Aim:In order to overcome data limitations on predicting future distributions of freshwater fishes at the lake scale, our aim was to build simple species distribution models for three focal fishes from warm-, cool- and cold-water guilds, using a small set of local environmental variables paired with climate information. Location:A total of 6715 inland freshwater lakes across Ontario, Canada. Methods:Multiple logistic regression techniques were employed to model historical species occurrence with seven candidate local environmental variables, including a climate index. Model accuracy was tested by: (i) validating with a subset of the historical dataset (centered around the 1970s), and (ii) predicting species occurrence in lakes using a contemporary dataset (centered around the 2000s), representing ∼30 years of climatic change. Projected climate data was then used to model species occurrence into the 2041–2070 time period. Results:The models for each thermal guild performed well (average AUC of 0.79) using the historical validation dataset as well as the contemporary dataset. The final set of predictors was unique to each species; however lake surface area, and the climate index were significant in all models. In all three models, the climate index was the most influential variable. Cold-water lake whitefish and cool-water walleye are predicted to be lost from a large number of lakes (1283 and 1792, respectively) by 2041–2070 across central and southern Ontario. Nevertheless, numerous hospitable lakes will be available to both fishes in Ontario’s far north (OFN), where 108 lakes are predicted to be suitable for lake whitefish and 247 are predicted to be suitable for walleye. Thousands of lakes across Ontario are predicted to be hospitable for rock bass in the future, including 35 locations in OFN. Main Conclusions:Modified logistic regression models, incorporating a minimum of local environmental and climate information, show promise in predicting the future distributions of fishes in Ontario lakes. This is an important first step in building conservation capacity for freshwater fishes in northern regions, where conservation value is very high, but data are particularly limited.
Read full abstract