Abstract 296 – Estimating bear density using non-invasive genetics: Importance of linking subsampling methods to modeling approaches

Nathan Hostetter, Caitlin K. Brett, Fabian Jimenez, Colleen Olfenbuttel, Daniel U.
Greene, Joseph D. Clark, Ben Augustine, Dana J. Morin

Non-invasive genetic sampling has revolutionized the collection of individual-level data to
investigate survival, abundance, and density of bear populations at large spatial scales. Non-
invasive genetic studies typically use mark-recapture study designs, where researchers place
multiple hair snares across a study area, then check and collect hair at regular intervals.
Subsequent genetic analysis of hair samples results in an encounter history for each individual
(i.e., where and when an individual was detected) that can be used for abundance and density
estimation. A large number of hair samples are often collected during these studies, which
creates a need to subsample and reduce the number hair samples for genotyping. Herein, we
investigate how different subsampling approaches lead to important modeling considerations
and inferences on density and spatial variation in density. Motivated by a large-scale study of
black bears in eastern North Carolina, USA, we use computer simulation studies to identify
conditions where subsampling methods can lead to biased density estimates and modeling
approaches to reduce those biases. Subsampling methods can include selecting a fixed number
of hair samples per site per occasion (e.g., 1 sample per site per week), random selection, and
selections weighted towards areas with greater samples; all of which can have a fixed total
number of samples for genotyping based on costs. Preliminary results indicate important
tradeoffs among subsampling methods that involve cost, statistical power, bias, and modeling
considerations. Importantly, selection of subsampling methods should consider study
objectives and inform the modelling approach. Results of this work provide important
information for researchers and managers interested in using non-invasive genetic sampling to
investigate landscape-scale abundance, density, and spatial variation in these processes.