and 50% from the other. Experienced bumble bees (those with 250 or more visits) tended to prefer the 20% over the 10% (x = 0.82, SD = 0.090, n = 12 bees) more strongly than they preferred the 50% over the 40% (x = 0.61, S.D. = 0.071, n = 12 bees), the results qualitatively predicted by the perceptual scale (Fig. 3.6).
Waddington & Gottlieb (1990) investigated the choice behavior of honeybees by giving them a binary choice between tubular blue and yellow artificial flowers. The flowers contained different volumes of sucrose solution. The degree of preference for the flower type having more solution depended on the relative difference in volume, rather than absolute difference between the flower types. Such behavior is consistent with decisions based on a non-linear relationship between subjective evaluation and objective nectar volume.
Foragers make decisions and choices based on the quality and quantity of pollen. Harder's (1990) work indicated that bumble bees make more flower visits per inflorescence visit when the flowers contain more pollen; they also revisit pollen-rich flowers more frequently. In some way, the bees gather information on the volume of pollen they have harvested, then base decisions on it. Buchmann & Cane (1989) also found clear empirical evidence that bumble bees and bees of the genus Ptiloglossa assessed the amount of pollen collected from individual Solanum flowers. Schmidt (1982, 1984) demonstrated that honeybees prefer pollen from certain plant species over others; these choices may be based on odor or taste of the pollen. Schmidt & Johnson's (1984) work suggests choices may be based on the pH and the percentage of crude protein in the pollen.
K.D. Waddington, C.M. Nelson & R.E. Page (unpublished data) quantified honeybees' choices between pollens of different quality to see if the choices were consistent with a non-linear subjective evaluation of pollen quality. We manipulated pollen quality by diluting pure pollen with various proportions of alpha-cellulose (vol./vol.) (see above). Bees were presented with choices between two dishes of pollen: pure pollen and a 1:1 dilution, pure pollen and a 3:1 (pollen:cellulose) dilution, or a 1:1 dilution and a 3:1 dilution. The difference in quality between the 1:1 and 3:1 mixtures is the same as that between the 3:1 mixture and pure pollen. The number of foragers counted at each dish of pollen of a pair was used to gauge indifference (equal number of foragers) or a preference for one dish of the pair. The bees were indifferent between pure pollen and the 3:1 mixture, but they preferred the 3:1 mixture (and pure pollen) over the 1:1 mixture. The results further confirm that bees base their choices on pollen quality. The choices observed suggest that the honeybees evaluated pure pollen and the 3:1 mixture equally but the 3:1 mixture was evaluated as greater than the 1:1 mixture; these choices are consistent with the hypothesis that evaluation of pollen quality is a non-linear function of pollen quality.
All these results are consistent with the hypothesis that bees base choices on subjective evaluations that are non-linearly related to objective nectar volume or concentration and pollen quality.
Choice behavior: variation in nectar concentration
Within and among species of plants, flowers vary in the concentration and volume of their nectar. The question is: do bees make choices based on these sources of variation.? For example, if two types of flowers provide the same expected nectar concentration, but the predictability of the concentration is high in one and low in the other, does it matter? Behavioral ecologists have recognized the effects of variability on choice, and incorporated reward variation into optimality models of foraging behavior. Response to variation, called "risk sensitivity," has become central to studies on choice behavior (Caraco 1980; Stephens & Charnov 1982). Risk sensitivity has been examined experimentally using many kinds of animals, including several species of nectarivorous invertebrates and vertebrates (see Kacelnik & Bateson 1996 for references).
Elsewhere, summaries can be found of risk-sensitive foraging models (Stephens & Paton 1986). However, Harder & Real (1987) began a discussion of the mechanisms underlying risk-sensitive foraging behavior which has been reviewed elsewhere (Perez & Waddington 1996; Waddington 1997). Here I will restrict the discussion by relating some recent data on choice behavior in relation to variation in nectar concentration to the ideas derived from measurement of subjective evaluation of concentration in honeybees.
We have studied response to variation in sugar concentration in three genera of bees: Apis mellifera (honeybees; Banschbach & Waddington 1994), Xylocopa micans (carpenter bees; Perez & Waddington 1996), Bombus impatiens and Bombus fervidus (bumble bees; Waddington 1995, K.D. Waddington, S. Lamenta & M. Jordan, unpublished data). Bees were given a choice between two types (colors) of artificial flowers which were equivalent in expected concentrations of sugar solution, but the types differed in the distributions of concentrations provided. The experimental paradigm follows the early study of response to variation in volume (Waddington et al. 1981). The bee always found the same concentration in the low-variance type and found a random sequence of two concentrations in the high-variance type. We obtained two results. Honeybees, Bombus fervidus, and carpenter bees were indifferent to (showed no preference between) the two types when the low-variance type provided 20%
sucrose and the high-variance provided 10% or 30% equally frequently in random sequence. However, when Bombus impatiens experienced a mean concentration of 30% and the high variance flower provided 10% or 50%, the bees preferred the low-variance flower (30%, p = 1.0), i.e., they became risk-averse (Fig. 3.7a).
Both indifference and the preference for the low-variance flower are consistent with the rule: visit the flower type with the higher perceived expected concentration. In Fig. 3.7b, concentrations are projected onto the perception axis using the relationship between concentration and honeybee dance rates (Fig. 3.2). A constant 30% has a higher perceived expected value than the lottery of 10% and 50% (p = 0.5); bees appear to be maximizing perceived expected concentration.
In the studies of three other species ofbees, a constant 20% was not preferred to the variable 10% and 30% presented equally frequently. A return to Fig. 3.7b, provides an explanation for the indifference. If these concentrations (10%, 20%, and 30%) were each projected onto the perceptual axis, the perceived expected concentrations would be nearly equivalent (dance rate about 22.5) for the high- and low-variance flowers, because the evaluation is linearly related to concentrations over the range of 10%-30% (Perez & Waddington 1996).
These are remarkable matches between the relationship describing perception of concentration and response to variance in concentration, especially considering that I used the honeybee's scale to predict choice in three other species ofbees. Of course, the actual relationship describing perception over the whole dynamic range of concentration in bumble bees and carpenter bees may differ from that of honeybees.
I suggest that we can link a bee's choice behavior to that bee's subjective evaluations of the quantities and qualities of pollen and nectar. These evaluations can be quantified - independently of choice behavior - using the honeybee's round dance. This technique is analogous to asking human beings to place the magnitude of a stimulus on a scale (Carterette & Friedman 1974). Thus far, we have found that the value perceived by bees is a non-linear function of nectar concentration. The non-linear relationship is generally described by the Weber-Fechner law of perception, which is a very general property of animals; the weighting of losses
Was this article helpful?