The dynamics of five grasses and white clover in a simulated mosaic sward.

(1) Two types of simulated swards were constructed to investigate spatiotemporal changes and the importance of species interactions in creating and maintaining pattern among six grassland species: Lolium perenne, Agrostis capillaris, Holcus lanatus, Poa trivialis, Cynosurus cristatus and Trifolium repensv. (2) Neighbour effects were studied in mosaic swards consisting of twenty hexagonal patches. Each hexagonal patch was sown with a single species. A species never had itself as a neighbour. To study the colonization of empty space, each of six trapezia, surrounding an empty central hexagon, was sown with one of the same six species. After twenty-seven months, the swards were destructively harvested and the contents of each hexagon separated into species. (3) Each species had its own rate and scale of change. Trifoliurn was by far the most mobile species. Only 20%/o of its biomass was in native hexagons, i.e. in which it had been sown. The stoloniferous grasses Agrostis and Poa divided their biomass about equally between native hexagons and those in which they were not sown. The stoloniferous species were most effective in colonizing empty space. Cynosurus was the least mobile species with almost 90%/o of its biomass in native hexagons. (4) The hypothesis that the spread of species into adjacent hexagons is independent of the specific identity of neighbours was falsified. Evidence for cyclic regeneration on the scale of the patch as postulated by Watt was not found. In contrast, each species was invaded to the same relative extent by all the others. The decreasing order of preference of neighbours was established as: Trifolium > Poa > Lolium > Cyno,surus > Holcus > Agrostis. It is suggested that pattern in plant communities cannot be viewed on a single spatial or temporal scale.