Occurrence of Bull Trout in Naturally Fragmented Habitat Patches of Varied Size

-Bull trout Salvelinus confluentus and other salmonids in the Pacific Northwest are believed at risk of local and regional extinctions because of ongoing habitat loss and fragmentation . Biologists have focused on defining and protecting critical stream channel characteristics, but there is little information regarding the scale or spatial geometry of habitat that may be necessary for the species' long-term persistence . We investigated the influence of habitat patch size on the occurrence of bull trout by determining the presence or absence of fish in naturally fragmented watersheds of the Boise River basin in Idaho . We defined patches of potential habitat for bull trout as watersheds above 1,600 m elevation, a criterion based on the presumed restriction of local populations by stream temperature . We used logistic regression to investigate the possible influence of patch size as well as stream width and gradient on the occurrence of bull trout at reach, stream, and patch scales of analysis . Both stream width and patch size were significant in the models, but individual effects could not be clearly resolved because of collinearity . The predicted probability of occurrence based on patch size alone was less than 0.10 for patches smaller than about 1,000 ha and more than 0.50 for patches larger than about 2,500 ha . Our results support the hypothesis that area of available habitat influences the distribution of disjunct populations of bull trout . An area effect is consistent with the predictions of island biogeography and metapopulation theory, and our work suggests that larger-scale spatial processes may be important to the persistence of species like bull trout . Bull trout Salvelinus confluentus is a recent addition to the growing list of salmonid fishes considered under the Endangered Species Act (Office of the Federal Register 59[June 10, 1994]:30254) . Populations have declined throughout much of the range ; some local populations are believed extinct, whereas others are viewed as remnants isolated in shrinking patches of suitable habitat (see papers in Howell and Buchanan 1992 ; Rieman and McIntyre 1993) . As with other salmonids, declines can be attributed to habitat degradation (Fraley and Shepard 1989 ; Howell and Buchanan 1992), displacement by exotic species (e .g ., Leary et al . 1993), increased mortality caused by fishing, or dams and diversions that influence migratory corridors . Fishery managers are attempting to minimize mortalities and to preserve habitat . Biologists are attempting to characterize critical habitat primarily through the more-or-less traditional measures of stream channel structure studied at the scale of habitat units or stream reaches (e .g ., Hawkins et al . 1993) . Effective conservation of sensitive populations, however, may require more than just reducing mortality and maintaining as much of the remaining critical habitat as possible . The larger-scale spatial geometry of habitat is now viewed as important to the persistence of many species . Effective conservation may imply maintaining or restoring a critical amount or mosaic of habitat as well (Simberloff 1988). Spatially influenced processes have received considerable attention in study of the dynamics and distribution of a variety of species. Island biogeography (MacArthur and Wilson 1967) and the emerging theory of metapopulation dynamics (Hanski 1991 ; Hanski and Gilpin 1991) hold that the distribution of populations among patches or "islands" of suitable habitat will be a function of local extinction and of dispersal and colonization processes (Hanski 1991, 1994). In essence, smaller, more isolated populations are less likely to persist because (1) small populations face higher risk of extinction through demographic and environmental stochasticity and Allee effects (Leigh 1981 ; Gilpin and Soulé 1986 ; Simberloff 1988 ; Saunders et al . 1990 ; Boyce 1992) ; and (2) isolated populations have a lower probability of demographic support or recolonization through dispersal from surrounding populations (Brown and KodricBrown 1977 ; Pulliam 1988). Metapopulation concepts have been strongly embraced in conservation biology. The spatial geometry of habitat reserves is a central issue in the conservation of species and populations (Simberloff 1988). But are these ideas relevant to the management of salmonids? Recent discussions on conservation of seriously depressed and fragmented western salmonid stocks have invoked spatial concepts and metapopulation theory (Nehlsen et al . 1992 ; Reeves and Sedell 1992; Frissell et al . 1993 ; Rieman and McIntyre 1993). There is a notable lack, however, of empirical evidence or spatially explicit models necessary to guide such management . Watershed or basin geometry may influence the diversity in fish communities (Sheldon 1988 ; Schlosser 1991 ; Osborne and Wiley 1992), but little information is available regarding the dynamics of individual species. Area effects appear important across a wide range of taxa (Gilpin and Diamond 1981 ; Hanski 1991, 1992; Kindvall and Ahlen 1992), but the nature of the response may also vary widely among species (Gilpin and Diamond 1981 ; Hanski 1991 ; Taylor 1991). Some salmonid populations have persisted for extended periods in small habitat patches isolated by natural barriers (Northcote 1992) . Are salmonids generally resistant to chance extinctions? Or, like other animals, are some species predictably sensitive to local extinction while others are not (Cutler 1990)? These questions are relevant for many salmonids that face increasing habitat loss . Although fish population declines can be tied to reduced habitat quality, introduction of exotics, and overfishing, it is not clear how population and species losses are aggravated by habitat fragmentation and isolation. Hanski (1991, 1992, 1994) and others (Gilpin and Diamond 1981) have shown that the distribution of animals structured as metapopulations in fragmented systems may reflect the underlying dynamics of local extinction and recolonization through dispersal . Incidence functions that relate the presence or absence of animals in local populations to the amount or other characteristics of available habitat have been widely used to draw inferences about factors likely to influence the persistence of populations (Adler and Wilson 1985 ; Taylor 1991 ; Hanski 1994) . Incidence functions have been viewed as particularly important where long-time series of detailed information on population demographics and dynamics are unavailable or unlikely in the future (Taylor 1991 ; Hanski 1994). Such analyses are a logical first step for addressing questions about the spatial dynamics of salmonids in fragmented habitats . Bull trout have patchy distributions within watersheds throughout their range (Rieman and McIntyre 1993). Although distributions are probably influenced by habitat loss, dams, diversions, and exotic species, bull trout also appear to be naturally restricted to colder stream temperatures (Fraley and Shepard 1989 ; Rieman and McIntyre 1993) . Because of the association with temperature, elevation should define habitat patches or "islands" in the headwaters of many larger watersheds that are at least partially isolated by distance across warmer waters . Such isolation could be particularly important near the southern limits of the species' range, where fragmentation of habitat by temperature may be more evident (Flebbe 1993) . In this paper we define suitable habitat patches for bull trout from the observed relationship of fish distribution with elevation (and presumably temperature) . We examine the association of patch size and other habitat features (stream width and gradient) with the distribution of bull trout throughout a large river basin . We hypothesize that area of available habitat is important to the persistence of local populations . If that is true, patch size should have a significant influence on occurrence of bull trout throughout the basin.