Understanding an animal’s response to its environment and predicting its behavior is a common goal of animal movement ecology and conservation. Advances in technology for studying animal movement have led to an increase in the quantities of movement data, and updated methods for analyzing this data are needed.
Classifying Animal Behavior States from Environmental Features
Supervised learning methods are used for looking at relationships between predictors such as environmental features and animal movements. Researchers reviewed four different supervised learning methods and used them to predict risk to bald eagles from colliding with wind turbines in various environments. The method XGBoost yielded the most accurate and efficient results, and the authors concluded that this method may work best for large data sets. If animal behaviors can be associated with landscape features, there is potential to identify landscapes where collision risk is more or less likely. This information can be used to guide wind turbine siting and wildlife conservation efforts.
Bergen, S., Huso, M.M., Duerr, A.E., Braham, M.A., Schmueker, S., Miller, T.A., and Katzner, T.E., 2022, A review of supervised learning methods for classifying animal behavioral states from environmental features: Methods in Ecology and Evolution, Online. https://doi.org/10.111/2041-210X.14019