In a tropical cloud forest at Monteverde, Costa Rica, three understory shrub species overlap greatly in flowering seasons and share hummingbird pollinators (Lampornis calolaema). We conducted two field experiments to determine if a plant's pollination and subsequent reproductive output reflect local densities of conspecific and heterospecific flowers. We controlled floral composition in the neighborhoods surrounding 12 focal plants of self-compatible Besleria triflora (Gesneriaceae) and of self-incompatible Palicourea lasiorrachis (Rubiaceae), producing prescribed mixtures of flowers of the focal plants' conspecifics and of Cephaelis elata (Rubiaceae). Each experiment subjected each focal plant to four different treatments: a floral neighborhood of 10 conspecific flowers and 10 of Cephaelis within a radius of 20 m (Treatment A); 90:10 (B); 10:90 (C); and 50:50 (D). Focal plants themselves always had 10 open flowers. We assigned the sequence of treatments to each focal plant in a Latin square crossover design using three balanced blocks of four plants each. Dependent variables were: frequency of hummingbird probes per flower; pollen tubes in styles; and seed output per flower. From standard models of competition for pollination, we predicted that pollination levels and seed production would be ranked among treatments: B(90:10) > D(50:50) > A(10:10) > C(10:90). These overall predictions were supported for both Palicourea and Besleria, based on tests for directionality showing that the rank order among adjusted treatment means did not differ from the order predicted. Separate a priori treatment contrasts indicated that in both species increasing flower density above base levels (Treatment A, 10:10) by adding conspecific flowers alone (B, 90:10) or along with Cephaelis (D, 50:50) increased the rate at which hummingbirds visited focal plants' flowers. In Palicourea, both the number of pollen tubes in styles and seed output per flower increased in the same fashion. In Besleria, the only effect was increased pollination (but not seed output) from A (10:10) to B (90:10). In contrast, enriching the basic neighborhood (A, 10:10) with Cephaelis flowers (C, 10:90) did not alter frequencies of hummingbird probes to either species of focal plant. Heterospecific enrichment did, however, decrease both pollination levels and seed output in Palicourea, and seed output in Besleria. In general, altering the floral neighborhood affected Palicourea more strongly than Besleria. Apparently the intensity of density dependence, at least in terms of female reproductive success, varies among plant species sharing pollinators. Seed outputs in self-incompatible plants such as Palicourea may closely reflect flower densities of neighbors, but interspecific effects on fertilization in self-compatible plants such as Besleria may be density vague.
