Fig. 2

Top factors determining the effect of alien efficient foragers (n = 21,600) on the persistence and density of native pollinators. In a, the alien’s mean Jaccardian index (i.e., diet overlap with native pollinators; x axis) and the connectivity of the most-generalist plant species (i.e., fraction of total native links connected to the most-generalist plant species; grayscale normalized as percentile) contributed 62% and 13%, respectively, to the CART model (five-folded R² = 0.89, see Supplementary Table 2) predicting the alien effect on native pollinator persistence, and were also highly significant factors in generalized linear models (Table 3). In b, the fraction of native plant species visited by the alien contributed 99% to the CART model (five-folded R² = 0.99, see Supplementary Table 2) predicting the alien effect on native pollinator density, and was also a significant factor in the generalized linear model (Table 3). Effects were calculated as the normalized difference of native persistence and density after and before the alien introduction. Zero and close-to-zero effects correspond to unsuccessful (1%) and naturalized (2%) efficient foragers. All effects are negative or zero