DISCUSSION
The goals of this study were to examine the population structure, gene flow, and signatures of sex-biased dispersal in a fungus-gardening ant species across several ecological gradients in Florida, USA. Overall, our results suggest that most long-distance dispersal is conducted by males. Specifically, females appear to have more limited dispersal as the mtDNA haplotypes are strongly geographically clustered (Figure 4). There were two COI clades found primarily in the Florida panhandle (Ts Clade I and Ts Clade II) and one that was restricted to the Florida peninsula (Ts Clade III). There were only three ants collected on the peninsula that somewhat clustered with panhandle ants: one with Ts Clade I (however, this specimen was separated from the Ts Clade I by 32 mutation steps) and two with Ts Clade II, separated by 12 mutation steps. These three ants were notably collected in North Central Florida, near Gainesville, geographically closer to the panhandle than the samples collected in Central Florida near Orlando. In contrast to mtDNA data, biparental microsatellite markers suggest significant gene flow across Florida (approximately 360km, linear distance between the Central region in Orlando and St. George Island, Florida) and minimal spatial structure (Figures 2 and 3). These results indicate considerable admixture of microsatellite alleles across the range of this study, which considering the geographic clustering of mtDNA COI sequences, likely arise from male movement. Together, these results suggest that males are responsible for most long-distance dispersal while females (and concomitantly, their co-dispersed fungal symbionts; (Tesson et al., 2015)) do not disperse very far. This result was surprising since T. septentrionalisqueens are not relatively large nor especially endowed with fat stores, which could impact their flying ability (Helms, 2018; Seal, 2009; Seal & Tschinkel, 2007b).
It is not clear how much of a barrier that rivers present to females since we did find evidence of some trans-river female dispersal; however, the distance that females seem to move across rivers is much shorter relative to the distance males appear to disperse. The Suwannee River may be an important barrier to female dispersal, though not impenetrable since two peninsular haplotypes (in three individuals) were found clustered with the two panhandle clades (Figure 4). Conversely, the Ochlocknee River in the panhandle may not be a significant dispersal barrier to either sex considering the extensive admixture of microsatellite alleles (Figure 2) and shared haplotype clades (Ts Clades I and II; and even identical haplotypes in some cases) in both the ARD and WRD (Figure 4). The latter finding is surprising considering the differing ecologies and environments (i.e., frequently flooded flatwoods in the ARD and dry, xeric sandhills in the WRD). That being said, the finding of reduced genetic diversity in the ARD relative to the WRD (Table 2) could suggest recent expansion in the WRD. As a result, on small scales (10s of kilometers), T. septentrionalis appears to be a very mobile species, capable of rapid population growth and extensive dispersal capabilities, but there are limits to their expansion abilities across larger scales (>100s of kilometers). Possible explanations for this conclusion could be related to Pleistocene bottlenecks and then subsequent expansion, and a subsequent time lag in the expansion of COI haplotypes. Therefore, it would appear likely that males have a greater dispersal capability than females. Field studies measuring the variation in flight distance within and between sexes could further inform our results.
Evidence is currently lacking as to whether male-biased dispersal is the general rule in the tribe Attini. This is surprising considering how important female dispersal is for the range expansion and ultimately evolution of the fungal symbiont (Mueller et al., 2001). For example, the basal neoattine Mycetophylax simplex exhibited relatively minor mtDNA (COI) variation across its range in Brazilian Atlantic Forest (Cardoso et al., 2015), which suggests that females are capable of long-distance dispersal. However, as a lower attine (i.e., an early branching lineage of attini), Mycetophylaxlikely has smaller queens than Trachymyrmex sensu lato ants and other members of the so-called ‘higher attini’ (Seal, 2009); thus, the energetic cost of dispersal for Mycetophylax compared to Trachymyrmex sensu lato could possibly be lower. As another example, Mycocepurus smithii indicated stronger gene flow and little spatial structure in populations across the Panamanian isthmus (inferred from microsatellites) unlike their fungal symbionts that were more spatially structured, though the study did not also employ mtDNA markers like the present study (Kellner et al., 2013). While spatially structured fungal symbionts could point to limited female dispersal and long-distance male dispersal like we found withT. septentrionalis , males are rare if not absent in M. smithii , which exhibit thelytokous parthenogenesis in Panama (Kellner et al., 2013). Thus, M. smithii female movement (and some level of disruption to vertical fungi transmission) likely explains the patterns in central Panama. While lower attines such as Mycetophylax and Mycocepuruscultivate fungi that are likely capable of independent life, fungi grown by higher attini such as Trachymyrmex and Atta are not (Schultz & Brady, 2008). Solomon et al. (2008) reported mtDNA (COI) clusters in three Atta species across continental scales, which suggests limited female dispersal, but did not examine whether males were capable of dispersing longer distances. Interestingly, ddRADseq (i.e., diploid markers) in Atta texanashowed evidence of spatial structure and isolation by distance across a north-south gradient in Texas (850km). Though fungal symbionts also illustrate significant north-south differentiation in this species, the patterns are not concordant with their host ants (Mueller, Mikheyev, Solomon, & Cooper, 2011; Smith et al., 2019), which could indicate independent/differential dispersal patterns of males, females, and fungal symbionts via unknown mechanisms.
Our results support greater dispersal abilities in male T. septentrionalis than females. Consequently, this suggests that the dispersal abilities of the vertically transmitted symbiotic fungus (and further associated microbial symbionts (Ishak et al., 2011; Ronque, Lyra, Migliorini, Bacci, & Oliveira, 2020)) are likely also limited and thus also exhibits spatial structure, unless the fungus also has the ability of independent dispersal as suggested in A. texana(Smith et al., 2019). Limited female and symbiont co-dispersal could represent a significant bottleneck to fungal diversification (and associated microbes). Bottlenecks are a common feature among vertically transmitted symbionts, which generally exhibit eroded genetic variation and reduced genomes compared to horizontally exchanged relatives (Bennett, McCutcheon, MacDonald, Romanovicz, & Moran, 2014; Douglas, 2010; Helms, Ijelu, & Haddad, 2019; Nikoh, Hosokawa, Oshima, Hattori, & Fukatsu, 2011). Bottlenecks may not only influence population demographics but also the adaptive abilities of co-dispersed symbionts under varying environments. Consequently, the overall coevolutionary patterns and associations observed in the fungus-farming ant symbiosis may be constrained by limited female dispersal especially in the higher attini that are characterized by obligate symbionts and large-bodied, fatter queens.
The approximately 49 ant species in the genus formerly known asTrachymyrmex (now split into 3 genera (Solomon et al., 2019)) grow conservatively 4-5 phylotypes of fungi (Ješovnik et al., 2017; Luiso, Kellner, Matthews, Mueller, & Seal, 2020; Solomon et al., 2019). One possible explanation is that ant host diversification in these derived lineages has happened at a faster rate than their fungal symbionts because of limited female ant dispersal. Whether attine ants and their fungal symbionts have different evolutionary (or expansion) rates is currently unknown. The most recent genome-level examinations suggested that attine fungal genomes have lower diversity of metabolic genes compared to free-living fungi; however, this was based on transcriptomes (measures of gene expression) as we lack fully annotated attine fungal genomes because attine fungi are functionally polyploid (Kooij, Aanen, Schiott, & Boomsma, 2015; Kooij, Poulsen, Schiøtt, & Boomsma, 2015; Nygaard et al., 2016). Alternatively, since neither ant nor fungi have to evolve at similar rates, reduced fungal lineage diversity could be due to higher evolutionary and subsequent extinction rates among the fungi, such that fungal diversification may occur more rapidly with ants adopting novel fungal strains and discarding others as climate and parasite pressure change the outcome of the interaction (Mehdiabadi, Hughes, & Mueller, 2006; Seal & Mueller, 2014; Seal, Schiøtt, & Mueller, 2014; Seal & Tschinkel, 2007a). Furthermore, some phylogenetic analyses suggest that the fungal lineages typically grown by leaf-cutter ants (i.e., Clade A fungi grown by Atta andAcromyrmex ) (Mueller et al., 2018) are younger than the ant lineages (Mikheyev, Mueller, & Abbot, 2010; Nygaard et al., 2016) which suggests a recent domestication event. However, analyses of more recent datasets have called this conclusion into question since some non-leaf-cutting ants grow Clade A fungi (Mueller et al., 2017; Mueller et al., 2018; Schultz et al., 2015), indicating that the fungi may have been around for as long as the less derivedTrachymyrmex ants. In conclusion, a greater understanding of the dispersal biology of these species could vastly improve our ability to understand and ultimately predict how host and symbiont populations expand and evolve across larger geographic and macroevolutionary scales.