SOURCES OF INACCURACY IN BLACK-TAILED DEER HERD COMPOSITION COUNTS

Herd composition counts to establish sex and age structure are commonl y used to assess population status for deer (Odocoileus spp.) and othe r large ungulates. Such counts are frequently biased, which compromise s their usefulness. We evaluated composition counts of black-tailed de er (O. hemionus columbianus) for biases by comparing known behavior of a sample of radio-telemetered animals with the outcome of systematic dawn and night spotlight herd composition counts. Three sources of err or (differential use of habitats, activity schedules, and deer behavio r relative to observers) varied (P less-than-or-equal-to 0.036) by die l period; only activity schedules did not vary (P greater-than-or-equa l-to 0.521) by sex and month. Variation in sources of error was comple x because of interactions between monthly and diel periods, month and sex, and diel period and sex. We used stepwise regression to test for the contribution of the 3 sources of error to numbers of deer observed and bias in the sex ratio from herd composition counts. Differential use of habitats accounted for most of the bias with numbers of deer ob served, and differential alarm behavior was the only variable that exp lained a significant amount of variation with sex ratio. Unless herd c omposition counts are standardized (by route, season, etc.) for deer p opulations in dense habitats, biases in demographic estimates may fluc tuate because errors are not systematic.