Oecologia (Berl.) 46, 68-74 (1980) Oecologia 9 by Springer-Verlag 1980 Pollen Carryover, Nectar Rewards, and Pollinator Behavior with Special Reference to Diervilla lonicera James D. Thomson and R.C. Plowright Department of Zoology, University of Toronto, Toronto, Ontario M5S 1A1, Canada Summary. Pollen carryover was measured in three species of bum- ble bee pollinated plants by counting the numbers of foreign grains applied to the stigmas of a series of flowers by bumble bees. Deposition declined with the number of flowers visited in a roughly exponential fashion; most grains were deposited on the first few flowers, but some grains went much farther, the maximum carry- over being 54 flowers. Variation in deposition was very high. In Diervilla lonicera, bees desposited significantly more grains on flowers which contained large amounts of nectar than on drained flowers. The implications are discussed in terms of plant strategies for optimizing pollination. Introduction The extent of pollen carryover affects important processes in the fertilization of entomophilous plants, e.g., neighborhood size and genetic isolation of strains (Levin and Kerster, 1967, 1968; Levin and Berube, 1972), assortative mating (Kiang, 1972; Levin, 1970), the relative effectiveness of different pollinators (Primack and Si- lander, 1975; Motten et al., 1979), competition between plants for pollinators (Levin and Anderson, 1970; Straw, 1972; Waser, 1978a, b; Thomson, 1978, 1980), optimal nectar secretion (Hart- ling, 1979; Hartling and Plowright, 1979), optimal inflorescence architecture (Pyke, 1978), optimal mate selection (Janzen, 1977), and, of course, seed set in many-flowered self incompatible plants (Hartling, 1979; Frankie et al., 1976) and the ratio of selfed to outcrossed seed in self compatible plants (Bateman, 1956; Hard- ing, 1977). It has hitherto usually been assumed that carryover is low, i.e., that pollen picked up at one plant gets no farther than the next plant (Levin and Kerster, 1969) or even the next flower (Frankie et al., 1976). However, the evidence for such as- sumptions is limited to a small number of studies, mostly by Levin and his colleagues. The recent empirical demonstration by Hartling (1979; Hartling and Plowright 1979) that pollen carryover is unexpectedly high in Trifolium pratense suggests that more cases should be examined before accepting low carryover as a general phenomenon. We present additional data for three plant species. While the functional form of pollen carryover is of most imme- diate interest, the flower-to-flower variation in grain deposition is also of practical and theoretical importance, as an indicator of pollination reliability. Furthermore, the existence of this varia- tion suggests hypotheses regarding the factors which cause more grains to be deposited in one visit than another. One such hypothe- sis, examined empirically below, is that a positive correlation exists between the volume of nectar in a flower and the number of grains placed on its stigma. Such a relationship would have obvi- ous implications for the evolution of reward rates of flowers, but to our knowledge has not been suggested before. Methods and Materials Three species of bumble bee-pollinated flowers were examined: Erythronium americanum, Clintonia borealis, (both Liliaceae), and Diervilla lonieera (Caprifoliaceae). All were locally common in areas near our study site in central New Brunswick. During May- July 1979, we collected foraging bumble bees and large numbers of flowers and buds, and brought them to the laboratory where the bees were refrigerated and the flowers held indoors in water pending use in pollen deposition trials, which were conducted outdoors, in screened enclosures. The exact procedures employed varied with plant species as follows: 1. In Erythronium americanum, we made use of a striking di- morphism in pollen color. Most populations had a majority of yellow-pollen forms with a minority of deep red-pollen individuals. The usefulness of this dimorphism is that red grains may be easily counted against the light background of the stigma. Thus, by introducing a single red flower into a bee's foraging sequence, and examining successively visited yellow flowers, the extent and functional form of pollen carryover can be assessed. For these experiments we used Bombus and Psithyrus queens which had been caught in areas dominated by yellow-pollen flowers. Bees were examined microscopically to insure that they were carrying no red pollen before their first trials. We placed the chilled bees on a bouquet of yellow-pollen flowers in the screened enclosure. Most began nectar feeding as soon as they had warmed sufficiently. Some did not forage, or foraged only briefly before taking flight and buzzing about the enclosure. These bees were allowed to fly to deplete their energy reserves, then captured, chilled, and again given the opportunity to forage. We found it easier to induce foraging by warming up chilled bees on the flowers than when already warmed bees were introduced. After a bee was warmed up enough to make short flights, we induced it to switch from the warming-up bouquet to a rectangular array of 50 experimental flowers in water vials, spaced ~ 8 cm apart. Forty-nine of these were yellow-pollen flowers, most of which had indehiscent anthers, although the corollas were fully 0029-8549/80/0046/0068/$01.40