Oh deer, what can the matter be? Evaluating deer impacts on forest regeneration in New York State

Background/Question/Methods

Deer (Cervidae) have been implicated as a major factor in suppressing forest regeneration across temperate regions of the globe.  Deer browse may alter forest composition by removing significant proportions of seedlings from the understory, thus changing successional pathways and ultimately canopy composition. However, the majority of studies have considered deer impacts on forest regeneration only over fine spatial scales and results vary substantially at broader spatial scales due to variation in habitat conditions and deer concentrations.  Determining how local effects scale up to broad spatial scales relevant to deer management is challenging due to the complexity and interactions of the many factors involved.

To tease apart deer effects over large spatial scales, we used the Forest Inventory and Analysis Database (FIA) to model forest regeneration data from 1652 plots across New York State. For ten selected tree species, we modeled seedling abundance as a function of deer browse pressure (DBP as indexed by deer harvest data), forest type, and canopy abundance of potential seed source trees along with relevant climate, stand level, and land cover variables. Models were compared based on DAIC scores. Final predictions were based on weighted averaged values from all supported models (DAIC <=2)

Results/Conclusions

All but two of the ten species had multiple well-supported models. Generally, supported models included climate variables (annual mean temperature and annual precipitation). For six of the ten species, seedling abundance decreased with increasing mean annual temperature, and five species also showed a negative response to increasing annual precipitation. For almost all tree species we observed a negative effect of the interaction of DBP with land cover variables on seedling abundance, specifically with respect to increasing forage area and edge density, and greater perimeter for forest, crop and developed areas within a 2-km radius of the FIA plot. This indicates that localized land cover patterns play a strong role in mediating how regional deer density estimates are impacting species abundance. Certain land-cover classes potentially attract more deer than others (i.e. forage land cover), while greater amounts of edge (perimeter) between land cover classes also appears to facilitate deer use, thus generating greater negative impacts on local seedling abundance.  Overall, these results demonstrate the importance of considering landscape context in conjunction with regional deer population estimates when assessing deer impacts on forests.