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

Ecology, March, 1996 by Carolina Murcia

INTRODUCTION

An animal pollinator that visits flowers of two or more plant species during a single foraging bout may lose some pollen grains of one species while visiting another. Such "interspecific pollen transfer" (Waser 1978b) may reduce the reproductive success of the plants involved (e.g., Waser 1978a, Galen and Gregory 1989, Armbruster and McGuire 1991, Feinsinger et al. 1991, McGuire and Armbruster 1991, McGuire 1993). For example, heterospecific grains arriving on stigmatic surfaces might interfere chemically or physically with the germination of conspecific grains (Waser 1978b, Sukhada and Jayachandra 1980, Kohn and Waser 1985, Ganeshaiah et al. 1986, Waser and Fugate 1986). Alternatively, when some pollen grains are lost to other species en route, fewer are available to fertilize conspecific ovules, such that donors' chances to sire seeds and recipients' chances to set seeds are both reduced (Waser 1978b; see, e.g., Campbell 1985a, b, Campbell and Motten 1985, Feinsinger and Tiebout 1991).

Floral architecture of the plants concerned should bear on the extent of interspecific pollen transfer. If pollen is deposited at distinct species-specific sites on animal visitors, interspecific pollen transfer is expected to be insignificant (cf. Feiusinger 1983, Waser 1983, Feinsinger et al. 1986). Presumably, the site where pollen is placed on and picked up from animals is dictated by the architecture of a flower's reproductive structures, in particular, the exact location of anthers and stigma. It follows that, in flower mixtures, plant species with divergent sexual architectures are expected to experience less interspecific pollen transfer, and fewer obtrusions on male or female components of reproductive success, than plants with similar sexual architectures. By this reasoning, which we term the "Sexual Architecture Hypothesis," natural selection and ecological sorting should both favor divergence in the architecture of reproductive parts (locations of anthers and stigmas) among closely related and sympatric plants, as an isolating mechanism, or among any plants commonly growing in mixtures and sharing pollinators, as a competition-reducing mechanism (Feinsinger 1983, 1987, Waser 1983, Armbruster 1985, 1986, Armbruster et al. 1994; cf. Howell 1977). Indeed, the Sexual Architecture Hypothesis in various guises is central to theories on floral evolution and the adaptive radiation of animal-pollinated angiosperms (cf. Grant 1949, Stebbins 1970, 1974, Levin 1978).

Validation of the Sexual Architecture Hypothesis requires data that relate interspecific pollen transfer to interspecific variation in sexual architecture. Previously, empirical studies have shown that intraspecific variation in sexual architecture can affect pollen loads on animal visitors and, hence, pollen transfer among con-specific plants (e.g., Lertzman and Gass 1983, Waser and Price 1984, Murcia 1990, Campbell et al. 1991, 1994, Podolsky 1992, 1993). To date, however, this approach has not been widely applied at the interspecific level.

Tropical hummingbird-pollinated plants provide an excellent testing ground for the Sexual Architecture Hypothesis and related questions. Individual hummingbirds often encounter and forage quite indiscriminately among mixtures of several flowering species (Stiles 1975, Feinsinger 1983, Feinsinger et al. 1986). As a result, birds frequently carry pollen of several plant taxa (Stiles 1975, Kodric-Brown et al. 1984, Feinsinger et al. 1986). Stigmas of the flowers involved often receive heterospecific grains (Kodric-Brown et al. 1984, Feinsinger et al. 1986, 1987), and fertilization rates of flowers in mixed-species arrays sometimes decline relative to those of flowers in pure-species clumps (Feinsinger et al. 1986, 1991). Nevertheless, stamen and style lengths can vary widely among the plants concerned (Feinsinger et al. 1986, Murray et al. 1987, although see Kodric-Brown et al. 1984), and sometimes the different pollen-grain species are spatially segregated on the bird (Stiles 1975, 1979; K. G. Murray and P. Feinsinger, unpublished data).

In cloud forest at Monteverde, Costa Rica, the shrub Palicourea lasiorrachis (Rubiaceae) shares its short-billed hummingbird pollinators with many other plant species displaying a great variety of floral architectures (Feinsinger et el. 1986, 1987, Murray et al. 1987). Through laboratory experiments, we examined how five of these species influenced pollen transfer between Palicourea flowers, also detailing various steps along the way: where donor Palicourea flowers deposited pollen on birds, how visits to other flower species altered that deposition pattern, what role the other species' anthers and stigmas played in removing Palicoutea pollen from birds, and how much heterospecific pollen landed on stigmas of Palicourea recipients. Among other goals, we set out to evaluate the Sexual Architecture Hypothesis with the following prediction: flower species most similar to Palicourea in anther and stigma locations and in sites of pollen placement on hummingbirds will interfere the most with loads of Palicourea pollen already on hummingbirds and with subsequent pollen transfer to recipient Palicourea flowers.