Fig. 11.2. Patch use by 12 Lasius pallitarsis ant colonies in response to relative food quality and predation risk. The rich-food patch had food concentration 2-16 times higher than the poor-food patch (the x axis). Experimental sessions either included a predator at the rich-food patch or had no predators at either patch. The ants varied the proportion of food taken from the rich-food patch in response to its relative quality and the predation risk. The effects of relative food quality and predation treatments were statistically significant (p < 0.01). (Data from Nonacs & Dill 1991.)
1991) as well as other insects (Quinn et al. 1974; Dukas 1999; Liu et al. 1999). Nevertheless, I highlight Gould's data, because further elaboration of his protocol may illustrate that bees can learn to avoid landing on one plant where they encountered a predation attempt while they keep visiting neighboring safe plants.
Morse (1986) concluded that bees in his study failed to either bypass flowers containing crab spiders or leave flowers with non-attacking spiders faster than spider-free flowers. The low resolution of the insect eye (Land 1981) and limited attention (Dukas 1998a; Dukas & Kamil 2000) can probably explain pollinators' failure to notice motionless, cryptic crab spiders or other ambushing predators. Still, the studies on crab spiders, bee-eaters, and bee-wolves reviewed above indicate that the success rates of these bee predators range between 10% to 30%. This means that bees can at least potentially respond to failed predation attempts by altering behavior in an attempt to reduce subsequent attacks.
Recently, Grostal & Dick (1999) reported that herbivorous spider mites
('Tetranychus urticae) avoided leaves where prédation on conspecifics had occurred, and Dixon & Agarwala (1999) found that pea aphids (Acyrthosiphon pisum) responded to odor tracks left by predatory ladybird larvae (Adalia bipunctata). Bees and other flower visitors may also perceive predator activity indirectly through sensing predator odor, an attack on another individual, or injured conspecifics. In honeybees, components of the sting and mandibular gland pheromones deter conspecifics (Free 1987; Balderrama et al. 1996). Honeybees may employ such pheromone-based information to avoid locations of high predator activity.
The ability to perceive predators, however, does not imply that pollinators must always alter their behavior in response to predator presence. The response to predation risk should reflect long-term costs and benefits, which I discuss below.
How should pollinators respond to perceived prédation.?
Anti-predatory adaptations have the potential to increase fitness, either through increased lifetime reproduction in solitary species or increased worker's lifetime contribution to colony growth in social insects. Bees seem to have some obvious anti-predatory traits: most species possess stings and many have aposematic coloration (Schmidt 1990). Sympatric bumble bee species even segregate into groups with similar color patterns, which are better explained by geographical than taxonomic affiliation, suggesting Mullerian mimicry (Plowright & Owen 1980). Similarly, many noxious butterflies have aposematic coloration; these butterflies and their non-poisonous mimics appear in complexes of Mullerian and Batesian mimicry (Gilbert 1983). In addition to morphological adaptations, bees and other pollinators probably possess less apparent behavioral traits that help them decrease predation risk.
Many times, alternative feeding options with different rates of food intake also have characteristic mortality rates. Such foraging-mortality tradeoffs can be evaluated with models that consider the lifetime outcomes of the available alternatives. The foraging-mortality tradeoff faced by a social bee can be approximated by the ratio g/x, where g is the rate of food collection and ¡1 is the mortality rate during foraging (Clark & Dukas 1994; see also Gilliam & Fraser 1987; Dukas & Edelstein-Keshet 1998). Maximization of this ratio would result in maximizing the lifetime contribution of the bee to her colony growth. Using this approximation, we can conclude that a social bee collecting food in flower patch 1 at the rate of 1 unit per trip and facing an expected predation rate of 0.01 per trip should prefer the less rewarding flower patch 2 with half the expected predation rate (p = 0.005) if the reward rate in patch 2 is larger than half the reward rate in patch 1. For example, with the values given, the expected lifetime contribution in patch 1 is 100 food units, so if patch 2 (with ¡x = 0.005) offers 0.6 food units per trip, it should be preferred because the expected lifetime contribution of a forager in that patch is 120 food units. Although the bee gains food in patch 2 at only 60% the rate of patch 1, her lifetime contribution is 20% higher due to increased expected lifespan. (For more elaborate models, see Clark & Dukas 1994; Dukas & Edelstein-Keshet 1998.)
To illustrate the power of a simple anti-predatory response, consider the following example based on Morse's (1986) data. Suppose a bumble bee had just successfully escaped a crab spider attack. Should she return to the same patch on subsequent foraging trips, or switch to another area.? The bumble bee's probability ofbeing attacked on the same inflorescence containing the crab spider may be as high as 0.08 compared to an attack probability by crab spiders of only 0.0004 on a randomly chosen inflorescence (Morse 1986, Table 5). Given the 200-fold difference in attack probability by crab spiders, the bee should probably attempt to avoid the inflorescence where she has been attacked even if this implies switching to a patch with somewhat lower reward rates.
Although sensitivity to predation risk should be ubiquitous among pollinators, one can readily imagine cases where short- or long-term fitness considerations might lead animals to ignore predators (Ydenberg & Dill 1986). For example, Cartar (1991) compared the response to a model predator of several bumble bees that foraged for a colony with either ample or depleted honey supply. The workers from honey-rich colonies were three times more likely to flee from the model predator than the bees from the honey-depleted colonies.
So far, I have assumed that the predators cannot alter their behavior as well. In reality, however, the interactions between pollinators and predators should be analyzed with a game-theory approach, where the predators are allowed to respond to the prey and vice versa (reviewed by Sih 1998). For example, Craig et al. (1996) suggested that the spider Nephila cla-vipes produces golden webs that attract more bees than any other web color (see also Craig & Bernard 1990; Craig & Freeman 1991; Craig 1994a; Blackledge & Wenzel 1999). The arms race between predators and pollinators would not lead to pollinators ignoring predators, but it would likely render anti-predator measures less effective due to the counter-adaptations by the predators.
Anti-predator behavioral adaptations of pollinators may be expressed all or most of the time, or expressed only in response to immediate predation risk (Dukas 1998b). Only the latter requires a capacity to assess levels of predation risk in space and time. That is, even if a pollinator cannot assess predation risk, due to sensory or cognitive limitations (Dukas 1998a), it can still possess evolved behaviors that reduce predation. The ability to respond selectively to perceived predation may be more efficient than a relatively fixed behavioral adaptation, because the former allows the expression of potentially costly anti-predator tactics only when necessary. However, selective response requires a pollinator to perceive the presence of a threat in appropriate circumstances.
Three types of direct response to a perceived heightened predation risk are increasing vigilance, avoiding subsequent visits to the same plant species or location, or adjusting foraging parameters in a way that can reduce the probability of subsequent attack or capture. Increasing predator vigilance can be achieved even through a fundamental sensitization mechanism, which most organisms possess (Jennings 1906; Papaj & Prokopy 1989). This means that after an attack, a pollinator would more readily initiate escape in response to stimuli such as sudden movement. I know of no systematic study documenting increased vigilance due to sen-sitization or learning in pollinators after exposure to threat. However, evidence presented by Armitage (1965) is rather suggestive.
Armitage compared wing wear of bumble bees captured by the bumblebee-wolf Philanthus bicinctus to wing wear of bumble bees he collected at the same flowers that were used as hunting grounds by the wasps. In two bumble bee species, the wasps' prey was strongly biased in favor of less worn (presumably younger) bees (Fig. 11.3). The wasps probably preferred or had higher success rates with younger, inexperienced bees. Gradual accumulation of experience may enhance foraging success of honeybees (Dukas & Visscher 1994), so such long-term experience might increase specific antipredatory behavior as well.
Suggestive field evidence for the ability of insects to avoid an area where they have encountered a predation attempt comes from two butterfly studies (Singer & Wedlake 1981; Mallet et al. 1987) that looked at the
Was this article helpful?