Predicting Walleye Recruitment Success in Northern Wisconsin Lakes

Predicting recruitment is a long-standing goal in fisheries management. To guide management of Wisconsin’s Walleye (Sander vitreus), we predicted the probability of successful recruitment, defined as fall young-of-year Walleye catch rates greater than 10/mile, based on 2,210 surveys in 303 lakes conducted from 1989-2012 in Wisconsin’s Ceded Territory. Recruitment was predicted from an initial set of 25 predictor variables describing lake morphometry, thermal characteristics, water chemistry, and watershed land cover. We used a random forest model and an aggressive model selection procedure to maximize predictive ability while minimizing the inclusion of spurious variables. Recruitment was predicted with 79% accuracy based on lake area, growing degree days, maximum depth, shoreline development factor, and conductivity. Walleye recruitment success was highest in large, cool, low conductivity lakes of moderate maximum depths.  Interactions between the effects of lake area and growing degree days indicate that successful Walleye recruitment was most likely in cooler lakes across a range of lake sizes, and in warmer lakes only in very large (> 730 ha) lakes. Results were used to evaluate the potential for lakes throughout Wisconsin to support natural reproduction and as a management tool to prioritize stocking in lakes with a stated management goal of restoration.