Abstract:
Although threespine stickleback (Gasterosteus aculeatus L) are known to inhabit a wide range of habitats, their distribution in lakes across Southeast Alaska is not known. Threespine stickleback are an important prey item for many consumers in freshwater ecosystems. Additionally, isolated populations may be genetically unique and thus important from a conservation perspective. This study focused on identifying' landscape factors and models useful in predicting the presence of threespine stickleback in small (0.5-5 ha) lakes of Southeast Alaska. Stickleback occurrence was assessed via snorkeling and minnow trapping in 54 lakes, which were divided into calibration (n=36) and prediction (n=18) data sets. A number of models representing four methodologies-generalized linear models, generalized additive models, classification trees, and artificial neural networks-were built based on the calibration set, cross-validated, and evaluated by prediction to the test set of lakes. Lake elevation, distance from saltwater, and slope of lake outlet stream were the most useful predictors of stickleback occurrence. Results suggest that the likelihood of stickleback presence is highest in low elevation lakes near the coast. Human development and recreational activity also tends to be common in these areas, and so land-use planning should account for the high potential occurrence of threespine stickleback here
