Predicting Walleye Total Length from Head and Mandible Measurements

Abstract

To supplement harvest information on walleye Sander vitreus collected in Lake Erie creel surveys, we assessed whether walleye size structure and growth could be estimated by using total lengths (TL) predicted from measurements of head length (HL) and mandible length (ML). Regression models were developed by using measurements taken from 124 tournament-caught walleyes sampled in late May 2003. In most cases, HL and ML explained more than 90% of the variation in TL (r ² > 0.90); however, residuals (absolute values) were frequently 10 mm or more. Despite estimation errors, the length-frequency distributions predicted from HL and ML models were not significantly different from observed distributions. Tests of the models with an additional 72 walleyes collected in June 2003 produced similar results. Predicted mean lengths at age for June walleyes ages 2–4 did not differ significantly from mean lengths at age estimated from observed TL. Use of an HL model developed for Minnesota walleyes yielded results similar to those of the Lake Erie models, suggesting that development of population-specific models may not be necessary. Given the accuracy and ease of measurement, we suggest that HL be used to predict walleye TL rather than ML; however, prediction errors indicate that age–length keys constructed by using TL predicted from HL should be validated before use.