Non-random recruitment of invader species in experimental grasslands

To assess potential effects of seed limitation, characteristics of invader species and characteristics of established plant communities on recruitment success, we conducted a split-plot experiment factorially combining three weeding treatments corresponding to increasing successional age (regular weeding < cessation of weeding after three years < never weeded since sowing) with two seed limitation treatments (control vs sowing a seed mixture of all experimental species = internal invaders) in experimental grasslands varying in species richness (1, 2, 4, 8, 16) and in functional group richness and composition (1, 2, 3, 4 functional groups: presence/absence of legumes x grasses x tall herbs x small herbs). Seed addition increased internal invader seedling densities and the probability of successful colonization per species. Legumes, tall herbs and species with large and long-lived seeds or a requirement for specific vectors for pollination and dispersal benefited from seed addition most. The number of successfully established internal invader species was highest in plots with low initial species richness and in the regularly weeded treatments and lowest in plots with high initial species richness and in the never weeded treatments, indicating decreased recruitment with increased successional age. Resident plant communities with legumes had mostly negative (legume and smalland tall-herb internal invaders, external invaders) or neutral (grass invaders) effects on seedling density and colonization probability of invader species whereas resident communities with grasses had positive effects on the colonization probability of invader species except grasses themselves. These results show that seed limitation, invader characteristics, and community characteristics all can affect recruitment success in predictable ways, suggesting non-random community assembly during succession starting from different initial species pools.