Discussion
As hypothesized, sexual dimorphism covaries with worker polymorphism. This pattern is evident when comparing the size-shape space of individual species as well as at the level of entire genera. While quite logical, that male and female characters do not evolve independently, even in the absence of sexual selection, males have largely been absent from empirical and theoretical studies of the hymenopteran social insect castes.
While this study documents how sexual dimorphism relates to worker polymorphism in two genera of highly social and very derived ants, it is logical to think that sexual dimorphism was important for the origin of the worker caste and all of its subsequent elaboration/modification. Because sexual dimorphism is likely the ancestral state of the Hymenoptera (Stubblefield and Seger 1994), the ancestral size-shape space in the wasp ancestor of the ants was the phenotype space available for selection that ultimately resulted in the worker caste. After all, the differences among castes is primarily the difference in the timing, location, and dosage of a shared set of genes, and so the worker was/is both part male and part female. Variation in mating and social systems among ant species varies the degree of sexual selection and thus sexual dimorphism. Thus it is likely that there is a correlation between mating systems and worker polymorphism. It may even be that the observed correlation between worker caste variation and genetic diversity within nests (Fjerdingstad and Crozier 2006, Anderson et al. 2008, Smith et al. 2008b, Hughes and Boomsma 2008, Smith et al. 2008a), that results from mating systems like polyandry, is mechanistically confounded by the phenotypic results of sexual selection (e.g., sexual, or natural, selection on male body size to increase mating probability).
The causal arrow in the relationship between sexual and worker dimorphism likely points in both directions. As above, sexual dimorphism may have provided the raw material (size-shape) for the evolution of the worker caste, but selection on worker characteristics, and on the population of workers in colonies (colony-level selection)(Oster and Wilson 1979), likely facilitates/inhibits the phenotypic variation available to sexual selection. Strong natural selection on worker size variation, for example, increases the size-shape space and may facilitate the production of novel sexual forms.
Importantly, the causal direction of the relationship is far less important than recognizing the non-independence of the two processes. Selection on worker polymorphism, for example, increases the phenotypic space available to sexual selection, not necessarily whether that selection occurs. Selection for increased variation in the size-shape of any caste increases the raw material for selection – it increases opportunity. As previously theorized, the colony environment buffers its members from individual-level selection and can be an incubator of sorts for novelty, especially via modifications of the social environment (Rajakumar et al. 2012, Molet et al. 2012).
Not surprisingly, queen-worker dimorphism was a better predictor of worker polymorphism than was sexual dimorphism because female castes are developmentally more similar than are the sexes. Previous studies have documented the correlation between queen-worker and worker polymorphism across species (Fjerdingstad and Crozier 2006), though these factors are not always correlated (Lecocq de Pletincx et al. 2021). The development of workers and queens is more similar than that of workers and males (Anderson et al. 2008). For example, queen-worker differentiation takes place in early development while worker caste differentiation occurs later; note, the most upstream mechanisms of female caste determination can occur prior to fertilization or in early embryogenesis (Schwander et al. 2008), but most differentiation (Lillico-Ouachour and Abouheif 2017), and canalization (Qiu et al. 2022), occurs later in larval development. Hymenopteran males and females differ genetically where male-destined eggs are typically unfertilized, and at least in honeybees the mechanism involves heterozygosity at a single locus, CSD, that then feeds into the conserved sex differentiation pathway of insects (Beye et al. 2003, Roth et al. 2019). Many studies are now demonstrating that sexual determination and differentiation and caste determination and differentiation are not mechanistically independent, both pathways utilize many of the same genes (Klein et al. 2016, Johnson and Jasper 2016, Warner et al. 2019, Roth et al. 2019).
One of the most surprising results of this study is the size-shape axis in both Pogonomyrmex species. In P. barbatus there is a size-shape relationship among workers, and in P. badius this relationship is the axis that differentiates both worker castes and both sexes (though workers and sexuals have different intercepts). The major worker caste in P. badius is an evolutionary novelty in the sense that it is a unique and derived character for the species, completely absent in all other extant North American Pogonomyrmex . Based on the measurements taken, the major worker is simply an extension of the minor worker along the size-shape axis. The major caste, in other words, is simply a size increase of the minor worker using the same allometric growth rules – and the gyne is, in a sense, a size extension of the male using the same allometic growth rules – and surprisingly, workers and sexuals vary on parallel axes only differing in their starting size. This result suggests that, despite appearances, all castes differ only slightly in basic developmental scaling.
Clearly, including more genera and more body measurements is a logical next step to further evaluate the relationship between sexual dimorphism and worker polymorphism. Looking at more subfamilies of ant will likely also reveal how generalizable this pattern is and how other aspects of biology constrain or facilitate caste evolution. Fortunately, the growing database of Antweb.org is an amazing resource and it seems that the maintainers of the database have done a commendable job of representing species-level variation. Given the high quality and taxonomic breadth of the database it is sure that it will be useful for the evaluation of additional macroevolutionary patterns in ants. Additionally, the availability of three-dimensional data for ant morphology and anatomy promises to add significantly to our knowledge of how castes differ in complex elements of internal anatomy and shape.