Fullmer, Jacob

Are remote cameras and mark-recapture estimators a viable method to monitor mule deer populations?

Faculty Mentor: Dr. Randy Larsen, Department of Plant and Wildlife Sciences

Introduction

Current methods for estimating mule deer populations are costly and difficult to complete with accuracy. Biologists traditionally use ground or aerial surveys, model-based simulations, or a combination of these two methods to estimate population sizes. Remote cameras may provide a more efficient method to reliably and precisely estimate populations. A recent BYU study successfully used remote cameras to accurately estimate a low-density mule deer population, but the applicability of this method to higher-density populations remains unclear. We wanted to apply this new method in a high-density area to determine the practicality of using remote cameras to estimate mule deer populations at a broad scale.

We teamed up with wildlife managers from the Church-owned Deseret Land and Livestock (DLL), a 200,000 acre (80,937 Ha) ranch in north eastern Utah. Approximately 52% of the ranch is suitable habitat for mule deer and this species is of primary concern and focus for ranch managers. Current methods used on the ranch to generate population estimates for mule deer require about 80 field hours and 20 office hours. Thus, we also made a comparison of time associated with use of remote cameras to generate population estimates.

Methodology

DLL biologists deployed remote cameras at 13 locations over a 3-month time frame during the early summer of 2015. The cameras were set up at a salt/mineral supplement used at the ranch, and were monitored and collected at varying intervals by DLL biologists. The memory cards with pictures were then delivered to us for analysis. Remote cameras collected 121,618 images including 18,988 depicting mule deer.

For this study, we evaluated the male population using unique antler characteristics which allowed us to “mark” and “recapture” individuals in photos and use traditional mark-recapture estimators. We selected two weeks during the summer when DLL biologists deployed the highest number of cameras. We found usable images of male mule deer from eight of thirteen sites. Each individual identified was assigned a unique ID number and “marked” from the initial 4,840 photos taken July 1-7. Then, during the second week July 8-14, 3,752 pictures were processed to determine how many marked individuals were “recaptured”. We then estimated population size using a Lincoln Peterson mark-recapture estimator. We used published information from the Utah Division of Wildlife Resources regarding herd composition (buck: doe ratios) to then produce an estimate of the overall population size.

Results

We identified 77 individual male (buck) mule deer between the mark and recapture periods. Of these, 61 individuals were marked during the initial week, 40 of which were recaptured during the second week. We identified 16 new individuals in week 2 that were not present on images obtained during week 1. Lincoln Peterson estimates indicated there were 85 male mule deer in the capture area. After adjusting for herd composition and then applying estimated densities to the rest of the DLL ranch, we estimated 3,420 mule deer compared to previous estimates of between 2,329 and 3,571 mule deer.

Discussion

The size of wildlife populations including those of mule deer vary in both time and space. Moreover, population sizes are notoriously difficult to estimate accurately. Our approach using remote cameras and a traditional (Lincoln-Petersen) mark-recapture estimator provided some traction and produced reasonable estimates in line with count-based approaches used by biologists on the DLL Ranch. Use of remote cameras is applicable to high-density mule deer populations.

That said, we spent in excess of 120 hours sorting photos, conducting statistical analyses, etc. compared to an average of 100 hours for traditional counts. Thus, this method, which produced reasonably good estimates, may be more time intensive than others. This technique may be most practical in remote areas that are difficult to access or for low-density populations. Traditional counts are difficult in areas of low density where limited detection results in low accuracy associated with estimates of population size. Remote cameras combined with mark-recapture estimators can overcome these difficulties of detection in these settings.

Conclusion

Our mark-recapture analysis results were reasonably accurate and precise. The application of remote cameras and mark-recapture techniques to estimate population size in a high-density mule deer population proved possible, although time consuming. Compared to the current method employed on the ranch, our approach required more time. Managers will need to weigh the increased precision against the time associated with use of remote cameras and mark-recapture estimators.