AbstractUnraveling the diverse forces controlling the abundance and distribution of fish across a landscape has been challenging, in part because few techniques exist to address multiple factors simultaneously in a single analysis. We used spatially explicit, varying‐coefficient generalized additive models to relate the summertime abundance and distribution of yellow perch Perca flavescens to a variety of predatory, water quality, landscape, and density‐dependent factors. The model used 25 years of fishery‐independent trawling data from southern Green Bay, Lake Michigan, an area that supported major fisheries for yellow perch until their decline in the 1990s. Local catch per unit effort (CPUE) of both age‐0 and age‐1 and older yellow perch was affected by the abundance of double‐crested cormorants Phalacrocorax auritus, dissolved oxygen, water clarity, and bottom depth, but not water temperature. In addition, the local response of age‐0 yellow perch CPUE to most predictor variables, including their own global density, had a unique spatial structure. For instance, increased cormorant abundance was related to declines in local yellow perch CPUE, especially near cormorant nesting islands, while increased dissolved oxygen levels were correlated with increased local yellow perch CPUE in shallow nearshore areas of southern Green Bay. In addition, local yellow perch CPUE increased near the mouth of the Fox River during years with higher water clarity, suggesting that water quality may be limiting to yellow perch in this region. Our results suggest that it is important to explicitly account for the different ways in which predictor variables influence fish abundance across the aquatic landscape.
Read full abstract