Abstract
Large herbivores may affect ecosystem processes and states, but such effects can be difficult to quantify, especially within multispecies assemblages. To better understand such processes and improve our predictive ability of systems undergoing change, herbivore diets can be studied using controlled feeding trials (or cafeteria tests). With some wildlife, such as large herbivores, it is impractical to empirically verify these findings, because it requires visually observing animals in forested environments, which can disturb them from their natural behaviors. Yet, in field-based cafeteria trials it is nearly impossible to differentiate selection between herbivore species that forage on similar plants and make very similar bite marks. However, during browsing ungulates leave saliva residue which includes some buccal cells and DNA that can be extracted for species identification. Here we used a newly developed eDNA-based method (biteDNA) to test the browsing preferences of four sympatric ungulate species in the wild. Overall, food preferences varied between species, but all species strongly preferred deciduous over coniferous species. Our method allows the study of plant-animal interactions in multispecies assemblages at very fine detail.