| Abstract: | A pervasive problem in invasion ecology is the limited recovery of native communities following removal of invaders. Little evidence exists on the causes of variation in post-invasion recovery. In a 4-year experiment using 65 sets of matched plots, we imposed an invader removal treatment (with control) on heterogeneous grassland plots invaded or uninvaded by an aggressive recent arrival, Aegilops triuncialis (barb goatgrass). We tested the validity of plot matching using transplants and soil analyses. We analyzed the community-level correlates of invader impacts, removal treatment side effects, and community recovery, each defined in two ways: by compositional similarity to uninvaded plots, and by relative native species richness. Recovery of native species richness in invaded and treated plots was high (approaching 100 %) although recovery of composition was not high (median 71 % Bray–Curtis dissimilarity to uninvaded untreated plots). We measured resilience as the residuals of community recovery in models that controlled for invader impacts and removal treatment side effects. Compositional resilience was highest where the uninvaded communities had the least cover by other invaders in the same functional group as the focal invader. Richness resilience was highest where the uninvaded communities had the lowest native species richness. Our study suggests that the recovery of native species per se may be a more relevant goal than the recovery of the exact pre-invasion species composition of particular sites, particularly in cases where pre-invasion species composition included exotic species other than the focal invader. |
