In animal-pollinated plant species, behavior is known to affect pollen dispersal (Bateman, 1947; Wright, 1951; Levin, 1981; Handel, 1983b). These observations have led to the prediction that behavior can have important effects on the genetic structure of plant populations. This prediction has been validated indirectly both by broad surveys of many plant taxa pollinated by a variety of animals with differing foraging patterns (Hamrick et al., 1979; Loveless and Hamrick, 1984) and by surveys of foraging patterns and resulting pollen dispersal in individual plant taxa (Levin and Kerster, 1974; Levin, 1981; Handel, 1983b). Still needed are comparative analyses of closely related taxa in order to isolate and understand the effects of specific features of life histories upon such structure (Loveless and Hamrick, 1984). In this study of two species of closely related tropical herbs, we provide evidence that differences in hummingbird foraging behavior and associated gene flow generate differences in levels of inbreeding of the plant populations concerned. Razisea spicata and Hansteinia blepharorachis (Acanthaceae) are perennial understory herbs in the cloud forests of Monteverde, Costa Rica; the two taxa are probably congeneric (L. McDade, Duke Univ., pers. comm.), and we found naturally occurring hybrid populations at the study site. Both species grow in similar habitats and often grow in mixed stands at similar densities (see also Feinsinger et al., 1986). They have similar flower life spans (2-3 days), flowering seasons (Oct.-Apr.), ovules per flower (4), shapes and sizes of pollen grains (indistinguishable under light microscopy), seed dispersal (dry capsules which dehisce to release 1-4 similarly sized disc-shaped seeds), and general physiognomies. Neither species is apomictic. Both species can grow clonally. Based on growth of individual plants, interplant differences in flower color, and preliminary electrophoretic evidence of stems 3-8 m apart (N = 5 stems in Razisea and N = 3 stems in Hansteinia), individual clones appear to be mostly 15 m in diameter. They differ in the following features: mean corolla length (Razisea, 46 mm; Hansteinia, 22 mm) and pollen grains per flower (Razisea, 3,094.4 + 125.7, N = 11 flowers; Hansteinia, 5,696.3 + 306.3, N = 7). Control pollinations suggest that Razisea is more self-compatible than Hansteinia. In Razisea, 24/ 89 cross-pollinated flowers and 21/67 self-pollinated flowers developed into fruits; these proportions are not significantly different (2 x 2 contingency test, G = 0.14, P > 0.50). In Hansteinia the proportions were 29/83 for crossand 10/94 for self-pollinated flowers, a significant difference (G = 15.15, P < 0.001). Finally, the species differ in their pollinators. Hansteinia is pollinated primarily (85% of visits) by the hummingbird Lampornis calolaema; in this short-billed species, males are often territorial and females are generalist foragers (Feinsinger et al., 1986). Both sexes tend to feed at adjoining flowers in systematic fashion before moving to another patch (Stiles, 1985; Feinsinger et al., 1986). In contrast, Razisea is pollinated mostly (89% of visits) by the long-billed Phaethornis guy which shows no territoriality and feeds at a few flowers in a patch, even when many are available, before moving to another patch. Both our observations (Feinsinger et al., 1986) and those of Stiles (1981, 1985) indicate that Lampornis are relatively sedentary, whereas Phaethornis travel about a good deal, in classical traplining pollinator (Janzen, 1971) fashion. These differences in foraging behavior are expected to generate different mating patterns in the plant populations involved. Specifically, the effect of systematic foraging on flowers of adjacent individuals (nearestneighbor foraging) has been shown both experimentally and theoretically (Wright, 1951; Levin and Kerster, 1974; Turner et al., 1982; Handel, 1983a, 1983b; Wyatt, 1984) to produce significant inbreeding, homozygosity, and patchiness in plant populations, whereas long-distance foraging and resulting gene flow can be expected to produce greater levels of outcrossing in the populations concerned. For these reasons, we predicted that Hansteinia should show evidence of higher levels of inbreeding than Razisea. Because of the similarities noted above, Hansteinia and Razisea are very convenient for testing this prediction. The salient feature in which they differ is their corolla length and the associated pollinator. Consequently, any differences that may be detected in levels of inbreeding are most likely to reflect differences in gene flow resulting from behavior.
Read full abstract