Potential loss and rehabilitation of stream longitudinal connectivity: fish populations in urban streams with culverts

Riverine connectivity is important to the persistence of fish communities, but culverts may impede fish movements to varying degrees and in both directions. Baffles can be installed in culverts to mitigate upstream connectivity loss; however, evaluation of their effectiveness is limited. To examine the potential impacts of culverts and their potential rehabilitation with baffles, we sampled fish populations in 26 streams that contained either (i) nonbaffled culverts or (ii) baffled culverts or (iii) lacked culverts (reference streams) in Metro Vancouver, British Columbia, Canada. Using mixed effects models, we compared fish responses across these three stream types to infer effects at the whole-stream scale (i.e., over both upstream and downstream positions equally), the within-stream scale (i.e., upstream versus downstream of culverts), and the interaction of scales. Densities (n.m(-2)) of coastrange sculpin (Cottus aleuticus) and prickly sculpin (Cottus asper) were significantly lower in nonbaffled and baffled stream types than in reference stream types, while densities of cutthroat trout (Oncorhynchus clarkii) and rainbow trout (Oncorhynchus mykiss) were significantly lower in reference stream types, indicating whole-stream differences. Using multivariate statistics, we similarly found that fish community compositions were significantly different across stream types. For our various fish responses, we found no interaction between stream type and upstream or downstream position. Further, we found reaches directly downstream of baffled culverts had greater fish biomass and that overall species richness increased with age of baffles. These data suggest that culverts may drive changes in fish populations at whole-stream scales, and restoration of these effects with baffles may take decades.