Geographic distributions and patterns of co-occurrence among black-bellied and shovel-nosed salamanders (Desmognathus spp.) in the Great Smoky Mountains National Park

Introduction

National parks and other protected areas are tremendous public resources and critical reservoirs of biological diversity in a time of unprecedented habitat destruction (Hobbs et al., 2010). Millions of visitors benefit from recreation and educational opportunities in U. S. National Parks every year (Martin et al., 2011). National Parks also serve as invaluable laboratories for natural science research (Rodhouse, Sergeant & Schweiger, 2016) and repositories of archeological and recent history (Vukomanovic & Randall, 2021). The native species inhabiting parks and protected areas are an essential dimension of their value.

To understand the value of parks for biodiversity, it is critical to know which species occur in protected lands. This is an ongoing research need, in part because the scientific understanding of species boundaries and taxonomy is continually evolving. For example, recent changes in salamander taxonomy have resulted in a marked increase in the number of recognized species in southeastern North America (Tilley et al., 2013; Camp & Wooten, 2016). The Great Smoky Mountains National Park (GSMNP) is a hotspot of salamander diversity (Dodd, 2004), but the number and identity of the species occurring there is uncertain after recent taxonomic revisions of the genus Desmognathus (Pyron & Beamer, 2022, 2023). Here we attempt to clarify the distributions and patterns of co-occurrence of three recently described species with respect to the GSMNP.

‘Black-bellied’ and ‘shovel-nosed’ salamanders (BB and SN) are the two most aquatic ecomorphs in Desmognathus (Petranka, 1998; Bruce, 2011). Their ecological syndromes are evident in morphological comparison (Fig. 1). Black-bellied salamanders are the less aquatic of the two. They have round, bulging eyes and a heavily keeled tail that terminates in a point. They are the largest ecomorph in the genus, with adults reaching >100 mm SVL (Valentine, 1974). Shovel-nosed salamanders are the more aquatic ecomorph and are characterized by lower-profile, almond-shaped eyes and a heavily keeled, spatulate tail (Martof, 1962). Shovel-nosed salamanders are also large compared with other Desmognathus spp., with adults reaching >70 mm SVL (Martof, 1962). Both ecomorphs have darkly pigmented venters and dorsal patterns vary from nearly black to a greenish speckling common to SN or a light orange-brown common to BB (Martof, 1962; Niemiller & Reynolds, 2011). Larvae are difficult to identify at early life stages but become more easily distinguishable near time of metamorphosis (Niemiller & Reynolds, 2011). Large BB larvae are often olive green in color with paired orange-yellow dorsal spots, while large SN larvae are usually a significantly darker, black-velvet color with distinctly white gills and spatulate tails (Martof, 1962).

Historically, BB and SN were recognized as two species (D. quadramaculatus Holbrook 1840 and D. marmoratus Moore 1899); however, recent molecular systematics has mapped the two ecomorphs to two deeply divergent clades within Desmognathus (Jackson, 2005; Jones & Weisrock, 2018) and subsequently split each ecomorph into multiple species (Pyron et al., 2022; Pyron & Beamer, 2022, 2023). Jackson (2005) recognized a northern clade containing BB and SN from the northern extent of their range in Virginia south through the Blue Ridge to roughly midway through Tennessee and North Carolina, and a southern clade ranging from contact with the northern clade through North Georgia and South Carolina (see Fig. 2 for previous sampling of each clade). Subsequent studies (Jones & Weisrock, 2018; Pyron et al., 2022) have denoted these clades as the Pisgah (northern) and Nantahala (southern) clades. Both Jackson (2005) and Jones & Weisrock (2018) detected introgression between BB and SN ecomorphs in the Pisgah clade but no evidence of introgression between ecomorphs in the Nantahala clade, and it seems clear that there is no gene flow between Pisgah and Nantahala lineages (Pyron et al., 2020).

Co-occurrence of divergent clades of the same ecomorph has rarely been documented in previous studies, with only one site in each of Jackson (2005), Jones & Weisrock (2018), Pyron & Beamer (2022), and Pyron et al. (2025) where both clades of black-bellied salamanders occurred together or in the same local stream system. There have been no documented sites where shovel-nosed salamanders from both clades co-occur. This pattern supports the hypothesis that co-occurrence and divergence of BB and SN ecomorphs has been facilitated by ecological niche partitioning. Based on our understanding of the ecological roles of BB and SN salamanders, competitive exclusion ought to preclude extensive co-occurrence of different species of the same ecomorph (Hardin, 1960; Hairston, Nishikawa & Stenhouse, 1987; Bruce, 2011). Divergence between species of each ecomorph is strongly influenced by watershed connectedness, especially in the more obligately aquatic SN, with apparently ancient isolation between populations on either side of the Eastern Continental Divide (Voss et al., 1995; Jackson, 2005).

The GSMNP includes the type locality of one newly described Nantahala clade BB: D. gvnigeuswotli (Pyron & Beamer, 2022). Based on limited genetic sampling, D. gvnigeuswotli is thought to occur throughout the mountain range along with one Pisgah clade SN: D. intermedius (Pyron & Beamer, 2023). One record indicates the presence of one Pisgah clade BB (D. mavrokoilius) in the northeastern part of the park (Jackson, 2005). Here we use expanded sampling in the GSMNP to document co-occurrence between black-bellied salamanders from two deeply divergent clades (D. mavrokoilius and D. gvnigeuswotli), corroborate previous reports of a lack of genetic differentiation between co-occurring black-bellied and shovel-nosed ecomorphs within the Pisgah clade (D. mavrokoilius and D. intermedius), and present preliminary data indicating the presence of an additional clade of shovel-nosed salamanders in Cades Cove, GSMNP.

Methods

Sampling

We sampled BB and SN salamanders throughout the GSMNP starting in summer 2023. As part of an ongoing research program to genetically evaluate previously documented populations, we visited sites with known or suspected occurrences of BB or SN, with an emphasis on sites with co-occurrence of both ecomorphs (Dodd, 2004). In addition, many BB and some SN were sampled opportunistically in smaller seeps and creeks throughout the GSMNP. At six sites, we obtained large population samples as bycatch during three-pass electrofishing brook trout surveys carried out by park service scientists. We released each individual alive after removing roughly 4–10 mm of tail tissue for DNA analysis. We identified each adult as BB or SN based on eye, head, and tail shape and we identified larvae by color and tail shape (Martof, 1962; Fig. 1). We also photographed most individuals for reference. Tissues were suspended in either ethanol or salt lysis buffer and stored at −20C until used. Due to our non-lethal sampling technique, we usually sampled every individual caught, except for very small larvae (less than 2.5 cm total length) and individuals with prior tail damage. This provided large sample sizes from multiple major creeks and watersheds throughout the GSMNP (Fig. 3). Tissue sampling was carried out in compliance with the United States Animal Welfare Act [7 U.S.C. 2131 et seq.] and according to University of Tennessee Institutional Animal Care and Use Committee (IACUC) protocol 2,710 and National Park Scientific Research Collection Permit GRSM-2023-SCI-2209. See the methods appendix in the Supplemental Material for more information on sampling.

Results

Here we report results from mtDNA data obtained from a sample of 270 individuals from 27 sites in the GSMNP (Fig. 3). Sample sizes for each site ranged from n = 1 to 48 individuals, $\overline{\mathrm{x}}$ = 10.0. We collected DNA from 163 BB, 83 SN, and 24 individuals that were not identified in the field. We identified salamanders in the field using eye, head, and tail shape as well as larval color as described above. Morphologically ambiguous individuals were described qualitatively or marked as ‘mixed features.’ Individuals marked as ‘mixed features’ most often had BB-like faces and SN-like spatulate tails, all of which had Pisgah clade mtDNA.

Phylogenetic analysis

Processing of sequence data resulted in an alignment of 375 bp which was combined with the (Jackson, 2005) mtDNA population set for phylogenetic analysis. The topology of major clades within Desmognathus was consistent between MrBayes and BEAST analyses and with those reported by previous research. Clade assignment of our samples was consistent across both models, and largely consistent with field identifications and expected geographic ranges (Fig. 4).

Six individuals from two sites in Cades Cove, GSMNP (denoted “CC” in Fig. 3) that were identified in the field as SN grouped as sister to the Nantahala clade containing D. folkertsi and D. aureatus with strong support in both analyses (0.99–1.0). Of these six SN, three were sampled from Mill Creek at the western end of Cades Cove and three from Anthony Creek at the eastern end (see Fig. 3). Neither the dwarf black-bellied salamander, D. folkertsi, nor the Southern SN (D. aureatus Pyron & Beamer, 2023) have been reported in the Great Smoky Mountains or anywhere in Tennessee. Our Cades Cove SN clade is strongly supported as the sister group to the clade containing D. aureatus and D. folkertsi (Fig. 4), but are not included within D. aureatus, also with strong support (0.90–0.99). One sample from Cades Cove (from Mill Creek on the west end of the Cove) was uncertain in its relationship to other Cades Cove samples. We represent this with a polytomy that we are unable to resolve with our limited sequence data. The branch lengths estimated by BEAST indicate divergence between the Cades Cove clade and the clade containing D. aureatus and D. folkertsi being similarly ancient to that between the two described Nantahala BB species (D. gvnigeusgwotli and D. amphileucus).

All other SN sequences clustered unambiguously with the Pisgah clade, and all BB sequences clustered unambiguously with either the Pisgah clade or D. gvnigeuswotli within the Nantahala clade, except for seven samples that were excluded from D. gvnigeuswotli with weak support (0.3) but still grouped with Nantahala BBs (Fig. 4). These seven grouped with strong support (0.99) with a BB from Jackson’s (2005) dataset that was collected along US 441 in the GSMNP, and the topology of the Nantahala BB clade reflects that of the cyt B tree reported by Jackson. Within the Pisgah clade, our sequences clustered with specimens that would be classified as D. mavrokoilius (if BB) or D. intermedius (if SN) according to the taxonomy of Pyron & Beamer (2022, 2023). As with previous results, there was no consistent mtDNA differentiation between BB and SN ecomorphs within the Pisgah clade (Fig. 5).

Discussion

The Great Smoky Mountains harbor some of the most diverse assemblages of salamanders anywhere in the world (Petranka, 1998; Dodd, 2004). Understanding the historical and contemporary processes responsible for the evolution and maintenance of such diversity depends on clear documentation of species distributions and patterns of co-occurrence. Desmognathus salamanders are a noted example of ecological niche partitioning, with morphological differences among species associated with differential use of aquatic and terrestrial microhabitats (Bruce, 2011). But they are also notorious for cryptic diversity and fuzzy species boundaries (Tilley et al., 2013; Camp & Wooten, 2016). Here we describe extensive co-occurrence of genetically distinct but morphologically indistinguishable BB salamanders, interbreeding between morphologically distinct but genetically similar BB and SN salamanders, and an undescribed mitochondrial lineage of SN salamanders that might be restricted to a small watershed in the northwestern corner of the Great Smoky Mountains.

Our results largely corroborate data from Jackson (2005), Jones & Weisrock (2018), and Pyron et al. (2022) with respect to phylogenetic relationships between the many cryptic lineages of BB and SN. Our finer-scale sampling throughout the GSMNP suggests that co-occurrence of BB from divergent clades is more extensive than previous data indicated. Frequent co-occurrence of cryptic lineages in a system where co-occurrence is understood as being facilitated by ecological niche partitioning is a challenge to the prevailing dogma (Hairston, 1987). Three potential explanations for this pattern are (1) that BBs from divergent clades partition habitat more finely than previously understood, (2) that some external disturbance or biotic interaction prevents competitive exclusion of one clade by the other, or (3) co-occurrence is a temporary phase in the invasion and displacement of one species range by the other. Developing more detailed descriptions for BB microhabitat occupancy and resource use could help to address the first possibility. Pyron & Beamer (2022) showed subtle statistical differences in body proportions that might be associated with resource use. In future sampling efforts, care should be taken to record specific information such as stream size, cover object types, and water flow. For example, Pierson, Fitzpatrick & Camp (2021) found a difference in Eurycea wilderae and E. cirrigera microhabitat use by categorizing sections of stream as ‘pools,’ ‘riffles,’ or ‘runs’ and genotyping the salamanders that were caught in each flow type. A similar approach involving BB and SN ecomorphs could be helpful in building a higher resolution understanding of ecological strategies used by each ecomorph and cryptic lineage.

The lack of mtDNA differentiation between Pisgah BB and SN (D. mavrokoilius and D. intermedius) also corroborates previous data, adding to the need for population genetic analyses to resolve questions of species delineation and hybridization. Jackson (2005) concluded that mitochondrial and nuclear variation were associated with geography but not morphology within the Pisgah clade. Pyron et al. (2020) suggested that the Pisgah clade could be subdivided into several BB and SN species with repeated parallel evolution explaining the lack of correspondence between genomic and phenotypic similarity. Subsequently, Pyron et al. (2022) supported this interpretation with a larger dataset, but also presented evidence of mixed ancestry across several geographically adjacent groups within the Pisgah clade, consistent with hybridization between ecomorphs. Pyron et al. (2025) interpreted the lack of correspondence between ancestry and phenotype as the result of relatively rare, ancient hybridization coupled with a hypothetical quantitative genetic threshold mechanism determining expression of discrete phenotypic syndromes.

The above studies were broad in geographic scope, with rarely more than two specimens per locality. Our analysis of population samples found no mtDNA differentiation between ecomorphs living in the same 100 m reaches of Cosby Creek, Pretty Hollow Creek, and Rough Fork. The variation in our data is shared between ecomorphs (Fig. 5), with different ecomorphs in the same stream consistently more similar to each other than to their corresponding ecomorphs in other streams as close as 4.5 km away (Rough Fork to Pretty Hollow; Table 2). We fail to reject the null hypothesis of no differentiation between Pisgah clade phenotypes within streams. However, any single locus might be misleading, and our sample sizes are too modest to detect correlations between ancestry and phenotype much weaker than about 0.64 (Cohen, 1988). Our future work will characterize potential admixture between BB and SN and measure dispersal and geneflow between local populations. These data will be essential for understanding the ecological mechanisms that maintain BB and SN phenotypes and/or facilitate hybridization between the two.

Our detection of a previously undescribed SN lineage in Cades Cove, GSMNP highlights the pervasiveness of cryptic diversity in Desmognathus and warrants further study to determine its relationship to known southern shovel-nosed salamanders (D. aureatus), describe any distinctive morphological characteristics, and determine the extent of its geographic range. Six individuals from two sites in Cades Cove grouped as a sister clade to D. aureatus and D. folkertsi, and BEAST estimated relatively deep divergence, similar to that between distinct Nantahala BB species. Moreover, the morphological and genomic distinctiveness of D. folkertsi (Camp et al., 2002) argues against classifying the Cades Cove clade as D. aureatus if additional molecular data corroborate its phylogenetic position as sister to D. aureatus and D. folkertsi (Fig. 4).

Our mtDNA analysis was designed for efficient categorization of samples into previously described clades (Jackson, 2005), not for de novo phylogenetic analysis. With only 375 base pairs, we replicated the major features of the previous analyses, but with less certainty and a few relatively minor differences in tree topology. From the 270 samples included in the phylogenetic analysis, a subset of 91 identified as Pisgah clade BB or SN were included in the haplotype network and genetic distance analyses.

Of the 270 individuals included in our phylogenetic analysis, only the six Cades Cove clade samples and one Nantahala BB were not within 93% sequence similarity to previously published haplotypes. Additionally, within our six Cades Cove SN samples, sequences were highly variable. The five samples that formed the monophyletic clade displayed up to 7% pairwise divergence, and GSM23123 was 9.8% different from the next closest Cades Cove sample. This high degree of variation and their being only 84% similar to the next closest species (D. folkertsi) on GenBank casts doubt on them being valid cytochrome B sequences. It is possible that these irregular sequences represent nuclear-mitochondrial insertions (NUMTs) rather than valid cytochrome B sequences. However, BLASTing our sequences against an existing genome assembly for D. fuscus returns no significant similarities, arguing against this possibility (GenBank acc.: GCA_050004315.1) Therefore, our diagnosis of the Cades Cove clade remains highly provisional until more comprehensive molecular analyses are completed.

In contrast to our findings of co-occurrence of BB ecomorphs from divergent clades, we found no evidence here of co-occurrence of Pisgah SN (D. intermedius) and our ‘Cades Cove’ SN (D. cf. aureatus). No Pisgah clade haplotypes were present in any samples from either Mill Creek or Anthony Creek, our two sites within Cades Cove. Both of our sampled creeks drain north into Abrams Creek, which flows west out of the Great Smoky Mountains (Fig. 3). The nearest documented SN (D. intermedius) was collected by Jackson (2005) in a creek that joins Abrams Creek lower in the watershed several miles below Abrams Falls. We also documented D. intermedius in the Twentymile Creek watershed, which is south of Cades Cove and hydrologically connected to Abrams Creek via the Little Tennessee River. This apparent lack of co-occurrence of SN species is consistent with the hypothesis that competitive exclusion impedes the coexistence of identical ecomorphs of divergent clades and further emphasizes the need for a better understanding of what might facilitate coexistence of distinct BB species. The presence of a Nantahala clade SN in a tributary of the Little Tennessee River in the GSMNP is especially shocking considering the relatively ancient divergence between SN on either side of the Eastern Continental Divide (Voss et al., 1995; Jones et al., 2006). Jones et al. (2006) suggested, based on a molecular clock analysis and geologic history of the region, that a geologic event isolated populations of SN, which are less likely than BB to disperse over land, before the Pleistocene. We hope that further sampling in Cades Cove and surrounding areas will clarify the geographic extent and evolutionary history of this disjunct lineage.

Conclusions

Our results reveal a more complex and diverse assemblage of BB and SN Desmognathus in the Great Smoky Mountains National Park than previously recognized, emphasizing the value of parks for preserving biodiversity, both known and not yet described. The Cherokee Black-bellied Salamander (D. gvnigeuswotli Pyron & Beamer, 2022) appears to occur throughout the mountain range with the possible exception of some of the easternmost streams flowing directly into the Pigeon River. The Blue Ridge Black-bellied Salamander (D. mavrokoilius Pyron & Beamer, 2022) is common in the Cosby and Cataloochee watersheds in addition to smaller drainages such as Tobes Creek on the eastern end of the Great Smoky Mountains. It might occur sporadically as far west as Deep Creek. These two BB species cannot be distinguished with certainty by known visual characteristics (Pyron & Beamer, 2022). The central shovel-nosed salamander (D. intermedius Pyron & Beamer, 2023) occurs across the Great Smoky Mountains from the easternmost streams to the Twentymile Creek area in the southwest. Hybridization between D. intermedius and D. mavrokoilius might be common where they co-occur. We observed several individuals with Pisgah clade mtDNA with the head and eyes of the BB morph but the spatulate tail of the SN morph. We also detected a potentially novel mitochondrial lineage of SN salamanders in Cades Cove. Further research is needed to determine the phylogenetic composition of SN salamanders in the north-western quadrant of the Great Smoky Mountains.

Supplemental Information