Traditional models of animal-mediated pollen dispersal (Bateman 1947) assume that the transported pollen available for pollination is completely mixed and experiences a fixed rate of decay. This model overestimates initial, and underestimates tail deposition probabilities. The high skew of empirical pollen dispersal functions coupled with the poor fit of traditional models may indicate that the amount of pollen "carried over" from one flower to the next may not be a constant fraction. The interactions between the pollinator, floral reproductive organs, and the transported pollen load may result in portions of the pollen load experiencing differential transport conditions. This may lead to highly variable donor and recipient specific deposition patterns. The high skew and systematic deviations from earlier pollen carryover models common to empirical pollen deposition data have led to the development of both modified single-geometric (Morris 1995), and double-geometric models for changing carryover (Harder & Wilson 1998). The double geometric form of the Harder & Wilson models directly allows for multiple pathways of pollen removal from the pollinator's body. This critical assumption allows for the potential effects of pollen donor morphology, pollinator identity and pollinator behavior on pollen dispersal to be tested empirically. As a preliminary step in examining these relationships in Mimulus ringens, pollen deposition was quantified for linear arrays of M. ringens visited by three sympatric species of Bombus. Fluorescent dye was readily detectable over an average of 17.4 flowers, with an observed maximum of 27 flowers receiving pollen in a given foraging sequence. Deposition patterns were highly variable with respect to pollen donor identity and pollinator species. The amount of initial pollen deposition is correlated with herkogamy (R=0.774). Also, the mean transport distance is correlated with initial pollen deposition (R=0.636). The double exponential regression explains the empirical pollen deposition function far better than the single geometric model.

Key words: mating systems, Mimulus ringens, pollen dispersal, pollination