Introduction

The California Biodiversity Project is designed to examine the impacts of projected urbanization on natural habitats using models of biodiversity and spatially explicit urban growth. This combination of models is intended to predict which species and habitats will be most stressed over the short term (10 - 15 years) and the medium term (15 - 40 years). The project operates at the county scale, with portions of the analysis functioning over larger, physiographically defined planning regions (e.g. Davis and Stoms, 1996). Other biodiversity research is being conducted at the state level, (e.g. Gap Analysis) and much work has been conducted at the local level. The emphasis of my research is on scale independence, and spatial portability of the biodiversity and urban growth models. Compatibility with the Gap Analysis products is also an integral component of this research. With these goals in mind, Santa Cruz County in central California is serving as a first pilot area (Figure 1). This area is adjacent to Monterey Bay and the Pacific Ocean, allowing a variety of ecological factors to be modeled including coastal influences, agriculture, steep terrain, rivers, forests, and open grasslands. As the Santa Cruz pilot project nears completion, additional areas are being considered for study to refine the models under different ecosystems and urban trends.

Research Scope and Design

As researchers study species and habitat use, a frequent goal of the investigation is to assess present or future ecosystem health. Biodiversity is one metric for health, functioning at particular spatial scales and thematic levels. Biodiversity is also integral to conservation biology, pertaining to small and declining populations and a variety of factors including habitat change as well as genetic and demographic alterations (Soulé and Mills 1992, Soulé 1991, Caughley 1994). With this perspective in mind, species level biodiversity over hectare to kilometer sized areas is the primary focus of my study, though my methods are not restricted to a specific sampling domain. In this research, I use a series of indicators to form a composite valuation of each study area, building upon methods developed over the last two years (Cogan 1996). Compared to the single species approach, the composite method I use has the disadvantage of lowering end product precision, however I also receive the beneficial trade-off of increasing relevance to biodiversity.

Indicators of Biodiversity

My ecological model consists of five weighted components which combine to assess land value:

These five ecological components are quantified as individual sub-models. By attempting to maintain each sub-model as in independent analysis, I promote flexibility in the model, enable open dialog between model users and stakeholders, and clarify my final results. After completion of each sub-model, my preliminary results are additively combined to comprehensively assess each land area, prescribing a measure of biodiversity value. Each of the five sub-models can be weighted to adjust its overall contribution to the analysis. For example, some model users may wish to de-emphasize restoration potential, while other users may prefer to make it a major factor in the analysis. Table 1 is a summary of each sub-model.

1) Ecoregional analysis is the first indicator I use, and is designed to evaluate course scale processes which function across large areas and decadal time periods. Using ecoregions (Figure 2) as defined by Jepson (1993), I combine information on land management, land cover, and area into a single quantitative assessment. This sub-model is based on three explicit assumptions:

Based on these three assumptions, I have assessed the habitat types found in Santa Cruz County. The graphical result of this ecoregional analysis is a map of the county with low, medium, and high importance areas identified (Figure 3). The results of the ecoregional analysis are not intended to be absolute, as the model is designed to incorporate several user adjustable parameters, such as weightings for land ownership which will alter the model output.

Example of Ecoregional Importance:
Redwood forest in Santa Cruz County. Most of the redwoods in the central west ecoregion are contained within Santa Cruz County. This gives this habitat type a greater importance in the county than would otherwise be considered if we used county data in isolation.

2) County special features are assessed in the second sub-model. At this step, areas of current special concern are mapped and assigned preliminary weights based on conservation importance. As with the ecoregional model, these weights are user adjustable. Well known special features in Santa Cruz County are unique soils areas, and old-growth redwood forest (Figure 4). These data were obtained from interviews with local experts, who often have remarkable insights and information on county special features. This portion of the analysis is very fine scale, and operates with information about known critical areas at the present time. While this type of data is not predictive, having information on present problem areas is an extremely valuable component of biodiversity analysis.

Example of County Special Features:
The Sandhills soil type in Santa Cruz County. Sandhills soils support several local species of plants which are currently threatened. This soil type is an indicator of important biodiversity areas which are currently (not just predicted to be) important for biodiversity.

3) Species habitat models are used in the third phase of the biologic assessment. Relying on the California Gap Analysis data models, each vegetation polygon is associated with a list of vertebrate species (Hollander et al. 1994, Scott et al. 1993). In the standard approach to Gap Analysis, each polygon can be summarized by species richness and compared to existing and potential biodiversity management areas (Scott et al. 1993). Because the land cover data is designed to be used at fairly course scales this approach is somewhat problematic at finer county scales. I am presently working on a method to use fine-scale county data with state level Gap vertebrate models. There are several difficulties with this, including minimum range areas for each species, inherently high error rates, and general species-area relationship issues. If successful, this method will be of interest to those working with Gap models at fine scales.

Example of Species/Habitat Relations:
Species often load on habitat in ways that are predictable from moisture gradients. Wetland habitats can be shown to support large numbers of species and thus carry a higher importance for biodiversity.

4) Landscape factors are another component of biodiversity analysis. As polygons of habitat change size, shape, and adjacency, habitat quality will also change. There are opportunities at several stages of the biodiversity analysis for landscape factors to be assessed.

Example of Landscape factors:
Distance to conflicting landuse. A habitat patch which is adjacent to an urban area may be of less biological value that one that is not.

5) Restoration opportunities are the last component of the five-part biodiversity model. I define areas of potential restoration as areas capable of increased biodiversity value with feasible levels of management change. In many cases, urban areas have low biodiversity value, and it is often not practical to improve these areas. Farmlands may also have low value, however landuse practices can often be modified to improve habitat, without financial impact to the farmer.

Example of Restoration opportunities:
Agriculture areas are capable of minor modification which can increase biodiversity value. Examples are larger hedge-rows between fields.

Urban Growth Models and Biodiversity Analysis

The five ecologic sub-models present a synopsis of biodiversity values for county lands. These areas are also modeled for future urbanization potential using growth models from Landis (1996) and Clarke (1996). These models differ in their assumptions on the nature of urban growth and will accordingly differ in their predictions over various time periods. The overall effect of these differences on biodiversity analysis is unknown, and is part of my work in progress.

The Clarke urban growth model, as modified for this analysis, requires several types of data to be resampled and formatted as raster images for input. These include data on roads, digital elevation models, and current land cover areas designated urban according to the Anderson level one land cover classification system (Anderson et al. 1976). The resulting product is a series of pixels representing probability of potential urbanization (Figure 5). By performing a query of each landuse polygon impacted by projected urbanization, I obtain a list of species impacted. Minimum size habitat requirements and decision rules for degree of impact are also calculated at this stage. Comparing the ratio of impacted vs. existing habitat in the county results in an impact assessment for each vertebrate species under each urban growth scenario.

Potential Applications of the Research

In summary, the combination of multiscale biodiversity assessment and spatially explicit urban growth models have the potential to help guide land use planning, and provide an early warning of possible development conflicts with species and habitats. With such direct relevance and importance, this area of research is likely to expand in the near future. While writing this paper, I have learned of a similar study in Pennsylvania (White et al. 1997) and undoubtedly other efforts will soon be underway. The time frame for my own analysis is 40 years in the future, with emphasis on scale independence, spatial portability of the biodiversity and urban growth models, and compatibility with nationally produced data such as the Gap Analysis products.

This research has several possible applications:

· Identify when and where projected urbanization will adversely impact the natural environment.

· Predict specific types of conflict between projected urbanization and the natural environment, i.e. agriculture vs. wetlands.

· Permit landuse planners to consider cumulative conflicts over several areas, identifying species and habitats which would otherwise be overlooked. This has advantages over current parcel based planning.

· Simulate development scenarios to assess biodiversity impact.

The application of Gap Analysis data and urban growth models to local scale biodiversity assessment is not a trivial process, and many opportunities for error exist. Multiple spatial and thematic ranges, scale dependant model artifacts, and uncertainties with future predictions are just a few of the challenges in this research. My efforts include the documentation of these potential errors, as I demonstrate the techniques possible for long range proactive biodiversity management.

Acknowledgments

This project is a collaborative effort by many individuals and organizations. Special thanks are given to Santa Cruz County Planning, and the many experts from the California Department of Fish and Game. I am also grateful for the assistance of Frank Davis, David Stoms, and Allen Hollander of the California Gap Analysis Project, Blair Csuti, Len Gaydos, Keith Clarke, Jim Quinn, J. Michael Scott, and Michael Jennings.

References

Anderson, J.R., E. Hardy, J. Roach, and R. Witmer. 1976. A land-use and land cover classification system for use with remote sensor data. U.S. Geological Survey professional paper 964.

Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology, 63:215-244.

Clarke, K. C., S. Hoppen, and L. Gaydos, 1996. Methods and Techniques for Rigorous Calibration of a Cellular Automation Model of Urban Growth In Proceedings of the Third International Conference/Workshop Integrating GIS and Environmental Modeling. Santa Fe, New Mexico January 21-25, 1996.

Cogan, C.B. 1996. The California biodiversity project: Selection and use of biodiversity indicators. In 1996 ESRI User Conference Proceedings , Redlands CA.

Davis, F. W., and D. M. Stoms, 1996. A spatial analytical hierarchy for Gap Analysis, in J. M. Scott, T. H. Tear, and F. W. Davis, editors. Gap Analysis: A Landscape Approach to Biodiversity Planning. American Society for Photogrammetry and Remote Sensing, Bethesda, MD, pp. 15-24.

Hollander, A. D., F. W. Davis, and D. M. Stoms, 1994. Hierarchical representation of species distributions using maps, images, and sighting data, in Mapping the Diversity of Nature, R. I. Miller, editor, Chapman & Hall, pp. 71-88.

Landis, J.D. 1995. Imagining land use futures - applying the California urban futures model. Journal of the American Planning Association, 61:438-437.

Soulé, M.E. 1991. Conservation: Tactics for A Constant Crisis. Science 253: 744-750.

Soulé, M.E. and L.S. Mills. 1992. Conservation genetics and conservation biology: a troubled marriage. In Conservation of Biodiversity for Sustainable Development, Sandlund, Hindar, and Brown, Eds. Scandinavian University Press, Oslo, pp. 55-69.

Scott, J. M., F. Davis, B. Csuti, R. Noss, B. Butterfield, C. Groves, H. Anderson, S. Caicco, F. D'Erchia, T. C. Edwards, Jr., J. Ulliman, and R. G. Wright, 1993. Gap analysis: A geographic approach to protection of biological diversity. Wildlife Monographs, 123: 1-41.

White, D., et al. In press. Assessing risks to biodiversity from future landscape change. Conservation Biology 11:1-13.