The control of community composition by distance, environment and history: a regional-scale study of the mountain grasslands of southern New Zealand

Aim To examine the roles of physical and biotic environment, distance, direction and dispersal in determining the composition of plant communities at a regional scale. Location Grassland and shrubland at 266 sites in the mountains of southern South Island, New Zealand. Methods Species abundances of vascular plants, bryophytes and lichens were measured by height-frequency. Predictors were soils, climate, mammal grazing, physical distance and direction between plots, and geological history represented by occurrence on the same or different mountain ranges. Relationships were evaluated with quadratic multiple regression and Mantel tests. Results Spatial autocorrelation was present for both vascular and lower plants. However, distance explained only a minor part of the variation. Distance relationships were anisotropic, with the vegetation differences being greatest in the direction of the prevailing south-westerly wind. Even using all the environmental information, much of the species composition remained unexplained. For species presence, distance alone explained 7% of the variation, environment alone 25% and both together 28%. The abundance of species was even less predictable: distance 2%, environment 11% and both 12%. With vascular species, climate-related factors contributed the most to prediction, especially altitude. Surprisingly, soil chemical factors were more important for lower plants that are not rooted in the substrate. Species likely to disperse further showed a tendency towards weaker distance differentiation. Vegetation composition seemed unrelated to contemporary grazing, although historical grazing/burning may have been causal. There was little evidence for a relationship of species composition with biogeographic/geological history. Main conclusions Difference in community composition in these grasslands is only weakly related to distance, and only half of the distance effect can be explained by the environmental factors measured. Dispersal and geological history do not seem to explain the discrepancy. Explicability of community differences was comparable to that in some previous case studies, although weaker than that in others, but all of the previous studies examined only part of the flora, such as 'abundant trees' or palms - none examined even all the vascular species. The species complement of a site seems more conservative than the abundance of the species, much of which was not explicable from current information and may reflect past management history.