Year of Award


Document Type

Thesis - Campus Access Only

Degree Type

Master of Science (MS)

Degree Name

Systems Ecology

Department or School/College

Division of Biological Sciences

Committee Chair

Len Broberg

Commitee Members

Bob Crabtree, L. Scott Mills, L. Mark Elbroch


GPS clusters, prey selection, cougar, predation, individual specialization


University of Montana


The interactions between predators and prey are a fundamental component of ecology and have direct relevance to the management and conservation of ecosystems around the World. Advances in global positioning system (GPS) collar technology have enabled researchers to gain insight into predator behavior, identify predation events in the field, and also build predictive predation models. Using GPS data from 26 collared cougars across three study areas, I investigated: 1) the mechanisms driving individual specialization in cougars and, 2) the utility of cluster models to predict predation events within and across study systems. In addressing the former, I used a subset of data from only a single study area including 13 cougars. I identified one specialist individual (P06) as having a unique diet relative to the population resulting from the continued selection of beaver (Castor canadensis). P06 actively hunted beaver by selecting for streams and creeks within his home range disproportionality to their availability and also traveled significantly slower while within beaver habitat – indicative of the slow, stalk and ambush cougar hunting strategy. When predation by specialist cougars targets sensitive or rare species, targeted (rather than broad) management actions will be more effective in reducing unwanted predation on sensitive species. To address the utility of predictive models, I used logistic regression to discriminate between kill and no-kill GPS clusters and modeled the binary response as function of multiple spatiotemporal variables. I generated within study area estimates of predation using a top model selected from a candidate set using an information criterion (AIC), and estimated predation across study areas using simple models with only temporal variables. Within study area estimates of predation were ≥91% accurate, while across study area estimates averaged 81% (SD = 6%) accuracy. Cluster models serve as a valuable tool to estimate general predation within and across study areas, although there are a number of instances when their use is not recommended. When prey species of interest rare or endangered, occur near human activity, are relatively small, or have range overlaps with other similarly sized cougar prey, rigorous field efforts will be required to produce accurate estimates of predation.

This record is only available
to users affiliated with
the University of Montana.

Request Access



© Copyright 2014 Blake Lowrey