Abstract
Local density and sexual composition are two aspects of floral neighborhoods thought to influence pollination and seed output of recipient plants. I characterized the floral neighborhood of 436 flowering ramets of Alstroemeria aurea, a southern Andean perennial, distributed among three sites. On each ramet, I measured total pollen receipt and seed output. The long-lived, bumblebee-pollinated flowers of A. aurea are synchronously protandrous with a given ramet being either all male or all female and thus incapable of self or geitonogamous pollination at the ramet level. Even though each ramet changes sex over time, A. aurea forms floral neighborhoods that remain stable with respect to density and sex ratio during the span of a focal ramet female phase. Contrary to expectation, under field conditions neither local density nor sexual identity explained significant amounts of variation in pollen receipt. Density of neighboring flowering ramets marginally affected pollen receipt in two of the three populations but in opposite directions. Despite the absence of strong effects of neighborhood sexual composition on pollen receipt, the sexual identity of neighbors affected seed output which suggests effects on the quality of pollination due to changes in patterns of pollen flow. I also compared pollen loads on the stigmas of artificially isolated ramets (6 m) with those on experimental focal ramets surrounded by six close neighbors (20 cm) that were either all male or all female. Here, pollen receipt by focal ramets in all-male neighborhoods was 1.3 times greater than in isolated ramets, and 3.8 times greater than in ramets in all-female neighborhoods. In these artificial neighborhoods, stigmatic pollen deposition increased significantly over time. In nature, rates of bumblebee visits were higher in female-biased (early-flowering) than in male-biased (late-flowering) co-occurring floral patches. Thus, spatio-temporal shifts in visitation frequencies associated with the sexual composition of floral neighborhoods might compensate for spatial variability in pollen availability within populations and explain the discrepancies between empirical and experimental results.