Evidence for constraint on species coexistence in vegetation of the Park Grass experiment

Repeated patterns, of a type that would be expected to result from limitations to species coexistence (i.e. assembly rules) were sought in the Park Grass experiment. This classical grassland experiment was sampled in two years, using replicated biomass samples. Variance in a number of measures was examined, and compared to the variance expected under appropriate null models, the latter based on assumptions of no interactions between species. In each case, an assembly rule would result in low variance. Examining variance in species richness between quadrats within a treatment, there was no indication of constraint on species co-occurrences; variance in richness was actually greater than expected under the null model, attributable to environmental variation or perhaps positive interactions between species. However, there was control on biomass, evidenced by variance in total biomass (i.e. over all species) within a treatment being significantly lower than expected under the null model. There was no indication of community structure based on guilds (i.e. functional types). Although there was in 1991 some, non-significant, indication of a constant proportion of species from the legume guild, there was no sign of such an effect in 1992. Searches for intrinsic guilds failed to converge. There was no indication at all of constancy in the proportional representation of guilds by biomass. Thus, there is good evidence for competitive control on plant growth, but none for control of species occurrences. There is no convincing evidence for guild structure in this community at the scale sampled. Possible conflict is discussed between the existence of evidence for temporal stability but the absence of evidence for spatial uniformity. It is concluded that most of the mechanisms proposed for temporal stability will not necessarily lead to control on spatial variation. For many mechanisms, this would depend on the spatial scale examined.