THE POPULATION-DYNAMIC FUNCTIONS OF SEED DISPERSAL

We summarize some of the population-dynamic consequences of the mosaic structure of plant populations for the evolution of seed dispersal. A fairly elaborated set of theoretical ideas exist regarding the evolut ion of dispersal and we have synthesized some of them in an attempt to make them more accessible to field ecologists. We consider the relati onship of these general theoretical ideas to our understanding of frui t and seed dispersal. We develop three related models to describe the similarities and differences in how dispersal functions for risk reduc tion (bet hedging), escaping the negative consequences of crowding, an d escaping high concentrations of relatives. We also briefly discuss d irected dispersal as a fourth population-dynamic aspect of dispersal. Dispersal can have a risk-reducing function only when there is global (metapopulation) temporal variance in success. Dispersal to escape the negative consequences of crowding requires only spatial and local tem poral environmental variation. Dispersal for escaping high concentrati ons of relatives requires no environmental variation, but does require genetic population structure. Directed dispersal, defined as non-rand om into particular patch types contingent on the expectation of local success, is always valuable when possible and represents an advantage independent the others which can occur with random dispersal. In an ef fort to accommodate for the differences between simple mathematical mo dels and the behavior of complex natural fruit and seed dispersal syst ems we have discussed the following issues: actual patterns of patch s tructure and dispersal distance; the implications of plant cosexuality , perenniality, and allocation costs of dispersal structures; and the impact of the detailed nature of density dependence, breeding systems, and genetic structure. We briefly compare the population-dynamic func tions of dispersal presented here with the widely cited functions of c olonization, escape, and directed dispersal. Finally, we suggest how t he theoretical models can be used with field data to estimate the fitn ess consequences of dispersal.