• Alvar;
  • Carousel model;
  • Facilitation;
  • Guild;
  • Niche limitation;
  • Öland;
  • Spatial dependence
  • Tutin et al. (1964–1980)

Abstract. In this study we make clear that the significance of deviations from an expected variance in species richness as demonstrated in alvar grassland, is a function of spatial pattern at the scale richness was measured, i.e. 10 cm2. If corrected for spatial dependence, more than half of the significant cases become nonsignificant. As regards cases of significant variance deficit, which has been interpreted as an indication of niche limitation, we suggest that there may be a simple physical limitation to the number of species on the scale of observation in the form of a low number of plant units which can find a place because of the modal plant size of the species involved. In most cases of significant variance deficit the modal size of the plant units involved was probably bigger than average.

Insofar as a significant variance deficit can be demonstrated, the resulting species richness distribution curve should be analyzed and we propose a skewness test, enabling us to differentiate between significantly left-skewed curves (‘niche limitation’), significantly right-skewed curves (‘niche facilitation’) and symmetrical curves. We present results obtained with the G-test, a log-likelihood ratio goodness of fit test. Only few cases of significantly left-skewed curves and a majority of symmetrical curves were found.

Attempts to demonstrate guild proportionality in grasslands suffer from the heterogeneity of usually distinguished guilds, such as annuals vs. perennials or graminoids vs. dicotyledons.

We observe that niche limitation may occur in alvar grassland, but indications for niche facilitation are stronger. Finally, we conclude that deviations in species richness variance are interesting indications of community structure, but only of spatial structure. Niche structure resulting from assembly rules should be investigated through experiments.