A dispersal model for airborne pollen based on assumptions about wind
directionality, gravity, and a wind threshold at which pollen is taken
by the wind is developed, using a three dimensional diffusion approxi
mation. The bivariate probability distribution of pollen receipt by fl
owers at the same height as the pollen source is derived. Gravity, ver
tical random movements, and vegetation density turn out to have simila
r effects on this distribution. Maximum likelihood methods far estimat
ing the combined parameters from data with multiple point or continuou
s pollen sources, and one or move plant varieties, are developed. Usin
g an example data set from the literature, it is shown that our model
gives a better fit than more traditional descriptive dispersal models
of the form e(-arb). We also show that estimates of important properti
es of the dispersal distribution, such as the variances, become consid
erably smaller using our model than for the more traditional models. F
inally, we discuss potential extensions and evolutionary implications
of these types of models. (C) 1997 Academic Press.