Partitioning between pollinator taxa
Innate preferences by different pollinator taxa may produce consistent selection on floral traits. Selection may shape suites of characters that work together (such as flower shape, size, color, scent, and timing of bloom) to attract and reward certain pollinator types as the interaction becomes more specialized. Such coordination among traits may make it difficult to switch to a different type of pollinator, as a mutant in one floral character probably would not do well in the context of the other characters, especially in the presence of the original pollinators. Thus the traditional scenario of floral divergence being driven by adaptation to different pollinators probably requires geographic isolation between plant populations served by different pollinators (Wilson & Thomson 1996; Chittka et al. 1999). However, if secondary contact occurs following allopatric divergence, and both pollinator types are present in the zone of contact, then assortative mating via partitioning of the incipient plant species between pollinator taxa may be important for the maintenance of the two distinct lineages (e.g., Fulton & Hodges 1999; Schemske & Bradshaw 1999). If there is no selection against hybrids, such assortative mating probably will not suffice to prevent hybridization and introgres-sion, as it is unlikely that pollinators will be completely constant. The more divergence there has been between populations, the greater the probability that there will be selection against hybrids, through reduced viability, pollinator attraction, or fecundity. This scenario resembles an advanced case of divergence in sympatry, as outlined earlier, in that there is already a strong genetic correlation between mating characters and traits responsible for low hybrid fitness. The combination of the genetic correlation, assortative mating, and selection against hybrids may suffice to keep the divergent lineages genetically distinct. Modeling of such situations is in order, to determine the strength and consistency of selection and assortative mating needed to maintain distinct lineages, along the lines of the animal-based models of Servedio (2000).
It seems very unlikely that partitioning among pollinator taxa would be important for initiating divergence within a population experiencing disruptive selection. Plants within a population have much less opportunity to develop different combinations of floral traits that appeal to innate preferences of different pollinators, compared to an allopatric situation. It is conceivable that variation in one key trait, such as length of a nectar spur, could initiate divergence by effectively restricting which visitor taxa transfer pollen (Fulton & Hodges 1999) and by causing a correlation in trait value (spur length) among mates, i.e., assortative mating. When spur length is correlated with trait(s) under disruptive selection, additional mechanisms that strengthen assortative mating and enable divergence, such as shifts in flowering phenology, would be favored. However, even in this case with variation in a key mating trait, competitive exclusion of alternative floral morphs by the preferred morph of the most effective or frequent pollinator seems the most likely outcome. Different populations with dissimilar pollinator communities may have different competitive outcomes, and thus diverge in allopatry, but gene flow and competition make such divergence within populations unlikely.
Partitioning of floral resources between pollinator taxa may serve to eliminate some members of a plant community from a pollen-transfer pool, but seems unlikely by itself to isolate single species. Even in relatively specialized plant-pollinator communities in the Cape region of South Africa, distantly related plants share pollinators, but often place pollen on different parts of pollinator bodies, thereby furthering assorta-tive mating via mechanical isolation (Dobzhansky 1937; Grant 1949, 1994). For example, Lapeirousia silenoides (Iridaceae) and Pelargonium sericifo-lium (Geraniaceae) are both pollinated by long-tongued flies of the genus Prosoeca (Nemestrinidae) and exhibit remarkable convergence of floral form (Goldblatt et al. 1995). The iris deposits pollen on the head and dorsal side of the flies, whereas the geranium places it ventrally. Flowers of both species have zygomorphic symmetry, which apparently helps to orient the flies "correctly" on the flowers (Goldblatt et al. 1995).
In summary, partitioning of floral resources among pollinator taxa may help to maintain distinct lineages when plant populations that have diverged in allopatry come into secondary contact, providing reinforcement of any post-pollination barriers to hybridization. However, this form of pollinator-mediated assortative mating is unlikely to aid in the initial divergence of lineages in sympatry, as gene flow and competition make it difficult to maintain multiple co-adapted gene complexes in a population. In general, partitioning among pollinator taxa seems much more likely to be only part of a suite of mechanisms of assortative mating, rather than to provide by itself complete reproductive isolation of sym-patric plant species.
Relatively short-term specialization by individuals has quite different consequences than fixed preferences according to pollinator type. Variation in a single floral trait may suffice to cause assortative mating via behavioral constancy or labile preference, if the trait is used as a recognition cue by foragers; differences in groups of traits are likely to induce stronger constancy (see Gegear & Laverty, this volume). Thus, evolution-arily labile traits such as petal size or color can induce assortative mating without appealing particularly to one kind of pollinator over another. One forager might prefer darker flowers because the first flowers it visited were dark and had plentiful nectar. Another forager might visit the same flowers and find them unrewarding relative to light ones and consequently develop a preference for light flowers. Over several seasons of experiments with freely foraging bumble bees (primarily Bombus appos-itus) visiting randomized arrays of two colors of snapdragons (yellow and white or yellow and red Antirrhinum majus: Scrophulariaceae), I found a great deal of heterogeneity of preference among foraging bouts and assortative mating with respect to flower color (Jones 1997; Jones & Reithel 2001).
When the same kinds of pollinators visit incipient plant species, there is little opportunity for disruptive (in sympatry) or divergent (in allo-patry) selection directly on floral traits. Plants may diverge in traits that are more or less neutral for pollinator attraction, such as petal color pattern in Clarkia (Jones 1996). This idea may help to resolve Ollerton's "paradox of plant-pollinator systems," which hinges in part on the questionable supposition that "specialization to a taxonomically narrow array of pollinators would appear to be a prerequisite for the evolution of floral novelty" (Ollerton 1996). Divergence in floral recognition cues may be an effective way to improve pollen targeting and assortative mating regardless of changes in pollinator communities, when the cues are not specifically tuned to particular pollinator taxa.
To the extent that foragers in an area are systematic, tending to return to patches at regular intervals, the relative rewards of different flower morphs in a patch fluctuate somewhat predictably (Possingham 1988). Extremely systematic foraging - trapline foraging - is known in several kinds of bees (Heinrich 1976; Kadmon 1992; Thomson 1996; Thomson & Chittka, this volume) and hummingbirds (Gill 1988). Labile preference and behavioral constancy should thus tend to "even out" pollinator service, as previously under-exploited flower types are discovered and preferentially visited for awhile. Short-term specialization therefore would be more likely to maintain multiple floral morphs in a population than would fixed pollinator preferences. Following the chain of logic through to the speciation process, individual specialization should be more likely than partitioning among pollinator taxa to help initiate divergence in a plant population under disruptive selection.
In cases of secondary contact of plant populations following divergence in allopatry, individual pollinator specialization seems at least as likely as partitioning among pollinator taxa to maintain genetically distinct lineages. Both types of non-random pollinator foraging provide the necessary assortative mating, but the individual specialization mechanism does not require the presence of multiple kinds of pollinators in order for each incipient plant species to be competitive. For example, an
Ipomopsis hybrid zone appears to fit an "advancing wave" model, in which traits characteristic of hummingbird-favored I. aggregata have an overall advantage due to the far greater and more reliable abundance of hummingbirds than hawkmoths; therefore I. aggregata is predicted to spread at the expense of I. tenuituba (Campbell et al. 1997). If some hummingbirds occasionally favored I. tenuituba, it would seem to have a much better chance oflocal persistence, but there is no evidence of constancy by hummingbirds foraging among these Ipomopsis species and hybrids (see Waser, this volume).
Short-term pollinator specialization seems especially relevant for the establishment of new hybrid or polyploid species in sympatry with parental species. Such lineages - allopolyploids, for example - typically occur in initially low frequencies, and thus face the disadvantage of being minority cytotypes (Levin 1983). Persistence of a new hybrid or polyploid lineage is very unlikely, unless it has some means of reproductive isolation from parental species (Thompson & Lumaret 1992). Assortative mating via pollinator specialization is a possibility when hybrids have distinct flowers, as is the case for allopolyploids such as Tragopogon mirus (Cook & Soltis 1999) and the autopolyploid Heuchera grossulariifolia (Segraves & Thompson 1999). For example, most pollinators distinguished between diploid and tetraploid H. grossulariifolia plants, with several insect species visiting the two floral types at significantly different frequencies, thus providing incomplete partitioning with respect to pollinator taxa (Segraves & Thompson 1999). Whether individual pollinators within species exhibited further specialization is not clear from the study; such specialization could promote stronger assortative mating than suggested by the species preference differences. Because labile preferences and behavioral constancy can result in occasional specialization on generally undervisited floral types, and because hybrid or polyploid flowers are unlikely to be different enough from both parental types to be visited by different pollinator taxa, partitioning within rather than between pollinator taxa seems more likely to aid in the establishment of new hybrid or polyploid species. Polyploidy is considered a prominent mechanism of speciation in plants (Soltis & Soltis 1993) and hybrid formation is quite common, especially in outcrossing perennials: naturally occurring hybrids make up 5%-22% of the species in five biosystematic floras, according to a recent survey (Ellstrand et al. 1996).
Pollinator specialization may be harder to come by when hybrids are present and serve as a bridge between species that otherwise are different enough to induce some level of pollinator-mediated assortative mating (Hodges et al. 1996; Rieseberg & Carney 1998). This is the case in hybrid swarms of Baptisia in Texas, where pollinator constancy to the parental species is strong enough to depress production of Fi hybrids, but where pollinators move freely between hybrids and parental species, at least in randomized arrays (Leebens-Mack & Milligan 1998). Other mechanisms of assortative mating must maintain the distinct lineages, such as poor competitiveness of pollen from hybrid plants. Pollen competition has been documented in at least two cases (Carney et al. 1996; Klips 1999) to be a partial isolating mechanism between hybridizing lineages (a form of intrinsic reproductivebarrier, discussed more fully byWaser, this volume).
In summary, when selection favors assortative mating, the easiest way to accomplish this may be to diverge enough in a floral recognition character, such as petal pigmentation pattern, to induce individual pollinator constancy. Short-term specialization may help to maintain a low-frequency floral type or one that is relatively undervisited overall. Pollinator-mediated assortative mating, together with selection against backcrossing (usually very strong in polyploids), may suffice to isolate a new hybrid or polyploid lineage from parental lineages, thus increasing the chance of speciation (Grant 1971; Rieseberg 1997; Wolfe et al. 1998; Milne et al. 1999); its importance in this regard remains to be established. This task will be facilitated by studies testing for individual pollinator specialization (i.e., constancy in the narrow sense and heterogeneity of preference) as well as partitioning among pollinator taxa with respect to the different plant lineages; the former behavioral mechanism is more likely and generally less studied by botanists than the latter.
Was this article helpful?