Date Approved

4-2020

Graduate Degree Type

Thesis

Degree Name

Biology (M.S.)

Degree Program

Biology

First Advisor

Dr. Alexandra Locher

Second Advisor

Dr. Paul Keenlance

Third Advisor

Dr. Joseph Jacquot

Fourth Advisor

Mr. Robert Sanders

Academic Year

2019/2020

Abstract

To accurately manage for the sustainability of wildlife populations, managers must first determine how to locate individuals of a population with great consistency. Determining the efficiency of detection techniques is a challenge, especially when the study species are scarce or elusive. Such is the case with the American marten (Martes americana), a small mustelid found across the Northern United States and Canada. In the lower peninsula of Michigan, marten are considered a species of concern, and the full extent of their range remains unknown. My goal was to test the efficacy of motiontriggered cameras for detecting the presence of American marten in the Manistee National Forest. Marten were live trapped (n=9), fitted with radio collars, and tracked from May 2018 to September 2019 to determine the extent of their home ranges. I then set two, separate, motion-triggered cameras, facing bait sites, within each home range for four weeks to test the probability of detection. Some cameras were set near microsite features (course woody debris, trees with cavities/nests, brush piles), while some cameras were set in randomly generated locations within the home range. Cameras were moved and rebaited after each four-week period and were deployed for a calendar year to test for any differences in detection rates due to seasonality or camera placement. Despite a hypothetical expectation of an approximately 1.0 detection probability, actual probabilities were much lower (0.63), supporting the idea that non-detections should not always be associated with an animal’s absence. Additionally, I found marten were more likely to be detected during the winter (Oct-Mar – 0.68, Apr- Sept – 0.56), when cameras were located in the core of the individuals home range (in core – 0.66, outside core – 0.55), and when camera locations were selected based on microsite characteristics rather than randomly placed (selected – 0.69, random – 0.39). This study reveals that camera traps alone are not efficient enough to consistently confirm a marten’s presence, even when the cameras are placed in the home range. I recommend high camera densities (4 per 3.27 km²) and selective camera placement, as well as additional detection techniques such as track and scat surveys for increasing the overall likelihood of detecting marten presence. This research should be used to strengthen any inferences made using camera trap studies in the future and will ideally provide wildlife managers and biologists with a framework for better detecting American marten in Michigan’s lower peninsula.

Share

COinS