The foregoing examples argue that we often will need to look beyond the strict ethological isolation paradigm in attempting to understand how animal pollinators contribute to angiosperm speciation. So long as the pollinator fauna contains even a minority of generalists, disruptive selection on floral phenotype in sympatry is likely to be weakened. If individual plant species are served by a diversity of pollinators, the distribution of pollinator phenotypes may be uni- rather than multi-modal at any site; sympatric divergence of phenotype is still theoretically possible (via an "adaptive dynamics" process; Dieckmann & Doebeli 1999), although this is unexplored to date. In allopatry or parapatry, relatively generalized pollination often may mean that divergent selection, if it occurs, stems from quantitative differences among populations in the relative abundances of different pollinators, rather than from qualitative turnover in pollinators. This view of selection by a pollinator fauna rather than a single pollinator species is not new (e.g., Grant & Grant 1965, pp 162-163), but has not been stressed sufficiently (Waser 1998; Dilley etal. 2000).
Opportunistic and generalized pollinators pose a more central challenge to the ethological isolation paradigm, however. Even if selection on floral phenotype is disruptive (in sympatry) or divergent (in parapatry), there may be enough pollinator infidelity to prevent strong reproductive isolation from this source alone. In this case, the pleiotropic connection between the evolution of floral diversity and of reproductive isolation is weakened. If this proposal is correct - if pollinators do not automatically provide strong or complete reproductive barriers via behavioral responses that by definition are extrinsic to the plants - then the focus in angiosperm speciation should expand to include more study of barriers that are clearly intrinsic to the plants. These barriers are those expressed in the success of parental crosses, and in the viability and fertility of hybrids produced from them. Several questions arise immediately: on what scale of physical, ecological, genetic, or taxonomic separation do such intrinsic reproductive barriers first arise within angiosperm species.? Are incipient intrinsic barriers elaborated eventually into complete reproductive barriers, and by what stages. Or is final reproductive isolation a mixture of ethological isolation and what I term intrinsic barriers, and if so in what proportions? What forms do intrinsic barriers take, and what does this imply about their genetic and ecological mechanisms? Finally, what role do animal pollinators play, however indirectly?
Insect biologists appear to be far ahead of botanists in answering such questions (Oliver 1972; Coyne & Orr 1997). Still, botanists can offer tentative answers, based on studies of crossing relationships that were popular until only a few decades ago (and should be revived). For example, Kruckeberg's (1957) studies of Streptanthus, Grant & Grant's (1960) of Gilia, and Vickery's (1978) of Mimulus, all indicate that intrinsic crossing barriers are scattered at all taxonomic levels both within single species and among species within a genus (see also Levin 1978). Conversely, morphologically and ecologically distinct species within an angiosperm genus often are interfertile (indeed, distinct genera are sometimes interfertile). Strong crossing barriers are sometimes found over scales of tens of km or less, which can be assumed to correspond to an early stage of genetic differentiation among populations (see also Waser et al. 2000). The nature of the barriers is variable, encompassing sterility or reduced productivity of the parental cross, inviability or reduced viability in Fi hybrids, sterility of the hybrids, and similar breakdown in later-generation hybrids (see also Levin 1978). In these regards partial barriers within species resemble stronger barriers among species or genera. Finally, the distribution of barriers, and their strength, appears somewhat haphazard within species, correlating weakly if at all with geographic, ecological, or phenotypic separation of populations, or other estimators of their degree of evolutionary differentiation such as subspecific status (e.g., Hughes & Vickery 1974, 1975). These observations are concordant with the view that intrinsic reproductive barriers are evolving partly independently of the pheno-typic traits, for example floral traits, by which angiosperm taxa are usually recognized.
The role of pollinators in this scenario of the evolution of reproductive isolation is much more passive than in the strict ethological isolation paradigm. The converse of predicted opportunism of a relatively unconstrained forager in a mixture of flower species is that this forager should not travel farther than necessary between food items, especially if travel is energetically costly. Flying pollinators endure high costs of travel and tend to move short distances between flowers (e.g., Pyke 1984; Wolf et al. 1989). This behavior sets up conditions of local mating and genetic isolation by distance which foster divergence of populations as a function of their physical separation, due either to adaptation to local environments or to random genetic drift. Crosses between such populations will ultimately begin to exhibit reproductive barriers (Hughes & Vickery 1974; Waser et al. 2000), and these barriers may be reinforced upon secondary contact (e.g., Paterniani 1969; Crosby 1970; Levin 1978; Waser & Price 1993) or may strengthen as a pleiotropic effect of further genetic divergence. Thus pollinators may indirectly facilitate microevolutionary differentiation and incipient reproductive isolation which eventually becomes elaborated into complete isolation between higher angiosperm taxa.
A very different role of animal pollinators must be given a brief mention. By their very act of generalization and opportunism in flower visits, these animals appear certain to contribute to another important source of genetic and phenotypic novelty in plants (see also Webber 1960). There is evidence both old and new for an important role of hybridization in angiosperm macroevolution (e.g., Lewis & Epling 1959; Riesberg 1995). By fostering hybridization, animal pollinators stand to play a central role in the "instantaneous" generation of species via allopolyploidy and homoploid hybrid speciation (see Arnold 1997). From this, and by causing gene exchange among taxa even when no new species arise, they should contribute centrally to patterns of reticulate evolution in angiosperms.
Was this article helpful?