Invasibility of alien Impatiens parviflora in temperate forest understories

• Boosted Regression Trees was used to model cover of invasive small balsam.
• Soil nutrients accounted for largest part of explained variance.
• A model with interactions explained twice as much variation as the additive model.
• Negative association between herb-layer species richness and small balsam was weak.

Closed-canopy forests had been assumed to be a highly resistant to plant invasions but recently it has been found that several invasive plants are also effectively adapted to shaded forest understories. One of them is small balsam, Impatiens parviflora, one of the most widespread invasive plants occurring in temperate and northern regions of Europe. Certain discrepancies in autecology of small balsam or in its interaction with native species in invaded communities have been published previously. These can be accounted for by habitat-related biases (e.g., restricted length of analysed environmental gradients), size of sampling units and geographical range of sampling. Therefore to achieve better insight into the success of small balsam, we used dataset of vegetation plots from various types of temperate forests in Western Carpathians, Slovakia. To identify principle environmental and vegetation-related determinants controlling cover values of small balsam in temperate forests, and to model small balsam’s response to particular gradients, we used Boosted Regression Trees (BRT), technique that can model complex curvilinear relationships and their interactions. While additive BRT-model accounted for 24.1% of variation in small balsam cover, model with up to five-way interactions explained 46.0%. Nutrients (Ellenberg indicator value) were the most influential predictor accounting for 21.6% of variance. We observed only weak negative association between herb-layer species richness and cover of small balsam. The highest cover of small balsam was observed at nutrient-richest sites, along with lowest summarised cover of other species in herb layer. Trade-offs between mean annual precipitation and nutrients were also observed for cover of small balsam; species cover was higher at nutrient-poorer sites that received lower mean annual precipitation, while opposite was true for nutrient rich sites. The present work is the first study showing the positive association between nutrients availability and small balsam cover in forest understories, at a large spatial scale.