specialists for whom habitat loss is most crucial – those species whose survival depends
critically on particular, usually rare, habitats for survival and successful reproduc-
tion. Habitats supply numerous attributes: food, protective cover from predators,
denning sites, shelter from inclement weather, and access to mates. This means that
habitat needs are probably unique for every species. Nonetheless, there are models
that predict the effects of dwindling supply, size, and spatial distribution of habitat
patches in a metapopulation framework (see Chapter 6).
One approach uses incidence functions to characterize probabilities of extinction
and recolonization for specific patches (Hanski 1994, 1998). Because extinction is
often negatively related to population size and small patches tend to hold small
populations at the best of times, extinction is usually modeled as a negative function
of patch size. Colonization rates tend to be low when patches are widely spaced, so
distance among patches is often a critical variable in incidence functions. Data on
the sequence of local extinction and recolonization events allow one to estimate
incidence functions across a matrix of possible sites. These functions can then be
solved, either using matrix techniques or via simulation, to evaluate the long-term
probabilities of persistence (Hanski 1998). Data for the Glanville fritillary (Melitaea
cinxia) show that this approach has high predictive capability, at least in well-
studied species (Hanski 1998; Lindenmayer et al. 2003).
An alternative approach is the software package ALEX(Possingham and Davies 1995).
This software provides a flexible structure for modeling successional change and other
temporal variation within patches, as well as accommodating a variety of methods
to model movements among patches, age structure, and demography within and among
patches. ALEXhas been successful in predicting some key metapopulation attributes
of a wide range of vertebrate species in fragmented forest patches in southeastern
Australia (Lindenmayer et al. 2003).
Some biologists have suggested that a particularly useful way to predict the effects
of habitat loss may be to explicitly link such processes with behavioral models of
patch use and resource selection (Goss-Custard and Sutherland 1997). By understanding
the factors guiding resource use, one can predict how changes in habitat availability
and population density might translate into alterations in birth and death rates across
the population (Stillman et al. 2000). This kind of approach has been most thoroughly
developed for shorebirds (Sutherland and Anderson 1993; Sutherland and Allport 1994;
Goss-Custard and Sutherland 1997; Percival et al. 1998). For example, humans com-
pete with oystercatchers via commercial harvesting of the large bivalves that oyster-
catchers prefer. Using behaviorally based models, Stillman et al. (2001) predicted the
demographic impact of bivalve fisheries on oystercatcher populations.
In principle, this kind of behaviorally based approach could be applied to other
wildlife species. For example, behavioral modeling of patch selection by Thomson’s
gazelles has been used to predict patterns of movement across the Serengeti plains
and demonstrate the importance of unrestricted access to large areas of rangeland
for the long-term viability of these grazers (Fryxell et al. 2004, 2005).A population may, by chance, be forced to extinction by year-to-year variation in
weather or other environmental factors. When the population is small it may exhibit
a random walk to extinction because its dynamics at low numbers are determined
by the unpredictable fortunes of individual members. Genetic drift and inbreeding
depression may also operate at low numbers to reduce fitness and thereby lower310 Chapter 17