The origins and detection of plant community structure: Reproductive versus vegetative processes

The exchange of ideas and information between vegetation ecology and pollination ecology is relatively restricted, yet both fields have devised methods to detect the structure of species assemblages and communities. To promote the exchange of ideas between fields I compare approaches, concepts, and problems faced by researchers working in each area Both vegetative and reproductive interactions may generate assemblage structure through ecological sorting or through character displacement. Vegetative interactions may lead to assemblage organization more often by ecological sorting and reproductive interactions more often by character displacement. Vegetative interactions generally operate over shorter temporal and smaller spatial scales than reproductive interactions and may be affected more strongly by temporal and spatial heterogeneity in abiotic and biotic environments. These differences affect how the concept of ecological niche should be applied to plants. The Hutchinsonian concept of niche needs to be significantly modified before it can be usefully applied to plants. Null models are a valuable tool for investigating both vegetative and reproductive structuring of plant assemblages; however, the procedures followed in the application of null models need further refinement. The appropriate formulation of the null model may require information that is unavailable, hence multiple models may have to be employed to ''bracket'' conclusions. The literature on pollination community ecology demonstrates that difficult decisions must be made about the likely processes that have generated the structure being tested, the relevant definition of sympatry, how guild membership should be defined and employed, and what constraints should be incorporated into the null model to impose realism. Differences in these decisions will affect the outcome of the analysis. While top-down studies of pattern have numerous advantages, they usually cannot identify the process(es) that have generated the patterns. Bottom-up, experimental studies can be useful for identifying the processes, but they can rarely be used to assess the structure of an entire natural assemblage. The optimal approach to studying assemblage structure is to detect patterns with top-down analysis and use experiments to identify the processes that generate and maintain the patterns.