The phlox family (Polemoniaceae) takes center stage in the ethological isolation paradigm. Grant & Grant (1965) argued that specialization to different pollinators has been critical to adaptive radiation within the family, by leading to ethological isolation. In addition these authors argued that divergence in floral phenotype would affect mechanical isolation (placing of pollen on separate parts of the same pollinator's body), the second facet of Grant's (1949, 1952, 1994) "floral isolation".
Grant proposed the congeners Ipomopsis aggregata and I. tenuituba as a good example of floral isolation. These species occupy broadly overlapping geographic ranges in the western USA. Although they often exhibit different habitat affinities, situations of contact or near contact (i.e., parental populations growing within a few km of each other) are not uncommon. In some but not all of these contact situations one finds obvious hybrids (Grant & Wilken 1988). Normally, hybridization is considered to be precluded or limited by floral isolation (along with habitat differences), in the form of predominant hummingbird pollination of the red, trumpet-shaped tubular flowers of I. aggregata and predominant hawkmoth pollination of the longer, narrower, pale flowers of I. tenuituba (Grant & Grant 1965; Grant 1994).
My colleagues and I are studying a hybrid zone in Colorado, and find a more complex role of pollinators. At our site, parental populations of I. aggregata and I. tenuituba are separated by about 3 km and 300 m elevation, with hybrids in between displaying clinal variation in floral phenotypes. The predominant pollinators are hummingbirds, which select for shorter, wider, more darkly pigmented flowers (Campbell et al. 1997). In experimental mixtures, hummingbirds undervisit but do not absolutely shun I. tenuituba, relative to I. aggregata and hybrids. Further experiments (Melendez-Ackerman et al. 1997) show that this discrimination stems mostly from inferior nectar rewards of the former relative to the latter types of flowers. In other words, hummingbirds are making an "economic" choice as predicted by foraging theory. The birds exhibit no detectable flower constancy. Aviary experiments indicate that mechanical isolation is weak; i.e., hummingbirds transfer substantial pollen between flowers of the two parental species, and the presence of hybrids facilitates this gene flow (Campbell et al. 1998). Finally, we have found that hawk-moths are rare; when present they visit all flower types, but are most common in populations toward the I. tenuituba side of the hybrid zone.
When hawkmoths are present, they and the hummingbirds select different floral phenotypes that most resemble the parent species (Campbell et al. 1997).
In years when hawkmoths visit, then, pollinators appear to cause disruptive selection on floral phenotype in sympatry, and divergent selection in parapatry. But there is a major caveat: hawkmoths are absent or very rare in most years. This situation may conceivably be ancestral, or may be a result of recent anthropogenic change in the western USA. Whatever the explanation, the present selection regime exerted by pollinators in most years strongly favors floral phenotypes that resemble those of I. aggregata, and does so throughout the hybrid zone (Campbell et al. 1997).
In terms of pollinator-mediated gene flow, furthermore, the Ipomopsis system fails to conform to a strict construction of the ethological isolation paradigm. Although hummingbirds and hawkmoths each prefer different floral phenotypes, the preferences are far from absolute. Hence both pollinators affect some genetic connection between the parental plant species, rather than isolating them completely. Independent genetic evidence agrees (Wolf et al. 1991, 1993; Wolf & Soltis 1992) in suggesting past gene flow between these taxa in various locations (and between other species in the genus as well), and also in suggesting multiple origins ofI. tenuituba, perhaps from surrounding I. aggregata populations (i.e., contact situations just as likely represent primary divergence as they do secondary contact; Wolf et al. 1997).
Notice that this turns the classical question on its head. Rather than ask, "How does hybridization occur given the barriers imposed by pollinators.?", we instead might profit from asking, "What keeps hybridization from happening more often, given overlapping pollinator preferences?" (Leebens-Mack & Milligan 1998 raise the same issue). In the Ipomopsis system we are just beginning to explore this latter question. For example, Alarcon & Campbell (2000) have shown that competitive superiority ofconspecific pollen does not occur in our hybrid zone. But such a block to hybridization might occur in other Ipomopsis contact situations as it does in other plant species (e.g., Arnold et al. 1993; Riesberg et al. 1995); indeed there is evidence for strong reproductive barriers, based in part on conspecific pollen precedence, in a contact situation between I. aggregata and another species, I. arizonica (Wolf et al. 2001). Similarly, known hybrid individuals grown from seeds transplanted across our hybrid zone survive well on average (Campbell & Waser 2001). But this does not preclude the possibility of low or zero hybrid viability in other contact situations.
Unfortunately we have few detailed studies in hand of hybrid zones or other situations in which evidence for the ethological isolation paradigm might be obtained. However, those few studies now in hand do suggest that the Ipomopsis system is representative of many pollination systems. For example, Werth (1955), whose title is quoted at the beginning of this chapter, showed that the behavior of a few generalist insects suffices to connect several pairs of related species in the German flora, and so answered his own question in the negative. Leebens-Mack & Milligan (1998) found that bumble bees and carpenter bees exhibited preference and constancy in experimental mixtures of two species of Baptisia and their hybrids, but these insects still caused substantial gene flow, which was enhanced by the presence of hybrids. Galen (1996) showed that bumble bees select for large flowers of Polemonium viscosum, resembling the phenotypes at high elevation where they are the main pollinators. At lower elevations both flies and bumble bees pollinate, and flowers are smaller, suggesting possible disruptive selection in parapatry, for which there is some evidence (Galen et al. 1987). However, visits by bumble bees at all sites suggest the likelihood of substantial gene flow. Goulson & Jerrim (1997) used allozymes, pollinator observations, and the movement of fluorescent dye powders to determine that floral isolation between two species of Silene in Britain is insufficient to prevent hybridization, and that hybrid individuals serve as a bridge for gene flow. Similarly, Wesselingh & Arnold (2000) found that neither hummingbirds nor bumble bees were strongly specific to different parental phenotypes in mixtures of two Iris species and their hybrids. Bradshaw et al. (1995) conversely implied that differences in flower color and morphology would confer virtually total reproductive isolation on two interfertile species of monkeyflower (Mimulus) via distinct preferences of bumble bee and hummingbird pollinators. However, Hiesey et al. (1971) and Sutherland & Vickery (1993) had earlier shown that floral traits of monkeyflowers in no case appear to erect absolute barriers to either bumble bees or hummingbirds. Recent experiments by Schemske & Bradshaw (1999) with mixtures of parental species and hybrids do indicate that bumble bees and hummingbirds prefer different trait expressions, but these preferences are not absolute, in keeping with the previous findings. Hence some ethological isolation exists, but it does not seem to be enough to explain the absence of observed hybrids between the parental species.
These examples are not meant to suggest that the ethological isolation paradigm will never apply. For example, Fulton & Hodges (1999) reported that hummingbirds show strong fidelity to Aquilegiaformosa, and hawk-moths show complete fidelity to A. pubescens, in experimental arrays of the two species. Our experience with Ipomopsis suggests that these studies should be extended across years and sites to determine how consistent and complete the apparent ethological isolation is, and to carefully examine the role of other flower visitors such as bees and flies. Furthermore it would be interesting to know how pollinators respond to arrays that include hybrids between the two Aquilegia species, because this presumably mimics an initial stage of evolution within a single ancestral population (whereas the use only of parental species in arrays may mimic a situation of secondary contact after differentiation in allopatry). In this regard, the experiments of A. Ippolito & T. Holtsford (1999; personal communication), which yield preliminary evidence for distinct specialization by hummingbirds vs. hawkmoths within single hybrid populations of Neotropical Nicotiana, may be the best example of pollinators imposing virtually complete ethological isolation within a unimodal set of floral phenotypes.
Was this article helpful?