A framework for investigating fish population density responses to habitat fragmentation

Globally, fish populations that rely on rivers are being fragmented by in-stream barriers, but it is unclear how pervasive any negative population responses to fragmentation are. We developed a modeling framework for estimating density responses of fishes upstream and downstream of in-stream barriers and applied it to learn how low-head dams have influenced fish assemblages in 24 Great Lakes tributaries. Of the 29 species modeled, we found strong evidence for increased (n = 6 species) and decreased population densities (n = 6 species) in habitat upstream of low-head dams, while we only found strong evidence for population increases downstream of dams (n = 16 species). The time required for 50% of the dam-induced change to fish population density to occur, t₅₀, was on the order of decades; mean t₅₀ was 41 years in habitat downstream of low-head dams and 21 years in upstream habitat. Estimates of t₅₀ suggest that negative effects occurred faster than positive effects. However, the power of this comparison remains limited because evidence for the logistic population growth model used to derive t₅₀ was strong for only 1 of 29 fish species. We propose that different mechanisms account for the responses observed, with positive responses driven by changes in assemblage structure and negative responses driven by decreased immigration. Our research demonstrates how frameworks considering metapopulation dynamics and the potential for time lags between barrier construction and fish population responses can advance understanding of how in-stream barriers structure the distribution and abundance of fish populations in rivers.