Fine-scale community structure of lawns

1. Sites on 12 lawns were sampled by a grid of contiguous quadrats, each 160 mm(2) or 360 mm(2). Six lawns were examined in Fiji and six in New Zealand, the latter including one natural 'lawn'. The results were examined for indication of niche limitation, i.e. a lower variance in the number of species per quadrat than that expected from the assumption of random species distribution. 2. A previous method, based on species frequencies over the whole site (designated 'site model'), indicated significant niche limitation at the smallest block size in all sites. It is suggested that this result could arise from micro-habitat differentiation if the pools of species occurring in different micro-habitats were similar in size--these can be termed 'limited species pools'. 3. A new method of testing for niche limitation is described - the 'patch model'--which overcomes the problem of limited species pools by using a null model that predicts the species richness of each quadrat from the local frequency of each species in a three-by-three cluster of quadrats centered on that quadrat. Using this method, some sites, but not all, no longer exhibited significant niche limitation. 4. Examination of the frequency distribution of quadrat richnesses reveals that the departure from the null model comprises a greater-than-expected frequency at or near the mode, with a deficit of both species-poor and species-rich quadrats, all features being significant for most sites examined, with no indication of skewing. 5. Species richness was significantly greater in New Zealand than in Fiji, both at the whole-site and at the individual-quadrat level. There was no indication of a difference in the degree of niche limitation between the countries.