Classification of Texas Reservoirs in Relation to Limnology and Fish Community Associations

I used cluster analysis to examine associations among 20 fish species to develop a classification scheme for 132 large Texas reservoirs. Five major groups of reservoirs were identified by cluster analysis based on species associations. Of 29 reservoirs surveyed previously, 76% were classified into the same species associations from one survey to the next. When 19 environmental variables were used in canonical correlation analysis of the five reservoir groups, I found a general east-to-west separation of species associations by water quality and a northwest-to-southeast separation by surface elevation and growing season. A discriminant functions model based on a reduced set of nine environmental variables had an unbiased error rate of 18% for predicting the species association in unclassified reservoirs. A stratified sampling scheme based on the classification model decreased the variance of statewide electrofishing catch per effort up to 43% for bluegill Lepomis macrochirus and 23% for largemouth bass Micropterus salmoides over a simple random sample of reservoirs.