Rape, Brassica napus L., sunflower, Helianthus annuus L, and sesame, Sesamum indicum L. pollens were collected in 1989 in the People's Republic of China as corbicular pellets removed from the pollen baskets on the bees' hind legs as they passed through pollen traps attached to honey bee hives. The pollen was kept in a refrigerator (approx. 4'C) for several months until taken to the United States, where it was stored at -20'C until used. The pollens pellets were sorted to 99% purity by hand, then acetolyzed (O'Rourke and Buchmann 1991) and identified by light microscopy. Rape and sunflower pollen grains were readily recognized from standard reference sources; sesame pollen identification was achieved after comparison with pollen grains from a sesame specimen in the herbarium of University of Arizona. A pollen mixture consisting of the pollen from 15 species of spring-blooming Sonoran desert plants as described in Schmidt and Johnson (1984) that had been stored at - 20'C since initial collection, was used as the standard mixture. This mixture was collected from colonies in natural areas around Tucson during the seasonal peak of pollen abundance and at the time of maximal honey bee colony population increase. We assume this pollen represents a typical normal diet of bees because they had an abundance of pollen available at the time and could readily chose among pollen sources. All pollen materials were sealed in glass jars and stored at -20'C shortly after collection until use. Young worker honey bees obtained from an apiary in Tucson, AZ , were used because they, unlike the older forager bees, are the primary consumers of pollen.


The feeding preference procedures were modified from those of Schmidt and Johnson (1984). Several frames of preemergent pupal brood (no adult bees) were taken from the colonies and maintained in an emergence box in a constant dark environmental room at 31-35'C and 70% RH. Three days later, 15 g of bees (approx. bees) were taken from the emergence box and placed in acrylic plastic and screen cages (9 by 6 by 15 cm) and provided 40% sucrose solution ad libitum (Fig. 1). Inside each cage was a piece of beeswax comb foundation as a clustering platform for the bees. The bees were then allowed to choose between pollen diets consisting of test pollen and the 15-pollen standard mixture. Each diet was made by thoroughly mixing the pollen with distilled water until a moist, kneadable and stable texture was obtained. Diets not used during the experimental setup (day 1) was stored at -20'C for use on day 3. Equal amounts of test and 15-pollen standard mixture pollen diets were weighed into separate clean plastic hollow stoppers (feeders) and placed at random in the right and left side feeder slots of the cage (Fig. 1). Tests were conducted at least in triplicate for each pollen. Two days later the weight of the diet remaining in the feeders was recorded and fresh diets were provided to replace residual diets. At this time the right and left positions of the test and control mixed diet feeders were reversed. Four days after initiation, the remaining weight of pollen diet was again recorded and the bees released into nearby hives. All experiments were conducted under red light in the environmental room. For each test, an extra feeder of each diet was made and placed in a cage without bees in the environmental room as a control for evaporation. The percentage of weight loss of this evaporation control was used to correct for water loss of the diets during the tests. The relative consumption of the test pollen was calculated according to the formula:


% consumed= (weight test diet consumed)/(weight test pollen consumed) + (weight control pollen consumed) X 100%.

The data from this experiment can be assumed to be normally distributed because they are measures on a continuously variable scale and do not fall near boundary areas. Thus, the t-test was used to determine if the means differed significantly from 50%, which is the expected value if the test and standard pollen are equally acceptable to the bees.

The longevities of the bees fed various diets were determined using procedures modified from Schmidt et al. (1987). The tests were conducted in the environmental room described previously using cages and basic procedures similar to those for the preference studies with the following 5 modifications: (1) 60 randomly selected 1-d-old bees were placed in each cage; (2) only I test diet was available, with no diet for the control group; (3) every 2 or 3 d, pollen consumption was measured, dead bees counted and removed, and residual diet was replaced with freshly thawed diet; (4) 40% sucrose solution was replaced weekly; and (5) the pollen diet was eliminated when food consumption became essentially zero, normally around day 20. Tests were replicated 4 times and were terminated when the last bee died. Results were analyzed by the Tukey multiple-comparison test (Zar 1984).

The results of the feeding preference comparisons between the 15-pollen standard mixture and each of the other pollens are shown in Table 1. The preference was measured as relative consumption of the test diet. Compared with the 15-pollen standard, sesame pollen was the least preferred (P < 0.05) by young bees workers, with an relative consumption of 34%; sunflower pollen was about equally preferred, with an average of 58.5%; and rape pollen was the most preferred (P < 0.01), with an average consumption of >73%.

The cumulative mortality curves (percentage of cumulative mortality versus the number of days) of bees fed test pollens during the longevity test are shown in Fig. 2. Sesame pollen fed bees exhibited low mortality for approx. 30 d, after which mortality increased dramatically. The shapes of the cumulative mortality curves for the sunflower and 15-pollen groups were more gradual and similar to each other

The rape pollen group was long and gradual. All 4 pollen types prolonged honey bee life-spans when compared with the control (no-pollen) group. In the control group, all bees died by day 28. In contrast, at that time, the cumulative mortality was only 25% in the rape pollen group, 26% in the sesame pollen group, and 41% in both the sunflower and 15-pollen standard groups. The median life spans of the bees were 19 d for the controls, 31 d for sunflower, 33 d for sesame, 33 d for the standard pollen, and 51 d for rape pollen (Fig. 2). The increase in life span of bees fed pollen diets compared to control (no pollen) bees is shown for each quartile of mortality in Table 2.
No differences were observed among the pollen diets in terms of survival time to the 25% mortality level. By the 50% mortality level, bees consuming rape pollen survived significantly longer than those consuming the 15-pollen standard or sunflower. By the 75% mortality level, bees fed rape pollen were clearly surviving longer than those fed the other 3 pollens.

The cumulative pollen consumptions during the longevity tests are shown in Fig. 3. In all pollen groups, the consumption increased dramatically during the first 8-10 d, then decreased until about day 22, when pollen ingestion was no longer measurable. By the end of this pollen feeding period, the cumulative consumption of sunflower pollen was 73 mg per bee, a value significantly (P < 0.01) higher than that of bees fed sesame pollen (35 mg per bee), 15-pollen standard (44 mg per bee), or rape pollen (47 mg per bee).

We thank Phil Jankins, University of Arizona Herbarium, for assistance with species identification of sesame pollen, Emily Miller, Charles Shipman, Steven Thoenes, Stephen Buchmann for experimental assistance, Gary Richardson for statistical assistance annd James Hagler, Al Hook and Eric Erickson for manuscript reviews.