Application of Discrete Choice Models to Predict White Crappie Temperature Selection in Two Missouri Impoundments

Abstract

AbstractTemperature selection by white crappie Pomoxis annularis over 24 h was determined in two Missouri impoundments, Little Dixie Lake and Rocky Fork Lake, on six and five dates, respectively, during June‐August 2001 using temperature‐sensitive radio transmitters. Temperature and dissolved oxygen (DO) profiles showed that the habitable portions of the upper water columns associated with fish locations varied by as much as 6°C in temperature. The habitable portions of water columns were subdivided into three strata (warm, moderate, and cool) such that each stratum covered an equivalent temperature range (≤2°C). Discrete choice models (DCMs) were developed for each impoundment to predict the relative probabilities of white crappies selecting the different strata in response to stratum height, DO concentration, average temperature throughout the habitable water column (AVG), and fish weight. For each impoundment, a distinct DCM containing AVG as the sole predictor variable was most supported by the data. Cross‐verification indicated that the DCMs accurately predicted white crappie thermal strata selection within the impoundments for which they were developed but that they were inaccurate when applied across the impoundments. The DCM for Little Dixie Lake predicted that white crappies mostly select warm strata at AVGs less than 27°C and mostly select cool strata at AVGs of 27°C or more. In contrast, the DCM for Rocky Fork Lake predicted that white crappies continue to select warm strata until AVGs reach 30°C, select moderate strata at AVGs of 30–32°C, and select cool strata only when AVGs exceed 32°C. The predicted mean daily thermal experiences of the white crappies in each impoundment, obtained by multiplying the relative selection probabilities for the different strata by the average temperatures in those strata, remained within 0.5°C of the mean of observed temperatures experienced by white crappies. These findings indicate that discrete choice modeling provides a novel and effective approach for predicting the daily thermal experiences of fish.