Plant diversity has been shown to significantly influence ecosystem functioning at both the species and genetic levels, but it remains unclear how these components of diversity interact to influence ecosystem properties in species-rich communities.  We created experimental microcosms using eight genotypes each of eight plant species clonally propagated from a 10x10 m area of ancient limestone grassland in Derbyshire, England.  Treatments consisted of factorially varied species diversity (SD; 1, 4, and 8 species) and genotypic diversity (GD; 1, 4, and 8 genotypes per species) grown in plots of heterogeneous soil depth and pH.  Our objectives were to quantify the relative contribution of GD and SD to the overall productivity and stability of the communities over three years, and to assess the nature and extent of interactions between GD and SD in this oligotrophic system of long-lived, outbreeding plants.
Results/Conclusions

Genotypes of most species exhibited considerable differences in performance over the three years of our study, despite all being derived from mature, proximate individuals in the field.  Effects of GD on overall productivity were highly dependent on SD and vice-versa.  Four-species communities were no more productive than monocultures of the dominant grass Festuca ovina, but Festuca communities planted with four genotypes were on average more productive than one- or four-species communities containing a single genotype.  Eight-species mixtures that included the robust herb Succisa pratensis were among the most productive communities overall, but were significantly more productive when planted with four genotypes, as opposed to one or eight genotypes.  GD effects were less significant in communities on shallow soils, except for eight-species mixtures where productivity increased by nearly 20% when planted with eight genotypes per species compared to one.  Performance of Festuca genotypes in genetic monocultures did not predict their performance in genetic mixtures, suggesting that simple dominance relationships related to genotype size could not explain GD effects.  Instead, genotype performance appeared to depend strongly on neighborhood composition, relating to both the specific and genetic identity of proximate individuals.  Overall, our results suggest that genetic diversity can influence ecosystem productivity by mediating interactions at the species level, and can significantly alter dominance relationships between species.  For communities dominated by perennial outbreeding plants where local GD is often substantial, GD has the potential to significantly alter the way that plant species regulate ecosystem functioning.