Evolutionary Ecology, 1995, 9, 304-317 Mechanisms of pollen deposition by insect pollinators WILLIAM F. MORRIS', MARC MANGEL and FREDERICK R. ADLER § Centerfor Population Biology, Universityof California, Davis, CA 95616, USA Summary Studies of pollen dispersal in insect-pollinated plants have often documented highly leptokurtic patterns of pollen deposition that can increase the likelihood of long-distance mating. To examine potential causes of highly leptokurtic deposition, we introduce four functions that arise when (1) the duration of pollinator visits to pollen sources is limited, (2) the rate of pollen deposition varies randomly among pollinators and/or among visits, (3) the rate of pollen deposition changes monotonically over time or (4) pollen is carried in layers or compartments on the pollinator's body that differ in deposition rate. Maximum likelihood techniques were used to fit deposition functions to data obtained from honey bees (Apis mellifera L.) visiting mustard plants (Brassica campestris L.) that contained a marker gene. Each of the alternative leptokurtic functions fit the experimental data better than a simple exponential function and the best-fit function predicted a mean pollen dispersal distance more than three times greater than the exponential. We argue that studies of pollen deposition need to test a broader range of deposition models to assess outcrossing distance in plant populations accurately. Keywords: Apis meUifera; Brassica campestris; isolation by distance; maximum likelihood; pollen deposition Introduction For insect-pollinated plants, the pattern in which pollen is deposited as a pollinator visits a sequence of flowers strongly influences pollen dispersal distance (Bateman, 1947; Levin and Kerster, 1974; Schaal, 1980). As a result, pollen deposition patterns have been quantified for a diverse array of plant-pollinator systems (Thomson and Plowright, 1980; Price and Waser, 1982; Waser and Price, 1982, 1983, 1984; Galen and Plowright, 1984; Campbell, 1985; Geber, 1985; Svensson, 1985; Thomson, 1986; Thomson et al., 1986; Waser, 1988; Roberston, 1992). A general feature that has emerged from these studies is that patterns of deposition rarely follow simple exponential curves (such curves would be expected if the fraction of pollen remaining on the pollinator that is deposited at each visit were constant). Rather, deposition patterns are characteristically more leptokurtic (Morris et al., 1994), that is, more pollen is deposited on early and late flowers in a visitation sequence and less is deposited on intermediate flowers, than a simple exponential curve fit to the same data would predict. Henceforth, we will refer to such patterns as 'highly leptokurtic'. Highly leptokurtic deposition may be evolutionarily significant, because it increases the opportunity for gene exchange between widely separated individuals and may alter both the potential for local genetic differentiation (Wright, 1969) and the speed with which novel alleles spread through plant populations. Given the evolutionary implications of highly leptokurtic pollen deposition, it is important that we examine the ecological mechanisms giving rise to it. In this paper, we explore four mechanisms that are likely to operate in many pollination systems. Specifically, we consider scenarios in which (1) the duration of pollinator visits to the pollen source is limited, (2) pollen Present address: Department of Zoology, Duke University, Box 90325, Durham, NC 27708--0325, USA. § Present address: Departments of Biology and Mathematics, Universityof Utah, Salt Lake City, UT 84112, USA. 0269-7653 © 1995 Chapman & Hall