BODY-MASS AND COMPOSITION INDEXES FOR FEMALE BARREN-GROUND CARIBOU

Monitoring body mass and composition is important for detecting declining vigor in caribou (Rangifer tarandus granti) populations. We developed regression models to predict body mass and composition of female barren-ground caribou, using easily measured variables, including those available under 4 constraints faced by managers: access to harvested carcasses but not to measurements that would destroy meat, no access to pre-dressing measurements of harvested carcasses, no access to specialized laboratory techniques, and access only to external measurements of live animals. We evaluated 38 continuous variables for predictability of body mass and composition. Heart, gastrocnemius, and liver masses were best (P = 0.001) all-seasons indices of body mass. Peroneus tertius mass correlated (P < 0.001) with body mass in June, September, and November. Predicted body mass generally did not differ (P > 0.01) from observed body mass for simple and multiple regression models. Gastrocnemius and kidney fat masses were best (P < 0.001) all-seasons indices of carcass plus viscera mass. Best (P < 0.001) monthly indices of carcass plus viscera mass were femur and peroneus tertius masses in March-April, peroneus tertius mass in June, kidney fat mass or kidney fat index in September, and heart and peroneus tertius masses in November. Water content in indicator muscles and left kidney fat index were the only all-seasons indices of percent body fat. Best (P < 0.001) monthly indices of percent body fat were femur marrow water and kidney fat mass in March-April, tibia marrow fat mass in June, gastrocnemius fat content and chest girth in September, and metatarsus marrow water in November. Marrow fat mass correlated (P < 0.001) linearly with percent body fat over all observed body fat values and was an alternative to percent marrow fat, which provided estimates of body fat in lean animals only. Single indices of percent body protein were found only in March-April. There were more indices of carcass plus viscera mass and percent body fat in September than in any other sampling month, and fewest in June. Superior (P < 0.001) correlations in multiple regression models of all body components justified use of combined indices, but use of meat-destroying pre-dressing, or complicated measurements were not always warranted. Exclusion of meat-destroying measurements from multiple regression models did not reduce (P < 0.001) correlations with body or carcass plus viscera masses appreciably, but reduced (P < 0.001) correlations with body fat in all seasons and with body protein in September and November. Further exclusion of pre-dressing measurements and those requiring sophisticated laboratory techniques reduced (P < 0.001) correlations with body mass and precluded use of multiple regression models in predicting body protein.