Dario Fernandez-Bellon, fRI Research

A common approach to monitoring bear populations is the use of non-invasive DNA sampling through hair snags to estimate population sizes. Although successful in acquiring population size estimates, genetic capture-recapture lacks the demographic insights needed to identify drivers of population dynamics. Here we test whether it is possible to use grizzly bear (Ursus arctos) hair hormone concentrations to determine key demographic parameters, specifically sex, age class, and in the case of females whether they were pregnant, lactating, or accompanied by cubs. We measured hormone profiles (16 steroid and thyroid hormones relating to reproduction, stress, and nutrition) from 130 grizzly bear hair samples collected during live-capture events in Alberta, Canada, from 2008-2019, for which we were able to determine sex, age class, and measures of reproductive status in females. We used random forest models to predict demographic parameters based on different combinations of hormone values, with a predictive accuracy ranging from 53% to 94%. Our best performing models were those developed to predict sex (80% accuracy when applied to all bears and 94% accuracy when subset to adult bears). Age class models performed better on male bears (86% accuracy) than on female bears (73% accuracy). Our analyses of female reproductive status were constrained by sample size limitations but resulted in a predictive model able to determine whether an adult female was accompanied by cubs of the year with 66% accuracy. By testing this methodology on hair samples collected from captured grizzly bears where these demographic parameters were known, we demonstrate its potential applicability to non-invasive monitoring approaches. Pairing hair hormone concentration analysis with genetic capture-recapture surveys has the potential to provide multi-dimensional population data to wildlife managers, better informing evidence-based decisions.