Genome Evolution in Social Insects

Abstract

Sociality is a complex phenotype which has evolved multiple times across the tree of life, particularly in the insects. Insect sociality likely arises via the evolution of the genome, with changes in specific genes, gene families, and gene regulatory mechanisms linked to social traits. In turn, sociality itself can influence patterns of genome evolution through the effects of smaller effective population sizes, longer generation times, and reproductive division of labor. Moreover, particular social insect lineages have some noteworthy genome features which may influence both trait and genome evolution, including elevated recombination rates, haplodiploidy, and supergenes. This review explores the genomic causes and consequences of social traits in insects, emphasizing the importance of using evolutionary perspectives to study these processes across diverse social lineages.

Significance.

Researchers have long sought to understand both the causes and consequences of sociality. Beginning with the sequencing of the honey bee genome in 2006, genomic perspectives on these topics have been widely explored, leading to a deeper understanding of how social traits arise and how they may impact the evolution of the genome. As the genomic basis for sociality is increasingly identified, it appears that a combination of shared and distinct mechanisms give rise to social traits across different insect taxa. Likewise, the impact of sociality on the genome is complex. Here, we highlight recent discoveries that expand our understanding of how insect sociality both arises from and then influences genome evolutionary processes, highlighting both gaps in our understanding of these topics and important areas for future study.

Introduction

The emergence of obligate sociality is considered a major evolutionary transition, in many ways analogous to the transition from unicellular to multicellular life (Szathmáry and Smith 1995). As such, social organisms and particularly social insects have been the focus of a tremendous amount of research, including in genomics, beginning with the sequencing of the honey bee genome (Honeybee Genome Sequencing Consortium 2006). Genomic approaches have been used to ask countless questions about sociality in insects, including those focused on identifying the underlying causes of social lifestyles, as well as those centered on understanding how sociality itself impacts the evolution of the genome.

Sociality can be conceptualized as a spectrum, running from fully solitary to eusocial, wherein species exhibit some combination of defining social traits (Fig. 1). That spectrum has been further subdivided in diverse ways by different researchers in the field. Historically, eusocial species have been defined as those exhibiting obligate group living and cooperative brood care; reproductive division of labor between reproductive and nonreproductive or less-reproductive castes or morphs; and overlapping generations of adults present together in a nest (Michener 1969). Less-social groups—variously referred to as social, subsocial, incipiently social, or primitively social—possess some subset of those characteristics, which may be expressed only facultatively or polymorphically across populations of a single species (Wilson 1971; West-Eberhard 1996; Rehan and Toth 2015). Here, we primarily focus on genome evolution in fully eusocial insects, except where otherwise noted.

Fig. 1.

Presence of defining social traits across insect lineages of varied sociality. Here, we outline the distribution of key social traits across various insect lineages and species, covering all five orders in which eusociality has evolved, and with a particular focus on lesser-known taxa. Species are typically considered to be eusocial if they have overlapping generations in a single colony, display cooperative brood care, and have reproductive division of labor. Some social groups also produce multiple worker morphs. The distribution of these traits varies: some taxa, like the vespid wasp Parischnogaster nigricans (Cameron, 1902), have overlapping generations and cooperative brood care, but any individual is capable of reproducing (Turillazzi 1989), whereas other wasps, some bees, the ambrosia beetle Austroplatypus incompertus Browne, 1971, the ants, and the termites display all three eusocial traits. Social thrips and social aphids are typically considered eusocial, although their nonreproductive morphs do not exhibit brood care in a canonical sense, and instead perform defensive roles (Abbot and Chapman 2017). Likewise, some social Hymenopterans and thrips do not exhibit a canonical division of reproductive labor; for example, the “nonreproductive” dwarf eldest daughter morphs of the bee Ceratina calcarata Robertson, 1900 may be capable of reproduction, but do not survive overwintering and therefore do not have the opportunity to reproduce (Rehan and Richards 2010), while some social thrips species have subfertile soldiers (Abbot and Chapman 2017).

Eusociality has evolved at least nine times independently in insects (Fig. 2). Eusociality is present in many Hymenoptera, with a single origin in the common ancestor of the ants and between four and fifteen origins in various bee and wasp lineages (Gibbs et al. 2012; Kocher and Paxton 2014; Peters et al. 2017; da Silva 2021). The Blattodea (cockroaches and termites) also have a single origin of eusociality, at the origin of the termites (Inward et al. 2007). Finally, some wood-boring beetles, particularly the ambrosia beetle Austroplatypus incompertus Browne, 1971, are eusocial (Choe and Crespi 1997). Some thrips and some aphids are also typically considered eusocial, although they lack cooperative brood care (Aoki 1977; Choe and Crespi 1997; Kirkendall et al. 1997).

Fig. 2.

Phylogenetic distribution of sociality in insects. a) Phylogeny of all insect orders, with triangles at the tips indicating the approximate number of species per order on a square-root scale; bolded tip labels and blue-and-green hatched triangles indicate that some taxa in the order are eusocial, while nonbolded tip labels and blue triangles indicate that the order contains no eusocial taxa. b–f) Phylogenies of the orders Thysanoptera, Hemiptera, Coleoptera, Hymenoptera, and Blattodea. Lineages that contain eusocial taxa are indicated in bold, and lineages that contain exclusively eusocial taxa are underlined. Order-level phylogeny modified from (Misof et al. 2014); species count estimates are from (Stork 2018); the Thysanoptera phylogeny is modified from (Tyagi et al. 2020); the Hemiptera phylogeny is modified from (Cryan and Urban 2012); the Coleoptera phylogeny is modified from (Zhang et al. 2018); the Hymenoptera phylogeny is modified from (Peters et al. 2017); and the Blattodea phylogeny is modified from (Wang et al. 2017).

How Have Genomes Evolved to Produce Sociality?

Theoretical Predictions for the Genomic Basis of Sociality

Ultimate Genomic Causes: The Haplodiploidy Theory of Social Evolution

Many social insects are haplodiploid, including bees, wasps, ants, and eusocial thrips. In these groups, females develop from fertilized, diploid eggs, while males develop from unfertilized, haploid eggs via arrhenotoky (Heimpel and de Boer 2008). Haplodiploidy has long been theorized to play a causal role in the evolution of eusociality, due to its effects on patterns of relatedness, kin selection, and inclusive fitness (Hamilton 1964; Trivers and Hare 1976; Fromhage and Kokko 2011; Johnstone et al. 2012). Initially, Hamilton argued that, because haplodiploidy results in females that share three quarters of their alleles with their sisters, it promotes the evolution of reproductive altruism and thus eusociality (Hamilton 1964). However, haplodiploidy also causes females to share only one quarter of their alleles with their brothers, disfavoring altruism; this pattern thus requires factors like female-biased sex ratios or male-biased dispersal in order for reproductive altruism to actually be favored (Trivers and Hare 1976; Gadagkar 1991; Johnstone et al. 2012). Further weakening the link between eusociality and haplodiploidy is the fact that eusociality is found in nonhaplodiploid groups like the termites and the ambrosia beetle A. incompertus (Nowak et al. 2010), while some haplodiploid groups, like sawflies, parasitic wasps, and most thrips, are not social. Thus, the degree to which haplodiploidy actually promotes the evolution of eusociality has been an area of ongoing debate (Nowak et al. 2010; Joshi and Wiens 2023; Suresh and Linksvayer 2026).

This uncertainty may in large part be due to statistical and analytical challenges posed by the distribution of eusociality and haplodiploidy on the insect phylogeny. Whereas the Hymenoptera comprise a large proportion of the transitions to sociality in insects (Fig. 2), they represent just a single origin of haplodiploidy (Bonifacii et al. 2026). As highlighted by Bonifacii and colleagues (2026), this pattern creates pseudoreplication that is problematic for many of the phylogenetic comparative methods used to assess correlated trait evolution (for example, Pagel's test for correlated evolution) (Maddison and FitzJohn 2015; Gardner and Organ 2021; Boyko and Beaulieu 2023). A recent test of the haplodiploidy hypothesis, which used methods designed to avoid this statistical issue, failed to find evidence that haplodiploidy facilitates the evolution of eusociality (Bonifacii et al. 2026). At present, it appears unlikely that further phylogenetic analyses will find strong support for the haplodiploidy hypothesis, barring updates to our understanding of the insect phylogeny or the discovery of eusociality and haplodiploidy in lineages where these traits have not yet been detected.

Proximate Genomic Causes: Rewiring Traits Found in Solitary Ancestors

In addition to exploring ultimate genomic explanations for the evolution of eusociality, researchers have sought to identify proximate mechanisms that underlie this complex trait (ie specific evolutionary changes in patterns of gene expression or in the activity of gene networks). Three widely invoked, nonmutually exclusive hypotheses make predictions about the proximate causes of eusocial phenotypes: the ovarian groundplan hypothesis (West-Eberhard 1996), the maternal heterochrony hypothesis (Linksvayer and Wade 2005), and the reproductive groundplan hypothesis (Amdam et al. 2004; Page and Amdam 2007). All three hypotheses propose that eusocial phenotypes evolve by reorganizing and redeploying components of ancestral, solitary life cycles.

The ovarian groundplan hypothesis posits that reproductive division of labor evolves by shifting the regulation of the reproductive cycle, such that pre- or post-reproductive traits, genes, and signaling pathways are expressed in workers while reproductive traits, genes, and signaling pathways are expressed in queens (West-Eberhard 1996). For example, workers exhibit traits like inactive ovaries and brood-care behaviors, while queens have active ovaries and engage in egg-laying. Investigations of this hypothesis often specifically focus on ovarian development and juvenile hormone signaling (West-Eberhard 1996; Grozinger et al. 2007).

Because it is impossible to access ancestral solitary taxa before the transition to sociality and compare them to extant social lineages, tests of the ovarian groundplan hypothesis are necessarily indirect, and have typically not involved direct comparisons of gene expression between solitary and social taxa. For instance, in early work on the ovarian groundplan hypothesis, Grozinger and colleagues predicted that honey bee nurses engaged in brood care should have patterns of gene expression that are similar to those of queens, while foragers should be less similar to queens (Grozinger et al. 2007). Using microarrays and published gene expression data, they found that nurse-biased genes are also upregulated in queens and reproductive workers, while forager-biased genes are not, a pattern consistent with their predictions (Grozinger et al. 2007). Later, Saleh and Ramírez (2019) leveraged the primitively eusocial orchid bee Euglossa dilemma (Bembé and Eltz, 2011) to test the ovarian groundplan hypothesis. In this species, solitary females initially forage, then found a nest, lay eggs, and guard their nest; when offspring emerge, the foundress establishes dominance over one daughter, who remains at the nest and primarily forages (Saleh and Ramírez 2019). Based on the ovarian groundplan hypothesis, the researchers predicted that patterns of gene expression which underlie the solitary foundresses' transition from foraging to guarding might also characterize differences between dominant reproductive foundresses and subordinate nonreproductive daughters (Saleh and Ramírez 2019). While patterns of gene expression in the brain did recapitulate this pattern, ovary gene expression did not, providing mixed support for the ovarian groundplan hypothesis (Saleh and Ramírez 2019).

Researchers have also attempted to test the ovarian groundplan hypothesis by assessing whether reproductive division of labor involves the differential expression of genes broadly linked to reproduction and diapause. In the derived, secondarily queenless ant genus Diacamma, dominant reproductive females upregulate genes involved in nutrient mobilization, which can be important for reproduction in insects, while subordinate non-reproductive females upregulate storage proteins, which have been associated with reproductive diapause (Okada et al. 2015). Likewise, the differential expression of vitellogenin, a storage protein involved in reproduction, and its paralogs across castes in Formica ants has been interpreted as supporting the ovarian groundplan hypothesis (Morandin et al. 2019).

A second framework frequently used to explore the evolutionary and genomic basis for sociality is the maternal heterochrony hypothesis, initially formulated by Linksvayer and Wade (2005). One key feature of eusociality is sib-social care, with workers caring for siblings rather than reproducing. The maternal heterochrony hypothesis posits that this sib-social care results from maternal care genes becoming expressed ontogenetically earlier (ie pre-reproductively) in workers (Linksvayer and Wade 2005). Linksvayer and Wade also specifically posited that this heterochronic shift is likely to result from evolutionary changes to a small number of genes. Investigations of the maternal heterochrony hypothesis typically test the prediction that gene expression patterns in post-reproductive individuals providing maternal care should be most similar to those of workers providing sib-social care, rather than to those of pre-reproductive individuals who do not yet engage in maternal care behaviors. This prediction has been supported in some taxa (the primitively eusocial wasp Polistes metricus Say, 1831 [Toth et al. 2007], the facultatively social carpenter bee Ceratina calcarata Robertson, 1900 [Rehan et al. 2014], and the primitively eusocial orchid bee E. dilemma [Saleh and Ramírez 2019]), but not in others (the eusocial bumble bee Bombus terrestris [Linnaeus, 1758] [Woodard et al. 2014]).

Finally, the reproductive groundplan hypothesis predicts that task specialization among social insect workers emerges by co-opting the genes and signaling pathways that regulate the reproductive cycle in solitary ancestors, such that expression of pre- or post-reproductive genes characterizes workers of different behavioral roles (Amdam et al. 2004; Page and Amdam 2007). In its initial formulation, the hypothesis specifically aimed to explain the mechanisms that cause foraging specialization on either pollen or nectar in worker honey bees (Amdam et al. 2004; Page et al. 2006; Roth et al. 2014). The reproductive groundplan hypothesis has since been applied to other forms of division of labor among eusocial insect workers.

Investigations of the hypothesis tend to focus on vitellogenin (Ihle et al. 2010; Corona et al. 2013), as well as ovarian (Graham et al. 2011; Pamminger and Hughes 2017), juvenile hormone, and ecdysone signaling processes (Amdam et al. 2004; Page and Amdam 2007). Findings from honey bees were interpreted as early support for the reproductive groundplan hypothesis: Graham and colleagues (2011) found that some quantitative trait loci (QTL) associated with variation in foraging behavior are also associated with variation in ovary size, indicating that genes which regulate the reproductive cycle might also cause differences in foraging preferences among honey bee workers. However, the QTL identified are large and contain many genes (Graham et al. 2011), and so based on this work it remains unclear whether foraging preference and ovary size are actually underlain by the same genes.

One group of proteins which are differentially produced across the reproductive cycle is vitellogenins, precursors of the major yolk storage protein vitellin (Tufail and Takeda 2008). Consistent with the reproductive groundplan hypothesis, vitellogenins appear to play a role in task division among social insect workers. For instance, altering the expression of vitellogenin using RNA interference (RNAi) provided manipulative evidence that, in some honey bee genotypes, expression of this protein influences workers' foraging preferences (Ihle et al. 2010). Though causality has yet to be demonstrated, a similar mechanism may function in other eusocial taxa: vitellogenin paralogs are differentially expressed between nurses and foragers in the harvester ant Pogonomyrmex barbatus (Smith, 1858) (Corona et al. 2013), and between builders and foragers in the termite Macrotermes bellicosus (Smeathman, 1781) (Elsner et al. 2021).

From each of these hypotheses, we can make specific predictions about shared patterns of expression for the key subset of genes that underlie particular behaviors. For example, under the maternal heterochrony hypothesis, we might specifically predict that the genetic mechanisms which cause females from solitary lineages to engage in brood care should be the same as those which cause workers in social lineages to care for brood. Crucially, such predictions cannot be effectively tested simply by comparing overall patterns of gene expression between solitary females, social workers, and social queens, because overall gene expression patterns reflect many differences in behavior, physiology, and other processes, not solely those which underlie variation in propensity to care for brood. In other words, the maternal heterochrony hypothesis does not predict that solitary females and social workers will share global patterns of gene expression, only that they have common patterns of expression for genes involved in care-related behaviors. Until those mechanisms are rigorously characterized, it will be challenging to meaningfully test these hypotheses. Moreover, most studies that are framed as tests of these hypotheses are limited by the fact that they assess patterns of gene expression only within social species, even though the hypotheses are predicated upon comparisons between solitary and social insects. Ideally, these hypotheses would be tested by contrasting relevant genetic mechanisms in ancestral (ie extinct) solitary taxa and in extant social groups, but this approach is, of course, impossible. As an approximation, we would suggest that researchers leverage socially polymorphic species, so as to compare patterns of gene expression between closely related social and solitary lineages through ontogeny. Examples of such species include some halictid bees, like Halictus sexcinctus (Fabricius, 1775) (Richards et al. 2003) and Lasioglossum albipes (Fabricius, 1781) (Plateaux-Quénu et al. 2000). In these species, social populations have a life cycle in which mated queens found a nest and then rear an initial brood of workers, which in turn help rear a second brood of reproductives that mate and found new nests. In contrast, in solitary populations mated females found nests and directly rear a brood of reproductives. These polymorphic species thus provide an excellent opportunity to first characterize and then compare the genetic basis for reproductive physiology and care behavior across a set of closely related solitary and social lineages.

Another framework, which seeks to explain the repeated evolution of sociality, the social genetic toolkit hypothesis draws on concepts from evolutionary developmental biology to posit that the activity of shared sets of genes with conserved roles give rise to division of labor across independent origins of sociality (Toth et al. 2007, Toth and Robinson 2009). In Toth and Robinson's original formulation, the hypothesis focused specifically on shared patterns of brain gene expression, but subsequent investigations of the hypothesis analyzed expression patterns in other tissues as well. Toth and Robinson also suggested that the genetic toolkit would involve genes associated with feeding and reproduction (Toth and Robinson 2009), though subsequent work has not necessarily focused on these processes.

Reported support for the genetic toolkit hypothesis is mixed. A number of studies have found minimal or nonsignificant overlap between the sets of genes that are differentially expressed between queen and worker morphs across independent origins of sociality (Ferreira et al. 2013; Berens et al. 2015; Taylor et al. 2021), though this pattern is not universal (Warner et al. 2019) and overlap has also been detected at the level of gene modules (Favreau et al. 2023).

This mixed support across studies is perhaps not surprising, for both methodological and conceptual reasons. From a methodological standpoint, studies which aim to test the genetic toolkit hypothesis have often focused on patterns of gene expression in the head or brain (Woodard et al. 2014; Berens et al. 2015; Standage et al. 2016; Taylor et al. 2021). While the original formulation of the hypothesis did focus on brain gene expression, it is possible that gene expression in other tissues plays an important role in establishing social phenotypes (Kapheim 2017); for instance, the ovaries play a role in ecdysteroid synthesis (Hartfelder et al. 1995), and the fat body synthesizes vitellogenin (Corona et al. 2007). Thus, studies focused on the brain would not be able to detect relevant and potentially shared patterns in other organs. Moreover, bulk RNA sequencing can obscure important processes which occur at the level of tissues or cells.

Likewise, many studies have either sampled only adults (Woodard et al. 2014; Standage et al. 2016; Taylor et al. 2021), which have already completed the process of queen-worker differentiation, or have compared different developmental stages across taxa (Berens et al. 2015). Relatively few studies have either explored patterns of gene expression in multiple body parts and in standardized developmental stages across taxa (but see [Warner et al. 2019], which investigated expression patterns in pooled whole eggs, larvae, and pupae and in pooled heads, mesosomas, and metasomas from adult honey bees and Monomorium pharaonis [Linnaeus, 1758] pharaoh ants).

From a conceptual standpoint, the genetic toolkit hypothesis does not make specific, testable predictions about the extent to which patterns of gene expression are expected to be shared across independent origins of sociality (Toth and Robinson 2009), a problem which has been identified by others in the field (Mikheyev and Linksvayer 2015;  Weitekamp et al. 2017). If, theoretically, social phenotypes are caused by the activity of a small set of genes, then shared activity of just this small set of genes across lineages would be in line with the genetic toolkit hypothesis. If, on the other hand, social phenotypes result from the differential expression of a large proportion of a species' genes, then a much greater degree of overlap in expression patterns across taxa would be needed to support the genetic toolkit hypothesis. Because the genetic basis for sociality remains far from completely understood, it is impossible to say which of these two scenarios best reflects the mechanistic basis and evolutionary history of sociality across insect lineages.

Perhaps in part because of this lack of clarity, researchers have also interpreted an enrichment of highly conserved pathways or functional categories (ie Gene Ontology and Kyoto Encyclopedia of Genes and Genomes terms) among differentially expressed genes as evidence in support of a genetic toolkit (Woodard et al. 2014), even for instances in which patterns of gene expression were found not to be shared across taxa (Berens et al. 2015; Taylor et al. 2021) or when patterns of gene expression were not compared across taxa (Standage et al. 2016). This pattern led Berens and colleagues to propose the “loose genetic toolkit” hypothesis, wherein independent origins of sociality are posited to involve the deployment of common biological pathways and functions, rather than shared sets of genes (Berens et al. 2015). These key functions tend to include the regulation of behavior, metabolic processes, reproduction, and transcription regulation (Woodard et al. 2014; Berens et al. 2015; Standage et al. 2016; Warner et al. 2019; Taylor et al. 2021).

On the other hand, independent origins of sociality may be the result of distinct sets of genes (Ferreira et al. 2013; Berens et al. 2015); these may even be novel, lineage-specific genes (Feldmeyer et al. 2014; Kapheim et al. 2015). However, identifying truly novel genes, rather than genes that have lost detectable homology to other genes due to rapid sequence evolution, is challenging (Weisman et al. 2020), and we expect that improved methods for identifying novel genes will clarify the extent to which these loci are involved in the evolution of sociality (Casola 2018; Arendsee et al. 2019; Weisman et al. 2020). Ultimately, both shared and lineage-specific genes may play important roles in producing social phenotypes (Mikheyev and Linksvayer 2015; Warner et al. 2019).

Empirical Evidence for the Genomic Basis of Sociality

In attempting to understand the genomic basis for sociality, researchers have identified a wide variety of specific loci, pathways, and selective mechanisms that may underlie social traits. While a detailed description of all such findings is beyond the scope of this review, here we discuss key methods used to tackle the topic and highlight some important results.

Using transcriptomic approaches in social species, researchers have identified specific candidate genes and pathways that may underlie social phenotypes. Across lineages, some genes have been repeatedly linked with social traits by differential gene expression studies. These genes include foraging, which encodes a cGMP-dependent protein kinase that has been associated with diverse sensory and cognitive functions, and which is expressed in the nervous system, among other tissues (Lucas and Ben-Shahar 2021). It has been implicated in generating both age-based behavioral transitions in bee and ant workers, and distinct morphological castes in ants and termites (reviewed in [Lucas and Ben-Shahar 2021]). Similarly, multiple studies indicate that the gene vitellogenin, which encodes a yolk protein precursor, and in some taxa, its homologs, may play a role in generating trait differences both between reproductive and nonreproductive individuals, and across different nonreproductive castes (Corona et al. 2007, 2013; Feldmeyer et al. 2014; Patalano et al. 2015; Morandin et al. 2019; Awde et al. 2020). Likewise, a body of gene expression research has implicated several interrelated molecular pathways in generating social traits, including insulin-like growth factor signaling (Chandra et al. 2018; Séité et al. 2022), target of rapamycin signaling (Korb et al. 2021), juvenile hormone signaling (Jongepier et al. 2018; Ju et al. 2023), and Hippo signaling (Barkdull and Moreau 2026).

Other studies have leveraged intraspecific comparisons based on the natural variation present in socially polymorphic or facultatively social species to characterize the genomic architecture of sociality. For instance, in the socially polymorphic bee Lasioglossum albipes, a genome-wide association study found eight regions of the genome that are significantly associated with sociality, indicating that sociality in this species is a complex, polygenic trait with both regulatory and protein-coding underpinnings (Kocher et al. 2018). Methods which directly assay the activity of regulatory loci can also be applied in an intraspecific context, and have been used to characterize differences in enhancer activity between social and solitary populations of L. albipes, linking selection on cis-regulatory elements with the evolution of social behavior in this species (Jones et al. 2024). In another socially polymorphic bee, Ceratina australensis (Perkins, 1912), Harpur and Rehan (2021) used genome-wide associations and haplotype-based approaches to show that loci linked to sociality have evolved under positive selection, suggesting that the evolution of social nesting in this species may be driven by the fitness benefits of sociality. Using gene expression and population genetic data, Kapheim and colleagues also found sociality-associated genes evolve under positive selection in the facultatively social bee Megalopta genalis Meade-Waldo, 1916, and identified many mutations in gene regulatory regions which may underlie the evolution of social behavior in this species (Kapheim et al. 2020a).

Finally, more than a decade of comparative genomics research in social insects has described broad patterns of selection that are associated with insect sociality, and which may have played a causal role in its evolution. Comparative genomics studies have demonstrated the importance of natural selection on both protein-coding genes, and on noncoding loci which are likely to have gene regulatory functions. For instance, in a study which focused on social and solitary bees, Shell and colleagues (2021) found that positive selection on protein-coding genes is associated with increased social complexity. Evidence from sweat bees, which have repeatedly gained and lost sociality, indicates that genes in social species also experience a reduction in relaxed selection relative to solitary species (Jones et al. 2023). Comparative studies have also identified patterns of evolution in putative regulatory elements among social species. For example, early work which contrasted the genomes of seven ants and the honey bee against genomes from 22 solitary insects found that sociality is associated with consistent differences in the abundance of transcription factor binding motifs in putative gene promoters (Simola et al. 2013a). Similarly, work in sweat bees found that putative regulatory elements in social species' genomes are enriched for some transcription factor binding motifs (Jones et al. 2023). Finally, comparative studies have also identified key gene functions that have evolved under selection linked to sociality, including functions linked to the nervous system, gene regulation, and chemical communication (Simola et al. 2013a; Kapheim et al. 2015; Harrison et al. 2018; Rubin et al. 2019; Shell et al. 2021; Jones et al. 2023).

Across approaches and scales, these studies indicate that sociality is a complex, polygenic trait which has, at least in part, a unique mechanistic basis across different evolutionary origins, with some overlap of genomic substrates at a broad scale (ie evolution of regulatory loci; involvement of common signaling pathways).

Identifying the specific molecular mechanisms which produce eusocial phenotypes has been and continues to be extremely challenging. In part this is because eusociality is a complex phenotype which emerges from the activity of many loci. Moreover, though the basis for eusociality has been intensely studied in obligately eusocial lineages like the ants, honey bee, and termites, each of these groups evolved eusociality tens of millions (many bees, and wasps) to more than 100 million years ago (ants and termites) (Peters et al. 2017; Harrison et al. 2018). This long evolutionary history presents challenges for identifying the earliest genomic changes which lead to social phenotypes. These groups also lack natural variation in the presence of eusociality, precluding researchers from using closely related but socially variable lineages to untangle the molecular basis for sociality. Studies focused on socially polymorphic lineages or lineages that have more recently evolved social phenotypes may help address some of these challenges.

A Role for Gene Family Evolution in Producing Social Traits

One molecular evolutionary mechanism that has been hypothesized to underlie some social traits is gene family evolution, ie evolutionary changes in the size of gene families which occur due to gene duplications, deletions, and pseudogenization (Hahn and Kern 2005). Eusocial insects display the remarkable capacity to produce differentiated adult forms (often referred to as castes or morphs) from the same genomic blueprint. This developmental plasticity necessitates genomic flexibility; gene duplications have been predicted to play an important role in generating this flexibility during the evolution of social insects (Gadagkar 1997). Duplications reduce selective constraint, and can permit the evolution of new gene functions by neofunctionalization or subfunctionalization of one or both gene copies. Consistent with this prediction, across the entire honey bee genome, duplicated genes are more likely to have caste- and sex-biased patterns of expression than nonduplicated genes (Chau and Goodisman 2017). Extensive gene duplications, with duplicates displaying caste-biased expression, are also apparent in the termites (Shigenobu et al. 2022), and duplicated genes tend to have caste-biased expression in bumble bees (Xu and Colgan 2025).

Specific gene pathways and functional classes appear to experience repeated duplications or gene family expansions across social insect genomes. Many of these are linked to the production of different adult castes via developmental plasticity. For instance, the development of caste differences across social taxa involves differential nutrition, with downstream signaling likely being transduced in part via insulin signals (Chandra et al. 2018; Séité et al. 2022). Hymenopterans typically have two insulin-like peptide genes, ILP1 and ILP2; differential expression of ILP2 is associated with caste identity in ants (Chandra et al. 2018). In Blattodea, including termites, there is an additional insulin-like peptide duplication, and one copy exhibits caste-biased expression in termites (Kremer et al. 2018). Juvenile hormone signaling is also repeatedly implicated in social insect caste differentiation. In termites, there are duplications of multiple juvenile hormone synthesis genes; these duplicates are caste-biased in their expression patterns (Jongepier et al. 2018). Likewise, in ants the size of the juvenile hormone acid methyltransferase gene family is positively correlated with the degree of queen-worker dimorphism exhibited by a species (Vizueta et al. 2025). Another gene repeatedly linked to social insect caste differences is vitellogenin (vg), which encodes a lipid-carrying protein that plays a variety of biological roles including yolk formation, antioxidant activity, and immunity (Morandin et al. 2014). All insects appear to have three vg homologs (vg, vg-like-A, and vg-like-B). An additional homolog, vg-like-C, is Hymenoptera-specific (Morandin et al. 2014). The four vg homologs have caste-biased patterns of expression across multiple social Hymenopteran lineages, although the exact patterns of caste bias appear to be evolutionarily labile (Nelson et al. 2007; Morandin et al. 2019; Awde et al. 2020). Ants have an additional four ant-specific copies of vitellogenin, which again have caste-biased patterns of expression (Corona et al. 2013; Feldmeyer et al. 2014; Kohlmeier et al. 2018).

Genes involved in chemical communication also appear to have experienced substantial gene family evolution in eusocial Hymenoptera. Across insect lineages broadly, odorant receptor gene families expand and contract through evolutionary time in a birth-death process (Legan et al. 2021). There are many duplications and gene family expansions among odorant receptors in social Hymenopterans (Karpe et al. 2016; Legan et al. 2021; Vizueta et al. 2025), and these expansions may facilitate complex nestmate recognition and other social behaviors in these groups. Ant genomes also contain Formicidae-specific chemosensory protein gene duplicates, which display a tendency to evolve rapidly and experience further duplication events (Kulmuni et al. 2013). However, a broad comparative analysis across three independent origins of sociality in Hymenoptera failed to find evidence that odorant receptor gene families in social lineages are either repeatedly expanded or consistently evolve at elevated rates (Gautam et al. 2024). Thus, it may be the case that gene family evolution in odorant receptors does not contribute substantially to the evolution of sociality in Hymenoptera; alternatively, it may do so in very lineage-specific ways that would not be detected by methods which test for convergent evolutionary patterns.

Like lineage-specific genes, gene duplicates remain technically challenging to identify. They are often missing from genome assemblies due to assembly errors which collapse multiple paralogs into a single, misassembled sequence. Downstream, the use of genomes produced using varied sequencing and assembly methods in comparative projects can exacerbate this issue, although the adoption of long-read sequencing for genome scaffolding may ameliorate these technical challenges. Finally, accurate expression analysis and functional manipulation of duplicated genes is challenging when paralog sequences remain highly similar, as both gene copies may be targeted by antibodies, CRISPR guide RNAs, etc, hindering paralog-specific investigation.

Whether gene family evolution plays a particularly important role in the evolution of social insects relative to other groups remains an open question. Aspects of social insect biology still make this hypothesis plausible; for example, many gene duplication events result from unequal crossing over during meiotic recombination, and so it is possible that the elevated recombination rate found in aculeate Hymenoptera causes a higher rate of gene duplication in social bees, wasps, and ants. Moreover, while most gene copies are lost from the genome following duplication events, it is possible that the developmental plasticity necessitated by reproductive division of labor acts to selectively favor the retention of duplicates in social insect genomes as a mechanism for relieving intralocus conflict (Xu and Colgan 2025). However, efforts to link gene family evolution with trait evolution are strongly affected by biases in taxon sampling (Couto et al. 2023; Gautam et al. 2024). To rigorously test hypotheses regarding the relationship of gene family evolution to sociality, we urge researchers to maximize taxonomic sampling so as to mitigate the impact of, for example, lineage-specific outlier changes in gene family size (eg the significant expansion of odorant receptors in the ant Ooceraea biroi [Forel, 1907] relative to other ants [Gautam et al. 2024]). Additionally, even when biologically plausible patterns of gene family evolution are detected in social lineages relative to solitary relatives, these comparative studies are inherently correlative. While it is impossible to directly test the evolutionary causality of events, including gene family expansions or contractions, which occurred in deep time, we encourage researchers to deploy manipulative experiments like gene knock-outs and knock-ins, where possible, to test the role of gene duplicates or gene losses in producing social traits.

The Role of Gene Regulatory Mechanisms in Generating Plastic Social Phenotypes

The reproductive division of labor that characterizes eusocial insects results from differential expression of genes and proteins throughout ontogeny. A variety of mechanisms regulate gene expression, including the activity of regulatory elements; changes in chromatin accessibility; DNA methylation; and the activity of noncoding regulatory RNAs (Fig. 3). Evolution of these gene regulatory mechanisms is likely an important process underlying sociality.

Fig. 3.

Genome features underlying gene regulation that have been implicated in social insect phenotypic plasticity. Social insects can exhibit several forms of phenotypic plasticity, including a) reproductive division of labor between reproductive and less- or non-reproductive morphs; b) age-based polyethism, in which workers progress through distinct behavioral states with age; and c) worker polyphenism, in which multiple, morphologically distinct worker morphs are produced. Multiple pre- and post-transcriptional gene regulatory mechanisms have been associated with these forms of plasticity, including cis-regulatory elements, histone modifications, DNA methylation, noncoding RNAs, and alternative splicing. These mechanisms are not mutually exclusive: multiple mechanisms may act to generate plastic phenotypes.

Cis-regulatory elements are noncoding sequences that regulate gene expression by facilitating transcription factor binding. Cis-regulatory elements can be identified computationally, for instance, by identifying transcription factor binding motifs or by finding noncoding regions with high sequence conservation through evolutionary time (ie conserved noncoding elements or noncoding alignable regions). Cis-regulatory elements can also be identified empirically, using approaches like ATAC-seq, CHIP-seq, or STARR-seq. Computational approaches tend to be more commonly applied in nonmodel species, including most social insects (but see, for example, Simola et al. 2013b; Jones et al. 2025).

Cis-regulatory elements may evolve in a number of different ways between social and solitary lineages. For instance, natural selection favoring the capacity to regulate gene expression in ways that underlie social insect caste differences can manifest as shifts in relative rates of sequence evolution. Consistent with a complex landscape of selective pressure on regulatory elements in social species, putative regulatory elements exhibit both more increases and more decreases in evolutionary rate in social bees relative to solitary bees than would be expected by chance (Rubin et al. 2019). Regulatory elements also evolve more quickly following secondary losses of sociality in sweat bees (Hymenoptera: Halictidae) (Jones et al. 2023). Regulatory element evolution can also underlie traits which represent elaborations of eusociality. For instance, in ants, lineages with novel worker morphs that develop via phenotypic plasticity have greater selective constraint on regulatory loci compared to lineages that produce only a single type of worker (Barkdull and Moreau 2026).

Increases in the rate of evolution of cis-regulatory elements are typically interpreted as a signal of relaxed selection; however, we strongly caution that this is not guaranteed to be true. Rate increases are also consistent with positive selection toward new functions; moreover, regulatory elements can maintain conserved functionality even after undergoing substantial sequence evolution, to the point of being nonalignable across taxa (McDonald and Reed 2024). Particularly because most studies in social insects identify putative cis-regulatory sequences computationally, additional empirical work will be needed to confirm both the functional role of these sequences, and the functional implications of any evolutionary changes to their sequences.

Selection on cis-regulatory elements can also manifest in ways beyond changes in evolutionary rate. For example, among cis-regulatory elements that are shared between social and solitary groups, there may be turnover in the identities of transcription factor binding motifs within the regulatory sequence. This process may have been involved in the evolution of sociality in ants and honey bees, where there is some evidence for convergent binding motif turnover relative to solitary Hymenoptera (Simola et al. 2013a).

Finally, social lineages may also have changes in the counts of transcription factor binding motifs within regulatory elements, with increases typically interpreted as conferring greater ability to flexibly regulate gene expression. Indeed, there are more transcription factor binding motifs in the putative, computationally identified gene promoters of social bees compared to both ancestrally solitary (Kapheim et al. 2015) and secondarily solitary bees (Jones et al. 2023). In particular, there appear to be more binding site gains in the promoters of genes whose coding sequences are under stronger selective constraints in social bees, indicating concurrent selection for regulatory flexibility and coding sequence conservation.

Gene expression is regulated by many mechanisms in addition to the activity of cis-regulatory sequences. In eukaryotes, including insects, nuclear DNA is tightly compacted around histone proteins to form chromatin. For genes to actually be transcribed, they must be unwound from histones and made accessible to transcription factors and other proteins. This accessibility is mediated in part by epigenetic modifications to histones, and so these modifications may help regulate developmental plasticity in social insects. Across eusocial insect lineages, histone modifications and chromatin state have been linked to plastic development, including via experimental manipulations. For instance, honey bee queen and worker larvae have distinct chromatin states, and these distinct states are correlated with differences in gene expression (Wojciechowski et al. 2018). In carpenter ants, experiments using pharmacological manipulations directly implicate histone acetylation in the maintenance of normal foraging behavior in differentiated, adult workers (Simola et al. 2013b; Glastad et al. 2020). Likewise, silencing of histone-modifying genes impairs normal soldier development in termites (Suzuki et al. 2019).

DNA methylation, wherein methyl groups are added to DNA and change gene expression without changing gene sequence, is another epigenetic mechanism hypothesized to be involved in social insect plasticity. Studies have suggested that methylation may play a critical role in queen-worker differentiation in honey bees (Kucharski et al. 2008; Herb et al. 2012); caste-specific methylation has also been found in termites (Glastad et al. 2016) where it is linked to caste-specific alternative splicing (Harrison et al. 2022). However, most studies that interrogate the role of methylation in social insect phenotypes have faced major technical challenges which limit researchers' ability to confirm its involvement, and there is currently no clear relationship between level of sociality and the extensiveness of DNA methylation across lineages (Oldroyd and Yagound 2021b). There is no association between methylation and reproductive status in the clonal raider ant O. biroi (Libbrecht et al. 2016), and experimental knockdown of methylation genes does not impact soldier-worker differentiation (Suzuki et al. 2019). Thus, the role of methylation in social phenotypes remains equivocal.

Gene expression can also be regulated post-transcriptionally, for instance by alternative splicing or through the activity of microRNAs and long noncoding RNAs. Each of these mechanisms may play a role in social insect plasticity (Ashby et al. 2016; Shields et al. 2018; Gao et al. 2020; Du et al. 2024; Luo and Zhou 2025), although studies sometimes fail to link post-transcriptional regulation with social phenotypes (Kapheim et al. 2020b). Additional investigation, aided by advances in sequencing technologies, will likely clarify the role for these mechanisms in generating social traits.

While cis-regulatory elements have been studied in evolutionary and comparative contexts, other regulatory mechanisms have yet to be considered through this lens. For instance, how have patterns of histone modification evolved across social insects, and what are the molecular mechanisms responsible for any evolutionary change? Do social insect genomes contain distinct or novel sets of regulatory RNAs, compared to solitary insects? Continued attention to the evolution of these gene regulatory mechanisms will likely shed light on the evolution of important social traits across insects.

Genome Features Peculiar to Social Insect Lineages

Recombination Rates in Social Insects

Recombination, via chromosome segregation and meiotic crossing-over, is a fundamental process in sexually reproducing organisms which results in offspring that have different combinations of alleles than do either of their parents. By breaking up existing combinations of alleles and producing new ones, recombination increases the efficacy of natural selection (Hill and Robertson 1966; Felsenstein 1974). Recombination rates vary across taxa, and this variation may be driven by either neutral or adaptive processes (Wilfert et al. 2007).

Aculeate Hymenoptera (ants, bees, and wasps, including both social and solitary groups) have an extremely wide range of recombination rates (DeLory et al. 2024). Moreover, compared to solitary aculeates, social lineages have very high rates of recombination relative to genome size (Wilfert et al. 2007; Sirviö et al. 2011; Jones et al. 2019). Several hypotheses have emerged to explain this pattern. For instance, eusociality carries costs, like higher pathogen risk; these costs may be mitigated by mechanisms like increased recombination rate that increase genetic diversity within a colony (Gadau et al. 2000; Crozier and Fjerdingstad 2001). Recombination also facilitates the continued production of novel genotypes across immunity-related loci, which is favored by co-evolution of hosts and parasites under the Red Queen hypothesis (Sirviö et al. 2006; DeLory et al. 2024). Increased genetic diversity may also promote task specialization within a colony, allowing it to respond to its environment (Jones et al. 2019). Finally, eusocial Hymenopteran queens often exhibit extreme longevity, which is associated with genomic damage via oxidative stress and transposable element accumulation. These factors may act to favor the evolution of increased recombination rates (DeLory et al. 2024). The increase in genetic drift in groups with smaller effective population sizes may also select for increased recombination, as recombination can reunite fitness-increasing alleles (Otto and Barton 2001). Eusocial Hymenoptera have reduced effective population sizes due to their strong reproductive skew, and so demography may help explain the evolution of higher rates of recombination in these groups (DeLory et al. 2024).

However, recent work by Everitt and colleagues investigating recombination in termites has indicated that elevated recombination may not be a universal feature of eusocial insect genomes (although that study did not sample termites' closest nonsocial relatives, the nontermite cockroaches) (Everitt et al. 2025). This suggests that high recombination rates in eusocial Hymenoptera may be driven by another factor, like male haploidy or male-biased sex ratios among reproductives (Everitt et al. 2025). Additional work in non-Hymenopteran social lineages, like ambrosia beetles, thrips, or aphids, is needed to clarify the relationship between sociality and recombination rates.

Even if there is no causal link between sociality and elevated rates of recombination, in those social taxa which do have elevated recombination rates, this phenomenon may affect genome evolution. For instance, higher recombination rates increase the efficacy of selection by disrupting linkage between alleles with opposing impacts on fitness (Kent and Zayed 2013). Thus, high recombination rates in some social lineages could lead to selection—either positive or purifying—that is stronger and targeted to smaller genomic intervals. For the same reason, increased recombination reduces the extent of hard selective sweeps, genetic hitchhiking, and background selection, and therefore facilitates the maintenance of genomic variation (Nachman 2002). On the other hand, mismatch repair during recombination can cause gene conversion, wherein the sequence on the chromosome with the mismatch is replaced by the homologous allele on the other chromosome; this process can actually accelerate the rate at which genetic variants, even deleterious ones, become fixed (Galtier and Duret 2007; Webster and Hurst 2012). Biases in mismatch repair also favor transmission of guanine or cytosine alleles over adenine or thymine at polymorphic sites, resulting in GC-biased gene conversion; this phenomenon has been shown to occur at higher rates in the social honey bee compared to the solitary leafcutter bee Megachile rotundata (Fabricius, 1787) (Jones et al. 2019). Thus, in social lineages characterized by high recombination rates, like social bees, we may expect that genomes will evolve under stronger selection and may show evidence for biases in allele fixation as a result of gene conversion. Further research rigorously characterizing recombination rates in a diverse array of social lineages will be needed to test these predictions.

Haplodiploidy Facilitates Purging of Deleterious Variation

Haplodiploidy, which characterizes many (but not all) social insect lineages, has direct consequences for genome evolution. Because deleterious alleles exist in only a single copy in haploid males, they will be more directly acted upon by natural selection, even when partially or completely recessive. This phenomenon has been demonstrated in the Northern paper wasp Polistes fuscatus (Fabricius, 1793), where deleterious alleles are preferentially purged in haploid males (Miller and Sheehan 2023). This effect of haplodiploidy may more generally explain the reduced genetic variation and reduced inbreeding depression that has been observed in haplodiploid lineages (Hedrick and Parker 1997). It may have even led to the evolution of the very high recombination rates that are observed in the genomes of haplodiploid social insects, as a means of reducing linkage between favorable and deleterious alleles (Miller and Sheehan 2023; Everitt et al. 2025).

Supergenes Can Underlie Sociality-Associated Traits

Supergenes are sets of loci that are both inherited together as a unit due to strong genetic linkage, and that affect complex phenotypes (Turner 1967; Thompson and Jiggins 2014; Charlesworth 2016). By reducing or eliminating recombination and keeping co-adapted or favorable alleles together, supergenes can promote adaptation (Berdan et al. 2022). While no supergenes have been identified that underlie eusociality itself, more than a decade of research has demonstrated that supergene loci underlie social structure phenotypes in several ant lineages. The first social supergene was described in the red imported fire ant, Solenopsis invicta Buren, 1972 (Wang et al. 2013). The species exhibits a social polymorphism, although both morphs are eusocial: it produces both single-queen (monogyne) and multiple-queen (polygyne) colonies, which differ in several dimensions, including queen size, nest founding strategy, queen dispersal, and male fitness (Keller & Ross 1993; Lawson et al. 2012). The S. invicta social polymorphism is underlain by two alleles of a supergene, termed SB and Sb. Monogyne colonies contain only SB homozygous females, whereas polygyne colonies have SB/Sb queens and both SB/Sb and SB/SB workers. The supergene encodes more than 470 genes, including an odorant binding protein initially thought to underlie colony type. The Sb allele includes three large inversions, and transposable elements may have been involved in the initial formation of the supergene (Huang et al. 2018; Stolle et al. 2019; Yan et al. 2020). In addition, the S. invicta supergene has repeatedly introgressed into other Solenopsis species, where it also appears to regulate social polymorphisms (Stolle et al. 2022).

Social supergenes have also been identified in several Formica species, including Formica glacialis Wheeler, 1908 and F. podzolica Francoeur, 1973, where two linked supergenes underlie a sex ratio-bias phenotype and a queen number phenotype (Lagunas-Robles et al. 2021); F. cinerea Mayr, 1853, where a supergene produces a queen miniaturization phenotype (Scarparo et al. 2023); and F. selysi Bondroit, 1918, F. neoclara Emery, 1893, and F. francoeuri Bolton, 1995, where supergenes underlie a queen number phenotype (Purcell et al. 2014; Brelsford et al. 2020; McGuire et al. 2022; Pierce et al. 2022). Putative supergenes linked to social polymorphisms have also recently been described in the harvester ant Pogonomyrmex californicus (Buckley, 1867), where alternative colony structure phenotypes may be underlain in part by a supergene (Errbii et al. 2024), and in Myrmica ruginodis Nylander, 1846, where a supergene appears to produce polymorphism in queen size, colony founding strategy, and queen number (Sigeman et al. 2025).

While supergenes are typically thought of as being adaptive, these social supergenes can impose fitness costs and act as selfish genetic elements. In F. selysi, one of the queen number supergene alleles is a maternal effect killer, causing hatching failure for half of the offspring of heterozygous mothers (Avril et al. 2020). In S. invicta, the Sb supergene allele reduces reproductive potential in heterozygous queens and lowers sperm counts in hemizygous males (Lawson et al. 2012). Moreover, it acts as a selfish genetic element, biasing transmission in males (Hettesheimer et al. 2025) and underlying a green-beard effect in females, where workers carrying the Sb allele recognize and kill queens which lack the allele, making the SB allele effectively homozygous lethal in queens. These selfish genetic effects generate genomic conflict and may select for the emergence of compensatory suppressor loci (Ross and Shoemaker 2018).

Additional supergenes will likely be identified in social insects, and numerous outstanding questions about these supergenes remain. For instance, what are the specific mechanisms by which supergenes produce alternative phenotypes? Are these mechanisms shared across taxa with similar social polymorphisms (eg queen number), or do independent mechanisms underlie trait convergence? Leveraging developmental transcriptomics and, where possible, functional genetic approaches may provide insights into these questions.

While the effects of a supergene architecture on the molecular evolution of the supergene region itself are fairly well understood (Purcell and Brelsford 2025), how do supergenes impact genome-wide patterns of molecular evolution? The S. invicta supergene has significant trans-regulatory effects on the rest of the genome, influencing gene expression, chromatin accessibility, and regulatory element activity (Jones et al. 2025). Do other social supergenes exert such genome-wide effects? How might this mediate patterns of selection?

Finally, is there taxonomic bias in the distribution of social insect supergenes? With the exception of several inversions that may underlie local adaptation to altitude in honey bees (Wallberg et al. 2017; Christmas et al. 2019; Everitt et al. 2023), putative supergenes have not been described in bees, wasps, termites, or other social insects. Is this a reflection of real patterns of supergene distribution in social lineages, or is it because supergenes underlie readily noticed phenotypes in the ants? Additional investigation, particularly in non-Hymenopteran social insects, will be required to answer these questions.

Evolutionary Consequences of Sociality

Genome-wide rates of molecular evolution vary widely across taxa, potentially due to a range of mechanisms including the efficiency of DNA repair machinery, variation in generation times, and demographic traits (Bromham 2009). The presence or absence of sociality does not itself appear to influence rates of molecular evolution (Bromham and Leys 2005; Rubin 2022). However, eusocial species differ from solitary species in key traits that are expected to impact genome-wide rates of molecular evolution, and these traits can also vary across eusocial species (Fig. 4).

Fig. 4.

Social traits which may influence patterns of genome evolution. a) Reproductive division of labor is expected to reduce effective population size. Given equivalent census sizes, solitary species (top of panel a), in which all individuals may reproduce, will have larger effective population sizes than social species (bottom of panel a), in which only some individuals reproduce, while others are nonreproductive. In this example, both species have a census size of 21 individuals, with 21 reproductive individuals in the solitary species (top of panel a, light blue) compared to one reproductive individual (bottom of panel a, dark green) and 20 nonreproductive individuals (bottom of panel a, light green) in the social species. b) Larger colony sizes should lead to smaller effective population sizes, if carrying capacities are the same. In this example, at a carrying capacity of 15 individuals, the species with smaller colonies have three reproductive individuals (top of panel b, dark green), while the species with larger colonies has only one reproductive individual (bottom of panel b, dark green). c) Two opposing predictions relate longer generation times to patterns of genome evolution. If per-year mutation rates are equivalent across taxa (panel c, top), then social species (lower portion of top panel, dark green) will accumulate a greater number of mutations than solitary species (upper portion of top panel, light blue). If, however, species with longer generation times have lower per-year mutation rates (panel c, bottom), then social species (lower portion of bottom panel, dark green) will accumulate fewer mutations than solitary species (upper portion of bottom panel, light blue).

The Potential Impact of Demographic Shifts on Genome Evolution in Social Taxa

Some traits which underlie sociality are expected to cause demographic shifts which may shape patterns of genome evolution in social species relative to solitary ones, primarily via effects on effective population size. Given the premise that carrying capacities are similar for social and nonsocial groups, reproductive division of labor is expected to cause reduced effective population sizes in social insects (Fig. 4a). In species with smaller effective population sizes, natural selection acts less efficiently, while the effects of genetic drift are stronger (Ohta 1987). As a result, social species and particularly those with large colonies should evolve under weaker positive and purifying selection, show evidence for relaxation of selection, and have lower total genetic diversity, relative to solitary groups (Charlesworth 2009).

These predictions have been explored across a range of social taxa. Early work using population transcriptomic data from three social hymenopterans, two butterflies, and one mosquito found lower within-species genetic diversity, a higher ratio of nonsynonymous to synonymous substitutions, and a higher rate of slightly deleterious mutations in the social species, patterns which are consistent with reduced effective population sizes (Romiguier et al. 2014). Crucially, however, this study did not control for the effects of shared ancestry when testing correlations between sociality and patterns of molecular evolution, complicating interpretation of the results. Later, Imrit and colleagues (2020) used population genomic and transcriptomic data to compare patterns of molecular evolution in a larger set of social hymenopterans relative to two flies and one butterfly. They found that, when controlling for phylogeny, purifying selection is weaker in social species (Imrit et al. 2020). A subsequent comparative genomic analysis of 10 social and 159 solitary Hymenoptera also reported relaxation of selection in social lineages, though this pattern was detectable only when excluding Anthophila bees, which experience strongly relaxed selection across all species regardless of sociality (Weyna and Romiguier 2021).

The preceding studies exclusively focused on sociality in Hymenoptera; in Blattodea, Ewart and colleagues (2024) found mixed support for a relationship between sociality and reduced effective population sizes when comparing patterns of genome evolution in social termites relative to solitary cockroaches. Termite genomes exhibit an overall shift toward relaxed selection, but do not harbor a greater number of deleterious mutations, as would be expected if small effective population sizes reduced the efficacy of selection (Ewart et al. 2024). Ewart and colleagues propose that this pattern suggests that termites possess some kind of mechanism to more effectively purge deleterious mutations, a hypothesis that would require additional testing. On the whole, the available evidence does suggest that sociality is associated with patterns of genome evolution that are consistent with reduced effective population sizes, although this pattern varies across lineages (Weyna and Romiguier 2021) and is not consistently recovered for all genome evolution metrics (Ewart et al. 2024).

Among social lineages, larger colony sizes cause an increase in reproductive skew, potentially further reducing effective population sizes if carrying capacities are similar across species (Fig. 4b). Supporting this prediction, across seven social hymenopterans Imrit and colleagues (2020) found that the genomes of species with larger colony sizes evolve under the weakest purifying selection, consistent with the prediction that larger colony sizes result in smaller effective population sizes. However, a subsequent study that included many more social taxa found patterns in rates of synonymous and nonsynonymous substitution that are inconsistent with the predicted negative relationship between colony size and effective population size, suggesting that colony size may instead mediate genome evolution via other mechanisms, perhaps through an effect of elevated fecundity (Rubin 2022). Thus, additional efforts are needed to understand how colony size may influence genome evolution in social insects.

In addition to the effects of colony size, reproductive skew is also mediated by worker reproduction. Despite the fact that reproductive division of labor is considered a defining characteristic of eusocial organisms, the degree to which reproduction is actually partitioned among individuals varies across species. Eusocial insect workers may totally lack key reproductive organs and be obligately sterile, as in some ants (eg Berndt and Eichler 1987); they may lay unfertilized eggs which develop into males (in fact a substantial proportion of males may be produced by workers, as in some Meliponine bees [Koedam et al. 2007]); or workers may be capable of sexual reproduction and produce both male and female offspring, as in some Ponerine ants (Peeters et al. 2000). Any such direct reproduction by workers should mitigate the reduction in effective population size caused by sociality. A comparative genomic analysis of 22 ant species found support for this hypothesis: species with reproductive workers are less likely to evolve under relaxed selection than those with obligately sterile workers (Barkdull and Moreau 2023). Whether this pattern holds across other social clades with variable degrees of worker reproduction, however, remains to be seen.

Rigorously characterizing the impacts of reproductive division of labor, colony size, and worker reproduction on genome evolution is challenging. Studies assume that effective population sizes are universally smaller in social species than in nonsocial ones, but this is almost certainly a significant simplification given the significant ecological differences and long divergence times that separate many social taxa from their closest solitary relatives. For ecological reasons, some solitary insect species can reasonably be expected to have quite small effective population sizes (for example, solitary bees that are extreme dietary specialists). Even between closely related social species, effective population size has been shown to vary (Liu et al. 2026). Therefore, we argue that in an era of affordable DNA sequencing and abundant high-quality reference genomes, the field should go beyond testing hypotheses predicated on the assumption that sociality universally reduces effective population size. Instead, we encourage researchers to make attempts to directly quantify effective population sizes in social taxa relative to nonsocial lineages using modern approaches based on genetic data (reviewed in [Waples 2024] and [Waples 2025]). These inferences can even be conducted with historical museum specimens, as was recently done for the social bumble bee Bombus franklini (Frison, 1921) (Schweizer et al. 2025). While methods for genetic inference of effective population size are not themselves free of limitations (Waples et al. 2016; Gargiulo et al. 2024), we anticipate that they will allow researchers to more robustly test hypotheses about the relationship between sociality and effective population size, and the consequences that these traits have for ongoing genome evolution in insects.

Generation Time and Genome Evolutionary Rates

Social insect reproductive individuals are typically long-lived by insect standards, and so average generation times are longer. Opposing predictions have been made regarding the relationship between generation time and genome evolution in social insects (Fig. 4c). Typically, in species with longer generation times germline cells may go through fewer replications per unit time, leading to a lower mutation rate per unit time, a pattern that has been detected across many eukaryotes (Thomas et al. 2010; Bergeron et al. 2023; Lewin and Eyre-Walker 2025). Based on this pattern, some researchers predict that longer generation times should lead to slower rates of genome evolution in social insects (Rubin 2022; Ewart et al. 2024). On the other hand, continuous oogenesis in Hymenoptera may actually result in the opposite pattern for social bees, ants, and wasps: as females produce larger numbers of eggs over longer lives, more germline cell divisions will have occurred to produce each egg. Thus, long generation times coupled with high fecundity might instead result in faster rates of molecular evolution for social hymenopterans (Bromham and Leys 2005).

To date, few studies have explored these predictions about the effect of increased generation times on patterns of genome evolution in social insects, and reported patterns are inconclusive. In termites, Ewart and colleagues interpreted lower rates of synonymous substitution relative to cockroaches as the result of relatively longer generation times in termites (Ewart et al. 2024). In contrast, across two clades of ant and one clade of bees Rubin (2022) found elevated rates of synonymous substitution as colony size, a trait generally correlated with generation time, increased. Additional work is needed to characterize the relationship between generation time and genome evolution in social insects, and particularly to ascertain whether continuous oogenesis leads to a departure from the general pattern of slower evolutionary rates in longer-lived groups. Clarifying the relationship between generation times and genome evolution in social insects could be facilitated by directly measuring mutation rates in social taxa whose reproductives vary in lifespan. For example, this could be done by repeatedly sampling workers from marked colonies to quantify the rate at which offspring accumulate mutations as the colony reproductive ages, targeting taxa with both shorter (eg Odonotomachus [Haskins and Haskins 1980]) and longer (eg Pogonomyrmex [Tschinkel 2017]) lifespans.

Selection on Nonreproductive Individuals May be Weakened

Genome evolution in social insects may also be shaped by the fact that social taxa produce many nonreproductive or less-reproductive individuals, whose phenotypes are not acted upon directly by natural selection. While genes expressed in reproductives directly impact fitness and are thus directly exposed to selection, genes expressed in nonreproductives influence fitness only inasmuch as they affect the fitness of reproductive individuals, and are therefore only indirectly exposed to selection. This phenomenon is expected to result in weaker positive and purifying selection on genes whose expression is biased toward nonreproductives, and a shift to relaxation of selection on these genes.

Some studies have found patterns that may align with these predictions. For instance, over the course of development in the ant M. pharaonis, genes that are upregulated in queens also have a significantly higher proportion of amino acid substitutions inferred to have been fixed by positive selection than do genes that are upregulated in sterile workers or genes that are not differentially expressed, both of which evolve under positive selection at similar rates (Warner et al. 2017).

Other studies have also found differences in evolutionary rates between nonreproductive- and reproductive-biased genes, but are limited in their ability to test predictions related to indirect selection by their methodology. For instance, early work in honey bees found that genes with worker-biased expression in adult brains have slower evolutionary rates than other genes, whereas genes with queen bias evolve at faster rates (Hunt et al. 2010). However, a slower evolutionary rate can result from either a reduction in positive selection or an increase in purifying selection (Wertheim et al. 2015), making it impossible to ascertain whether the findings of Hunt and colleagues support or contradict the hypothesized effect of kin selection on worker genes. Likewise, a population genetic study in a single termite species found that genes with soldier-biased expression have elevated nonsynonymous to synonymous mutation rates, compared to genes whose expression is biased toward workers or reproductives (Radford et al. 2022). While pattern does suggest that soldier-biased genes are characterized by distinct patterns of molecular evolution; however, Radford and colleagues were not able to infer the specific selective regimes that these genes experience, as dN/dS is not an appropriate metric of positive selection at the population level (Kryazhimskiy and Plotkin 2008).

In other instances, researchers have failed to find support for the predicted relationship between expression bias and selection. In the wasp Polistes canadensis (Linnaeus, 1758), whose workers retain the ability to leave their natal nest and become queens, Ferreira and colleagues found no evidence that worker-biased genes evolve at a different rate than queen-biased or unbiased genes (Ferreira et al. 2013). Likewise, a study of four termite species found no evidence that genes whose expression is biased toward sterile workers evolve under relaxed selection relative to other genes (Harrison et al. 2021). Finally, across an ant, a honey bee, and a wasp, population genomic analyses of negative selection indicate that worker-biased genes evolve no differently than queen-biased genes, except that in later-stage honey bee larvae queen-biased genes do evolve under negative selection more frequently (Imrit et al. 2020).

Thus, the prediction that reproductive division of labor weakens selection on genes primarily expressed in nonreproductives has to date received mixed support. There are a number of potential explanations for this pattern. Most studies of gene expression bias quantify expression patterns at only one or a few developmental time points, and so may not fully capture important patterns of gene expression that are involved in generating and maintaining differences between reproductive and nonreproductive individuals. Moreover, despite early predictions that the evolution of distinct reproductive and nonreproductive castes would require that each caste express a highly distinct set of genes (Gadagkar 1997), gene expression studies to date have found that very few genes are expressed exclusively in nonreproductive individuals (Morandin et al. 2017). This pattern suggests that the majority of genes have the potential to experience direct selection when expressed in reproductives, although we also note that most of these studies used bulk- or organ-level RNA sequencing, which can fail to detect important gene expression differences at the cell or tissue level. Therefore, additional characterization of caste-specific patterns of gene expression at greater temporal and spatial resolution may clarify the relationship between caste, indirect selection, and genome evolution.

Sociality May Select for Genomic Imprinting

Sociality is expected to favor the evolution of parent-of-origin gene expression effects like genomic imprinting (ie the differential expression of maternal and paternal alleles). As outlined below, this prediction emerges from the kinship theory of genomic imprinting (Haig 2000).

In diploid organisms, selection will typically favor equal expression of both maternal and paternal alleles; this helps ensure that genes are expressed at optimal levels, can mitigate the impact of deleterious mutations, and in some cases causes heterozygote advantage (Oldroyd and Yagound 2021a). However, exceptions to this general rule are predicted to occur when there are differences in average relatedness between offspring and each parent (Haig 2000), as may frequently be the case in social insects. Asymmetric relatedness should favor differential expression of alleles from each parent, which can be produced by the evolution of a mechanism for genomic imprinting, like epigenetic marks that indicate which parent each allele was inherited from (Oldroyd and Yagound 2021a).

Among social insects, haplodiploidy is one prominent cause of asymmetric relatedness. In haplodiploid taxa with singly mated females (monandry), a mother's daughters share all of their paternal alleles, but only 50% of their maternal alleles on average. This pattern means that mothers accrue less of a fitness benefit from their daughters altruistically forgoing direct reproduction in favor of sibling care, relative to the benefit accrued by fathers (Oldroyd and Yagound 2021a). Therefore, in monandrous haplodiploid social insects, genes which increase selfish worker reproduction should be under selection to increase their expression level, when they are inherited from the mother (Oldroyd and Yagound 2021a).

This pattern should be reversed by polyandry, even in haplodiploid lineages. Polyandry occurs in many social insect lineages (Crozier and Fjerdingstad 2001), though it appears to be rare in termites (Wu et al. 2014). In polyandrous taxa, individuals share all of their maternal alleles with their siblings, but share paternal alleles with only some siblings (Queller 2003). In a social insect colony, this pattern of asymmetric relatedness should favor increased expression of paternal alleles that cause worker individuals to behave selfishly (Queller 2003; Oldroyd and Yagound 2021a). This can include paternal alleles of genes that help a worker obtain more resources during development, or that promote direct worker reproduction, even when these phenotypes impose a cost to the colony as whole (Oldroyd and Yagound 2021a).

Patterns of gene expression that are consistent with these predictions have been empirically described in both monandrous and polyandrous haplodiploid social Hymenoptera, albeit in only a limited number of species. For instance, under nest conditions that allow worker reproduction to occur in the monandrous bumble bee B. terrestris, many genes exhibit strong expression biases toward the maternal allele, particularly in genes with reproductive functions (Marshall et al. 2020). Likewise, in the polyandrous honey bee, reciprocal crosses across different genetic stocks reveal elevated expression of paternal alleles in reproductive tissues and in reproductively active workers (Galbraith et al. 2016). Moreover, developing honey bee queens display stronger expression bias toward paternal alleles than do developing workers, again a pattern consistent with the prediction that in polyandrous taxa paternal alleles should be expressed in a manner which favors individual reproductive capacity (Bresnahan et al. 2025).

Genomic imprinting may also exist in diploid termites, where it has been proposed as a mechanism for caste determination (Matsuura et al. 2018). Some termites, including Reticulotermes speratus (Kolbe, 1885), produce three different reproductive morphs: primary, alate kings, and queens, which develop from nymphs; neotenic kings and queens that develop from nymphs; and neotenic kings and queens that develop from workers (Matsuura 2020). Some species also produce a fourth reproductive morph, neotenic queens that arise via parthenogenesis of primary queens (Matsuura 2020). The caste fate of offspring is influenced by the combination of maternal and paternal morph (Hayashi et al. 2007; Kitade et al. 2011; Matsuura et al. 2018). While this was originally interpreted as evidence for genetic caste determination (Hayashi et al. 2007), further study has demonstrated that this cannot be the case (Matsuura et al. 2018). Instead, genomic imprinting has been proposed as a model to explain this phenomenon (Matsuura et al. 2018; Matsuura 2020). However, further study will be needed to actually demonstrate the existence of parent-of-origin expression effects in these termites; indeed, one study which analyzed a subset of genes in R. speratus embryos failed to find any such effects (Kamiyama et al. 2025).

Additional study is needed to fully assess the existence of genomic imprinting across social insects. To date, parent-of-origin effects on gene expression have only been characterized in two bee species (Galbraith et al. 2016; Marshall et al. 2020; Bresnahan et al. 2025). Phenotypic patterns consistent with genomic imprinting have also been identified in termites, as discussed above (Matsuura et al. 2018), and in the ant Linepithema humile (Mayr, 1868) (Libbrecht and Keller 2013). The existence of genomic imprinting in other social taxa remains an open question. In addition, in all of these cases the actual molecular mechanisms that underlie any imprinting are totally unknown (Pegoraro et al. 2017). While the mechanistic basis for imprinting in mammals is typically DNA methylation, that seems less likely to be the case in insects; other epigenetic mechanisms like histone modifications may instead play important roles (Pegoraro et al. 2017).

Future Directions

Leveraging Technological Advances

In 2008, 2 years after the sequencing of the honey bee genome, Smith and colleagues predicted that within 10 years the research community would have generated whole genome sequences for 10 to 20 social insect species and their relatives (Smith et al. 2008). Instead, by 2018 more than 35 social insect genomes had been submitted to NCBI, permitting an explosion in the number of questions about social insect genome evolution that have been asked and, in some cases, at least partially answered.

Now, 19 years after the sequencing of the honey bee genome, new technologies hold substantial promise to expand our understanding of the relationship between sociality and the genome. For instance, the increased use of long-read sequencing and other approaches for generating high-quality, highly contiguous genome sequences is fundamentally improving and accelerating our ability to conduct genomic research in social insects.

Moreover, technological advances are expanding the kinds of loci which can be feasibly identified and interrogated in the genome. To date, most social insect genomic research has focused on protein-coding genes, whose identification in the genome is relatively straightforward using computational and RNAseq-based annotation methods. When researchers have interrogated gene regulatory loci in social insects, they have typically identified these loci using computational methods, which are useful but imperfect (eg using approaches based on sequence conservation in noncoding regions of the genome) (Nelson and Wardle 2013). However, regulatory loci can instead be directly identified and characterized using sequencing-based approaches. These include ATACseq, which identifies regulatory elements by determining which regions of the genome are in an accessible chromatin state (Buenrostro et al. 2013); CHIPseq, which uses antibodies against a DNA-binding protein to identify the loci with which that protein interacts (Robertson et al. 2007); MOA-seq, which uses the nuclease activity of micrococcal nuclease to identify sites in the genome that are bound by proteins (Savadel et al. 2021); and STARR-seq, which assays the entire genome for enhancer activity by transfecting candidate sequences into cell culture (Arnold et al. 2013), among others.

Among social insects, these methods have primarily been applied in honey bees, where they have been used to explore the regulatory mechanisms underlying the honey bee queen-worker polyphenism (Wojciechowski et al. 2018; Lowe et al. 2022; Zhang et al. 2023; Bresnahan et al. 2025) and division of labor among workers (Jones et al. 2020; Fang et al. 2022; Iino et al. 2023), and to characterize patterns of histone methylation related to behavior and social context (Shpigler et al. 2017, 2019). They have also been used to address similar questions in a small number of ant species (Wang et al. 2020; Gospocic et al. 2021; Jones et al. 2025), particularly the carpenter ant Camponotus floridanus (Buckley, 1866) (Simola et al. 2013b; Glastad et al. 2020, 2021, 2023). Finally, ATACseq has been used to characterize regulatory loci in the genome of the bumble bee B. terrestris (Zhao et al. 2020), and STARRseq has been used to identify enhancer loci which differ between solitary and social populations of the sweat bee L. albipes (Jones et al. 2024). We expect that the increased adoption of these methods by social insect researchers will advance our understanding of the role that gene regulatory mechanisms play in generating social phenotypes, and the evolutionary changes to gene regulation which underlie transitions from solitary to social life histories.

Moreover, critics have noted that most genomic research in social insects has so far been correlational, associating social traits of interest with patterns of sequence evolution or gene expression (Heneberg 2026). It is our hope that the research community will increasingly apply techniques which permit the direct experimental manipulation of gene expression or regulatory element activity to questions of social insect evolution. Two prominent techniques are RNAi and CRISPR-mediated gene editing. In an RNAi manipulation, an organism is fed or injected with a double-stranded RNA that matches the gene to be knocked down; the organism's anti-viral immune response to double-stranded RNA then prevents either translation or transcription of the target gene (Fire et al. 1998). In CRISPR gene editing, a guide RNA and a Cas9 nuclease are introduced into an organism via techniques like injection or electroporation; this causes the creation and repair of double-stranded breaks in a target sequence in the genome (Gasiunas et al. 2012; Jinek et al. 2012). The process can produce knockdown, knockout, or knock-in mutations at the target locus; if introduced in germline cells these mutations are heritable, permitting the creation of stable mutant lines (Gasiunas et al. 2012; Jinek et al. 2012).

These manipulative techniques are challenging to implement in social insects (Salvesen and Dearden 2025). For instance, workers engaged in brood care frequently destroy manipulated eggs or reject larvae as they develop (Sieriebriennikov et al. 2021; McAfee et al. 2022; Konu et al. 2023). It is also impossible to establish mutant lines using CRISPR in taxa that do not reproduce in the lab, like almost all ants (Sieriebriennikov et al. 2021). Despite these challenges, researchers have deployed knockdown and knockout manipulations in honey bees, Polistes wasps, and a few ant and termite species. They have been used to study sex determination in honey bees (Cheng et al. 2023; Otte et al. 2023); queen-worker polyphenisms in honey bees (Patel et al. 2007; Wolschin et al. 2011; Cameron et al. 2013; Ronai et al. 2016) and wasps (Hunt et al. 2011); soldier differentiation in termites (Zhou et al. 2006, 2007; Tarver et al. 2012; Hattori et al. 2013; Masuoka et al. 2018; Yaguchi et al. 2018); and the role of specific genes in behavior in honey bees (Chen et al. 2021; Sommer et al. 2024; Köhnen et al. 2025) and ants (Trible et al. 2017; Yan et al. 2017; Kohlmeier et al. 2018).

Continued efforts to implement these techniques in nonmodel social insects will allow researchers to move from identifying correlations between social traits and particular genes to demonstrating a causal role. In particular, as techniques for genetic manipulation become available in additional social lineages—eg non-Apid social bees, non-Polistine social wasps, and social thrips, aphids, and beetles—researchers will be poised to assess the extent to which convergent molecular mechanisms do or do not underlie independent origins of social traits.

Expanding Taxonomic Scope

To date, genomic studies in social insects have overwhelmingly focused on bees, ants, and wasps, with growing attention paid to termites. There is just one recent genome assembly for a eusocial aphid, Pseudoregma bambucicola (Takahashi, 1922) (Zhang et al. 2024), and no published genome assemblies exist for the eusocial ambrosia beetle A. incompertus or for any eusocial thrips species. To fully understand both the genomic causes and consequences of eusociality, these neglected lineages must be included in studies of social insect genome evolution. Our understanding of social insect biology has also benefited tremendously from studies in subsocial, socially polymorphic, and facultatively social bees and wasps. Moreover, subsociality is found in groups outside the Hymenoptera—for instance, burying beetles (Coleoptera: Staphylinidae: Nicrophorus), which exhibit extensive biparental care and can breed cooperatively (Won et al. 2018), and earwigs (Dermaptera), which exhibit extensive maternal care and can live in cooperative groups with overlapping generations (Honorio et al. 2025). These subsocial taxa represent independent origins of social behaviors, and so studying them in a comparative framework will allow researchers to ascertain whether the same genomic mechanisms are convergently involved in care behavior across insects, or alternatively whether these different lineages have taken lineage-specific routes to sociality. For instance, are the genes expressed by an earwig interacting with other members of a social group the same genes expressed by an ant interacting with a nestmate? Are the mechanisms which regulate the transition to post-reproductive care behaviors the same in burying beetles and in Polistes wasps? Thus, in addition to shedding light on the evolution of sociality, this expanded taxonomic focus will contribute to a better understanding of fundamental questions about repeatability and convergence in the evolution of complex traits (Agrawal 2017).

Study Sociality as a Complex Phenotype Composed of Many Constituent Traits

Decades of research have demonstrated that, across lineages, insect sociality has complex genetic architectures which we are far from fully characterizing. Ultimately, because social life histories are complex, multitrait phenotypes, it will be challenging to rigorously test hypotheses regarding genomic convergence across lineages until we have more fully characterized the molecular basis for sociality's constituent traits. Research on other convergently evolved complex traits whose mechanistic underpinnings are better characterized show a mix of molecular convergence and molecular independence, particularly when considered across scales. For example, animal eyes are a fairly well-characterized complex trait that has repeatedly emerged across lineages (Oakley and Speiser 2015; Oakley 2025). Across independent origins, some components of eyes, like screening pigments, are ancient and represent instances of convergent gene co-option (Oakley and Speiser 2015), a pattern which suggests that photosensitivity in ancestral metazoans likely acted as a preadaptation in eye evolution (Washburn et al. 2016). Other eye components are not convergent: for instance, lenses in different eye-bearing lineages are formed by different sets of proteins with independent evolutionary histories (Oakley and Speiser 2015). Thus, the repeated evolution of animal eyes is the result of both convergent co-option and lineage-specific genetic mechanisms (Oakley and Speiser 2015; Oakley 2025).

As complex phenotypes, social life histories are composed of traits like aggregation, reproductive skew between individuals, and sibling-directed care behaviors. The genes, regulatory loci, and gene networks which give rise to these traits are not fully understood in any taxon, and many taxa have received little to no genetic study. Likewise, the ways in which sociality is conceptualized have shaped our ability to investigate its genomic basis (Richards 2019). Attempts to categorize diverse social life histories into discrete categories result in oversimplifications, because social life histories are composed of many traits that may or may not co-vary with one another (Richards 2019). For instance, the honey bee A. mellifera and the bumble bee B. terrestris are both considered eusocial; however, A. mellifera has substantially greater queen-worker reproductive skew than B. terrestris, but much lower queen-worker dimorphism (Holland and Bloch 2020). Thus, simplified categorizations obscure important variation, and this has implications for studying the genomic basis of social traits, particularly in comparative contexts. For example, because the genetic mechanisms which underlie reproduction and body size may be at least partially distinct, lumping B. terrestris and A. mellifera together as “eusocial” may hinder research which seeks to identify the genetic basis for sociality or specific social traits across taxa (Holland and Bloch 2020).

Going forward, we expect that a comparative understanding of social evolution across lineages will be advanced as researchers rigorously characterize the genetic basis for sociality's constituent traits across developmental trajectories in individual social lineages. We note that prematurely framing investigations into the genetic basis of sociality as testing hypotheses regarding convergence risks the streetlight effect, as researchers emphasize any evidence which supports the convergent use of well-characterized mechanisms in their results while not highlighting new, less obviously interpretable results. As new methods and study systems are adopted by the social insect genomics community, we anticipate progress toward understanding the relationships between social life histories and insect genomes.

Funding

This work was supported in part by the National Science Foundation through awards to CSM (NSF DBI 2210800 and NSF DEB 1900357). Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Data Availability

No new data were generated or analyzed in support of this research.