Fish Species Distribution Models for Lake Ontario

Thursday, August 21, 2014: 11:50 AM
2101 (Centre des congrès de Québec // Québec City Convention Centre)
James E. McKenna Jr. , Tunison Laboratory of Aquatic Science, US Geological Survey, Great Lakes Science Center, Cortland, NY
Maureen Walsh , Lake Ontario Biological Station, USGS Great Lakes Science Center, Oswego, NY
Robert Alexander , Tunison Laboratory of Aquatic Science, USGS Great Lakes Science Center, Cortland, NY
Chris Castiglione , Lower Great Lakes Fish and Wildlife Conservation Office, US Fish and Wildlife Service, Basom, NY
Habitat is a major determinant of species distributions. Broad sampling of habitats allows us to develop models that provide an indication of the “natural” distribution of a given species and their associated optimal habitat conditions. Those distributions can be modified by degrading factors and other influences that may limit the species’ distribution. Together, Species Distribution Models may be used to estimate biodiversity at multiple spatial scales throughout large areas. We used trawl- and seine-collected fish data to develop neural network models that predict the relative abundances and distributions of 25 fish species in Lake Ontario. There were 1,933 standardized samples available from 25 of 36 Aquatic Habitat Area types (representative of >97% by area) to train the neural networks. Each model explained >80% of variation in the data (using 10 – 15 predictors). These models show the distributions of potential optimal and marginal habitats for each species throughout Lake Ontario and can be used to measure biodiversity; optimal habitat conditions may be specified. Model predictions show distributions that generally conform to known broad-scale distributions of species like Yellow Perch, Spottail Shiner, and Alewife, but also reveal finer-scale areas of potential optimal and marginal habitats.