Speciation and phylogeography
Like most mygalomorphs, the Antrodiaetus unicolor species complex has certain life history traits (i.e. long generation times, limited dispersal capabilities, morphological conservatism) that drastically influence population structure (Coyle, 1971). This in conjunction with extensive molecular divergence would most likely indicate that the populations have been isolated from gene flow for an extended period of time, which would increase speciation potential (Barraclough, 2019). While that seems to be the case when comparing A. microunicolor , A.unicolorA, and A.unicolorB, lineages within A.unicolorB remain ambiguous.
Specifically, the STRUCTURE analyses for both All_M20 and several UniB datasets recover conflicting clusters (Figures 3a & 4; also see 3RAD section above). These clusters generated for all datasets were incongruent with monophyletic groups, which might be reflective of the use of unlinked SNPs for STRUCTURE versus using all variation for phylogenetic inference. While VAE for UniB_M30 provides some evidence of structure separating these lineages, there is still slight overlap between them (Figure 4a). Derkarabetian et al. (2019) used STRUCTURE and VAE for species delimitation, which revealed clear VAE and STRUCTURE clusters agreeing with multiple lines of evidence (i.e., COI clades, DAPC, morphology) for species-level divergence within the harvestmanMetanonychus , a group that is also known to have high population structure, conservative somatic morphology, with allopatry between species in the focal species group. Our study, however, includes a taxon with sympatry/parapatry across multiple lineages in addition to being morphologically homogeneous, with A. microunicolor being the exception. Given the more recent divergence in the Antrodiaetus unicolor species complex (~11.5 mya; Hendrixson & Bond, 2007) compared to Metanonychus (~25 mya; Derkarabetian et al., 2019), the underlying genetic patterns indicate speciation may not necessarily be complete and/or as clearly reflected in the VAE clusters.
Although the mixed cluster assignments of individuals suggest admixture between populations is occurring, it could also indicate incomplete lineage sorting (i.e., ancestral polymorphism) and/or recent range expansion (Avise, 2009). In addition, the species tree generated in ASTRAL (Supplemental Figure S2) showed considerable uncertainty for most of the relationships within A. unicolor B indicating an appreciable amount of gene tree discordance. There are several potential reasons for gene tree/species tree discordance (incomplete lineage sorting, reticulation, gene duplication, or horizontal gene transfer; see Maddison, 1997) but among these, incomplete lineage sorting (ILS) seems the most likely explanation due to the prevalence of ILS in taxa with long generation times, large effective population sizes, and/or low mutation rates (Degnan & Rosenberg, 2009). Additionally, mtDNA data in previous analyses hinted at the possibility of a recent range expansion (Hendrixson, unpublished data; see speciation scenarios below).
All spiders in this species complex appear to have similar habitat requirements (Coyle, 1971; Hendrixson & Bond, 2005a). However, morphology, behavior, and large-scale ecological (i.e., climatic) data offered some plausible insight into potential adaptive divergence. Morphological, geographical, and molecular data clearly demarcateA. microunicolor and A. unicolor A as distinct lineages, yet the niche similarity tests between these two lineages reveals no significantly divergent or conserved niches. Two of the background areas (75km and 100km) approach significant niche divergence when comparingA. microunicolor occurrence points to the background area ofA. unicolor A but not vice versa. It is possible that niche divergence important for speciation in this complex occurs in other ecological variables not tested in our study; however, our clustering analyses show minimal gene flow between these lineages and suggests a speciation model in which reproductive isolation accumulated in allopatry without much ecological differentiation (i.e., niche divergence may not have been a large driver in the speciation process). With subsequent sympatry between A. microunicolor and A. unicolor B these lineages potentially maintained reproductive isolation through premating barriers, specifically character displacement with disparities in size and breeding periods (e.g., Bond and Sierwald 2002).
The other lineages, largely sympatric, seem morphologically and behaviorally similar, thus we use ENM ecological data. A. microunicolor + A. unicolor A and A. unicolor B niche similarity analyses yielded conflicting results. The comparison ofA. microunicolor + A. unicolor A occurrences to the background area of A. unicolor B would suggest niche divergence (i.e., significantly more different than expected) whereas the comparison of occurrences of A. unicolor B to the background area of A. microunicolor + A. unicolor A suggests niche conservatism (i.e., significantly more similar than expected). For niche divergence, one possible explanation could be that A. unicolor B is much more widespread than both A. microunicolor and A. unicolor A and, therefore, has more potential for heterogeneous environmental conditions in the habitat available to them (e.g., more tolerance for lower elevation compared to A. microunicolor andA. unicolor A). Additionally, niche conservatism could be the result of A. unicolor B’s distribution mostly overlapping A. microunicolor and A. unicolor A’s distributions, suggesting thatA. unicolor B can not only easily inhabit the same environmental conditions but also likely prefers similar habitats as A. microunicolor and A. unicolor A.
The niche similarity analyses involving lineages within A. unicolor B yielded no obvious results for either niche divergence or niche conservatism. The niche similarity analyses for A. unicolor B1 and A. unicolor B2 show that their niches are no more different or similar than expected for any of the background areas tested, although the 100km background area hinted at niche conservatism (not significantly though). The Little Tennessee River may be a potential geographic barrier between these two lineages facilitating reproductive isolation, but not enough time has occurred for the accumulation of neutral divergence indicative of allopatric speciation. Numerous taxa in the southern Appalachians exhibit similar phylogeographical patterns, identifying either the Asheville or Little Tennessee River Basins as potential barriers to gene flow:Hypochilus araneomorph spiders (Keith & Hedin, 2012), harvestmenSabacon cavicolens (Hedin & McCormack, 2017) and Fumontana deprehendor (Thomas & Hedin, 2008), Dasycerus carolinensisbeetles (Caterino & Langton-Myers, 2019), Narceus millipedes (Walker et al., 2009), and plethodontid salamanders (Kozak & Wiens, 2010; Weisrock & Larson, 2006). Consequently, these lineages could prospectively be considered distinct species after sufficient time and divergence in allopatry. Additionally, the niche similarity analyses forA. unicolor B1 + A. unicolor B2 to A. unicolor B3, while not consistently significant for all background areas, does show signs of niche divergence. For the 100km and 75km background areasA. unicolor B3 occurrences were significantly different from the background of A. unicolor B1 and A. unicolor B2 but not quite significantly different for the reciprocal comparison. Also, the 50km background area hints at niche divergence for both comparisons. One explanation could be that the ancestor of A. unicolor B lineages had undergone range expansion where A. unicolor B3 later became geographically isolated from A. unicolor B1 and A. unicolor B2 for a significant period of time generating reproductive isolation, with subsequent secondary contact potentially driving niche divergence and, therefore, reinforcing reproductive isolation in sympatry.
One caveat for using large-scale ecological data like these is the potential for not having the resolution needed for detecting microhabitat differences, which are often important for divergent selection and driving speciation (Massatti & Knowles, 2014; Soberón, 2007). Mygalomorph spiders like Antrodiaetus are known for habitat specialization at small scales, such as preferring north-facing slopes, shaded ravines, and particular soil types (Hedin et al., 2015; Starrett et al., 2018). Therefore, it is possible that we could be overlooking significant microhabitat differences that potentially drive niche divergence, with speciation following. Another potential issue could be how we defined the background areas. Considering that these spiders have low dispersal capabilities, it may be that the regions with higher amounts of distance incorrectly show niche divergence due to incorporating potentially uninhabitable environments and, therefore, misinterpreting possible speciation mechanisms (Warren et al., 2008).
Species limits of the A. unicolor species complex
Species delimitation of the Antrodiaetus unicolor species complex has long been challenging due to both morphological stasis and variation across their distribution (Coyle, 1971; Hendrixson & Bond, 2005a). As already discussed, A. microunicolor is unambiguously recognized as a species based on morphological, behavioral, and genetic divergence from A. unicolor . However, the now paraphyletic assemblage ofA. unicolor lineages lacks distinctive features that could be used to distinguish them. Because no nomenclatural changes have been made to elevate these lineages to species status, the species-level diversity remains underestimated in this complex.
This study employed genome-wide SNPs for several clustering analyses and niche-based distribution modeling to evaluate genetic and ecological exchangeability, which further elucidated potential species boundaries. Clustering analyses with all individuals present (All_M20) as well as morphology and behavior coincide with the three species hypothesis (A. microunicolor , A. unicolor A, and A. unicolor B). Alternatively, the niche identity analyses and STRUCTURE/VAE analyses for A. unicolor B individuals only (UniB_M30) support the five species hypothesis (additional splitting of A. unicolor B into B1, B2, and B3). Although it is possible that these lineages were once allopatric and diverged before recent secondary contact in the Southern Appalachians, the tentative interpretation of potential incipient speciation events within A. unicolor B and less conservative niche identity test results remain too ambiguous for confident species delineation within the clade. Our integrative approach utilizing morphological, behavioral, and a substantial amount of molecular characters with ecological niche modeling provided ample evidence of an additional cohesion species within the complex for a total of three (A. microunicolor , A. unicolor A, and A. unicolor B), which potentially warrants distinguishing these two genetically and ecologically different lineages as separate species. In that case, one lineage would be designated as the true name-bearing Antrodiaetus unicolor and one with a new name. Considering that Hendrixson and Bond (2005a) designated a neotype for Antrodiaetus unicolor from Desoto State Park in northeastern Alabama and specimens included in this study from the surrounding areas cluster together in the lineageA. unicolor B, this lineage should be considered the true name-bearing Antrodiaetus unicolor. To our knowledge, none of the existing available names (junior synonyms of A. unicolor sensu lato) would be attributed to any of the A. unicolor A individuals based on geography. As such a new name will need to be proposed forA. unicolor A.
Overall, our study demonstrates the efficacy of genomic-scale data for recognizing cryptic species, suggesting that species delimitation with one data type, whether one mitochondrial gene or morphology, underestimates species diversity in taxa with low vagility and relative morphological stasis. Incorporating multiple lines of evidence (i.e., morphological, behavioral, geographical, and ecological diversity) in addition to genomic-scale data underscores the robustness of integrative species delimitation approaches across all organismal diversity despite differences in biological or ecological characteristics. We were able to resolve species-level paraphyly within the A. unicolor complex by delimiting an additional putative species despite morphological homogeneity. In addition, our study highlights another instance of cryptic speciation in the southern Appalachians, with phylogeographical patterns contributing to our understanding of the processes generating biodiversity in this rich, geologically and environmentally complex region.