The Park Grass Experiment (PGE) is the longestobserved set of experimental plant communities in existence. Although the gross composition of the vegetation was at equilibrium over the 60-yr period from 1920 to 1979, annual records show that individual species exhibited a range of dynamics. We tested two hypotheses to explain why some species initially increased and why subsequently some of these (the outbreak species) decreased again. The study was designed around eight phylogenetically independent contrasts (PICs), each containing related species with different dynamics. Our first hypothesis was that persistent increasers and outbreakers have higher intrinsic rates of natural increase than control species (species without trends), allowing them to spread when interspecific competition is reduced by drought. This was tested by measuring establishment and seed production of species in field experiments, with and without interspecific competition. Seed production in outbreak species responded more strongly to release from interspecific competition than it did in either of the other groups of species. Our second hypothesis was that outbreak species eventually declined because they lacked the genetic variation necessary to adapt to the novel habitats to which they had initially spread. We tested this by measuring mating systems and genetic diversity in persistent and outbreak species in the PGE. In seven out of seven PICs tested, the outbreak species was more selfing than its persistent relative. There was a significant positive correlation between outcrossing rate and gene diversity. These results support roles for both ecological and genetic traits in long-term dynamics.