Does land-use affect bird abundance?
University of New Brunswick
Over the last 50 years many species have seen negative population trends and scientists have identified habitat loss, change, and fragmentation as possible causes. I investigated the ability of random forest and boosted trees models to predict bird abundances based on land-use throughout the United States. Bird abundance data was acquired from the North American Breed Bird Survey (BBS) and land-use data was gathered by the Multi-Resolution Land Characteristics (MRLC) consortium. The models had a mixed transferability between the training and test data with 14 of 21 models better fit the training data. The models were mixed in their ability to accurately predict abundances based on land-use compared to the mean; 14 of 21 models more accuracy predict abundance compared to the mean suggesting an effect of land-use on abundance. Six models were classified as “good,” 11 as “cautious,” and 4 as “poor.”