Occupancy modeling of forest carnivores in Missouri

My research focused on patterns of distribution and co-occurrence of the carnivore guild in forested ecosystems of southern Missouri. Carnivores have the ability to structure and drive the dynamics of entire ecosystems, and these effects are disproportionate to their abundance on the landscape. I used infrared remote cameras, trackplate boxes, and scat transects to survey sites throughout the southern forested region of Missouri. My goals were to assess the validity of noninvasive techniques and use the ensuing survey data to document the distribution of forest carnivores, to analyze habitat associations and patterns of occurrence for forest carnivores, and to model carnivore species occupancy for the Ozarks region.;The study of carnivore communities across landscapes has become more feasible and economical with the advent of modern noninvasive technologies. To better understand the value and limitations of these methods and approaches I examined presence data collected non-invasively in 2005-06 from 53 forested sites. I estimated latency to detection (LTD) and probability of detection (POD) given presence for seven species, compared these metrics across detection devices, and evaluated the effects of altering the order of implementation of the method on detectability. I found that by combining methods a more accurate picture of the community composition is obtained, and that it is necessary to tailor survey methodology to target species over an extended period.;Relatively little is known about eastern spotted skunk ecology, a state endangered species in Missouri. Model results for it and the striped skunk revealed that forested habitat was a predictor of eastern spotted skunk occupancy, while urban features of a landscape were better predictors of striped skunk occupancy. The map for striped skunk revealed an association between areas of habitat transitions from forested to urban landcover type and a relatively higher occupancy rate across the region. One of the most significant findings was that incorporation of the other carnivore species into each species occupancy model (e.g. including striped skunk presence as a covariate in the spotted skunk models) resulted in an improved model. This suggests that there is some habitat component that is being accounted for indirectly by the incorporation of the other species, and may also imply the potential for competition between the two species.;The two most commonly detected mammalian mesopredators in my surveys were the opossum and raccoon. Raccoons were detected at 94% of sites and opossum at 89% of sites. Model averaging was also used for these two species due to weak resolution of a single important model set in occupancy model selection analyses. As was observed for the skunk species, model averaging revealed that by incorporating knowledge of the presence of one species (e.g. raccoon) into the modeling process for other species (e.g. opossum) the predictive ability of the resulting model is improved. Odds ratios revealed that the most important predictors of raccoon occupancy at a landscape level were knowledge of opossum presence (OR = 1.60) and distance to water (OR = 1.16). Core area index (OR = 4.37) and knowledge of raccoon presence (OR = 2.08) were the most important predictors of opossum occupancy.;In the past century the coyote has become an abundant apex predator in much of eastern North American temperate forest habitats. How coyotes may influence the distribution of other forest carnivores such as bobcat and gray fox is unclear. Again, due to a lack of a single important model set I used model averaging to predict occupancy of coyote, bobcat, and gray fox. The probability of actual occupancy (PAO) for coyotes was best predicted by measures of human disturbance and prey distribution. Bobcats were primarily influenced by coyote presence and prey availability. Gray foxes were the most heterogeneous of the species in their regional predicted distribution, and were best predicted by measures of coyote and bobcat presence as well as by landscape elements related to humans and settled areas. (Abstract shortened by UMI.)...