| Abstrakt: |
We directly observed the feeding behaviors of Yaku sika deer (Cervus nippon yakushimae), the smallest subspecies of Japanese sika deer, by following 6 wild, habituated, and individually identifiable animals in a natural warm‐temperate forest on the island of Yakushima, southern Japan, 2005–2006. Deer fed on various plant species and parts but predominantly on the fallen leaves, fruits, seeds, and flowers of woody plants (∼82% of food items) at 59 natural feeding plots. We analyzed 8 key plant constituents of eaten and uneaten items at the feeding plots, and examined the effects of each constituent on food selection by multivariate and univariate generalized linear mixed models. The multivariate analysis, which evaluated the effect of each plant constituent on food selection by controlling influences of the other constituents in the food items, showed that deer selected food items with higher contents of crude fat, crude protein, and total phenolics but with lower contents of lignin and condensed tannin from the available items at the feeding plots. Neither soluble nor structural carbohydrate, nor ash content affected selection of food items. It may be advantageous for these small ungulates with shorter gastrointestinal tracts to select foods that are relatively rich in fat because of the higher energy content of fat compared with carbohydrates. Furthermore, it appears that non‐tannin phenolics may have beneficial functions in this species. The univariate analyses showed different effects of crude protein and structural carbohydrate on food selection compared to those derived from the multivariate analysis. We demonstrate that accounting for influences of other plant constituents by using multivariate analyses is important to ensure that any effects of individual plant constituents are not overlooked or overstated. © 2018 The Wildlife Society. Wild deer selected food items with higher contents of crude fat, crude protein, and total phenolics and lower contents of lignin and condensed tannin at natural feeding plots in an evergreen broad‐leaved forest. To best understand the effect of each plant constituent on food selection by wild animals, direct observation of their feeding behaviors and multivariate approaches that can control for the influence of other co‐varying constituents in each potential food item should be considered to minimize biases. [ABSTRACT FROM AUTHOR] |
Plný text