Predicting trajectories of temperate forest understorey vegetation responses to global change

Predicting forest understorey community responses to global change and forest management is vital given the importance of the understorey for biodiversity conservation and forest functioning. Though substantial effort has gone into disentangling how global change will impact the understorey community, the scarcity of information on site-specific environmental drivers together with large temporal and spatial drivers has limited our understanding of how global change drivers affect understorey characteristics at specific forest sites. Here, using understorey resurvey data collected from 1363 plots across temperate Europe and applying a machine learning approach, we used Gradient Boosting Regression Models (GBM) to model and predict trajectories of four understorey characteristics (species richness, total understorey vegetation cover, proportion of woody species and proportion of forest specialists) to global-change and site-specific drivers (e.g. soil, overstory conditions). We applied the final GBM models to 8 forest sites in Austria to evaluate the effect of future scenarios for nitrogen deposition, climate change and forest management on the forest understory in the year 2030, and project the trajectory of understorey properties from year 1993 to 2030.  The trajectory results showed that increasing nitrogen deposition decreased species richness and proportion of woody species, but increased total understorey vegetation cover and proportion of forest specialists. The effect of climate warming on the proportion of forest specialists appeared to be limited but led to a decrease in species richness, total vegetation cover and proportion of woody species. Finally, a closed canopy could shift the community towards more woody species and forest specialists but may lower species richness and total vegetation cover. Our presented model allows the prediction of trajectories of understorey vegetation responses to global change and management interventions at specific forest sites.