A new model for caste development in social wasps

Abstract

Specialization into reproductive and non-reproductive castes is one of the defining traits of eusocial insects. Knowledge of the proximal causes of caste differentiation is therefore central to achieving an understanding of the evolution of eusociality. Castes are an example of a polyphenism, multiple, discrete phenotypes arising from a single genotype in response to differing environmental conditions. Here we focus on recent work in the social wasps to provide insight into how environmental conditions may trigger the development of caste across a range from independent- to swarm-founding social species. The amount of food larvae receive has long been recognized as a key input factor in the determination of caste, but that alone is insufficient to account for the range of combinations of size, development time, and caste among the female offspring of Polistes, an independent-founding wasp. Recent experimental work on P. fuscatus has shown that vibrations that are associated with the feeding of larvae are another essential environmental input in the determination of caste. We present a model of how vibrational signaling in the context of feeding larvae could interact with nutritional input to account for the developmental patterns seen in these wasps. Mapping the distribution of vibrational signaling onto a phylogeny of the social wasps suggests that this trait characterized the common ancestor of the subfamilies Vespinae + Polistinae, diversified in the independent-founding species, then was superseded by caste-determining mechanisms in the swarm-founding and vespine species that function more effectively in larger colonies. 

The most fundamental specialization of the eusocial insects is the division of colony members into two castes, workers (functionally sterile individuals) and reproductives.1 Understanding the proximal mechanisms leading to differentiation among individuals is therefore central to an understanding the evolution of eusociality itself.2 Because the wasp family Vespidae includes species representing a range of degrees of sociality from solitary to swarm-founding, the wasps can provide clues about what the early evolutionary steps toward caste-determining mechanisms might have been.3–6

The degree of differentiation between the queen (reproductive) and worker castes in the wasps ranges from none in the subfamily Stenogastrinae to major differences in size and morphology in the Vespinae (hornets and yellowjackets). In species with morphologically distinct castes, it is clear that developmental differentiation between gynes (future queens) and workers must begin in the larval stages. In many members of the large subfamily Polistinae, including the independent-founding genera Polistes and Ropalidia, the castes are morphologically identical, but manifest physiological differences that also must originate in the larval stages.7 This is supported by recent evidence that differences in the mRNA expression of several genes and in levels of a hexameric storage protein are already evident by the fifth (last) instar of larval development in Polistes metricus.5,8 There is no evidence for a genetic basis for caste differences in wasps. Instead, gyne and worker are examples of polyphenism, the epigenetic production within a single genotype of discrete forms of morphology, physiology or behavior in response to environmental differences.9 Here we review recent advances in our understanding of the nature of the environmental conditions triggering caste differentiation in the social wasps and propose a new model.

In the highly dimorphic hornets and yellowjackets, it is thought that food amount coupled with pheromones or oral secretions fed to the larvae by the workers combine to trigger the divergence of paths.10 For the non-dimorphic polistines such as Polistes, Ropalidia and others, the long-standing view is that differences in the quantity of nourishment received during the larval stage act as a “nutritional switch” to bias development toward one caste or the other.7,8,11–14 Female larvae fed ample food develop into gynes, whereas the less well-fed become workers.4 Food-supplementation studies have shown that colonies given nutrients ad libitum produce higher frequencies of female offspring with gyne-like traits, including larger size, more fat body, greater cold tolerance and higher diapause potential, than control colonies.15–19 Conversely, under-feeding produces smaller offspring.19 Workers, which are reared by the queen early in the colony cycle, are said to be the result of poor larval nutrition due to a low worker-to-larva (W/L) ratio at that time. The gynes, produced later, are the result of improved nutrition due to an increasing W/L ratio.5,13,20,21 Despite these pre-adult biasing effects, adults retain some behavioral and physiological plasticity to respond to the social context they find themselves in.7

Limits of the Nutritional-Switch Hypothesis

A mechanism based strictly on the amount of food received by larvae appears to be too simplistic, even for Polistes. This single-dimensional variable is insufficient to account for the multi-variate patterns of development seen in this and other independent-founding polistines.22 First, the rate of larval development, in insects generally positively correlated with resource availability,9 does not correlate well with size or caste. In Polistes, Mischocyttarus and Ropalidia, larval development times are shortest for the first few (worker) offspring, rise rapidly to a maximum for later-emerging workers, then gradually decline to intermediate durations for the remainder of the colony cycle, when gynes are produced.23–27 Second, the first-produced workers not only develop more rapidly than any subsequently produced females, but paradoxically they are also the smallest.24,26 Mead16 and Kudo28 concluded that the rapid growth of the first larvae in Polistes is due to more intensive feeding, but this is contradicted by studies providing supplemental feeding of larvae, which leads to both more rapid larval development and growth to larger, not smaller, adult size.22 Furthermore, the larger females produced later in the colony cycle include workers that do not differ measurably in size from gynes.21,24 In other words, size, development rate and caste vary semi-independently of one another to produce more combinations than can be explained by the single input variable of food quantity.

Possible Role for Vibrational Signals

It has recently been suggested that vibrational signaling provides a second source of input that may modulate the effect of nutrition quantity.22 Vibrational signals are especially conspicuous in the genera Polistes, Mischocyttarus, Ropalidia and Belonogaster. Females produce these signals either by rapidly shaking the body while standing on the nest, or by striking some part of the body against the nest. The movements are typically vigorous enough to produce audible sound and to cause the whole nest to shake or vibrate.29,30 Some categories of these signals are performed primarily by the dominant females while feeding the larvae. The best-known example is antennal drumming in Polistes, which is performed at the highest frequencies early in the colony cycle, when workers are being reared, and at much reduced rates later, when gynes are being reared.31,32 Recent experimental evidence for Polistes fuscatus has shown that larvae in nests subjected to simulated antennal drumming (pulses at 17 Hz) develop into adults with significantly less fat body than larvae given random-frequency vibration of the same intensity.33 Low body fat is a worker-like trait, whereas high body fat is a characteristic of gynes.34,35 That vibrational signals can have this effect is not surprising. Mechanical stressors are known to have dramatic biochemical and developmental effects on insects and other animals.22

We suggest the following scenario as one possible way nutritional and mechanical inputs could interact to lead to the observed patterns of caste, size and larval development time (Fig. 1). Early in the colony cycle, the first few queen-reared larvae are fed high amounts throughout their larval period,16,28 leading to their rapid development. The high frequency of feeding-coupled vibrations they are subjected to as larvae31,32 inhibits the synthesis of fat stores33 and other factors such as storage proteins that are essential for the capacity to undergo diapause,5,36 thus biasing their development toward a worker phenotype. Body size, which is loosely correlated with levels of fat and storage proteins, is reduced in these first emerging offspring.34,35 The next few cohorts of queen-reared offspring also emerge as workers24 due to the continuing high levels of vibrations.32 These workers have larger body sizes24 because of the higher nutritional input they receive amidst seasonally improving food availability21 and possibly to their greater total food intake due to their longer development time. Their longer development time as larvae may be due to low nutritional inputs received as 1^st- and 2^nd-instar larvae, when older larvae in the earlier cohorts were being preferentially fed. Later in the colony cycle, the worker-reared larvae experience high nutritional input due to the improved worker-to-larva ratio.21 These larvae experience low levels of vibrations compared to the earlier queen-reared larvae.31,32 The combination of high nutritional input and low vibrational input causes these worker-reared offspring to develop rapidly, have larger body sizes and higher fat stores and storage protein levels, i.e., a gyne phenotype. Workers continue to emerge, albeit in decreasing numbers,21 possibly due to uneven distribution of food along with individual differences in exposure to and sensitivity to vibrational stimuli.

Taxonomic Distribution of Vibrational Signaling

Mapping the taxonomic distribution of vibrational behavior in the context of feeding the larvae onto a phylogeny of the social Vespidae highlights its co-occurrence with independent-founding genera (Fig. 2). These genera are characterized by small colony sizes (typically well under 100 individuals) and absence of morphologically distinct castes.37 Whether vibrational signals in these other genera play a caste-biasing role similar to that reported in Polistes remains to be investigated, but the principle of parsimony suggests that they do. Wasps in the less-derived subfamily Stenogastrinae form even smaller societies, lack reproductive and worker castes,7 and also lack any evidence of vibrational signaling directed at the larvae.

The Vespinae present a more diverse picture (Fig. 2). Most form large colonies and show pronounced queen-worker dimorphism. The clearest account of vibrational signaling directed at the larvae is for Vespa tropica. When a founding queen returns to the nest with food for the larvae, she engages in prolonged tapping of the hind legs on the nest.41 This species is unusual in that colonies are among the smallest in the subfamily, with 5–15 workers at maturity, and gyne and worker are not morphologically different, although as in Polistes the gynes have much more fat body than workers,41 indicating caste differentiation originating in the immature stages. The more derived genera Vespula and Dolichovespula (Fig. 2) are the most dimorphic of the social wasps, with gynes up to 40% larger and morphologically and physiologically distinct from workers.37 Colonies are large and gynes are reared in specially constructed large cells, opening the possibility that workers could be cued to feed different amounts or qualities of food by the size of the cell a larva is in. Thus, a strictly nutritional switch could be acting, although pheromones are also suspected to play a role.10 Interestingly, vibrational signals have been reported for a number of species. Workers of Vespula consobrina occasionally drum their abdomens (gasters) vigorously on the nest during rounds of brood-cell inspection.42 This species also forms small colonies, with worker numbers of less than 100. Gaster drumming by founding queens or founding queens and workers has also been reported for Dolichovespula spp.41,43 and in young colonies of Vespula spp., although it is not clear whether it is associated with feeding the larvae.44

The four taxa that initiate colonies by means of swarms of queens and workers (Fig. 2) are also diverse, ranging from species with small colony size and lacking any queen-worker dimorphism or size difference45 to those with very large colonies and queens more than 20% larger than workers in some dimensions. Vibrational signaling directed at larvae has not been reported for any of the several hundred species in these groups, although most have not been studied at all. Queen-worker dimorphism in these species appears not to be derived from size-based allometry.37 In Apoica, for example, queens and workers overlap completely in body size, yet differ in body proportions.37 This means that the caste-differentiating mechanism cannot be triggered by a simple nutritional switch based on amount of food the larvae receive. Again, pheromones or oral secretions may be involved.

Evolutionary History of Mechanisms of Caste Differentiation

These patterns of occurrence begin to illuminate the evolution of environmental triggers of caste-determining developmental pathways. Vibrational signals appear to be a behaviorally applied mode of caste biasing that may be effective only in small colonies. They may have originated as unspecialized forms of mechanical stress, perhaps applied directly to the larvae. The more specialized signals characterizing extant species may be ritualized forms of this early behavior. One scenario is that vibrational caste-biasing characterized the common ancestor of the clade Vespinae + Polistinae (Fig. 2A), diversified in form and function among the independent founders, then was superseded in the more derived taxa by other mechanisms, such as pheromones or differences in food quality. These may well be more effective mechanisms of caste biasing, able to affect hundreds or thousands of larvae, rather than a few tens. These mechanisms, which have evolved independently in several social vespid lineages, may have been instrumental in enabling the evolution of the larger colony sizes seen in the more derived vespines and in the several clades of swarm founders (Fig. 2).

Abbreviations

W/L	=	worker-to-larva ratio
JH	=	juvenile hormone
IIS	=	insulin/insulin-like signaling pathways

Figures and Tables

Figure 1 Model for the interaction of vibrational signaling and nutrition in caste biasing in Polistes. (A) Hypothesized changes in the temporal pattern of vibration and levels of nutrition received per larva per hour across the colony cycle.32 (B) Correlated changes in the development time, body size and caste of emerging adults across the colony cycle.21,24 See text for explanation.

Figure 2 Cladogram of the eusocial groups of vespid wasps. VS?, occurrence or not of vibrational signals performed by adults in association with feeding the larvae. MC?, occurrence or not of morphological differences between queen and worker. Taxa shown in black are independent-founding wasps; those shown in red are swarm founders. Cladogram based on several authors.38–40

Acknowledgments

We are grateful to B.J. Taylor, T. Schueller and A. Toth for critically reading the manuscript. Research supported by the College of Agricultural and Life Sciences, University of Wisconsin-Madison.