Introduction
The presence of substantial population structure within many if not most taxa is well-documented empirically (Slatkin 1987; Palumbi et al. 1997), and delimiting species in such instances has become a focal topic in systematics (Hey and Pinho 2012; Sukumaran and Knowles 2017). A variety of processes can generate geographic genetic structure such as local environmental adaptation and isolation by distance (Duminil et al. 2007; Leffler et al. 2012), accumulating deep divergence between populations over time and resulting in multiple phylogeographic lineages within cohesive species (Rissler et al. 2006; Soltis et al. 2006). If these divergences persist and are not resorbed (Rosenblum et al. 2012), the eventual outcome is generally speciation (Dynesius and Jansson 2014), wherein populations diverge sufficiently to acquire independent evolutionary trajectories characterized by reproductive isolation and low rates of hybridization (Singhal and Moritz 2013). Completed speciation is often characterized by adaptive ecological, genetic, and phenotypic differentiation that reinforce species boundaries via reproductive isolation and selection against hybrids over time (Coyne and Orr 2004).
In contrast, substantial population structure can originate within species that is nonetheless held in check by high rates of gene flow (Bohonak 1999) and is consequently not indicative of incipient, incomplete, or ongoing speciation (see Huang and Knowles 2016), but is instead a long-term, stable, or persistent endpoint on its own (Yang et al. 2019). Furthermore, the degree and rate of population divergence may or may not be correlated to rate of completed speciation, depending on geography and connectivity (Kisel et al. 2010; Harvey et al. 2017; Singhal et al. 2018b, 2022; Burbrink et al. 2023). Secondary contact can even result in lineage fusion (Maier et al. 2019) and speciation reversal (Kearns et al. 2022), with subsequent gain of structure (Frei et al. 2022). Consequently, population structure may represent the outcome of various processes yielding geographic genetic diversity within species that nonetheless do not reflect speciation trajectories (Cutter and Gray 2016; Sukumaran and Knowles 2017).
Genetic structure in species may therefore consist of divergence that is neutral (e.g., isolation by barrier, distance, or hierarchy; Baptestini et al. 2013) or adaptive (e.g., ecological selection resulting in isolation by adaptation or environment; Nosil et al. 2008; Wang and Bradburd 2014). Such isolation (Zbinden et al. 2022; Moreno-Contreras et al. 2023) can consequently produce regional divergence that is bound together by migration over time and space (Garnier et al. 2004; Sexton et al. 2014). This may represent incomplete speciation (Nosil et al. 2009) or ephemeral differentiation that is ultimately resorbed (Rosenblum et al. 2012). Protracted scenarios of speciation-like outcomes can also occur wherein evolutionarily distinct lineages nevertheless exhibit gene flow in primary or secondary contact (Smadja and Butlin 2011; Pardo-Diaz et al. 2012; Burbrink et al. 2021). Complex landscapes may therefore produce substantial local differentiation (Brauer et al. 2018; Nali et al. 2020), interacting with climatic cycles to produce recurring episodes of divergence (Fitzpatrick et al. 2009) and secondary contact with ongoing gene flow (Schield et al. 2019; O’Connell et al. 2021).
Consequently, deep intraspecific divergence of genetic lineages can be generated by a variety of geographic and ecological processes that are either i) part of a generalized trajectory of divergence towards speciation, or ii) population structure that is nonetheless unified by gene flow (Avise 2000; Coyne and Orr 2004). Therefore, demographic model selection and tests of isolation-by-distance (Jackson et al. 2017) can be instrumental for differentiating structure versus speciation (Sukumaran and Knowles 2017). When the former is produced by landscape-scale processes that are held in check by ongoing migration (Seeholzer and Brumfield 2018), this is detectable by a variety of methods (Carstens et al. 2022). In contrast, speciation is demonstrated by reduced migration (below thresholds of isolation by distance or environment) between ecologically, geographically, and phenotypically divergent lineages (Burbrink et al. 2021).
Finally, integrated analyses of phenotype that test for the correlated ecomorphological signature of phylogeographic lineage divergence are also crucial for testing speciation (Zamudio et al. 2016), even in putatively cryptic taxa (Singhal et al. 2018a). Consequently, multi-modal analyses are needed to discern an accurate picture in many cases (Bertl et al. 2018); the data and methods needed to test structure versus speciation will often be the same as those needed to identify the mode, mechanism, and process of ecological, geographic, genetic, and phenotypic divergence (Pyron et al. 2023). The fundamental test is therefore whether the processes driving the accumulation of geographic genetic structure (e.g., IBD/IBE) are implicated in diversifying ecomorphological selection promoting divergence between the incipient phylogeographic lineages, or whether other mechanisms simultaneously foster high levels of migration between these populations sufficient to promote lineage cohesion and maintain species identity (Kisel et al. 2010; Seeholzer and Brumfield 2018; Freedman et al. 2023; Prates et al. 2023).
Here, we evaluate “structure” versus “speciation” in the Seepage Salamander (Desmognathus aeneus ) using tests of isolation-by-distance, automated demographic-model selection, species distribution modeling, genotype-environment and -phenotype association, and morphometric divergence. Seepage Salamanders are miniaturized (~3–5cm total length), terrestrial, direct-developing lungless salamanders (Plethodontidae) from the southeastern United States, primarily from the Blue Ridge of Georgia and North Carolina (Harrison 1992). Given their size, their home range and dispersal abilities are miniscule, and they are therefore highly susceptible to landscape-level genetic fragmentation over ecological timescales. They also exhibit strong ecological specificity for seepage habitats associated with the headwaters of high-gradient creeks and streams. Interestingly, they are also found in analogues of this microhabitat at low elevations in dispersed, isolated pockets across the Piedmont, Southeastern Plains, Ridge and Valley, and Southwestern Appalachians in central Alabama and western Georgia, some of which were not discovered until the 2010’s (Graham et al. 2012).
Recent range- and genome-wide results indicate up to 6.3% mitochondrial divergence between four deeply divergent but spatially adjacent phylogenomic lineages with extensive admixture (Beamer and Lamb 2020; Pyron et al. 2020, 2022). Given their broad but fragmented geographic range, we suggest that Seepage Salamanders occupied a continuous distribution across the southeastern U.S. during cooler glacial periods. Similarly, we hypothesize that their microhabitat specificity results in stabilizing ecological selection and species cohesion despite landscape genetic variation. Finally, we expect morphometric similarity to reflect this ecological uniformity. We find that the evolutionary history of the group is characterized by deep lineage divergence that is counterbalanced by episodes of increased migration and a lack of phenotypic differentiation; “structure” rather than “speciation.” It is difficult to pinpoint the exact nature of geographic genetic diversity (e.g., IBD/IBE) given the complex ecological nature of their microhabitat adaptations, apparently involving hydrological, edaphic, floristic, and climatic interactions. Nevertheless, the frequency of estimated migration and observed admixture in individuals suggests that reinforcement is not operating to select against hybridization between the genetically differentiated populations. Formal tests of the dichotomy between structure and speciation should be employed when this distinction is consequential in phylogeography.