Published online by Cambridge University Press: 09 August 2011
Freshwater fish and mussel diversity is declining at an alarming rate across North America. Human land uses and disturbances within watersheds have been implicated as the primary cause of declines. In this paper, we demonstrate the utility of land use/land cover (LULC) variables in species distribution modeling and conservation planning using a straightforward multiscale approach for prioritizing freshwater fish and mussel conservation areas in the upper Green River catchment (Ohio River basin, USA). We developed distribution and species richness models for 10 uncommon fishes and 14 rare mussels using multiscale landscape data and boosted regression tree (BRT) analyses based on LULC composition and pattern, geology composition, and soil composition data. We then used probability of occurrence, endemicity, prevalence, trend and range of individual species to estimate the conservation value of each stream reach. Conservation areas were defined for three spatial scales nested within the catchment management zone (focal areas, riparian management buffer and subcatchment management zone) using a simple optimization technique. Priority conservation areas were located primarily in the eastern (upper Green River) and southern (upper Barren River) portions of the catchment. We found that focal species richness is explained most by soil composition in the subcatchment. However, nested within the subcatchment scale focal species richness responded positively to percent forest and negatively to patch density of developed/exposed land in the reach buffer. For both the reach and riparian buffers, retaining forested tracts of land and limiting the level of development and fragmentation would benefit the focal species.