39
4
Quantitative Methods for Modeling
Species Habitat: Comparative
Performance and an Application to
Australian Plants
Jane Elith
Introduction
Habitat evaluation procedures are used to quantify the value of land as habitat for
a species. They result in maps of species habitats that are useful for applications
involving estimation of the impacts of management alternatives; ecological assessment; conservation planning; and the identification of steps that may be taken
to avoid habitat losses or to compensate unavoidable habitat losses due to a
proposed action (Van Horne and Wiens 1991; Gray et al. 1996; Rand and Newman
1998). For example, species that have narrow geographic distributions but are
abundant in at least one place would benefit from a protected area. Species that are
widely distributed within specialized habitat may benefit from targeted management prescriptions within a multiple-use landscape. Trade-offs may be identified
so that biodiversity objectives may be met efficiently. The effectiveness of such
management decisions depends critically on reliable knowledge of the current and
potential distribution of species.
Frequently, management decisions are based on “known locations.” These are
geographic coordinates that accompany field observations or specimens in museums and herbariums. Such records are generally used as the basis for the
construction of distribution maps, which are composed of polygons that enclose
all known locations, conditioned by available ecological knowledge and subjective extrapolations based on topography or the spatial distribution of environmental variables. These maps are rarely validated, and their reliability is rarely
reported.
The hard boundaries represented in most maps of species distributions are
abstractions (Gaston 1994). In reality, habitat suitability is determined by the
response of species to a set of spatially distributed, random variables. Hence the
definition of habitat requires an understanding of the way in which environmental
variables affect the likelihood that a species will be found at a given location (e.g.,
Austin 1985; Burgman 1989; Prober and Austin 1990; McIntyre and Lavorel
1994; Coates 1996). Species distributions are strongly influenced by processes
such as successional dynamics and disturbance (Main 1984; Attiwill 1994).
4
Quantitative Methods for Modeling
Species Habitat: Comparative
Performance and an Application to
Australian Plants
Jane Elith
Introduction
Habitat evaluation procedures are used to quantify the value of land as habitat for
a species. They result in maps of species habitats that are useful for applications
involving estimation of the impacts of management alternatives; ecological assessment; conservation planning; and the identification of steps that may be taken
to avoid habitat losses or to compensate unavoidable habitat losses due to a
proposed action (Van Horne and Wiens 1991; Gray et al. 1996; Rand and Newman
1998). For example, species that have narrow geographic distributions but are
abundant in at least one place would benefit from a protected area. Species that are
widely distributed within specialized habitat may benefit from targeted management prescriptions within a multiple-use landscape. Trade-offs may be identified
so that biodiversity objectives may be met efficiently. The effectiveness of such
management decisions depends critically on reliable knowledge of the current and
potential distribution of species.
Frequently, management decisions are based on “known locations.” These are
geographic coordinates that accompany field observations or specimens in museums and herbariums. Such records are generally used as the basis for the
construction of distribution maps, which are composed of polygons that enclose
all known locations, conditioned by available ecological knowledge and subjective extrapolations based on topography or the spatial distribution of environmental variables. These maps are rarely validated, and their reliability is rarely
reported.
The hard boundaries represented in most maps of species distributions are
abstractions (Gaston 1994). In reality, habitat suitability is determined by the
response of species to a set of spatially distributed, random variables. Hence the
definition of habitat requires an understanding of the way in which environmental
variables affect the likelihood that a species will be found at a given location (e.g.,
Austin 1985; Burgman 1989; Prober and Austin 1990; McIntyre and Lavorel
1994; Coates 1996). Species distributions are strongly influenced by processes
such as successional dynamics and disturbance (Main 1984; Attiwill 1994).