PARENT SESSION
Contributed Oral Session 35: Invasive Species: Great Lakes Area
Monday, August 8, 1:30 PM - 5:00 PM, Meeting Room 524 B, Level 5, Palais des congrès de Montréal

Predictive model of exotic rainbow smelt (Osmerus mordax) distribution in inland lakes of the Great Lakes region.

Mercado-Silva, Norman^*,1, Vander Zanden, Jake¹, Maxted, Jeffrey¹, Olden, Julian¹, ¹ Center for Limnology, Madison, Wisconsin, USA

ABSTRACT- Rainbow smelt (Osmerus mordax) is a non-native fish species currently spreading to inland lakes throughout the Great Lakes region (GLR). It is expected to have detrimental effects on native fish communities in lakes it invades. Currently we lack a method for predicting where invasions will occur. We use classification trees to develop a predictive model to identify lakes in the GLR that could be invaded by smelt. The model is based on environmental and biological data from lakes where the species is native. Six variables (lake area, maximum depth, length of shoreline perimeter, average Secchi depth, average pH, presence of Coregonids) used for model development were obtained for lakes in South - South Eastern Maine (USA). These were used to build a predictive model for regions where the species is non-native and where relatively few lakes have thus far been invaded (Northern Maine; Ontario, Canada; and Wisconsin, USA). The most important predictive variables were maximum depth, lake area, and Secchi depth, with other variables having little importance. This indicates that deep, relatively large, clear lakes are more prone to invasion. We validated the model using information from invaded lakes outside of the native range of the species in Maine. Here the model had 80% success in predicting invaded and non-invaded lakes. We then used the model to predict known smelt presence in Ontario and Wisconsin lakes. The model performed well in this test with 82% correct classifications in each region. Based on the robustness of the model, we predict 530 lakes in Wisconsin and 708 lakes in Ontario where conditions exist for new smelt invasions. These predictions are valuable for natural resource managers in the GLR to focus monitoring and invasion prevention efforts to a limited number of lakes.

Key words: invasive species, rainbow smelt, great lakes, prediction