Date of Completion


Document Type

Honors College Thesis


Wildlife and Fisheries Biology

Thesis Type

Honors College

First Advisor

Dr. James Murdoch

Second Advisor

Dr. J. Ellen Marsden

Third Advisor

Dr. Allan Strong


Abundance, Canis latrans, Coyote, Champlain Valley, Density, Occupancy Modeling, Vermont


Predators such as, coyotes (Canis latrans), have profound effects on ecosystems. Coyotes are recent arrivals in the northeastern United States of America, and in Vermont their ecology remains poorly understood. Even basic population characteristics remain largely unknown. I used a Royle-Nichols Abundance-Induced Heterogeneity Model to estimate coyote site abundance in northwestern Vermont. The model was developed by averaging the outputs of supported candidate models of detection/non-detection data collected from 71 camera traps in 2008, 2011 and 2017. The averaged model included the null model and the following covariates: the proportion of water/wetland, agriculture, coniferous forest, deciduous forest, mixed forest, development, shrub/scrub and the mean bobcat habitat suitability within the radius of an average coyote home range of a site. The candidate model with the strongest empirical support was the null model, followed by the water/wetland model, but all the candidate models assessed had strong empirical support (Δ AIC < 2). The covariates water/wetland, agriculture, shrub/scrub and mixed forest had a positive effect on abundance, whereas the other covariates had a negative effect. Abundance ranged from 0.078 coyotes/km2 to 0.089 coyotes/km2 and was greatest in the western part of the study area. Using model outputs, I estimated abundance in the state Wildlife Management Units (WMUs) in the study area: B, G, I, F1 and F2. WMU B had the greatest abundance estimate (148 coyotes), while WMU I had the lowest (77 coyotes). Across all WMUs abundance was 457 coyotes. Abundance values predicted from the model were lower than expected based on the state’s abundance estimate. One advantage of the model approach is that it incorporated the influence of landscape variables on abundance and resulted in a measure of precision (SE) for each parameter. The model provides managers a means of understanding how coyote abundance varies with features of the environment.

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License.