Tree regression analysis on the nesting habitat of smallmouth bass

Habitat conditions may determine locations of patchily distributed smallmou th bass nests in lakes, since young smallmouth bass are fragile and are the refore vulnerable to suboptimal physical characteristics in their typically variable nesting area, the littoral zone. Knowing which habitat conditions are important to nest locations would ultimately be useful in protecting o ptimal nesting areas from anthropogenic disturbances in lakes. To evaluate factors related to the nest distributions of a smallmouth bass (Micropterus dolomieui) population, physical habitat conditions were measured at 36 1-k m-long and 31 100-m-long sites in Lake Opeongo (Ontario, Canada). Both tree regression analysis (a recently devised and unique statistical tool) and s tandard multiple regression were used to determine the relationships betwee n nest density and four habitat variables. Tree regression analysis does no t require assumptions of linearity or homoscedasticity of variances, and it automatically identifies interactions among variables, Furthermore tree re gression results were more accurate and more precise than standard multiple -regression results, Mechanisms underlying the significant relationships be tween nest densities and both littoral zone temperatures and shoreline reti culation in 1-km-long sites and the nonsignificant results at the 100-m-lon g scale, are discussed. Cross-validation results quantify the difficulty in extrapolating sample findings to whole populations in ecological research.