Coping with chemical complexity in mammal-plant interactions: near-infrared spectroscopy as a predictor of Eucalyptus foliar nutrients and of the feeding rates of folivorous marsupials.

We investigated the utility of near-infrared reflectance spectroscopy (NIRS) as a means of rapidly assaying chemical constituents of Eucalyptus leaves and of directly predicting the intake of foliage from individual trees by greater gliders (Petauroides volans) and common ringtail possums (Pseudocheirus peregrinus). The concentrations of total nitrogen, neutral detergent fiber, condensed tannins and total phenolics could be predicted accurately by partial least squares regression models relating the near-infrared reflectance spectra of foliage samples to analyses performed using standard laboratory procedures. Coefficients of determination (r 2) for all four constituents ranged between 0.88 and 0.98, and standard errors of prediction between 0.80 mg g-1dry matter (DM) for total nitrogen and 5.14 quebracho equivalents g-1DM for condensed tannins. Near-infrared spectral-based models of food intake had r 2 values of 0.90 and 0.95 with a standard error of prediction of 3.4 and 8.3 g DM kg-0.75 day-1 for greater gliders and common ringtail possums respectively. We used the predictive model of food intake for greater gliders to examine the relationship between leaf palatability and documented food preferences of animals in the wild. Ranked differences in leaf palatability across four Eucalyptus species were consistent with documented food preferences of greater gliders in the wild. We conclude that NIRS provides a powerful tool to predict foraging behaviour of herbivores where forage choices are determined by compositional attributes of food.