GAP Home USGS Home

| GAP home | USGS home |

Volume No. 11, 2002

Animal Modeling

Prioritizing Conservation of Biodiversity Using a Multispecies Approach

Karen V. Root

Department of Biological Sciences, Bowling Green State University, Bowling Green, Ohio

Introduction

The conservation of ecosystems focuses on evaluating individual sites or landscapes based on their component species.  Building on successful single species evaluation methods (e.g., habitat suitability analysis and population viability analysis), I developed a method for estimating the value of a particular site based on its ecological components, i.e., species, and the threats facing it.  The method has two important features: it assigns an ecological value based on many species and facilitates multispecies assessments of ecological effects.  The goal is to minimize the extinction risk and maximize habitat quality for the component species; the viability of the populations, rather than just the species’ presence, is considered.  As a demonstration I applied my method to a set of California species that are included in the California Gap Analysis Project (Davis et al. 1998).

Methods

I combined habitat suitability maps for each species with the extinction risk faced by each species in a single map of multispecies conservation values (MCVs).  Using the risk of extinction as a weighting factor means the more imperiled a species is, the more priority is given to its habitat requirements.  A high value for the MCV (e.g., 9 or 10 on a scale of 1 to 10) represents the highest-quality habitat for the set of species with the highest risk of extinction or decline.

Many types of models can be used to estimate the risk of extinction and the contribution of each cell; the model choice depends on the species and the data available.  In this example I constructed spatially explicit population models, using RAMAS GIS (Applied Biomathematics, Setauket, New York; Akçakaya 1998), for a set of species in the 10 southern counties of California (Root et al. 2003).  This method, though, is not limited in the number of species or the size of the area that can be included; I chose only six species and a reduced area to simplify the computations.

The six species used for this analysis were California Gnatcatcher, Cactus Wren, California Spotted Owl, desert tortoise, Stephens’ kangaroo rat, and San Joaquin kit fox.  For each species, I created a raster map of habitat suitability based on the California Gap Analysis database with values of 0 through 5, with 5 being the most suitable (Davis et al. 1998).  These maps were imported into RAMAS GIS (Akçakaya 1998) and served as the basis of the spatial structure of the metapopulation or population.

Based on the available data, I constructed a female-only, stage-based, stochastic, spatially explicit model for each species (Root et al. 2003) using published data and models wherever possible (see Table 1 for references).  I assumed populations were limited by both the quality and the quantity of habitat, and dispersal and correlation among populations was distance-dependent.  The contribution of each cell to the overall risk of extinction was estimated as the difference between the risk of extinction with all populations included minus the risk with the population (that the cell belonged to) removed.  Therefore, the MCV for each cell j was the sum of the products of habitat suitability for species i at location j (Sij), risk of extinction in 50 years (Pi) and contribution of location j to the viability of species i (Cji) divided by the sum of all of the extinction risks:

            MCVj,  (Root et al. 2003).

For comparison, I also examined an alternative measure of risk, the risk of a 50% decline in abundance in 50 years.

Table 1.  Six species that were selected, their vulnerability status as assigned by the U.S. Fish and Wildlife Service (federal) or California Fish and Game Commission (state), and the sources for species-specific demographic data and models.

Common Name

Scientific Name

Federal
Status

State
Status

Sources

Coastal California Gnatcatcher

Polioptila californica californica

Threatened

None

Akçakaya 1997; Akçakaya & Atwood 1996, 1997; Bontrager 1991

Cactus Wren

Campylorhynchus brunneicapillus

None

None

Akçakaya and Atwood 1996

Northern Spotted Owl

Strix occidentalis caurina

Threatened

None

Call et al. 1992; Lahaye et al. 1994

Desert Tortoise

Gopherus agassizii

Threatened

Threatened

Doak et al. 1994; Luke et al. 1991; O'Connor et al. 1994; Root 1999; Turner et al. 1986

Stephens' Kangaroo Rat

Dipodomys stephensi

Endangered

Endangered

Price and Kelly 1994; Price et al. 1994

San Joaquin Kit Fox

Vulpes macrotis mutica

Endangered

Threatened

Disney and Spiegel 1992; White and Garrott 1997

Results

Based on these models, the MCV map (Figure 1a) revealed valuable habitat patches scattered across the 10 southern counties of California.  Nine percent of the polygons (5.3% of the total area) had a value in the top category of the MCV values, and 24.1% (12.4% of the total area) were in the top two categories of the MCV values.  Approximately 29% of the polygons (38.9% of the area) had a negative MCV value, which occurred when a particular location, if included, increased the overall risk of extinction (e.g., a sink population).

When the alternative risk measure, i.e., risk of a 50% decline, was used in estimating the MCV, the resulting map showed a few changes (Figure 1b).  In this case there were fewer negative MCV values.  Only 7.7% of the polygons (9% of the total area) had a negative MCV value.  The top two categories of the MCV values included 19% of the polygons (7.9% of the total area), and the top category included 18% of the polygons (7.7% of the total area).

 

Figure 1.  Multispecies conservation values maps of the southern 10 counties of California for 6 species based on their habitat suitabilities from the California Gap Analysis (Davis et al. 1998) weighted by (a) the probability of extinction for each species, or (b) the probability of a 50% decline in abundance for each species, estimated from population models.  The categories shown represent ten intervals of equal area; a larger value indicates a higher conservation value.

Discussion

The resulting MCV maps highlight regions of conservation importance for the six species that were included.  Sites with higher MCVs (e.g., 9 or 10 on a scale of 1 to 10) had, in general, higher habitat suitability for species with a higher risk of extinction or decline.  The regions where there was the greatest overlap among the six species were also where many of the highest MCV values were found.  The most valuable locations, in general, were along the eastern side of the state, which closely reflects the higher risk of extinction for the species found in these areas.  The valuable sites along the eastern side correspond with the highly endangered coastal sage scrub habitat included in reserves designs of Natural Community Conservation Planning Program (Akçakaya and Atwood 1997; Davis et al. 1998).

It is interesting that the map based on the risk of a 50% decline rather than risk of extinction shows a slightly different pattern.  Areas on the western side of the state had a higher MCV value under the risk of decline compared with the MCV value under the risk of extinction.  Species may have a very high risk of a large decline but a negligible risk of complete extinction and would be valued higher in this weighting.  The risk of a decline may provide an important early warning for species that are not currently considered threatened or endangered but may be quite vulnerable to changes in their environment.

Using a model to explicitly measure the risk of extinction or decline is generally preferable to using an index when data are available (Root et al. 2003).  The amount of data needed is driven by the choice of model for estimating the risk of extinction and the contribution of each location; a simple unstructured population model will require far less data than an individual-based simulation model.  There are also methods for estimating the risk of extinction of a species using only presence-absence or siting data (see Solow 1993a, b; Burgman et al. 1995, 2000).  An important advantage of models is that they can highlight which parameters have the most influence on the risk of extinction, warranting further study, and guide future research efforts.  For many of the species included in this example, the adult survival value had the greatest influence on the population growth rate and subsequent risk of extinction or decline.

The risk-based multispecies conservation value is also flexible in terms of scale.  Both the habitat suitability analysis and metapopulation models can be developed at a scale appropriate for the individual species.  Many more species can be readily accommodated with this method than the six I used in this test case.  Dynamic elements can be explored, such as the effects of fires, timber harvest, drought, and other factors.  One can incorporate a potential effect in the metapopulation model, estimate the risk, and compare the resulting MCV map to the map without an effect.  Potential changes also can be incorporated into the habitat suitability maps that reflect planning choices so that the outcomes of different plans can be compared.  This method (now implemented in software; Root 2002) provides a quantitative and spatially explicit conservation value useful for such applications as a multispecies recovery plan, a regional habitat conservation plan, or an evaluation of local management alternatives.

Literature Cited

Akçakaya, H.R.  1998.  RAMAS GIS: Linking landscape data with population viability analysis.  Version 3.0.  Applied Biomathematics, Setauket, New York.

Akçakaya, H.R., and J.L. Atwood.  1996.  A geographic extinction risk model for the management of multiple species reserves.  Technical report.  Southern California Edison, Rosemead, California.

Akçakaya, H.R., and J.L. Atwood.  1997.  A habitat-based metapopulation model of the California Gnatcatcher.  Conservation Biology 11:422-434.

Bontrager, D.R.  1991.  Habitat requirements, home range and breeding biology of the California Gnatcatcher in south Orange Country, California.  Santa Margarita Company, California.

Burgman, M.A., R.C. Grimson, and S. Ferson.  1995.  Inferring threat from scientific collections.  Conservation Biology 9:923-928.

Burgman, M., B.R. Maslin, D. Andrewartha, M.R. Keatley, C. Boek, and M. McCarthy.  2000.  Inferring threat from scientific collections: Power tests and an application to Western Australia Acacia species.  Pages 7-26 in S. Ferson and M. Burgman, editors.  Quantitative methods for conservation biology.  Springer Verlag, New York.

Call, D.R., R.J. Gutiérrez, and J. Verner.  1992.  Foraging habitat and home-range characteristics of California Spotted Owls in the Sierra Nevada.  The Condor 94:880-888.

Davis, F.W., D.M. Stoms, A.D. Hollander, K.A. Thomas, P.A. Stine, D. Odion, M.I. Borchert, J.H. Thorne, M.V. Gray, R.E. Walker, K. Warner, and J. Graae.  1998.  The California Gap Analysis Project.  Final report.  University of California, Santa Barbara.  Available from http://www.biogeog.ucsb.edu/projects/gap/gap_rep.html.

Disney, M., and L.K. Spiegel.  1992.  Sources and rates of San Joaquin kit fox mortality in Western Kern County, California.  Transactions of the Western Section of the Wildlife Society 28:73-82.

Doak, D., P. Kareiva, and B. Klepetka.  1994.  Modeling population viability for the desert tortoise in the western Mojave Desert.  Ecological Applications 4:446-460.

Lahaye, W.S., R.J. Gutiérrez, and H.R. Akçakaya.  1994.  Spotted owl metapopulation dynamics in Southern California.  Journal of Animal Ecology 63:775-785.

Luke, C., A. Karl, and P. Garcia.  1991.  Review of the emergency listing of the desert tortoise (Gopherus agassizii).  Report.  City of Ridgecrest, California.

O'Connor, M.P., L.C. Zimmerman, D.E. Ruby, S.J. Bulova, and J.R. Spotila.  1994.  Home range size and movements by desert tortoise, Gopherus agassizii, in the Eastern Mojave Desert.  Herpetological Monographs 8:60-71.

Price, M.V., and P.A. Kelly.  1994.  An age-structured demographic model for the endangered Stephens’ kangaroo rat.  Conservation Biology 8:810-821.

Price, M.V., P.A. Kelly, and R.L. Goldingay.  1994.  Distances moved by Stephens’ kangaroo rat (Dipodomys stephensi merriam) and implications for conservation.  Journal of Mammology 75:929-939.

Root, K.V.  1999.  RAMAS ecological risk model for desert tortoise.  Technical Report.  Southern California Edison, Rosemead, California.

Root, K.V.  2002.  RAMAS Multispecies Assessment: Estimating multispecies conservation values across the landscape.  Applied Biomathematics, Setauket, New York.

Root, K.V., H.R. Akçakaya, and L. Ginzburg.  2003.  A multispecies approach to ecological valuation and conservation.  Conservation Biology 17(1):196-206.

Solow, A.R.  1993a.  Inferring extinction from sighting data.  Ecology 74:962-964.

Solow, A.R.  1993b.  Inferring extinction in a declining population.  Journal of Mathematical Biology 32:79-82.

Turner, F.B., P. Hayden, B.L. Burge, and J.B. Roberson.  1986.  Egg production by the desert tortoise (Gopherus agassizii) in California.  Herpetologica 42:93-104.

White, P.J., and R.A. Garrott.  1997.  Factors regulating kit fox populations.  Canadian Journal of Zoology 75:1982-1988.

Return to Table of Contents