Large-scale drivers of fish biodiversity differ across an environmentally variable Great Plains watershed

Abstract

Understanding the empirical relationships between biotic diversity and components of the environment is crucial for effective research and management, particularly in highly disturbed watersheds. The Smoky Hill River is a semi-arid prairie stream with a historic native fish community that is adapted to the extreme and highly variable climatic and hydrological conditions characteristic of the Great Plains streams. Following a literature review on environmental variables, diversity responses, and analysis methods, I evaluated the importance of land use, flow, discontinuities (dams, confluences), and stream type (mainstem-tributary) variables in explaining fish richness using AICc model selection with multiple linear, Poisson and negative-binomial regressions. I then compared these results from 48 sites across three watershed regions to those from a long-term monitoring dataset (ST) using the same candidate variables. Finally, I examined phylogenetic patterns of the fish community using ordination analyses. Patterns and drivers of biodiversity differed with watershed region, land use, stream type, and flow. Fish species richness in the Smoky Hill watershed was negatively correlated with percent developed land in the Lower region of the watershed, but positively correlated with percent herbaceous grassland, the reference prairie condition, in the Upper region of the watershed. Summer mean flow was consistently and positively related to species richness in the Middle and Upper regions of the watershed where flow was limited. In the Lower region of the watershed, species richness was higher in the more flow-moderate tributaries relative to high-flow mainstem sites. In the Middle and Upper flow-limited regions, species richness was lower in the low-flow tributaries than main stem sites. Families of fish species were also related to region and stream type (mainstem vs. tributary). A comparison of two databases showed how different goals, questions, and methods result in different insights, emphasizing the need for establishing a priori goals before sampling.