What happens to the sown species if a biodiversity experiment is not weeded?

Studies in experimental grasslands have extensively documented the effects of sown plant diversity on the colonization of new species, but the responses of the sown plant combinations themselves have rarely been investigated. We established experimental grasslands differing in species richness (1, 2, 4, 8, and 16) and functional group number and composition (1-4; legumes, grasses, small herbs, tall herbs), and we studied the changes in the abundance of sown species (residents) in both weeded and non-weeded subplots over a period of five years after sowing. The accumulation of new species through spontaneous colonization in the non-weeded treatment did not affect the number of resident species, but had increasingly negative effects over time on the cover of resident species and their aboveground biomass production at community level. Temporal stability of resident populations was lower and year-to-year changes in resident species composition were larger in non-weeded than in weeded subplots. Compositional dissimilarity between weeded and non-weeded treatments increased through time. These negative effects of the colonization of new species on the abundances and stability of resident populations depended on resident species identity and not on additional variation between different functional groups. The colonization of new species did not change the number of resident species emerging from seeds, but reduced seedling densities of residents. Colonization did not affect the structure of resident communities as measured by species evenness, functional trait diversity and mean trait values suggesting that colonization can destabilize the species composition of residents in terms of abundance while leaving them unchanged in terms of functional characteristics. Generally, negative impacts of colonizing species on residents which accelerated through time decreased with an increasing number of sown species. Sowing more diverse grassland mixtures increases their predictability in terms of ecosystem characteristics, which is important for ecological restoration and sustainable agriculture.