Interspecific pollen loss by hummingbirds visiting flower mixtures: effects of floral architecture

Ecology, March, 1996 by Carolina Murcia

Likewise, even though some Palicourea stigmas received many pollen grains from intervening flowers (Table 5), usually numbers were insufficient to cover much of the stigmatic surface or to account for the results in Table 3A. In particular, few grains of Hansteinia or Dicliptera ever reached Palicourea recipients' stigmas, yet those two species had the strongest negative impact on pollen-tube numbers in the recipients' styles. In one case, the stigma of a recipient flower had nearly 69% of its surface covered by Satyria grains; in another case, 25% was covered by Besleria grains (Table 5). Percentages were usually far lower (on average, 3.3% of the stigmatic surface covered by Satyria grains if spread in a single layer, and [less than]1% covered by Besleria). As a further test for the role of "stigma clogging" (cf. Waser 1978b, Waser and Fugate 1986), we examined the relationship between the fraction of stigmatic surface covered by heterospecific (Besleria or Satyria) pollen and the number of Palicourea pollen tubes found in the style below. We restricted the analysis to those recipient flowers (first and second in the sequence) most likely to show a negative relationship if stigma clogging were indeed inhibiting pollen tube growth. In a General Linear Model analysis that factored out the identity of the intervening species (Besleria or Satyria) and recipient number (first or second in the sequence), there was no significant relationship between pollen tube numbers and the fraction of stigmatic surface obstructed ([F.sub.1,32] = 1.454, P = 0.237).

DISCUSSION

Common sense and parsimony are not always sufficient

Guilds of consumer species display mixtures of strong and weak competitive interactions (Connell 1983, Schoener 1983, Brown et al. 1986, Heske et al. 1994). The same should be true for guilds of plants sharing pollinators, among which the net result of pairwise interactions may range from strongly competitive to neutral or even facilitative (Waser and Real 1979, Rathcke 1983, Thomson 1983, Galen and Gregory 1989, McGuire and Armbruster 1991, McGuire 1993). The "Sexual Architecture Hypothesis" has provided a straightforward, parsimonious, and intuitively satisfying means of linking these variable proximate outcomes to interspecific variation in phenotypic traits. Increased divergence in the sexual architecture (position of anthers and stigmas) of plants sharing pollinators should reduce interspecific pollen transfer and lessen the negative consequences of flowering in mixtures (Feinsinger 1983, Waser 1983, Armbruster et al. 1994).

The Sexual Architecture Hypothesis has had special appeal to those working with hummingbirds. The hummingbird bill and head provide plants with an essentially linear "resource continuum" along which different species often segregate their pollen placement (e.g., Stiles 1975, 1979, Brown and Kodric-Brown 1979, Feinsinger et al. 1986, Murray et al. 1987; but see Kodric-Brown et al. 1984). Our detailed examination of hummingbird-pollinated plants, though, failed to corroborate the hypothesis. Among Palicourea "guildmates" that we examined, those expected to interfere the least with Palicourea pollen transfer, according to the Sexual Architecture Hypothesis, actually interfered more strongly than did other guildmates with architectures and pollen placements more similar to those of Palicourea. Furthermore, we found little evidence that interspecific pollen transfer to sexual parts per se, either the transfer of foreign pollen to Palicourea stigmas or transfer of Palicourea pollen to anthers and stigmas of intervening flowers, was biologically significant. Far too few foreign grains arrived to clog Palicourea recipients' stigmas, and few Palicourea grains were lost specifically to anthers and stigmas of intervening flowers.