Predicting Spawning Habitat for Lake Whitefish Coregonus clupeaformis and Cisco Coregonus artedi in the Lake Erie and Lake Ontario Regions using Classification and Regression Tree (CART) and Random Forest Models

Lake Whitefish;cisco;Great Lakes;School for Environment and Sustainability

Schaefer, HannahRoseman, Edward ;

University of Michigan

关键词: Lake Whitefish;    cisco;    Great Lakes;    School for Environment and Sustainability;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/148646/HSchaefer_Thesis.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

The Great Lakes region was historically populated by many different coregonine species, but much of that diversity has been lost. In Lake Erie and Lake Ontario, both the Lake Whitefish Coregonus clupeaformis and Cisco Coregonus artedi occurred in high numbers before habitat degradation, overfishing, invasive species, and other factors caused significant declines in their populations. To highlight areas of potential restoration in this region, two predictive models incorporating spawning habitat variables of fetch distance, ice cover duration, date of ice onset, distance from tributaries, and substrate were developed. The Classification and Regression Tree (CART) model predicted spawning habitat for Lake Whitefish with 77% accuracy and with 55% accuracy for Cisco. The random forest model predicted Lake Whitefish spawning sites with a 72% accuracy rate and a 63% accuracy rate for Cisco. Variables found most important for Lake Whitefish in the CART model were a fetch distance less than 51km and an ice cover duration less than 77 days, whereas the random forest model found fetch distance and date of ice onset to be the most important variables. The most important variables in the Cisco CART model were an ice duration greater than 56 days, a hard or clay substrate, and a fetch distance less than 45km. Sand and mud substrates also predicted presence when ice duration was between 56 and 70 days and the date of first ice was on or before January 13. Important variables from the random forest model included ice cover duration and fetch distance. The importance of ice cover for predicting spawning habitat suggests that climate change may play a significant role in the future sustainability of these species in the lower Great Lakes. This study provides an insight to important variables that may be considered for future Lake Whitefish and Cisco restoration projects while also highlighting important regions in the lower lakes where these projects could occur.