UCSB Biogeography Lab

Predicting Vulnerability of Regional Biota
Extending Gap Analysis

Principal Investigator: Dr. Frank Davis

Funding agency: University of Idaho-Gap Analysis Program
Project period: July 1, 1997 to December 31, 1998

EXECUTIVE SUMMARY
David M. Stoms, Allan D. Hollander, and Frank W. Davis

Report Date: 31 July 1999

Among several alternative approaches to predicting vulnerability from surrogate data, GAP uses the management profile. In this method, species and ecosystems with less land managed for biodiversity objectives are considered more vulnerable than those with more protected land. Conservation priorities are then ranked on the basis of the proportion of protected land. In practice, this initial prioritization is frequently tempered by qualitative assessment of the level of actual threat facing individual species and communities. The research described in this report evaluates the relationship between management profile and other indicators of biotic vulnerability. The objectives included identifying where and how GAP might improve the prediction of vulnerability with existing or new data sets.

We begin the report with a diagram that illustrated a general framework for thinking about the patterns and processes of land use and management that affect biotic vulnerability. The ideal of measuring vulnerability for all species directly is shown to be infeasible because of the complexity of existing and permitted land use activities, the spatial configuration of land management, and the uniqueness of species responses to these factors. The framework outlines a progression from simple to complex predictors of the vulnerability of species and communities based processes that affect biotic vulnerability and spatial patterns resulting from those processes. We ask whether GAP status profile gives a reasonable approximation of biotic vulnerability? We undertake two independent approaches to testing this hypothesis using data from the California Gap Analysis Project. In the first approach, land management status profiles for plant communities are compared to ecological indicators related to processes of permitted land uses. This analysis provides an indication of how well the pattern from management status classification represents these processes that indirectly affect biodiversity vulnerability. The second approach uses the trends in species abundance from the Breeding Bird Survey data to measure biotic responses directly and compares those trends to the level of protection as measured by management status in California. Our hypothesis states that declining species will tend to be those with the lowest levels of protection while those that are stable or increasing will have higher levels of protection.
Gap Management Status And Regional Indicators Of Threats To Biodiversity
Conservation assessment requires quantitative criteria for evaluating the relative degree of threat faced by species or ecological communities. Identifying appropriate criteria for communities is complicated because the species inhabiting them can have many different responses to land uses and other forms of environmental stress. The Gap Analysis Program (GAP) uses summary data on the proportion of the community that is protected as an estimate of its vulnerability. Management status from a gap analysis of California was compared with three ecological indicators (permitted land uses, human population growth, and the spatial extent of road effects) that more directly represent impacts on biodiversity. The classification of management status appears to provide a crude first approximation of these three indicators. Public and private lands that are not formally protected were susceptible to extensive land use conversion or resource extraction in both rural and urban settings. Some plant community types are more susceptible to future infringement by human population increases that were not well predicted by management status alone. Other community types are heavily roaded despite being moderately well protected. It is suggested that indicators such as future growth and current road effects could complement status in rating the potential vulnerability of plant communities and setting conservation priorities. The choice of indicators will depend on the threatening processes in a given region and the availability of spatial data to map or model them.

In Chapter 2, we compare gap management status profiles to independent measures of permitted land uses (zoning), projected human population growth, and the spatial extent of road effects, which are more directly related to biotic responses. That is, the pattern of land management profile is compared to measures of processes in the framework. The classification of management status appears to provide a crude first approximation of the three indicators. On national forest lands in the Sierra Nevada, half of the status 3 lands are allocated to timber harvest while the remainder is managed for lower intensity multiple uses. Virtually all of the private land (status 4) in rural El Dorado County is zoned for urbanization, low-density ranchettes, or resource extraction. In the highly urbanized southern California counties, a lower percentage of developable land remains. On the other hand, the proportion of protection in status 1 and 2 areas for community types did not predict future population growth or road effects very reliably. Although the community types with the highest projected growth also had the least representation in managed areas, the reverse was not necessarily true—that poorly represented types had high projected populations. Status 4 lands had nearly 10 times as much area affected by roads as status 1 as estimated by the roadedness index. Status 3 public lands had nearly 5 times as much as status 1. Some types with low levels of protection nevertheless were relatively unroaded, while others had a relatively large proportion affected by roads even with high levels of protection. Estimates of future population density and current roadedness could complement the level of protection in setting conservation priorities.

[Note: Chapter 2 was published in slightly different form in Landscape Ecology.

Stoms, D. M. 2000. GAP management status and regional indicators of threats to biodiversity. Landscape Ecology 15: 21-33.
[abstract at Kluwer]
Comparison Of Breeding Bird Survey Trends With Gap Predictions
In Chapter 3, we use the trends in species abundance from the Breeding Bird Survey data to measure biotic responses directly and compared those trends to the level of protection as measured by management status in California. Our hypothesis stated that declining species will tend to be those with the lowest levels of protection while those that are stable or increasing will have higher levels of protection. The comparison of GAP management profiles with trend data from the BBS found relatively little correspondence. Overall, there was little difference between the patterns of protection for species showing significant decline and those showing no decline. Adding other factors believed to be related to species declines (roadedness, human population density, and range size) did little to improve the success of a logistic regression model in predicting which species were declining. This general lack of correspondence was found for all bird species in the transects, those that were best modeled by CA-GAP, and for non-migrants. Changing the resolution of the trend data from statewide aggregations to the individual transects also made little difference.

The simplest explanation is that management status does not map closely onto biological impacts. That is, there is still plenty of good-quality habitat for many species in the landscape, even on private lands or on level 3 public lands. In other words, tabulating species by management status level is a statement more about policy than about biological vulnerability. Finding that a species is mostly on private land may argue for additional land for biological reserves, or creating incentives for conservation by private landowners, as an insurance policy to ensure long-term maintenance of habitat. There is no evidence, at least according to our analyses, that species are inherently more at risk because of management status.
One major problem is that human disturbance, which is the main thing indices of vulnerability try to represent, differentially affects species. Some species tolerate human-dominated landscapes; others do not. One direction that might find better correspondence of management status and population trends would be to distinguish species that are adapted to human-dominated habitats from those more intolerant. It is possible that the more adaptable species that would not decline with lack of protection may be masking a more meaningful relationship for the less tolerant birds. We did find some bias in the representation of management status levels and ownership by the BBS routes. We might expect that the BBS would tend to overestimate the likelihood of a population decline for species that largely inhabit higher elevation habitats in protected areas.

It is therefore important to expand the program of monitoring wildlife populations. We used BBS data because there are no other comparable data set for other vertebrate groups. We would argue that without such monitoring information, we will be unable to determine which species are truly at risk.
Recommendations
Ideally, prioritization of species and communities for conservation would be based on true measures of their individual vulnerability to extinction or to unacceptable levels of decline in abundance. As outlined in our framework in this report, this should be based on a combination of the actual and projected land use activities within the species’ range, the topology of land management, the ecological correlates of extinction proneness, and the species-specific responses to these factors at all life history stages. We have begun to address pieces of this framework in this report, but it remains a useful outline for guiding future GAP research needs. Our recommendations follow the Processes and Patterns of the conceptual framework, plus a recommendation about the integration of assessment with decision support for conservation planning.
Processes

* GAP should supplement management status mapping with information on land uses permitted within individual tracts of land.
* GAP could pioneer the synthesis of knowledge about road effects on biodiversity.
* GAP should collaborate with urban growth modelers to integrate the impacts of potential urbanization on biodiversity.
* A matrix of responses of species and communities to land use activities should be developed from literature review and expert opinion.

Patterns

* GAP should investigate the role of the configuration of reserves in the unreserved matrix in maintaining viable populations.
* GAP should not attempt to model "presettlement" vegetation to determine historical losses as a predictor of vulnerability. If historic loss is to be used, GAP should rely on qualitative estimates from the Heritage Program or similar sources.

Decision Support

* Reserve selection algorithms should be developed that minimize vulnerability of biodiversity elements, not just meet representation targets.

ACKNOWLEDGMENTS
We thank Michael Jennings and Patrick Crist for the continuing direction of the National Gap Analysis Program. Vincent Burke at the University of Missouri organized a special symposium on gap analysis at the 13th annual conference of the International Association for Landscape Ecology, United States Regional Association, held March 17-21, 1998, in East Lansing, Michigan. This symposium was the impetus for the analysis presented in Chapter 2 of this report. We are grateful to Chris Cogan, Tim Duane, and members of the Biogeography Lab at UCSB for stimulating many of the ideas presented in Chapter 2. Joe Walsh assisted with some of the data processing. We sincerely appreciate the helpful suggestions of three anonymous reviewers and Vincent Burke for the version of Chapter 2 that was accepted for publication in Landscape Ecology.



What's New Research & Projects Remote Sensing & GIS Landscape Ecology Conservation Planning Publications People Data Links	\| Home \| Contact \| UCSB \| Bren \| ICESS \|

	Predicting Vulnerability of Regional Biota Extending Gap Analysis Principal Investigator: Dr. Frank Davis Funding agency: University of Idaho-Gap Analysis Program Project period: July 1, 1997 to December 31, 1998 EXECUTIVE SUMMARY David M. Stoms, Allan D. Hollander, and Frank W. Davis Report Date: 31 July 1999 Among several alternative approaches to predicting vulnerability from surrogate data, GAP uses the management profile. In this method, species and ecosystems with less land managed for biodiversity objectives are considered more vulnerable than those with more protected land. Conservation priorities are then ranked on the basis of the proportion of protected land. In practice, this initial prioritization is frequently tempered by qualitative assessment of the level of actual threat facing individual species and communities. The research described in this report evaluates the relationship between management profile and other indicators of biotic vulnerability. The objectives included identifying where and how GAP might improve the prediction of vulnerability with existing or new data sets. We begin the report with a diagram that illustrated a general framework for thinking about the patterns and processes of land use and management that affect biotic vulnerability. The ideal of measuring vulnerability for all species directly is shown to be infeasible because of the complexity of existing and permitted land use activities, the spatial configuration of land management, and the uniqueness of species responses to these factors. The framework outlines a progression from simple to complex predictors of the vulnerability of species and communities based processes that affect biotic vulnerability and spatial patterns resulting from those processes. We ask whether GAP status profile gives a reasonable approximation of biotic vulnerability? We undertake two independent approaches to testing this hypothesis using data from the California Gap Analysis Project. In the first approach, land management status profiles for plant communities are compared to ecological indicators related to processes of permitted land uses. This analysis provides an indication of how well the pattern from management status classification represents these processes that indirectly affect biodiversity vulnerability. The second approach uses the trends in species abundance from the Breeding Bird Survey data to measure biotic responses directly and compares those trends to the level of protection as measured by management status in California. Our hypothesis states that declining species will tend to be those with the lowest levels of protection while those that are stable or increasing will have higher levels of protection. Gap Management Status And Regional Indicators Of Threats To Biodiversity Conservation assessment requires quantitative criteria for evaluating the relative degree of threat faced by species or ecological communities. Identifying appropriate criteria for communities is complicated because the species inhabiting them can have many different responses to land uses and other forms of environmental stress. The Gap Analysis Program (GAP) uses summary data on the proportion of the community that is protected as an estimate of its vulnerability. Management status from a gap analysis of California was compared with three ecological indicators (permitted land uses, human population growth, and the spatial extent of road effects) that more directly represent impacts on biodiversity. The classification of management status appears to provide a crude first approximation of these three indicators. Public and private lands that are not formally protected were susceptible to extensive land use conversion or resource extraction in both rural and urban settings. Some plant community types are more susceptible to future infringement by human population increases that were not well predicted by management status alone. Other community types are heavily roaded despite being moderately well protected. It is suggested that indicators such as future growth and current road effects could complement status in rating the potential vulnerability of plant communities and setting conservation priorities. The choice of indicators will depend on the threatening processes in a given region and the availability of spatial data to map or model them. In Chapter 2, we compare gap management status profiles to independent measures of permitted land uses (zoning), projected human population growth, and the spatial extent of road effects, which are more directly related to biotic responses. That is, the pattern of land management profile is compared to measures of processes in the framework. The classification of management status appears to provide a crude first approximation of the three indicators. On national forest lands in the Sierra Nevada, half of the status 3 lands are allocated to timber harvest while the remainder is managed for lower intensity multiple uses. Virtually all of the private land (status 4) in rural El Dorado County is zoned for urbanization, low-density ranchettes, or resource extraction. In the highly urbanized southern California counties, a lower percentage of developable land remains. On the other hand, the proportion of protection in status 1 and 2 areas for community types did not predict future population growth or road effects very reliably. Although the community types with the highest projected growth also had the least representation in managed areas, the reverse was not necessarily true—that poorly represented types had high projected populations. Status 4 lands had nearly 10 times as much area affected by roads as status 1 as estimated by the roadedness index. Status 3 public lands had nearly 5 times as much as status 1. Some types with low levels of protection nevertheless were relatively unroaded, while others had a relatively large proportion affected by roads even with high levels of protection. Estimates of future population density and current roadedness could complement the level of protection in setting conservation priorities. [Note: Chapter 2 was published in slightly different form in Landscape Ecology. Stoms, D. M. 2000. GAP management status and regional indicators of threats to biodiversity. Landscape Ecology 15: 21-33. [abstract at Kluwer] Comparison Of Breeding Bird Survey Trends With Gap Predictions In Chapter 3, we use the trends in species abundance from the Breeding Bird Survey data to measure biotic responses directly and compared those trends to the level of protection as measured by management status in California. Our hypothesis stated that declining species will tend to be those with the lowest levels of protection while those that are stable or increasing will have higher levels of protection. The comparison of GAP management profiles with trend data from the BBS found relatively little correspondence. Overall, there was little difference between the patterns of protection for species showing significant decline and those showing no decline. Adding other factors believed to be related to species declines (roadedness, human population density, and range size) did little to improve the success of a logistic regression model in predicting which species were declining. This general lack of correspondence was found for all bird species in the transects, those that were best modeled by CA-GAP, and for non-migrants. Changing the resolution of the trend data from statewide aggregations to the individual transects also made little difference. The simplest explanation is that management status does not map closely onto biological impacts. That is, there is still plenty of good-quality habitat for many species in the landscape, even on private lands or on level 3 public lands. In other words, tabulating species by management status level is a statement more about policy than about biological vulnerability. Finding that a species is mostly on private land may argue for additional land for biological reserves, or creating incentives for conservation by private landowners, as an insurance policy to ensure long-term maintenance of habitat. There is no evidence, at least according to our analyses, that species are inherently more at risk because of management status. One major problem is that human disturbance, which is the main thing indices of vulnerability try to represent, differentially affects species. Some species tolerate human-dominated landscapes; others do not. One direction that might find better correspondence of management status and population trends would be to distinguish species that are adapted to human-dominated habitats from those more intolerant. It is possible that the more adaptable species that would not decline with lack of protection may be masking a more meaningful relationship for the less tolerant birds. We did find some bias in the representation of management status levels and ownership by the BBS routes. We might expect that the BBS would tend to overestimate the likelihood of a population decline for species that largely inhabit higher elevation habitats in protected areas. It is therefore important to expand the program of monitoring wildlife populations. We used BBS data because there are no other comparable data set for other vertebrate groups. We would argue that without such monitoring information, we will be unable to determine which species are truly at risk. Recommendations Ideally, prioritization of species and communities for conservation would be based on true measures of their individual vulnerability to extinction or to unacceptable levels of decline in abundance. As outlined in our framework in this report, this should be based on a combination of the actual and projected land use activities within the species’ range, the topology of land management, the ecological correlates of extinction proneness, and the species-specific responses to these factors at all life history stages. We have begun to address pieces of this framework in this report, but it remains a useful outline for guiding future GAP research needs. Our recommendations follow the Processes and Patterns of the conceptual framework, plus a recommendation about the integration of assessment with decision support for conservation planning. Processes * GAP should supplement management status mapping with information on land uses permitted within individual tracts of land. * GAP could pioneer the synthesis of knowledge about road effects on biodiversity. * GAP should collaborate with urban growth modelers to integrate the impacts of potential urbanization on biodiversity. * A matrix of responses of species and communities to land use activities should be developed from literature review and expert opinion. Patterns * GAP should investigate the role of the configuration of reserves in the unreserved matrix in maintaining viable populations. * GAP should not attempt to model "presettlement" vegetation to determine historical losses as a predictor of vulnerability. If historic loss is to be used, GAP should rely on qualitative estimates from the Heritage Program or similar sources. Decision Support * Reserve selection algorithms should be developed that minimize vulnerability of biodiversity elements, not just meet representation targets. ACKNOWLEDGMENTS We thank Michael Jennings and Patrick Crist for the continuing direction of the National Gap Analysis Program. Vincent Burke at the University of Missouri organized a special symposium on gap analysis at the 13th annual conference of the International Association for Landscape Ecology, United States Regional Association, held March 17-21, 1998, in East Lansing, Michigan. This symposium was the impetus for the analysis presented in Chapter 2 of this report. We are grateful to Chris Cogan, Tim Duane, and members of the Biogeography Lab at UCSB for stimulating many of the ideas presented in Chapter 2. Joe Walsh assisted with some of the data processing. We sincerely appreciate the helpful suggestions of three anonymous reviewers and Vincent Burke for the version of Chapter 2 that was accepted for publication in Landscape Ecology.
	Email stoms@bren.ucsb.edu