Open Collections

Stewart, Laura Nicole

University of British Columbia

Monitoring animals in northern Canada is especially difficult, and there is a need for cost-effective methods and sampling designs. Camera traps and autonomous recording units (ARUs) are both promising tools, but they are rarely combined. This thesis focuses on two areas of integration: sampling design and habitat modeling. Data were collected in Ts’udé Nilįné Tuyeta, an Indigenous and Territorial Protected Area in the Northwest Territories. Hierarchical sampling designs, where sensors are deployed in clusters, are common for ARUs but rarely used for camera traps. I evaluated a hierarchical design for camera traps using resampling to determine the sample size required to estimate detection rate, habitat associations, and species richness across our study area. I found that 4 cameras per cluster across 35 clusters were sufficient for most of the metrics tested, but up to 13 cameras per cluster were needed to precisely estimate detection rate of rarer species. Cluster-scale species richness was unreliable even with up to 18 cameras per cluster, suggesting richness is best estimated at a larger scale. In my third chapter, I evaluated the use of ARUs and camera traps to create habitat models for sandhill crane Antigone canadensis. I tested for differences in the spatial scale of inference by performing model selection using two buffer sizes, and whether a model that integrated both data types would improve predictive performance. I found that ARUs made the best inferences at the landscape scale, while cameras made the best inferences at the home range scale. The integrated model did show evidence for hierarchical habitat selection by cranes, but did not improve predictive performance. Overall, I found that each data source had merits of its own, but that, in this case, integration into one model was not an improvement over modeling data separately. These two data chapters showcase two benefits of pairing camera traps and ARUs: first, improving response variables by carefully considering sampling design; and second, improving our understanding of habitat selection by using data from both sensors. Efforts should continue to apply and test these and other methods for effective wildlife monitoring in Canada’s rapidly changing northern environments.

Text

2023-09-30

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/85575

Forestry

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/