摘要：The Niangua darter (Etheostoma nianguae) is a threatened stream fish endemic to the Osage River basin in southcentral Missouri, USA. Knowledge of Niangua darter habitat relationships has been identified as an important component of recovery. We assessed the predictive performance of a previously developed summer microhabitat model for this species. The original model was developed with multiple logistic regression using Little Niangua River data, with water depth and substrate particle size serving as predictors of darter presence or absence. New data were collected in Little Niangua River and Maries River at microhabitats occupied (n = 16) and unoccupied (n = 72) by the darter. About one in four microhabitats was misclassified by the model and values of Cohen's kappa generally did not exceed 0.40, indicating that the model's performance could only be rated as fair. Performance was slightly better in Little Niangua River than in Maries River, with a substantial number of false positives in the latter stream. However, suitable microhabitats may be unsaturated due to relatively low Niangua darter densities, which generally confounds model assessment for a rare species.
