29 January 2018
We evaluated if steroid hormone concentration profiles in hair collected from brown bears could be used to discriminate between immature and adult bears. Classification accuracy was excellent for male age classes. We also had good success with accurately classifying adult females, but poor accuracy with classifying immature females.
Although combining genetic and endocrine data from non-invasively collected hair samples has potential to improve the conservation of threatened mammals, few studies have evaluated this opportunity. In this study, we determined if steroid hormone (testosterone, progesterone, estradiol and cortisol) concentration profiles in 169 hair samples collected from free-ranging brown bears ( Ursus arctos) could be used to accurately discriminate between immature and adult bears within each sex. Because hair samples were acquired opportunistically, we also needed to establish if interactions between hormones and several non-hormone factors (ordinal day, year, contact method, study area) were associated with age class. For each sex, we first compared a suite of candidate models by Akaike Information Criteria model selection, using different adult-age thresholds (3, 4 and 5 years), to determine the most supported adult age. Because hair hormone levels better reflect the endocrine state at an earlier time, possibly during the previous year, then at the time of sampling, we re-analysed the data, excluding the records for bears at the adult-age threshold, to establish if classification accuracy improved. For both sexes, candidate models were most supported based on a 3-year-old adult-age threshold. Classification accuracy did not improve with the 3-year-old bear data excluded. Male age class was predicted with a high degree of accuracy (88.4%) based on the concomitant concentrations of all four hormones. Female age class was predicted with less accuracy (77.1%) based only on testosterone and cortisol. Accuracy was reduced for females, primarily because we had poor success in correctly classifying immature bears (60%) whereas classification success for adult females was similar to that for males (84.5%). Given the small and unbalanced sample used in this study, our findings should be viewed as preliminary, but they should also provide a basis for more comprehensive future studies.