B. Butterfield, C. Groves, H. Anderson, S. Caicco, F. Derchia, T.C. Edwards, J. Ulliman, and R.G. Wright.
1993. Gap Analysis: A geographic approach to protection of biological diversity. Wildlife Monographs 123:1-41.
Stoms, D.M. 2000. GAP management status and regional indicators of threats to biodiversity. Landscape Ecology 15:21-33.
Theobald, D.M. 1997. Incorporating human disturbance in models of wildlife habitat suitability. Unpublished report. Natural Resource Ecology Lab, Colorado State University, Fort Collins.
Theobald, D.M. 2000. Fragmentation by inholdings and exurban development. Pages 155-174 in R.L. Knight, F.W. Smith, S.W. Buskirk, W.H. Romme, and W.L. Baker, editors. Forest fragmentation in the central Rocky Mountains. University Press of Colorado, Boulder, Colorado.
Theobald, D.M. 2001a.
Technical description of mapping historical, current, and future housing densities in the US using Census block-groups. Natural Resource Ecology Lab, Colorado State University. 31 May. http://www.ndis.nrel.colostate.edu/davet/dev_patterns.htm
Theobald, D.M. 2001b.
Land use dynamics beyond the urban fringe. Geographical Review 91(3):544-564.
Theobald, D.M., J.M. Miller and N.T. Hobbs. 1997. Estimating the cumulative effects of development on wildlife habitat. Landscape and Urban Planning 39(1):25-36.
Theobald, D.M., N.T. Hobbs, D. Schrupp, and L. O'Brien. 1998. An assessment of imperiled habitat in Colorado (poster). Annual Meeting of International Association for Landscape Ecology.
March 17, 1998, East Lansing, Michigan. http://www.nrel.colostate.edu:8080/~davet/co_assess/assessment.htm
Theobald, D.M., D. Schrupp, and L. O'Brien. 2001. Assessing risk of habitat loss due to private land development in Colorado. Final report for Cooperative Agreement No. 00HQAG0010, USGS/BRD Gap Analysis Program. 62 pp. http://www.ndis.nrel.colostate.edu/davet
U.S. Census Bureau. 2001.
Census 2000 SF1.