Modeling River Otter (Lontra canadensis) Habitat Suitability using Citizen Science Data
Date
2025-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
The North American river otter (Lontra canadensis) has undergone population
decline and recovery due to historical overharvesting, habitat loss, and recent
reintroduction efforts. My study models river otter habitat suitability using Maxent in
four Mid-Atlantic states—Pennsylvania, Maryland, Delaware, and New Jersey—using
citizen science presence-only data from iNaturalist. I incorporated environmental
variables (wetlands, forests, and stream density) and anthropogenic factors (harvest
rates and human population density) to assess their relative contributions. The full
model, which included all five covariates, performed best (AUC = 0.812), with an
increasing harvest rate as the strongest predictor. Environmental-only (AUC = 0.731)
and anthropogenic-only (AUC = 0.785) models performed less effectively, supporting
the hypothesis that both types of variables improve predictive accuracy. Wetlands and
moderate human population density positively influenced habitat suitability, while
detectability biases affected remote or heavily urbanized areas. These results
demonstrate the value of combining citizen science data with spatial modeling to
assess habitat use for semi-aquatic mammals like the river otter in human-dominated
landscapes.