Tinbergen (1960) suggested that common-morph preference arises because foragers can maximize feeding efficiency by concentrating on a single food type. Bumble bee workers, however, show common corolla color morph preference even when rewards and handling times are identical. Real (1990) suggested behavioral constraints as the cause: the subjectively perceived probability and the actual probability of encountering a morph may differ, such that low probability events are overestimated; alternatively, a bee may repeatedly switch its attention from one morph to the other (the shifting attention hypothesis sensu Dawkins 1971a). However, neither of these suggestions fully explains all the behavioral patterns observed. Chittka & Thomson (1997) trained bees on artificial flowers of different colors and morphologies. With respect to handling flowers, avoiding and correcting errors, and initial traveling times between flowers, bees trained on only one color-morphology combination outperformed bees trained on two combinations. I have noted a reduction in traveling time as the common-morph preference develops. Both observations suggest efficiency advantages for pollinator specialization even when floral morphologies are identical; further experiments are required.
Mechanistically, the link between visitation order and common-morph preference implies that short-term memory affects frequency dependence (Menzel 1999, this volume). The increase of frequency dependence with experience suggests that continued reinforcement on common morphs consolidates long-term memory, resulting in continued preference. Predators that develop "search images" while learning to distinguish cryptic prey from the background will display frequency dependence if those images interfere with each other (Dawkins 1971b). Although interference has been shown for short-term memory (Menzel 1979), the importance of interference for common-morph preference by pollinators is unclear.
Why do rare unrewarding morphs receive more pollinator visits.? Ultimately, this is expected as a consequence of efficient avoidance of empty flowers by pollinators. Bumble bees may learn to avoid unrewarding flowers by making a particular number of test visits (Heinrich 1975). If this mechanism applies, the task of learning to avoid all unrewarding morphs will increase as the number of morphs increases. A bee will have to make more sampling visits. Furthermore, the numbers of each unrewarding morph sampled should be independent of that morph's frequency in the population. Dukas & Real (1993) did not find the predicted increase in sampling visits with morph number. I found that the number of flowers sampled of unrewarding morphs did depend on the frequencies of those morphs (Spearman's rs = 0.60 and 0.52, p < 0.001, for two unrewarding morphs that varied in frequency). Further, the equal-sampling hypothesis does not predict disassortative visitation. Dukas & Real (1993) suggested that unrewarding morphs are not memorized during sampling, but if this is correct, rare-morph preference would not occur (Ferdy et al. 1998). An alternative hypothesis (Smithson & Macnair 1997b) suggests that sampling an unrewarding morph causes negative reinforcement stored in short-term memory, increasing the likelihood subsequently of sampling a different morph, thereby causing disassorta-tive visitation. Data from humans suggest that the time to search for a particular target type does not increase as the number of alternative stimuli increases, as long as the alternatives vary only in one way, e.g., color or shape. If the alternative stimuli vary in more than one way, e.g., both color and shape, it takes significantly longer to find the target (Treisman & Gelade 1980; see also Gegear & Laverty this volume).
The mechanisms behind pollinator sampling of unrewarding morphs and consequent negative FDS require further experimentation, particu larly with respect to interactions among several traits. In nature, many unrewarding species are rare orchids, so if increasing phenotypic variability within a population can increase the total number of pollinator visits to the species, this may have important conservation implications (Ferdy et al. 1998).
Was this article helpful?