Bees collect food from flowers that differ in morphology, color, and scent. Nectar‐seeking foragers can rapidly associate a flower's cues with its profitability, measured as caloric value or ‘net energy gain,’ and generally develop preferences for more profitable species. If two flower types are equally easy to discover and feed from, differences in profitability will arise from differences in the volume or the sugar concentration of their nectar crops. Although there has been much study of how bees respond to one or the other of these two kinds of nectar variation, few studies have considered both at once. We presented free‐foraging bumblebees with two different types of equally rewarding artificial flowers. After a period of familiarization, we made one type more rewarding than the other by increasing its nectar concentration, volume, or both. Bees responded more rapidly to a change in the reward's sugar concentration than to a change in its volume, even if the profitability differences were approximately equal. Sucrose concentration differences (40% vs. 13%) caused bees to virtually abandon the more dilute flower type, whether both types offered the same volume (2 μl) or the less concentrated reward offered higher volume (7 μl vs. 0.85 μl). When the two types of flower differed only in nectar volume (7 μl vs. 0.85 μl), the less rewarding type continued to receive 22% of the visits. We propose three different hypotheses to explain the stronger response of the bees to changes in sugar concentration: (i) their response threshold to sucrose concentration might change; (ii) less time is needed to assess the concentration of a reward than its volume; and (iii) a smaller sample size may be needed for reliable estimation of profitability when flowers differ in concentration.

Diapause is the key adaptation allowing insects to survive unfavourable conditions and inhabit an array of environments. Physiological changes during diapause are largely conserved across species and are hypothesized to be regulated by a conserved suite of genes (a 'toolkit'). Furthermore, it is hypothesized that in social insects, this toolkit was co-opted to mediate caste differentiation between long-lived, reproductive, diapause-capable queens and short-lived, sterile workers. Using Bombus terrestris queens, we examined the physiological and transcriptomic changes associated with diapause and CO2 treatment, which causes queens to bypass diapause. We performed comparative analyses with genes previously identified to be associated with diapause in the Dipteran Sarcophaga crassipalpis and with caste differentiation in bumble bees. As in Diptera, diapause in bumble bees is associated with physiological and transcriptional changes related to nutrient storage, stress resistance and core metabolic pathways. There is a significant overlap, both at the level of transcript and gene ontology, between the genetic mechanisms mediating diapause in B. terrestris and S. crassipalpis, reaffirming the existence of a conserved insect diapause genetic toolkit. However, a substantial proportion (10%) of the differentially regulated transcripts in diapausing queens have no clear orthologs in other species, and key players regulating diapause in Diptera (juvenile hormone and vitellogenin) appear to have distinct functions in bumble bees. We also found a substantial overlap between genes related to caste determination and diapause in bumble bees. Thus, our studies demonstrate an intriguing interplay between pathways underpinning adaptation to environmental extremes and the evolution of sociality in insects.

Abstract. ''Peak shift'' is a behavioral response bias arising from discrimination learning in which animals display a directional, but limited, preference for or avoidance of unusual stimuli. Its hypothesized evolutionary relevance has been primarily in the realm of aposematic coloration and limited sexual dimorphism. Here, we develop a novel functional approach to peak shift, based on signal detection theory, which characterizes the response bias as arising from uncertainty about stimulus appearance, frequency, and quality. This approach allows the influence of peak shift to be generalized to the evolution of signals in a variety of domains and sensory modalities. The approach is illustrated with a bumblebee (Bombus impatiens) discrimination learning experiment. Bees exhibited peak shift while foraging in an artificial Batesian mimicry system. Changes in flower abundance, color distribution, and visitation reward induced bees to preferentially visit novel flower colors that reduced the risk of flower-type misidentification. Under conditions of signal uncertainty, peak shift results in visitation to rarer, but more easily distinguished, morphological variants of rewarding species in preference to their average morphology. Peak shift is a common and taxonomically widespread phenomenon. This example of the possible role of peak shift in signal evolution can be generalized to other systems in which a signal receiver learns to make choices in situations in which signal variation is linked to the sender's reproductive success.

In this study we describe colony development, larval development, worker reproduction and mating frequency of the queen in laboratory colonies of Bombus impatiens Cresson. Comparison of our observations with data from B. terrestris -the best studied bumblebee -revealed both similarities and dissimilarities between the two. Colonies of B. impatiens have a high number of workers (374.5±108), but relatively low number of young queens (9.4 ± 20) and males (8.9 ± 20). As in B. terrestris, size differences between the castes in B. impatiens were pronounced and became conspicuous as differences in the molting weight of the larvae, starting at the second instar. Workers are able to lay eggs and about 9 % of the workers in old colonies had mature oocytes in their ovaries. Despite this, workers were almost never observed to lay eggs in the presence of the queen, and the "competition phase" which is known from B. terrestris was never observed. Mating frequency of the queen was examined using molecular techniques. Two out of ten queens had mated with 2 different males, suggesting some degree of polyandry in the population.