Potential drivers of speciation in the E. ruidum complex 
Allopatric speciation, which involves geographic isolation of segments of a formerly contiguous population, is firmly established as the primary mechanism by which new species evolve (Coyne and Orr 2004). Without spatial isolation among populations, it is difficult to have a scenario in which gene flow can be reduced or eliminated. However, it is known that speciation can be multifactorial with multiple axes of differentiation characterizing even incipient species (Mullen and Shaw 2014; Brodetzki et al. 2019). For instance, in Heliconiusbutterflies color pattern divergence, mate preference, host-plant use and microhabitat choice possibly are implicated in species diversification (Smiley 1978; Brown Jr 1981; Estrada and Jiggins 2002; Mallet et al. 2007). In the case of the species of the E. ruidumcomplex, we do not see any clear correlation between their distribution pattern and geographic barriers, nor possible events of microallopatry or niche specialization. For instance, in the state of Chiapas, Mexico, the populations of E. ruidum sp. 1 and sp. 2 occurred in quasi-sympatry, at less than 15 km from each other (Lachaud 1990). Even more puzzling is the geographic distribution of the three delimited species restricted to lowland areas of Oaxaca and Guerrero, in southeast Mexico, which we found to be separated from each other only by 5 to 30 km. Below we discuss three potential phenotypic and genotypic factors that could have promoted the speciation among the species of theE. ruidum complex: mitonuclear conflict, chemical cue divergence, and colony structure.
Mt performance affects every aspect of individual fitness and its correct function relies on mitonuclear compatibility (Angers et al. 2018; Zaidi and Makova 2019). This intimate interaction between the mt and nuclear products has led to a strong coevolution between both genomes (Blier et al. 2001; Burton et al. 2013; Chou and Leu 2015). It has been proposed that mitonuclear coevolution in isolated populations triggers speciation, since population-specific mitonuclear coadaptations lead to between-population mitonuclear incompatibility, thus precluding gene flow (Hill 2016). It is therefore plausible that the rapidly evolving mt haplotypes found in the three heteroplasmic species of theE. ruidum complex restricted to southeast Mexico played a role in their speciation if they were expressed and interacted with the nuclear products, consequently pressing for its evolution (Hill 2017). This could have been achieved through a marked population structure promoted by assortative mating, which created mitonuclear incompatibilities and barriers to gene flow between the taxa involved. However, since heteroplasmy does not seem to occur in the widely distributed E.ruidum spp. 1 and 2, other factors probably could also have promoted the species diversification in this complex.
Social behaviour and recognition cues could act as major traits that could drive assortative mating, limit gene flow and ultimately promote speciation (Hochberg et al. 2003). Sympatric speciation based on social trait divergence has been suggested for ants of the Cataglyphis niger complex, where three incipient species possess consistent differences in CHC composition, social structure and mtDNA sequence data (Brodetzki et al. 2019). Our results showed that the CHC distances among populations mirrored their phylogenetic relationships obtained from mt and genome-wide data. A similarly high correlation between CHC and genetic data has been observed in other social insects (Isoptera: Dronnet et al. 2006; stingless bees: Leonhardt et al. 2013; ants: Hartke et al. 2019). CHC divergence has been associated with nestmate/non-nestmate communication (van Zweden and d’Etorre, 2010) and may also play an important role in pre-mating isolation (Savarit et al. 1999; Smadja and Butlin 2009; Snellings et al. 2018).
Another important trait in social insects is colony structure. In some ant groups, shifts from monogynous to polygynous colonies and gyne morphological polymorphism (macro and microgynes) are also known to be involved in their speciation (Seifert 2010; Brodetzki et al. 2019). For instance, limited dispersal of females of polygynous species can promote differentiation between populations, and if male dispersal is also restricted this can also lead to their speciation (Pamilo and Rosengren 1984; Seppä and Pamilo 1995; Gyllenstrand et al. 2002). A previous study that focused on a population of E. ruidum sp. 2 from Rosario Izapa in the state of Chiapas, southeast Mexico, showed that macrogynes and microgynes can be adopted by their monogynous and polygynous mother colonies, leading to low dispersal (Lenoir et al. 2011). Moreover, queen size dimorphism and social polymorphism has been observed in laboratory colonies both in E. ruidum sp. 3 and sp. 4 (Peña-Carrillo, unpublished data), where few microgynes were produced. Further colony structure studies performed for species of the E. ruidum complex will reveal whether the presence of queen size dimorphism and social polymorphism was implicated in their diversification process.
Ectatomma ruidum has been the subject of a vast number of studies, including those on social structure (e.g. Corbara et al. 1989), foraging and diet (e.g. Lachaud 1990; Riera-Valera and Pérez-Sánchez 2009; Santamaría et al. 2009), macronutrient regulation (e.g. Cook and Behmer 2010), home ranges and nestmate recognition (e.g. Breed et al. 1990), queen dimorphism (e.g. Lachaud et al. 1999; Lenoir et al. 2011), communication behavior (e.g. Pratt 1989), and parasitoid interaction (Howard et al. 2001). Most of these studies, however, were based on specimens from different localities across the Neotropics. Our gathered evidence consistently shows that E. ruidum actually represents a species-complex of species, which would change the interpretation and the extension of the conclusions that were drawn in previous studies. Our study thus highlights the importance of using different sources of molecular data for species delimitation of morphologically conserved taxa. Moreover, it is important to consider the occurrence of heteroplasmy in systematic studies, since it can lead to incorrect estimates of phylogenies if it is not detected. The presence of extensive heteroplasmy within the E. ruidum complex highlights the necessity of being aware of the occurrence of this phenomenon in other insect groups and metazoans in general, since it appears that it is not as rare as previously thought (White et al 2008; Robison et al. 2015; Macey et al. 2021).