摘要：We developed analysis of covariance (ANCOVA) models to predict fish species richness as a function of two simple landscape-scale, binary variables (surface water connectivity to other aquatic systems and human riparian development), with lake surface area as the covariate. All three variables were significantly related to species richness and explained 61 % of the variance in a data set from 55 limnologically similar lakes in northern Wisconsin. We developed additional ANCOVA models to assess the relative effects of these three variables on four functional species groups, including gamefish, riverine, tolerant, and intolerant species. The number of gamefish species was greater in lakes with riparian development than in undeveloped lakes; this is probably related to introductions by management agencies and anglers. Connectivity increased the number of riverine species and intolerant species. Intolerant species may be more abundant because connecting streams provide colonization routes and decrease time between periodic extirpation and recolonization events. In addition, flowing water connections may provide a refuge from winter hypoxia and decrease extirpation rates of intolerant species. Tolerant fishes were ubiquitous on the landscape and unaffected by the measured variables. Because simple landscape variables can dominate assemblage dynamics of small lakes, and human-mediated introductions can increase species richness at low and moderate levels of disturbance, the use of fish as biological indicators for small lake assessment is not practical.
