COMPETITION AMONG PLANTS SHARING HUMMINGBIRD POLLINATORS - LABORATORY EXPERIMENTS ON A MECHANISM

At Monteverde, Costa Rica, the plants Besleria triflora (Gesneriaceae), Palicourea lasiorrachis, and Cephaelis elata (both Rubiaceae) share hummingbirds (Lampornis calolaema) as pollinators. To investigate the effect that intervening visits to Cephaelis flowers have on pollen transfer between plants of self-compatible Besleria or of self-incompatible Palicourea, we presented captive Lampornis with series of hand-held flowers: first, 2 or 15 flowers of either Besleria or Palicourea as pollen donors, followed by 0, 2, or 10 Cephaelis flowers, followed by 20 recipient flowers of the first species. In protandrous Besleria, we used male-phase flowers as donors and female-phase flowers as recipients. In distylous Palicourea, we used pin (long-styled) flowers as donors and thrum (short-styled) flowers as recipients. We assessed pollen transfer by counting pollen tubes in recipients' styles and, in Besleria, by counting pollen grains deposited on stigmas. Intervention of Cephaelis flowers strongly reduced pollen transfer among Palicourea flowers. Increasing the number of donor Palicourea flowers significantly increased the absolute amount of pollen transferred but did not overcome the negative effect of Cephaelis. Pollen transfer among Besleria flowers also declined with the intervention of Cephaelis flowers; the decline was less strongly significant than for Palicourea, perhaps because pollen loads on stigmas of some female Besleria flowers can contain self pollen left over from their male phase. Under controlled conditions, then, strong competition may exist where plants share pollinator individuals, a situation faced by the experimental species and many others at Monteverde. Other studies on these plants reveal that despite its demonstrability in the laboratory, however, competition among individual plants varies in its ability to explain density dependence in pollination among plant populations in the field. Furthermore, community-level patterns of coexistence among plant species at Monteverde feature few ''ghosts of competition past.'' The collective results demonstrate that studying the role of particular interactions in community structure must involve examinations of mechanisms at the level of individuals, processes at the population level, and patterns of coexistence rather than just one level of this hierarchy. Studying mechanisms alone could exaggerate their importance to community structure, whereas studying coexistence patterns alone could misjudge the strength of interactions among individuals.