ANALYSIS OF DIGESTION DATA - APPARENT AND TRUE DIGESTIBILITIES OF FOODS EATEN BY BEAVERS

Digestibility data generally are analyzed as ratio variables, but to b e useful this method requires that animals consume large quantities of food. If little is eaten, as may occur with novel or unpalatable food s, apparent digestibility estimates can be seriously biased and statis tical analyses may suffer from very low power to detect digestibility differences among foods. Moreover, ratio variables are undesirable for statistical analyses. We suggest that for many wildlife studies, true digestibility estimated from regression analysis is preferable to app arent digestibility estimated from ratio variables. True and apparent digestibilities were determined from the same data set. The regression method for estimating true digestibility was more applicable to wildl ife species, provided greater statistical power, and avoided the unrel iability of ratio variables used to estimate apparent digestibility. T he advantages of the regression method are illustrated using data on d igestibility of three woody forage species eaten by beavers (Castor ca nadensis).