Pollinator behavior on inflorescences

While foraging in the three-dimensional environment of an inflorescence, a pollinator must choose a starting flower, negotiate a route among visited flowers, and determine when to leave the plant. These decisions affect foraging benefits and costs by determining the number of flowers visited and the time and energy expended on flight. As we discuss in the next section, these decisions also establish the extent of between-flower self-pollination (geitonogamy) and pollen export to other plants.

Many pollinators visiting vertical spikes or racemes invariably start foraging on either lower (bees, wasps, and hawkmoths: e.g., Waddington & Heinrich 1979; Corbet et al. 1981; Dreisig 1985; Rasheed & Harder 1997&) or upper flowers (flies: Arista et al. 1999), thereby predetermining their subsequent movement direction within the inflorescence. For bumble bees, the proclivity to move upward apparently involves a functional constraint, as it persists on inverted inflorescences (Heinrich 1979) or when resources per flower increase or decrease along the inflorescence (Waddington & Heinrich 1979; Corbet et al. 1981). Given this constraint, bees respond to resource gradients in vertical inflorescences by altering their starting and leaving positions in ways that enhance their foraging economy (Pyke 1979; Waddington & Heinrich 1979; Rasheed & Harder 1997&). By generally moving upward, bumble bees seldom revisit flowers on vertical inflorescences (Pyke 1979; Galen & Plowright 1985). In contrast to insects, hummingbirds move less stereotypically on vertical inflorescences, starting on bottom or top flowers with roughly equal frequency (Wolf & Hainsworth 1986; Healy & Hurly, this volume).

Inflorescences with more three-dimensional structure than a raceme seem to complicate pollinator foraging. The only study to examine this effect (Hainsworth et al. 1983) compared the responses of hummingbirds to vertical, two-dimensional inflorescences and hemispheric, three-dimensional inflorescences. On three-dimensional inflorescences, birds probed fewer flowers, with proportionately fewer revisits than when they visited vertical inflorescences. In addition, flights between flowers lasted longer on hemispheric than on vertical inflorescences, even though the flowers were closer together. Hence, the spatial arrangement of surrounding flowers altered the cost of moving between two flowers separated by a specific distance. If this outcome applies more generally, different inflorescence architectures likely establish unique foraging environments for pollinators, and consequently influence pollen transfer within and between plants.

The effect of inflorescence architecture on pollinator behavior depends partly on whether pollinators modify foraging conditions predictably for subsequent visitors by depleting resources. On vertical inflorescences, the upward movement of bumble bees creates a positive correlation in nectar standing crop among flowers on the same inflorescence (Waddington 1981; Dreisig 1989), so the state of one flower provides information about that of higher flowers. In this environment, a bumble bee invariably continues feeding on its current inflorescence after visiting a rewarding flower, whereas it typically switches to another inflorescence after encountering a single empty flower (Dreisig 1989). In contrast, on the head-like inflorescences, such as Monarda fistulosa, bumble bees generally do not leave an inflorescence until encountering several empty flowers (Cresswell 1990). Presumably, this reduced responsiveness reflects the less stereotyped movement of bees on these less ordered inflorescences, thereby limiting correlation in reward availability among flowers (also see Kadmon & Shmida 1992; but see Wolf & Hainsworth 1986). The contrast between these responses indicates that the role of inflorescence architecture in modifying the economics of pollinator foraging (and pollen dispersal) extends beyond the effects of floral display on the actions of individual pollinators to include indirect interactions between all pollinators attracted to an inflorescence.

Geitonogamy and outcross siring success

The presence of multiple flowers permits pollen transport between a plant's own flowers (reviewed by Harder & Barrett 1996; Snow et al. 1996; also see Brunet & Eckert 1998; Rademaker & de Jong 1998). In general, geitonogamy increases as a pollinator visits more flowers on a plant. For example, consider the destinations of pollen removed from the first of five flowers visited by a pollinator on a plant (Fig. 15.1a). Geitonogamous pollen transfer from this flower occurs during the pollinator's next four flower visits. If, instead, the pollinator visited eight additional flowers on the same plant, geitonogamy would claim a larger fraction of the total pollen dispersed from the donor flower and the plant as a whole. Because pollinators tend to visit more flowers on larger inflorescences (reviewed by Ohashi & Yahara, this volume), geitonogamy generally increases with display size (reviewed by Harder & Barrett 1996; Snow et al. 1996).

In addition to increasing the number of matings susceptible to inbreeding depression, geitonogamy reduces the pollen available for dispersal to other plants (pollen discounting: Kohn & Barrett 1994; Harder & Barrett 1995; Emms etal. 1997; Harder etal. 2000). Lloyd (1992) proposed that geitonogamy always diminishes outcrossing opportunities when pollen transport between flowers on the same plant involves the same processes as transport between plants. Because pollen dispersal between a donor and recipient flower varies non-linearly with the number of flower visits that separate them, each additional flower that a pollinator visits adds a progressively smaller increment to total pollen export (see Fig. 15.1b) (Harder & Barrett 1996; Emms etal. 1997; Rademaker & de Jong 1998).

The significant mating cost of pollen discounting probably favors inflorescence designs that restrict geitonogamy, including limits on the number of flowers displayed simultaneously, segregation of the sex roles among flowers within inflorescences, and heterostyly (Kohn & Barrett 1994; Harder & Barrett 1995; Harder et al. 2000). However, the benefits of specific anti-geitonogamy mechanisms depend on pollinator characteristics. For example, presentation of male flowers above female flowers limits pollen discounting for vertical inflorescences when bees move upward (Harder et al. 2000), but would aggravate discounting if pollinators move downward, and would have little effect if pollinators move unpredictably among flowers within inflorescences. Hence, plant species pollinated predominately by animals with different movement patterns should exhibit different patterns of sexual segregation.


Many features that distinguish animal-pollinated species from abioti-cally pollinated species serve the signaling and reward functions that govern a plant's attractiveness to pollinators, including nectar and pollen availability, a showy perianth, and fragrance (reviewed by Mitchell 1993). Attractiveness generally increases with the number of flowers displayed simultaneously (reviewed by Ohashi & Yahara, this volume), and so depends on the aggregate signal perceived by pollinators and their expected foraging returns from a plant's entire floral display (e.g., Weiss 1991). Consequently, the signals and rewards of individual flowers must be considered in the context of their collective contributions to a plant's overall reproductive success.

Given the expense of attraction (reviewed by Morgan 1992), the benefits must be significant. Obviously, a plant must attract enough pollinators to engage much of its pollen in dispersal and to bring in enough pollen to fertilize most of its ovules. Furthermore, participation of many individual animals in pollination increases a plant's mate diversity when different pollinators follow different foraging paths. A less obvious, but significant, benefit of attracting many pollinators arises because of the diminishing returns associated with increased pollen removal by individual pollinators which accompany pollinator grooming, pollen layering, and geitonogamous pollen discounting. Because of diminishing returns, a pollinator that removes half of a plant's pollen will export more than half as much as another pollinator that removes it all (see Fig. 15.1b). Consequently, plants enhance their pollen export by restricting pollen removal by individual pollinators and involving many pollinators in dispersal (Harder & Thomson 1989; Iwasa et al. 1995). Indeed, optimal restriction of pollen removal could increase siring success by more than an order of magnitude when pollinators are abundant, although time-dependent processes, such as loss of pollen viability and competition among male gametophytes for access to ovules, counteract the benefits of restricted removal (Harder & Wilson 1994).

To appreciate the benefits of restricting pollen removal, consider the relation of total pollen export to the proportion of a plant's flowers visited by each pollinator (see Iwasa et al. 1995 for mathematical details). Because of the diminishing returns caused by geitonogamous pollen discounting (see Fig. 15.1b), two pollinators that each visited half of a plant's flowers would export more pollen overall than a single pollinator that visited all the flowers, even though the number of visits per flower is identical. Attraction of many pollinators further enhances pollen export, as long as each pollinator visits only a fraction of a plant's open flowers, thereby limiting pollen discounting (see Fig. 15.2). However, if pollinators visit too few flowers, pollen can remain in anthers (pollen-removal failure), thereby reducing the plant's total pollen export. Hence, as the solid curve in Fig. 15.2 illustrates, maximization of pollen export occurs when the proportion of flowers visited by each pollinator balances the risk of pollen-removal failure against the mating cost of geitonogamous pollen discounting. The appropriate balance depends on the number of pollinators attracted. Deviation from this optimum reduces total pollen export, particularly when many pollinators visit. However, total pollen export declines asymmetrically on either side of the optimum, so that plants lose less from erring towards too much pollen discounting than from having pollen left in anthers (Fig. 15.2).

Given that enhanced attractiveness increases pollen export only if each pollinator removes a limited amount of pollen, how do plants restrict pollen removal.? Two types of mechanisms serve this purpose: packaging

Fig. 15.2. Relation of expected pollen export by all pollinators to average pollinator attraction and the proportion of available flowers visited by each pollinator (based on Eq. 1 in Iwasa et al. 1995). The bold solid line depicts the proportion of flowers visited that maximizes expected pollen export for a specific average pollinator availability. Depression of pollen export around this optimum results from pollen-removal failure or geitonogamous pollen discounting, as indicated. The dashed lines illustrate expected pollen export for fixed expected intensities of visits per flower (m). This example involves the same parameter values as Fig. 15.1.

Fig. 15.2. Relation of expected pollen export by all pollinators to average pollinator attraction and the proportion of available flowers visited by each pollinator (based on Eq. 1 in Iwasa et al. 1995). The bold solid line depicts the proportion of flowers visited that maximizes expected pollen export for a specific average pollinator availability. Depression of pollen export around this optimum results from pollen-removal failure or geitonogamous pollen discounting, as indicated. The dashed lines illustrate expected pollen export for fixed expected intensities of visits per flower (m). This example involves the same parameter values as Fig. 15.1.

mechanisms control the amount of pollen exposed at one time, whereas dispensingmechanisms limit the amount of exposed pollen removed by each pollinator (Harder & Thomson 1989). Packaging mechanisms can be implemented in individual flowers through staggered anther dehiscence, or on the entire plant through staggered opening of flowers. These mechanisms enable strict management of pollen removal because they are completely under a plant's control. In contrast, many dispensing mechanisms adjust pollen removal to a plant's prevailing frequency of pollinator visits

(see Harder & Wilson 1994; Harder & Barclay 1994). Floral mechanisms that serve as dispensing mechanisms include anther position (Harder & Barrett 1993), poricidal anthers (Harder & Barclay 1994; King & Buchmann 1996), secondary pollen presentation (Yeo 1993; Harder & Wilson 1994), anther tripping (Armstrong 1992; Lebuhn & Anderson 1994), and nectar production.

Nectar production provides a unique means of dispensing pollen, because it allows plants to counteract diminishing returns on pollen removal from both individual flowers and inflorescences. Nectar volume influences pollen removal by positively affecting the duration of visits to individual flowers and the number of flowers visited per inflorescence (see above). Because nectar generally accumulates steadily (Burquez & Corbet 1991), individual pollinators remove less pollen from individual flowers (Jones et al. 1998) and visit fewer flowers per inflorescence (Kadmon & Shmida 1992; Hodges 1995) when pollinators are abundant and visits occur frequently. Therefore, the combination of nectar production rate and visit frequency enables restricted pollen removal in a manner that responds to pollinator availability.

In seeming contradiction to this proposal, the rate of nectar production varies considerably among flowers within inflorescences, with some flowers producing little nectar (Feinsinger 1978; Brink 1982; Marden 1984; Gilbert et al. 1991). Bell (1986) proposed that empty flowers allowed plants to save some of the expense of nectar production without forfeiting much pollinator service. However, nectar may not be expensive for many plants (reviewed by Harder & Barrett 1992), so that empty flowers may do little to reduce the cost of attraction. Instead, we propose that by maintaining a fraction of flowers that produce little nectar, plants encourage pollinators to leave inflorescences after visiting only a fraction of their open flowers, thereby restricting pollen removal per pollinator and enhancing the aggregate pollen dispersal provided by all pollinators that visit. According to this hypothesis, empty flowers should be most common in species pollinated by abundant pollinators, because restricted removal is most beneficial when pollinators visit frequently.

The preceding discussion of attraction focused on the benefits for male success, rather than female success, through pollen receipt. We adopted this emphasis because the needs of pollen receipt are often realized with fewer pollinator visits than are those of pollen dispersal (e.g., Young & Stanton 1990; Mitchell & Waser 1992; Aizen & Basilio 1998; Bell & Cresswell 1998). This asymmetry arises from the dissimilarity in mating opportunities through female and male roles. The opportunities for paternal success depend on the number of available ovules in the population as a whole. As a result, outcross siring success increases continuously with a plant's relative contribution of pollen to stigmas. In contrast, each pistil contains a limited number of ovules, so that female outcross success levels off as stigmas receive an increasing share of exported pollen (e.g., Snow 1982; Shore & Barrett 1984; Galen 1992). Indeed, receipt of too much pollen can cause interference between pollen tubes and reduce seed production (reviewed by Young & Young 1992). Because of this asymmetry, the considerable effort expended on attraction by many animal-pollinated plants seems to benefit male success more than female success, even though pollen export must equal pollen import at the population level.

We conclude by emphasizing two essential features of the selection of floral design and display. The first feature arises from recognition that floral and inflorescence characteristics create the environment within which pollinators tend to maximize a specific foraging currency. Because of this role, plant evolution could improve foraging benefits or alleviate costs; however, it will do so only to the extent that such changes promote plant mating (e.g., Harder & Cruzan 1990; Harder & Barclay 1994). Therefore, the evolutionary relevance of specific floral or inflorescence traits must extend beyond their impact on pollinator behavior to realized mating outcomes. The second feature deserving emphasis is that, despite the key role of individual flowers in controlling pollen exchange with pollinators, mating fundamentally involves entire plants. For example, contrary to the expectation that a plant's pollen export increases mono-tonically with the number of pollinator visits received by each flower (e.g., Harder & Thomson 1989; Harder & Wilson 1994), increasing visits per flower with no change in the average number of visits per plant eventually reduces export (Fig. 15.2). Because of such non-monotonic effects, selection of floral traits will often optimize pollination of individual flowers to maximize a plant's mating success.

Was this article helpful?

0 0

Post a comment