Last month, two papers on the evolution of eusociality were published in high-profile journals: one by Karen M. Kapheim and colleagues in Science, the other by Sandra M. Rehan and Amy L. Toth in Trends in Ecology & Evolution (TREE). Social and eusocial insects are an attractive system for studying major transitions, sharing some of the key features that make the volvocine algae so good for this purpose: multiple, independent origins of traits thought to be important to the transition and extant species with intermediate levels of sociality. These features make the social insects, like the volvocine algae, well-suited for comparative studies.

The TREE paper proposes that different evolutionary processes dominate at different stages during the transition from solitary to eusocial life. The authors seem to mean this in two somewhat related ways: first, that the early stages of the transition are driven primarily by individual and kin selection and the later stages by colony-level selection; second, that the molecular mechanisms underlying steps up the ‘social ladder’ change over the course of the transition. The focus of the paper is overwhelmingly on the latter of these shifts, and so for now I’m going to skirt the minefield of whether or not kin selection and colony-level selection are actually different processes.

A brief review of hypotheses for the evolution of sociality divides these ideas up into two major categories, which the authors acknowledge are artificial and overlapping: those related to changes in gene expression and those related to changes in genomic sequence. Among the hypotheses related to changes in gene expression, the ovarian ground-plan hypothesis predicts that

There’s an interesting parallel here with Aurora Nedelcu’s work on the regA gene in Volvox and Chlamydomonas, in which she suggests that temporal changes in gene expression in an ancestral unicell shifted were co-opted into differences between cells within a multicellular organism. Also in the gene expression category, the maternal heterochrony hypothesis

…posits that reproductive division of labor evolved via a reorganization of the timing of offspring-care gene expression…

so that maternal care was essentially co-opted into sibling care. Finally, the genetic toolkit hypothesis

…proposes that regulatory changes in specific genes, pathways, or networks with conserved roles across species are important in the evolution of novel phenotypes.

Hypotheses related to changes in genomic sequence are divided into those involving novel genes, changes in existing coding sequences, and shifts from gene-regulatory to protein-coding changes. The last of these is most similar to the authors’ view:

…our framework predicts that, early in social evolution, regulatory genomic changes, such as shifts in the timing and location of expression of conserved genes… are the primary drivers of social phenotypes.

Later in the transition,

The authors emphasize that what they have in mind is a shift in the relative importance of these two modes of change, not a dichotomous change from one to the other. They suggest a strategy for whole-genome based studies to test their hypothesis and suggest that such studies have great potential to increase our understanding of the evolution of eusociality.

Among the studies cited by Rehan and Toth as promising first steps is the Science paper by Kapheim and colleagues, a comparative genomic study of ten bee species that vary in their degree of sociality. This study correlated capacity for gene regulation, quantified as the number and binding strength of transcription factor binding sites, with degree of sociality and concluded that

More social species also had higher numbers of genes predicted to be methylated, and this was also taken as evidence for an increased capacity for gene regulation. Genes that evolved more rapidly in highly social species were enriched for gene-regulatory functions, again suggesting the importance of changes in gene regulation.

Origins and elaborations of eusociality were characterized by constrained protein evolution and increased capacity for differential gene regulation, but the specific genes involved were mostly different between different lineages. Some gene family expansions occurred in the evolution of complex eusociality, but again the particulars were lineage-specific. The authors take the lineage-specificity of many of the changes involved in the evolution of eusociality as evidence of a high degree of contingency in social evolution.

These results would seem, in general, to support the hypothesis advanced by Rehan and Toth: the inferred increase in gene-regulatory capacity is common to all of the eusocial species, while the expansions of gene families related to social behaviors were restricted to the lineages with ‘complex eusociality.’ I don’t really know what to make of the metrics used to quantify gene-regulatory capacity: number and strength of transcription factor binding sites, numbers of methylated genes, and enrichment of gene-regulatory functions in rapidly evolving genes. Are these measures commonly used to compare gene-regulatory capacities?

Regardless, it’s an interesting hypothesis. I wonder if it can be generalized to other transitions. In the volvocine algae, we know that both changes in gene regulation and expansions of a small number of gene families affected traits important to the evolution of multicellularity. The order of these changes seems to not quite match up with the hypothesis of Rehan and Toth, though. The changes in gene expression that establish cellular differentiation in the Volvox carteri lineage happened more recently than the establishment and expansion of the extracellular matrix, which may have involved an expansion of the Volvox matrix metalloproteinase (VMP) gene family. Of course, it’s possible that the  VMP family expansion is lineage-specific, and that the initial expansion of the extracellular matrix near the origin of the Volvocaceae involved changes in gene regulation or in existing coding sequences.

Answers to some of these questions may become available with the release of more volvocine genomes. It will be interesting to see whether the pattern of early gene expression changes and later genome-scale changes inferred for the evolution of eusociality in bees holds for the evolution of multicellularity. Is it even possible to compare such things across different transitions? Is it meaningful to ask what stage of sociality is equivalent to what stage of multicellularity?