Biological processes controlling the assembly of plant communities on cliff faces
Cliffs of the Niagara Escarpment, Canada, possess a flora that is predictable in species composition along the entire escarpment, although they are surrounded by forest communities that are not so. The goal of this study was to examine why the flora of cliff communities is so predictable when the flora of surrounding communities is not. The approach was first to determine the species pool (seed rain) and then determine at which stage (seed dispersal, seed bank, germination and seedling establishment) species were filtered from the cliff-face community. To examine the seed rain and seed bank, samples were collected from the plateau, cliff face and talus communities at two sites. Discriminant analysis of the seed rain separated the two sites and there was also a slight separation of the three community types. When the seed bank composition was examined, community types were completely separated but the seed bank composition did not precisely reflect the composition of mature plant communities. To examine the fate of seeds after dispersal, seeds were placed on cliff face ledges and crevices and censused to determine duration time and germination rate. More seeds remained on ledges than in crevices. On ledges, duration time was longer for species with either heavy or small seeds, but on crevices this trend was reversed. Seeds of a variety of tree species were able to germinate on the cliff face. To examine seedling establishment, newly germinated seeds of seven tree species were planted on a cliff face. 'Thuja occidentalis ' typically had the highest survivorship although seedlings of most species survived into their third year. Seedling survival was associated with short shoot height and an extensive root system; however, a low proportion of biomass was allocated to roots. The assembly of cliff face communities is a complex process of continual mortality that occurs over a series of developmental stages. The dominance of 'T. occidentalis' was not predictable when individual stages were examined, but it was inevitable when the cascade of filters acting on each of the species was considered.