Abstract
To sustainably manage forests, we must understand the effects of forestry practices on wildlife using a reliable index of their abundance. Here, we assessed the short-term responses of roe deer ( Capreolus pygargus ) and water deer ( Hydropotes inermis ) to two forestry practices – selective cutting and shelter-wood cutting – in a temperate deciduous forest in Pyeongchang, South Korea. From May to September 2016, we conducted repeated pellet-group surveys and recorded herbivory signs across one selective-cut stand, one shelter-wood stand, and one uncut control stand to compare deer relative abundance, habitat use, and feeding-site selection among treatments. To obtain reliable estimates of relative abundance, we evaluated the ability of zero-inflated models to account for faecal pellet disappearance and applied zero-inflated negative binomial models for further analysis. Relative abundance of the deer was consistent across the stands, whereas habitat use increased with overstory coverage. Feeding-site selection differed markedly among treatments, with the probability of feeding-site occurrence increasing from 6.6% in the control stand to 20.1% in the selective-cut stand and 36.6% in the shelter-wood stand. These findings demonstrate how short-term structural changes following harvesting affect deer abundance, habitat use, and feeding behaviour. Coupling pellet-group surveys with zero-inflated models provides a practical approach for accounting for pellet disappearance and improving inferences about herbivore responses to forestry practices.