The developmental genetics and physiology of honeybee societies

Abstract

Eusocial animal societies, as diverse as those found in the ants, bees, wasps, shrimp and naked mole-rats, are structured around one or few reproductive females. The remaining females are helpers called 'workers' that are mostly sterile. A paradigm in studies of eusociality is that worker sterility is a key to societal functions because advanced sociality cannot be achieved when there is conflict over reproduction. Yet, traits such as sensory responsiveness, foraging and hoarding behaviour that change between female reproductive life stages also vary between workers. This variation is central to worker division of labour, a complex social trait believed to be instrumental for the ecological success of animal societies. Thus, we took a step back from established views on worker sterility and societal functions, and hypothesized that division of labour can be better understood if adaptive variation in worker behaviour is seen as emerging from pre-existing mechanisms associated with female reproduction. In exploring this reproductive ground plan hypothesis (RGPH) in honeybee workers, we established that variation in foraging division of labour correlates with ovary size and is affected by expression changes in vitellogenin, an egg yolk protein precursor. Here, we explain and reconcile the RGPH with data on honeybee sensory sensitivity, genomic mapping, transcript and endocrine profiling, and link our discussion with Ihle et al. (2010, this issue, pp. xx-xx). The findings bring together mechanistic and evolutionary explanations of honeybee worker behaviour. This essay suggests that a broader view on worker reproductive traits can increase the understanding of animal social behaviour.

Figure 1

Variation in foraging behaviour correlates with differences in sensory and reproductive traits in worker honeybees. Trait associations in selected high (a) and low (b) pollen-hoarding strain bees. Horizontal arrows show the timeline of worker ontogeny (Age). High strain bees emerge as adults with a larger ovary (white line-drawings inside bees), elevated sucrose responsiveness measured by the proboscis extension reflex (PER, blue bars), and develop a higher peak titre of vitellogenin yolk protein (black bars) as young adults compared with low strain bees. The vitellogenin level of high strain bees then drops rapidly and workers initiate foraging earlier in life than bees with low strain genotype. This difference in ‘age at first foraging’ (AFF) is indicated by violet circles, panel (a) versus (b). As foragers, high strain workers bias their collecting towards pollen (‘P’, bee in (a)), while low strain workers are biased towards nectar (‘N’, bee in (b)). (c) The corresponding trait correlations in wild-type (unselected) worker bees, which show considerable phenotypic variation (illustrated by ovary, PER, and vitellogenin symbols of various sizes). Black, connecting triangles indicate positive (+) correlations between vitellogenin expression, PER and ovary size, and between ovary size and pollen foraging, and negative correlations (−) between PER and age at foraging onset, and between ovary size and foraging onset. The white triangle specifies that PER is also correlated with foraging choice directly. These associations in wild-type bees reflect the same relationships as those seen in selected pollen-hoarding strain workers.

Figure 2

The reproductive ground plan hypothesis (RGPH) outlines how female reproductive biology may have been co-opted in the division of labour between foraging worker honeybees. Top panel: in ancestors of honeybees, the progression of reproductive development was linked to changes in female food-related behaviour. During periods of no active reproduction (top, right), the ovary was undeveloped (inactive, not enlarged with yolk) and the vitellogenin titre (black bars) was low. During reproduction, ovary activity and vitellogenin protein expression increased prior to oocyte yolk deposition, before vitellogenin returned to baseline (top, left), and only this maturational stage, or phase of the reproductive cycle, would be associated with pollen collection (‘P’) for nest provisioning of young. Below: similar to phenotypic aspects of the reproductive physiology of ancestors, high pollen-hoarding strain bees show rapid modulation (up–down) of vitellogenin (bottom, left), while low strain bees have more moderate, constant levels (bottom, right). The RGPH proposes that this differential physiology is linked via a pleiotropic gene network to the corresponding behaviour of the ancestor, explaining how pollen foraging is more prevalent in high strain workers compared to low strain bees. Full-line arrows in the top panel specify that reproduction perhaps was maturational in the ancestor (i.e. with one reproductive period). Dashed-line arrow indicates the alternative; a reproductive cycle that could be repeated.

Figure 3

The phenotype of wild-type vitellogenin (vg) gene knockdown workers is consistent with trait associations in high and low pollen-hoarding strain bees (Nelson et al. 2007). (a) RNAi-mediated knockdown of the vg gene (red (−) arrow) results in repressed vitellogenin protein levels (black bars) from early life, lowering the bees’ age at first foraging (AFF). This behaviour is similar to high strain bees (b) that show early decline of vitellogenin and low AFF during ontogeny. Thereafter, knockdown workers bias their foraging effort towards nectar (‘N’), similar to low strain bees (c) that do not experience high vitellogenin exposure as young adults. Thus, the early-life vitellogenin level predicts the foraging bias of worker bees, as indicated by the arrows from vitellogenin to the nectar and pollen (‘P’) symbols in the three panels. The white (+) and (−) arrows in (b) refer to ovarian signalling. They suggest that the larger ovary of high strain bees may explain the rapid up–down modulation and high peak levels of vitellogenin in this genotype, but a connection between large ovary size and early vitellogenin decline (the white (- ?) arrow) is not experimentally confirmed.

Figure 4

The reproductive ground plan hypothesis (RGPH) underscores that vitellogenin (black bars) and juvenile hormone (JH, grey arrow-bars) modules are inherent to insect reproduction (Amdam et al. 2007). Reproductive biology is influenced by environmental factors such as food availability sensed by nutrient-sensitive pathways (green). These pathways include the insulin receptor substrate (IRS) and target of rapamycin (TOR); while PDK1 is a phosphorylation target downstream of both IRS and TOR. The response is integrated by endocrine signal transmission via systemic hormones like JH and nuclear hormone receptors such as USP (blue) and HR46 (46, purple). The double repressor hypothesis (DRH) (Amdam & Omholt 2003) proposes that transition from nursing (mid panels) to foraging behaviour (right panels) in honeybees is influenced by a feedback loop (red (−) arrows) between vitellogenin and JH. Vitellogenin inhibits JH synthesis, USP expression and insulin-like signalling (Guidugli et al. 2005; Hunt et al. 2007; Ament et al. 2008) that are increased in foragers. When vitellogenin levels drop, JH can be triggered to increase and further suppress expression of its own repressor. We propose that bidirectional selection on pollen-hoarding behaviour acted on these relationships. (a) In high pollen-hoarding strain bees, nutrient sensing and reproductive endocrine signal transmission is increased, resulting in developmental retention of a larger ovary (filament structures inside larva, top left panel) and dynamic interplay between the ovary, vitellogenin, and adult behaviour (top mid (nurse) versus top right (forager) panel). (b) In low strain bees, nutrient sensing and reproductive endocrine signal transmission is less active (shown as less intense green and red indicators), resulting in reduced ovary size (fewer filaments retained inside larva, bottom left panel) and less sensitivity of interplay between ovary, vitellogenin, and behaviour in adults (bottom mid versus bottom right panel). Increased HR46 expression (Wang et al. 2009) might be central to the low strain phenotype, as this receptor has potential to competitively inhibit the signal transmission of JH and USP.