New Zealand Journal of Ecology (1994) 18(1): 41- 50

Vegetation Texture as an Approach to Community Structure—Community-Level Convergence in a New Zealand Temperate Rain-Forest

Research Article
Benjamin Smith  
Simon H. Moore  
Philip B. Grove  
Neil S. Harris  
Samuel Mann  
J. Bastow Wilson  
  1. Botany Department, University of Otago, P.O. Box 56, Dunedin, New Zealand

Functional convergence of different communities in similar environments would be expected as an outcome of the operation of 'assembly rules'. At an ecological level, competitive exclusion would restrict the co-occurrence of species with similar niches. Repetition of competitive sorting on an evolutionary time scale might lead to character displacement. Either process would ultimately lead to species niches being more regularly arranged in ecological factor space than expected on a random basis, with the consequence that the niche structure of different communities in similar environments would converge. We assessed the applicability of this model of community structure by comparing vegetation between study sites spaced in altitude 20 m apart along a continuous gradient in South Westland low-altitude conifer/broad-leaved forest, with respect to seven variates of vegetation texture primarily concerning the morphology of the photosynthetic unit (PSU). We employed a null model that assigns observed species to sites at random, as would be expected in the absence of assembly rules for the communities, comparing observed variation in texture to variation under the null model to look for convergence or divergence and to determine statistical significance. Significant convergence between adjacent sites was found in all variates when species weighted either by percentage cover or cover rank were used to calculate site texture means, but convergence was less pronounced among groups of five or 10 consecutive sites. Significant divergence occurred at the five-site level (three variates) using cover rank as a weighting factor and at the two-, five- and 10-site levels (five variates) when no weighting factor was used. Overall, divergence was more pronounced among sets of sites spanning a wider range in altitude, which seemed consistent with the presence of an environmental gradient along the transect, although a DCA ordination of site floristics failed to reveal a simple altitudinal trend. This study is the first to seek community-level convergence within a local area and the first to find statistically significant convergence between vegetation patches.