Bark stripping damage by red deer (Cervus elaphus L.): assessing the spatial distribution on the stand level using generalised additive models

Bark stripping is a key topic in forestry because of economic losses due to associated fungal infections of wood, finally resulting in growth decrease and the loss of ecosystem services. Numerous studies identified factors influencing the spatial distribution of bark stripping damage between stands or at the landscape scale. However, patterns within single stands are not yet reported. In this research, we performed a terrestrial laser scanning supported census of nine stands in Austria (9026 trees in total). A generalised additive model with a binomial distribution (link = logit) and soap film smoother was fitted to the data. The probability of bark stripping on the single tree level depended on the following covariates: Spruce was more vulnerable than larch, damage probability decreased with DBH and the local slope and increased with the Epanechnikov Kernel (bandwidth = 15 m) estimate of tree density. At the nearest neighbour distance of two metres, there was a damage maximum. The spatial distribution of bark stripping damage was clumped, and its intensity decreased with increasing distance to forest roads. In 67.7% of the cases, the model predicted the right outcome for the total population (overall model accuracy). This percentage varied between 55.3 and 79.1% between stands. In conclusion, the spatial distribution should be considered in inventory designs for bark stripping damages to mitigate bark stripping effects on the forests.

Automated summary

Bark stripping is a significant issue in forestry as it leads to economic losses and fungal infections of wood, resulting in decreased growth and loss of ecosystem services. Previous studies have focused on factors influencing bark stripping damage at larger scales, but not within single stands. This research used terrestrial laser scanning to analyze nine stands in Austria and found that the probability of bark stripping at the individual tree level was influenced by tree species, tree size, slope, and tree density. The spatial distribution of bark stripping damage was clustered and intensity decreased with increasing distance from forest roads. Inventory designs should consider spatial distribution to mitigate bark stripping effects.