Abstract
Developing appropriate seed dispersal functions to quantify seed dispersal is one of the important themes in plant ecology, because the post-dispersal distributions of seeds translate spatial patterns of parent plants to offspring and result in significant ecological consequences. In this study, the best-fitting seed dispersal functions were identified for 4 tree species in the Kenting forest dynamics plot in Taiwan. Dispersal functions between animal- and wind-dispersed species were also compared. Seed rain data were collected weekly from 72 seed traps (0.5 m2) along 4 transects within a 10-ha permanent plot (400×250 m) in Kenting from August 2006 to April 2009. We chose 2 animal-dispersed species, and 2 wind-dispersed counterparts for this study. Applying the inversemodeling approach, we developed seed dispersal functions via maximum likelihood methods. Weibull, lognormal, and 2-dimensional t (2Dt) models with negative binomial errors were used. Akaike’s Information Criterion was used to determine the best-fitting model. The results indicated that the lognormal and 2Dt models were the best-fitting models for the studied species. The bestfitting models did not differ between the 2 dispersal modes since they were equally represented for animal- and wind-dispersed species. However, the level of goodness-of-fit was higher for winddispersed than animal-dispersed species. In the future, mechanistic models should incorporate the behavior of seed dispersers to improve model fitting.