Refining, testing and evaluating spatially explicit models for wind dispersed plants

Dispersal is the process by which plants expand their range and explore new habitats. When local habitats become inhospitable, dispersal ability becomes the key mechanism allowing species to evade extinction. Despite the efforts in obtaining empirical dispersal curves and developing sophisticated spatial models, the main issue that remains unresolved is that of scale. Although predictions at the local scale are better than those aiming to describe dispersal at greater distances, they remain too unrealistic to be used in subsequent models that govern growth, mortality and resource exploitation. My first chapter aims to improve predictions at the local scale by refining the parameters of a spatially explicit model. I determined the effect of substituting basal area for cone production as a proxy for seed output. The results showed that the r2 from the regression of predicted versus observed densities increased by 5% for seeds and 15% for seedling simulations. Next, I determined the effects of allowing the horizontal wind speeds to vary. The results showed that correlations of observed vs. predicted recruitment are a function of the assumed meteorological conditions used to drive them. My second chapter tested the ability of inverse modeling to predict recruitment both at the stand level and beyond. Using the maternally derived DNA from seed coats of the North American tree species Pinus strobus, I compared the most common approach (inverse modeling) with the newer but far more time-consuming method of using microsatellite markers. I showed that inverse modeling grossly underestimates seed dispersal potential in this species and thus caution against its continued use. With the aim to improve spatial models, this thesis would not be complete without an examination of the role of wind on seed abscission---the precursor to dispersal. Previous attempts to link the probability of abscission with meteorological phenomena were set within averaging times that exceeded the time frame of seed abscission (<1 second) by at least 15 times. Using 1-minute averaging times, I showed that seed release, for the wind dispersed tropical tree Ceiba aesculifolia, is proportional to the square of the horizontal wind speeds. Furthermore, the data showed that this relationship is highly time sensitive where a correlation is no longer evident at averaging times exceeding 25-minute intervals. This thesis is concluded by showing that updrafts are much more effective at causing seed release than all other wind directions (i.e. downdrafts and horizontal). What it can not show, however, is the frequency of upward abscission events within a forest environment and how these results can be implemented within spatially explicit models that predict the dispersal potential of seeds traveling horizontally or vertically. Indeed, this can be addressed in future work.