Hierarchical shifts in stream hydrology, changes in geology, and drainage boundaries can exist in lotic system as natural barriers to dispersal (Hughes et al. 1995; Fetzner and Crandall 2003; Nguyen et al. 2015). In addition, benthic, headwater-adapted species naturally exhibit hierarchical population structure due to inherent biological factors such as low dispersal potential and habitat specificity (Fluker et al. 2014; Hurry et al. 2015; Schmidt and Schaefer 2018a). Headwater species may stave off local extinction processes through dispersal, but human-mediated habitat fragmentation can disrupt dispersal and further isolate populations, reducing their long-term viability (Ward et. A. 1994; Alp et al. 2012; Paz-Vinas et al. 2015).
For conservation management, it is important to disentangle naturally occurring population structure from structure resulting from recent disturbances (Zellmer and Knowles 2009; Davis et al. 2014; Epps and Keyghobadi 2015). We expected to find signatures of long-standing isolation within C. pristinus attributed to reduced dispersal at three spatial scales: between the major drainage divides separating the Sequatchie River and Caney Fork River, among tributary systems of the Caney Fork River due to shifts in geology and topography associated with the physiographic break of the Western Escarpment of the Cumberland Plateau and among headwater tributaries due to shift in hydrology associated with larger order streams. Although, we identified patterns of historic isolation between the Caney Fork and Sequatchie forms of C. pristinus, the observed structure was not related to any of our hypothesized barriers. Furthermore, we found no evidence for long-standing geographic structure within either of the forms of C. pristinus as indicated by short branch lengths and a common haplotype shared among the majority of sites within each morphological form. This suggests that within each form (or Clade) some level of gene flow was maintained historically among populations.
In contrast, the results from our microsatellite analyses identified contemporary patterns of population structure within the Caney Fork form. Based on our EEMs analysis, some of the observed structure and areas of reduced gene flow were related to the physiographic shift from the Cumberland Plateau to the Western Escarpment and, to a lesser degree, and increase in stream size. This pattern was not reflected in our mitochondrial dataset, so we are unable to determine if the separation of the three upper Caney Fork clusters by the Western Escarpment from downstream clusters is due to long-standing isolation or related to contemporary patterns of human-mediated habitat alteration, or a combination of both. However, population structure observed within tributary systems not separated by these physiographic breaks, such as within clusters in the upper Caney Fork River, and the overall low levels of genetic diversity recovered across population clusters indicate recent declines and increased population isolation, likely associated with human habitat alterations that have negatively impacted the species. Low allelic richness, small Ne, evidence of population bottlenecks, and signatures of inbreeding observed across population clusters and for the Caney Fork form as a whole, are indicative of a declining species at an elevated risk of extinction (Spielman et al. 2004).
Phylogeography of Cambarus pristinus
Previous studies have shown that natural barriers to dispersal exist in lotic systems (i.e., hierarchical shifts in stream hydrology, changes in geology, and drainage boundaries) and that these barriers contribute to long-standing isolation in headwater-adapted aquatic organisms (Hughes et al. 1995; Fetzner Jr. and Crandall 2003; Nguyen et al. 2004; Hollingsworth Jr. and Near 2009; Kanno et al. 2011; Lamphere and Blum 2011; Hurry et al. 2015). However, the degree of population structure in headwater-adapted species varies with some like Etheostoma basilare, Corrugated Darter, and Creaserinus burrisi, Burrowing Bog Crayfish, exhibiting strong patterns of population structure attributed to historic vicariance (Hollingsworth Jr. and Near 2009; Clay et al. 2020), while other like Euastacus bispinosus, Glenelg Spiny Crayfish, and Etheostoma sagitta spilotum, Kentucky Arrow Darter maintained connectivity across river systems and physiographic shifts within their range (Miller et al. 2014; Blanton et al. 2019). The latter suggests that some headwater-adapted species may be less constrained by inherent dispersal potential or factors such as tributary density, stream size, drainage shape, or distance (Turner and Trexler 1998; Schmidt and Schaefer 2018b).
Based on the habitat specificity exhibited by C. pristinus we predicted that larger river habitats and geographic breaks would serve as filters or barriers to dispersal, contributing to reduced gene flow among populations separated by these features. We recovered two genetically distinct and divergent clades representing the two morphological forms of C. pristinus, suggesting long-standing isolation of these forms. However, the geographic distribution of these two forms does not correspond to existing geographic breaks examined, given that the Sequatchie form spans the Cumberland-Tennessee River drainage divide. It is unlikely that the current distribution is explained by long distant dispersal due to the extensive distance (including most of the Tennessee and Cumberland River mainstems) separating extant populations and the lack of records of either form in streams of > 4th order in size that they would need to traverse. More likely, an ancestral population present on the Cumberland Plateau experienced a vicariant event resulting in the isolation of the two forms either in their present locations or between the Sequatchie and Caney Fork rivers; in the latter case, the isolation event may have been followed by range expansion through underground, overland, or headwater transfer of the Sequatchie form into its present range in the upper Caney Fork River system (Finlay et al. 2006; Berendzen et al. 2008; Niemiller and Zigler 2013). Estimates of divergence times from closely related taxa endemic to the Cumberland Plateau suggests the two forms of C. pristinus diverged sometime during or after the Pleistocene (Crandall et al. 2015). Climate oscillations and hydrological changes occurring during that time have been linked to the speciation of several lineages of freshwater taxa within the region (Thornbury 1965; Near et al. 2001; Berendzen et al. 2003, 2008; Kozak et al. 2006). Estimating divergence times for the two lineages would help link the observed patterns to specific geological events in the area and to evaluate and test for potential alternative scenarios that may have led to divergence between the two forms.
Overall haplotype diversity within each morphological form was low, and two sites, 7 (Caney Fork form) and 10 (Sequatchie form), exhibited a single haplotype unique to each site. This may indicate a recent reduction in population size for each form and limited gene flow to sites 7 and 10 (Schrimpf et al. 2011). Alternatively, it is also possible that the low levels of haplotype diversity reflect the species’ ecological or evolutionary history. Similarly, low haplotype diversity is observed in the broadly distributed C. parvoculus and C. jezerinaci (Mountain Midget crayfish and Spiny Scale crayfish, respectively) of the Cumberland Plateau (Thoma and Fetzner Jr. 2008). Low haplotype diversity has been observed in species with small Ne and among species that have undergone demographic stochasticity or a historic range reduction (Bucklin and Weibe 1998; Matocq and Villablanca 2001), which are observed features of C. pristinus.
Contemporary genetic diversity and population structure
In contrast to our mtDNA results that implied population connectivity across the range of the Caney Fork form, our contemporary levels of population structure inferred from our microsatellite data indicate that the Caney Fork form is comprised of six genetically differentiated populations. Patterns of isolation by distance (IBD) due to the hierarchical, linear nature of streams are expected for freshwater organisms (Wright 1943; Frissell et al. 1986; Fetzner et al 2003; Sexton et al. 2013). However, we did not recover a significant relationship between geographic and river distances for the Caney Fork form (Fig. 3). We did recover a significant relationship between genetic structure and Euclidean distance but it only explained a small portion of the genetic differentiation observed. The lack of an IBD pattern using river distance and improbable occurrence of overland dispersal, despite a significant IBD association using Euclidean distance, suggests that something other than geographic distance contributed to population structure for the Caney Fork form. In headwater-adapted species with limited dispersal potential, the expected linear relationship between genetic and geographic distance may only occur at smaller geographic scales and become nonlinear at greater geographic scales. In such cases, habitat patch arrangement may be a more important predictor of genetic differentiation than geographic distance alone (van Strien et al. 2015). Under this scenario, a non-linear IBD relationship may occur at larger distances or spatial scales (Schmidt and Scahefer 2018a)). For example, IBD may be important in explaining population structure at finer spatial scales of the Caney Fork form such as for population structure observed between Clusters A, B, and C but landscape variables (i.e., Western Escarpment, shifts in hydrology of the mainstem Caney Fork River) may be a better explanatory factor for population structure among all the Caney Fork form clusters rather than IBD.
As noted, our microsatellite data indicate a potential relationship between the Western Escarpment and reduced levels of gene flow among populations separated by this feature. We also found fine-scale population structure among populations that crossed the Caney Fork River, suggesting reduced gene flow occurs as a stream size becomes > 4th order. These results suggest the shift from the Cumberland Plateau to the Western Escarpment may act as a natural barrier or filter to gene flow, and to a lesser extent, larger order rivers may also limit gene flow among populations within the Caney Fork form. However, the uplift of the Cumberland Plateau and subsequent incision of the Caney Fork tributaries into the Plateau likely occurred during the Pliocene and into Pleistocene (Sasowsky et al. 1995; Anthony and Granger 2007). If these features limit dispersal among populations, it is interesting that we did not see evidence of long-standing isolation around these features (Western Escarpment and > 4th order mainstem habitats) in our mtDNA results.
There are several possible explanations for why we did not detect geographic structure in the Caney Fork form in our mitochondrial DNA analyses. For example, nuclear and mitochondrial genes exhibit different inheritance patterns and are expected to exhibit contrasting patterns of population structure in species exhibiting sex-biased dispersal (Caparroz et al. 2009). Similar patterns of genetic structure to those herein were observed in Red grouse (Lagopus lagopus scoticus) where population structure was detected using microsatellite markers but panmixia was inferred from mtDNA (Piertney et al. 1998, 2000). The authors suggested that several processes could produce this pattern including recent divergence of a species exhibiting low haplotype divergence or selective pressure on the mitochondrial genome and hypothesized that strong female-biased dispersal created a situation where locally the effective population size of the nuclear genome was much smaller than that of the mitochondrial genome (Piertney et al. 2000). Accordingly, the nuclear genome would be more susceptible to genetic drift and result in stronger patterns of population structure. Female-biased dispersal has been documented in crustaceans (Stevens et al. 2006; Cannas et al. 2012), but the few studies of dispersal in crayfish have shown that dispersal is either non-biased (Bubb et al. 2006) or male-biased (Wutz and Geist 2013). However, we do see low haplotype divergence in the Caney Fork form and studies suggest divergence from the Sequatchie form may be as recent as the Pleistocene or later (Crandall et al. 2015).
Alternatively, an increasing number of studies have identified exceptions to long-supported assumptions regarding mtDNA (i.e., maternally inherited, neutral selection, constant mutation rate, non-recombinant; White et al. 2008; Galtier et al. 2009). One potential pitfall associated with mtDNA is heteroplasmy, or the presence of multiple mtDNA haplotypes in a single individual. Once thought to be rare, an increasing number of studies have reported heteroplasmic mtDNA for several animal taxa including crustaceans (White et al. 2008; Zuber et al. 2012; Rodríguez-Pena et al. 2020). Additionally, there is growing support for recombination events of the mitochondrial genome in animals (White et al. 2008). Heteroplasmy coupled with recombination events impacts phylogenic inferences and can reset the molecular clock (Posada 2001; Posada and Crandall 2002), erasing evidence of past isolation and divergence. Lack of congruence in phylogenetic trees from multiple genes can identify the potential for heteroplasmy and/or recombination events (Posada 2001). However, because of our focus on population-level structure, our phylogeographic estimation for C. pristinus was limited to a single mitochondrial gene. It is unclear what role heteroplasmy may play in the patterns we observed for the evolutionary history of C. pristinus.
Although we cannot rule out that some degree of the observed contemporary population structure reflects reduced dispersal due to long-standing, natural instream features (i.e., the Western Escarpment of the Cumberland Plateau, large mainstem habitats), it is likely that recent metapopulation dynamics of C. pristinus have been disrupted due to human-mediated habitat degradation that has contributed to observed declines in occurrence, population size, and abundance (Johansen et al. 2016). Small populations, such as the small census estimates for C. pristinus (Johansen et al. 2016), are more sensitive to genetic drift, which in the absence of gene flow, can reduce genetic diversity leading to the loss of beneficial alleles and the fixation of detrimental alleles, and ultimately to the loss of adaptive potential (Frankham 1995). Point estimates of Ne for each population and the Caney Fork form overall were below 100. The 100/1000 rule states that an Ne of 100 is required to prevent inbreeding depression and genetic drift while an Ne of 1000 is necessary to maintain evolutionary potential. This rule provides a useful benchmark to assess the role population size has in the current conservation status of a species (Frankham et al. 2014). Several populations of the Caney Fork form, particularly those in the southern portion of the range, show signs of susceptibility to the negative effects of genetic drift and inbreeding via low Ne estimates and high degrees of isolation.
Inbreeding is another potential threat to the long-term viability of isolated populations. Inbreeding exacerbates the effects of genetic drift and can interact with a small population size to push a species into a vortex that can rapidly push a species to extinction (Gilpin and Soulé 1986; Keller and Waller 2002; Frankham 2005; Wright et al. 2008). We did detect genetic signatures of inbreeding in several of our population clusters and for the Caney Fork form as a whole. This would be expected based on the small population size and isolation we observed in the Caney Fork form.
Additionally, we detected a bottleneck in population D, suggesting a recent drastic decline in population size, which could reduce genetic diversity through the random loss of alleles. Although we do not have direct evidence of what caused this decline, historic mining practices within the headwaters of Puncheoncamp Creek (population cluster D) and subsequent conversion to silviculture practices are a likely cause of population decline (Moore 1985; Withers and McCoy 2005; Schorr et al. 2006). Mining and silviculture activities negatively impact crayfish populations by increasing sedimentation and metal pollutants and altering water chemistry and riparian buffers (Allert et al. 2012, 2013; Helms et al. 2013; Loughman et al. 2015; Richman et al. 2015). The site for population D was different from other sites we visited for this study. The site consisted of a bedrock stream bed with few patches of large cobble substrate, a high degree of bank erosion, and a large amount of sediment deposition compared to our other sites. Anthropogenic disturbances are prevalent throughout the range of C. pristinus (Moore 1985; Withers and McCoy 2005; Schorr et al. 2006), and it is possible other populations may have experienced bottleneck events, but our method requires a reduction of 50–80% of the effective population to detect bottleneck two-thirds of the time (Hoban et al. 2013). It is also possible that other factors such as pre-bottleneck genetic diversity, bottleneck persistence, the timing of bottleneck event, and population growth obscured genetic signals needed to test for bottleneck effects (Williamson-Natesan 2005; Zachariah Peery et al. 2012).
Signatures of isolation, inbreeding, and population bottlenecks indicate a species at an elevated risk of extinction and our assessment of genetic diversity for the Caney Fork form supports this conclusion. The loss of genetic diversity often precedes the distributional and demographic indicators of a declining species (Brook et al. 2002; Spielman et al. 2004). While not directly comparable due to variation in alleles, values for these metrics for the Caney Fork form were similar to those for other imperiled crayfishes and relatively lower than values associated with non-imperiled crayfish (Gouin et al 2011; Li et al. 2012; Miller et al. 2014; Duncan et al. 2020; Table S2). Although it may be common for naturally rare species to maintain low levels of genetic diversity through gene flow; this is unlikely to be the case for C. pristinus. The pattern of population structure observed in the Caney Fork form indicates gene flow is limited and is likely contributing to our low genetic diversity estimates. In concert with our mitochondrial assessment, C. pristinus appears to exhibit reduced adaptive potential when compared to non-imperiled crayfishes. However, more work is needed to explicitly test the role of contemporary land uses on genetic diversity and population connectivity (Frankham 1995; Matocq and Villablanca 2001).
The observed morphological differences and genetic divergence observed between the two forms of C. pristinus, indicate that the forms are on independent evolutionary trajectories. These results also support their recognition as distinct species. We recommend that the Sequatchie form be provisionally referred to as Cambarus aff. pristinus. Due to the type locality residing within the upper Caney Fork River watershed (Hobbs 1965) and the holotype specimen representing a population of the Caney Fork form we recommend that only the Caney Fork form retains the Cambarus pristinus nomenclature. Management strategies should be developed independently for each lineage. This will be particularly important if translocation or captive propagation are utilized to supplement declining populations or re-establish gene flow between extant population (Vrijenhoek 1998).
Although our study indicates that IBD and natural in-stream features, primarily the shift to the Western Escarpment of the Cumberland Plateau and to a lesser degree stream size, contribute to the observed isolation observed in C. pristinus (Caney Fork form), they do not fully explain our genetic diversity results. Low genetic diversity measures combined with population loss, inbreeding, and bottleneck events suggest that contemporary anthropogenic impacts also negatively impact C. pristinus population connectivity. Additional work from a landscape genetics approach would provide a more robust test of effects of specific landscape and environmental factors that have contributed to the observed patterns of genetic diversity and isolation we recovered.
We identified patterns of low heterozygosity, low allelic diversity and inbreeding indicative of a species suffering the negative impacts of genetic stochasticity related to small population size and isolation (Brook et al. 2002). Although, C. pristinus may benefit from translocation of individuals among isolated populations, such actions would require a better understanding of the basic biology of this species (Ingvarsson 2001). For example, we do not know how contemporary population connectivity and effective population size estimates compare to historic patterns. Anecdotal evidence supports the idea that this species was never particularly abundant across its range (Hobbs 1965) and there is evidence that rare organisms possess traits that enable them to remain viable in a state of rarity (Kunin and Gaston 1993). Therefore, understanding the basic ecology (i.e., microhabitat requirements, territoriality, intraspecific competition, or sex-biased dispersal) and life-history traits will be vital to the success of conservation strategies utilizing translocation or captive propagation (Blundel et al. 2002; Butler et al. 2005; Johansen et al. 2018).
Regular genetic monitoring will be essential for successful conservation. Our estimates of Ne were low and illustrate that only 10% of individuals estimated for census size contribute genes to subsequent generations. Thus, mark-recapture census estimates may typically overestimate the health of the population. Additionally, populations will often exhibit genetic indicators of decline before population loss or size reduction (Brook et al. 2002; Frankenham et al. 2014); regular genetic monitoring will establish trends for the species and identify early warning signs for isolated populations and the species overall.
Although we could not assess genetic diversity and gene flow for C. aff. pristinus due to low sample size, our surveys and observations of habitat decline for this taxon highlight a need for increased management. Silviculture is prevalent throughout its range and many streams we visited had increased sedimentation and loss of interstitial spaces. Although not explicitly tested, it is likely that C. aff. pristinus has similar habitat requirements to C. pristinus, indicating silviculture may be detrimental to population persistence (Withers and McCoy 2005; Rohrback and Withers 2006). This is further supported by our repeated failures to detect individuals at several historical sites within streams surrounded by silviculture and indicates local population loss. We propose that additional surveys in concert with a genetic assessment be conducted to determine the current status of C. aff. pristinus. In addition, we recommend surveys that use an occupancy modeling framework to determine if this taxon is extirpated from historic sites or if individuals are simply difficult to detect.
In conclusion, we recommend continued recognition of Cambarus pristinus as a species of conservation concern, with elevated risks of extinction. We propose a similar ranking for C. aff. pristinus. Both have small ranges, are patchily distributed, occur in low densities, and have been potentially extirpated from parts of their ranges. Furthermore, we demonstrate that C. pristinus exhibits a suite of genetic traits that indicate it is susceptible to stochastic and long-term threats. The data we have provided here can serve as a guide to developing effective conservation strategies for both species. However, more work is needed to understand each species basic biology and ecology to guide specific conservation efforts that may reverse their statuses.