Abstract:
Accurately estimating the distribution of a species is important for managing sustainable
populations of fishes. The Yellow Perch Perca flavescens is an important sport fish in the Great Lakes region and one of the most abundant fishes in Great Lakes coastal wetlands, which they commonly use for spawning and nursery habitat. Many fisheries management decisions are based on results from sampling fish assemblages, but these methods rarely account for incomplete detection (i.e., presence of a species that is not detected by sampling), which could create biased results. We applied the method of occupancy modeling, which accounts for incomplete detection, to Yellow Perch presence/absence data from coastal wetlands across all five Great Lakes. We used occupancy models with environmental variables to predict the detection probability of fyke-net sampling and the occupancy of Yellow Perch under different environmental conditions. We found that both detection probability and occupancy of Yellow Perch varied among Great Lakes and with changes in other environmental variables. The best statistical model included sampling depth, specific conductivity, wetland hydrologic connection, and Great Lake basin. Yellow Perch occupancy was predicted to be highest in areas with greater depth, lower specific conductivity, and a riverine connection to a Great Lake. All naïve occupancy estimates were lower than the occupancy estimates predicted by our models. Our base model with no covariates predicted an occupancy of 0.68 and detection probability of 0.669
across all sites. Our results predict which coastal wetland habitats were preferred by Yellow Perch (i.e., those with low specific conductivity and greater depth) and emphasize the importance of incorporating detection probability into occupancy estimates. Our results can help provide support for the conservation of coastal wetlands with preferred Yellow Perch habitat, and guidance for future coastal wetland restoration projects.
