Simulating plant species loss in experimental grassland plots generally results in decreased ecosystem functioning, in particular primary productivity. Various statistical-biological explanations have been put forward for this positive biodiversity effect. What was not studied in these experiments was if the artificial plant communities were stable with regard to species composition. For example, it could be that artificial communities lose species, which are not well integrated in the community matrix, by extinction. Another possibility is that artificial communities are unstable because they may be easily invaded by new species. Obviously, if artificial communities are unstable with regard to species composition, they may also be unstable with regard to functioning. To investigate these possibilities we analyzed species abundances and aboveground plant productivity in a large grassland biodiversity experiment in Jena, Germany, under continuous weeding, spontaneous invasion and deliberate seed addition.

Results/Conclusions

Under continuous weeding artificial communities develop typical log-normal rank-abundance distributions but few species go extinct and the positive biodiversity-productivity relationship is maintained. When the artificial communities were no longer weeded, they were rapidly invaded by new species which were functionally different from the resident species. This process was speeded up slightly by deliberate seed addition, indicating a moderate level of dispersal limitation of species from the regional pool. As a consequence of the non-random invasion process the re-assembled communities after invasion converged towards similar levels of species richness, high functional richness and evenness and more or less constant productivity.

One explanation for non-random invasion could have been that invading species used resources that were not consumed by resident species. In this case, community productivity should have increased after invasion. However, this was only observed sometimes. In other cases, invasion led to “equilibrium” communities with sub-maximal productivity. We therefore sought an alternative explanation: probably, invasion was often made possible because residents had accumulated pathogens but left pathogen-free niches for invaders. We suggest that both explanations contributed to the observed results and conclude that re-assembly of artificial communities into more natural ones is commonly driven by availability of resources and absence of pathogens for invaders.

As a consequence, stable species compositions and ecosystem functioning may be difficult to achieve in artificial communities unless invaders and pathogens are strictly controlled. Furthermore, if natural community composition is more strongly driven by pathogen than by resource niches of plant species, some expectations about biodiversity effects on ecosystem functioning must be revisited.