In grassland communities plant species diversity is highest on nutrient-poor soils and biodiversity is progressively lost as nutrients (particularly N) increase. Genetic diversity is strongly correlated with mating system (because selfing lowers heterozygosity), but the direct relationship (if any) to ecological variables is unclear. We used the Park Grass Experiment (PGE), begun in 1856, to test three hypotheses about the relationship between genetic diversity in populations of the grass Anthoxanthum odoratum, resource supply, soil pH and the species diversity of plant communities. They were:

• The coexistence hypothesis: Genetic diversity within species favours coexistence among species and therefore species and genetic diversity are positively correlated;
  
• The niche breadth hypothesis: Interspecific competition constrains the niche breadth of individual species, thereby limiting genetic variation. Species diversity and genetic diversity should therefore be negatively correlated;
  
• The stress hypothesis: Genetic diversity is constrained by physiological stress (e.g. caused by nutrient limitation) and is not directly correlated with species diversity.
  

Results/Conclusions

We used AFLP markers to measure the genetic diversity of populations of A. odoratum growing in 10 plots that lie along nutrient and soil pH gradients in the PGE. Although species diversity is negatively correlated with nutrients along this gradient and is severly reduced by values of soil pH < 4.5, we found that genetic diversity in A. odoratum was highly significantly (P < 0.0001) correlated with the number of nutrients added (ranging from nil to three) to a plot and was not correlated with species diversity, soil pH or biomass of A. odoratum. These results allow us to reject the coexistence and niche breadth hypotheses and support the stress hypothesis.