Determinants of long-distance seed dispersal by wind in grasslands

Long-distance seed dispersal is an important topic in ecology, but notoriously difficult to quantify. Previous modeling approaches have failed to simulate long-distance dispersal, and it has remained unclear which mechanisms determine long-distance dispersal and what their relative importance is. We simulated wind dispersal of grassland plant seeds with four mechanistic models of increasing complexity and realism to assess which processes and which attributes of plants and their environment determine dispersal distances. We compared simulation results of the models to each other and to data from field dispersal experiments. The more complex and realistic models predicted short-distance dispersal more accurately and were the only models able to simulate long-distance dispersal. The model comparisons showed that autocorrelated turbulent fluctuations in vertical wind velocity are the key mechanism for long-distance dispersal. Seed dispersal distances are longest under high wind velocity conditions, when mechanically produced turbulent air movements are large. Under very low wind velocity conditions seeds are dispersed farther when there is more surface heating, but never as far as during strong wind events. Model sensitivity analyses showed that mean horizontal wind velocity, seed release height, and vegetation height are crucial determinants of dispersal potential and dispersal distances. Between plant species (but not within a species), seed terminal velocity is an additional important determinant of long-distance dispersal. These results imply that seed release height is the most important plant-controlled dispersal parameter for grassland plants, and that the structure of the local vegetation greatly affects dispersal distances. Thus, management plans for grasslands should take into account that changes in vegetation structure, e.g., due to eutrophication, can reduce the seed dispersal ability of wind-dispersed plant species.