The two vertical entrances to Shelta Cave are located within a large sinkhole in a residential area in northwest Huntsville, Alabama within the Highland Rim physiographic province and Tennessee River Watershed in northern Alabama. The 762-m cave system trends in an east-west direction under Cave Avenue, Pulaksi Pike, and several residential homes. Shelta Cave was purchased in 1967 by the National Speleological Society (NSS) to protect and preserve the diverse cave community for scientific research and conservation (Hobbs and Bagley 1989; Culver 1999). The cave is still owned and maintained as a preserve by the NSS. Shelta Cave is developed within the Mississippian-aged Tuscumbia Limestone and Fort Payne Chert undifferentiated (Rheams et al. 1992) and is characterized by three large rooms that are interconnected by short passageways (Fig. 1A). For detailed descriptions of Shelta Cave see Johnston (1933), Torode (1973), Cooper (1975), Hobbs and Bagley (1989), and Rheams et al. (1992). The largest room is Jones Hall measuring ~201 m east-west × 250 m north-south that consists of two main levels: an upper level with a substantial accumulation of breakdown slabs on the floor and a lower level ~9 m below the upper level with a mud-cover floor and some breakdown slabs. The western area of the lower level is known as Johnson Hall. Historically, a Grey Bat summer maternity colony roosted in the northwest section of Jones Hall in an area called Bat Lake. East Hall (also called East Room) is the smallest of the three main chambers measuring ~244 m × 61 m. Miller Hall (also called West Room) is the western most chamber and is accessed from Jones Hall through Cooper’s Crawl. This chamber measures ~135 m × 300 m with ceilings rarely exceeding 4.6 m and the floor characterized by large blocks of breakdown covered by thick mud. The 6.1-m pit entrances open into a high passage that connects Jones Hall with East Hall. During periods of low water levels in late summer and fall, isolated pools can be found in Jones Hall and East Hall that gradually diminish in extent until no water is present (Cooper 1975). Miller Hall contains the only permanent water in the cave year-round – West Lake which typically maintains 0.9–1.2 m of water and up to 0.6 m of mud and silt substrate during the dry season. Grey Bats also were known to roost in this section of the cave. During winter and spring, the local water table rises, and the lower levels of the cave completely flood with groundwater becoming inaccessible. Water levels may fluctuate 10 m or more between the wet and dry seasons. Dye tracing has demonstrated that water flows from West Lake in Miller Hall to Jones Hall likely to East Hall and ultimately to Brahan Spring 3.5 km to the southeast of the cave.

Figure 1. 

A Map of Shelta Cave showing the distribution of aquatic habitat during high water levels and the location of Shelta Cave Crayfish observations (black crayfish symbol) during the study including B a female in 2019 from North Lake and C a male in 2020 from West Lake. Map modified with permission of the Alabama Cave Survey.

We conducted visual encounter surveys of accessible aquatic habitats (isolated pools and phreatic lakes) on 20 occasions between October 2018 and July 2021. Surveys were conducted by 1–5 researchers for 0.75–6.5 person-hours depending on water levels. During low water levels, surveys were conducted on foot in the lower levels of Jones Hall, Miller Hall, and East Hall. During high water levels, surveys were typically limited to the upper levels of Jones Hall and East Hall. We searched for aquatic life with headlamps and handheld dive lights. During high water, we also conducted snorkel surveys in Jones Hall. We made a concerted effort to capture with handheld dipnets and examine all cave crayfish observed. Select crayfish were photographed before measuring total length, carapace length, chelae length and width, and examined reproductive condition. Species identification was based on examination of chelae, gonopods, and body size (as noted above in the Introduction) and aided by photographs of type material in the North Carolina State Museum of Natural Sciences by Guenter Schuster. We also removed and retained a walking leg as a tissue sample preserved in 100% ethanol for genetic analyses. Captured crayfish were released at their point of capture after processing.

To investigate whether cave crayfish abundance (i.e., direct visual counts) has changed over time at Shelta Cave, we compiled count data from literature sources spanning 1968–2012, including Cooper (1975), Lee (1987), Hobbs and Bagley (1989), Rheams et al. (1992), McGregor et al. (1994, 1997), Cooper and Cooper (1997), and Miller (2013). These data were combined with visual counts from our recent surveys. We employed generalized linear models (GLMs) with the census counts as the response variable and survey date as the explanatory variable for two time periods: 1) 1968–1975 during Cooper’s (1975) regular crayfish surveys for his dissertation work, and 2) 1985–2021. We examined two datasets: a dataset representing visual counts for O. sheltae and the second which includes visual counts of all cave crayfishes observed irrespective of species. Because count data often exhibit a Poisson or negative binomial distribution and also can be zero-inflated (Linden and Mantyniemi 2011), we explored the best fit of several different distributions, including zero-inflated and non-zero-inflated Poisson, negative binomial, and negative binomial with NB2 parameterization [variance = μ (1 + μ /k)], using the glmmTMB (Brooks et al. 2017) package in R. We developed zero-inflated models using a single zero-inflation parameter but also developed hurdle models that first modeled the binary likelihood that a 0 value is observed and modeled the non-zero observations using a truncated Poisson or negative binomial model. We determined the best fitting models using AICc using the bblme package in R (Bolker and R Core Development Team 2017). The best fitting model was used to estimate the overall trend during each time period. We also tested for differences in abundance of each dataset between the two time periods using a Mann-Whitney test after testing for deviations from normality with a Shapiro-Wilk normality test.

We extracted DNA from walking legs using the Qiagen DNEasy Blood and Tissue Kit according to the manufacturer’s protocol except for a few modifications. Walking legs were manually crushed using a small pestle after addition of Buffer ATL. This was followed by the addition of 40 µL of proteinase K. The sample was incubated overnight at 56 °C, with occasional vortexing while incubating to ensure adequate mixing. After the addition of Buffer AL, the sample was incubated at 70 °C for 10 minutes. Finally, the DNA was eluted using 125 µL of Buffer AE which had been preheated to 70 °C.

Polymerase chain reaction (PCR) was used to amplify two mitochondrial loci, 454 bp of 16S rRNA (16S) and 642 bp of cytochrome oxidase subunit I (CO1). Each 25 µL PCR reaction consisted of 12.5 µL of GoTaq Colorless MasterMix (Promega), 1.0 µL each of 10 µM forward and reverse primers (Table 1), 7.5 µL of molecular grade water, and 3.0 µL of DNA template. Gel electrophoresis was used to confirm successful PCR amplification using an Axygen gel documentation system. PCR products were cleaned using ExoSAP-IT (Affymetrix) and sequenced in both directions using BigDye chemistry at Eurofins MWG Operon (Louisville, Kentucky) using PCR primers.

Forward and reverse sequences were trimmed at the ends based on quality and assembled into contigs in ChromasPro v2.1.8 (Technelysium Pty Ltd, South Brisbane, Australia). Contigs were aligned using MUSCLE (Edgar 2004) in MEGA X version 10.0.5 (Kumar et al. 2018). We checked for the presence of premature stop codons and indels in CO1 sequences, which would indicate pseudogene sequences, by converting to amino acid sequences in MEGA X. Novel sequences generated in this study were accessioned into GenBank (accession nos. ON380874–ON380893 for CO1 and ON383885–ON383904 for 16S). We also included in our 16S and CO1 datasets sequences for other cave and surface cambarid crayfishes available on GenBank (Suppl. material 1: Table S1) to infer phylogenetic placement of O. sheltae within Cambaridae and with respect to other cave crayfishes, in particular. This approach has been used in other decapods (e.g., Varela and Bracken-Grissom 2021; Varela et al. 2021). The crayfishes Astacus astacus (Linnaeus, 1758) and Pacifastacus leniusculus (Dana, 1852) both in the family Astacidae were included as outgroup taxa. We constructed gene genealogies using maximum likelihood (ML) analysis using RAxML-HPC v.9.2.10 (Stamatakis 2014). Optimal models of nucleotide substitution for each locus, including first, second, and third codon positions for CO1 and 16S, were determined in PartitionFinder2 (Lanfear et al. 2017) using corrected Akaike’s Information Criterion (AICc). ML analyses were conducted under the GTRGAMMA model and rapid bootstrapping algorithm with 10,000 bootstraps. Partitions determined by PartitionFinder2 were included in the analyses. Trees were visualized in FigTree v1.4 (Rambaut 2014). We calculated mean uncorrected pairwise distances between O. sheltae and other cave crayfishes occurring in northern Alabama in MEGA X (Kumar et al. 2018).

We observed 20 cave crayfish (mean ± 1 SD: 1.3 ± 1.6 crayfish) during 12 of 20 surveys between October 2018 and July 2021. Eighteen crayfish were identified as O. australis. However, two individuals were identified as O. sheltae (Fig. 1). During a snorkel survey of North Lake in Jones Hall on 31 May 2019, MLN captured a small crayfish in ~3.7 m of water. The female crayfish measured 36 mm total length and 18 mm carapace length (Fig. 1B). The crayfish lacked first pleopods, possessed narrow and elongate chelae, and lacked prominent spines on the mesial margin of the carpus. All characters, in addition to smaller adult size, are consistent with O. sheltae. Developing ova also were observed internally. A second individual was captured in West Lake in Miller Hall on 28 August 2020 by MLN and NS. This form I male measured 30.5 mm total length and 14.4 mm carapace length, possessed narrow, elongate chela, and lacked prominent spines on the mesial margin of the carpus (Fig. 1C). In addition, gonopod shape was consistent with the original description of O. sheltae (fig. 1B, C, E, and F in Cooper and Cooper 1997).

We assembled cave crayfish count data for 122 surveys spanning from November 1968 through July 2021 including the current study (Suppl. material 2: Table S2). Total cave crayfish abundance averaged 140.4 ± 61.9 crayfish (maximum: 250; minimum: 34) during Cooper’s (1975) study over 18 surveys spanning November 1968–November 1973. Orconectes sheltae abundance averaged 6.2 ± 6.2 individuals (maximum: 18; minimum: 0) over the same period. Both total cave crayfish (U = 1818, P < 0.001; Fig. 2) and O. sheltae abundance (U = 1907, P < 0.001; Fig. 3) were drastically lower over 101 surveys spanning December 1985–July 2021 compared to the November 1968–July 1975 period. Total cave crayfish abundance averaged 2.8 ± 3.2 crayfish (maximum: 14; minimum: 0) and O. sheltae abundance averaged 0.03 ± 0.17 crayfish (maximum: 1; minimum: 0) over this latter period. Best fitting models (negative binomial and negative binomial with ND2 parameterization; Table 2) showed significant declines in abundance between 1968 and 1975 for all cave crayfishes (Fig. 2B) and O. sheltae (Fig. 3B).

Figure 2. 

A Trends in visual census counts of all cave crayfishes (Orconectes australis, Cambarus jonesi, and C. sheltae) at Shelta Cave from 1968–2021, with B trends during the study of Cooper (1975) and C since 1985 also visualized. Note that the scale on the x- and y-axes differ for B and C.

Figure 3. 

A Trends in visual census counts of Cambarus sheltae at Shelta Cave from 1968–2021, with B trends during the study of Cooper (1975) and C since 1985 also visualized. Note that the scale on the x-axis differs for B and C.

The 16S and CO1 ML analyses placed O. sheltae in a clade with other troglobiotic Cambarus endemic to the Interior Low Plateau karst region in northern Alabama with strong support (Figs 4, 5). This clade includes C. hamulatus, C. jonesi, C. laconensis, C. pecki, C. speleocoopi, and O. sheltae. Mean uncorrected pairwise distances between O. sheltae and other species in this clade ranged 5.1–7.8% and 3.0–6.2% for CO1 and 16S, respectively. The other cave-obligate Orconectes form a monophyletic group more closely related to other members of Cambarus (Figs 4, 5). Mean uncorrected pairwise distances between O. sheltae and cave Orconectes ranged 10.6–11.9% and 6.8–7.8% for CO1 and 16S, respectively.

Figure 4. 

Maximum-likelihood phylogram showing relationships among Orconectes sheltae (in green) and other cave (in blue) and surface (in black) cambarid crayfishes inferred from the mitochondrial CO1 locus. Bootstrap support >80% is shown to the left of the corresponding node.

Figure 5. 

Maximum-likelihood phylogram showing relationships among Orconectes sheltae (in green) and other cave (in blue) and surface (in black) cambarid crayfishes inferred from the mitochondrial 16S ribosomal RNA locus. Bootstrap support >80% is shown to the left of the corresponding node.

The discovery of two individuals of O. sheltae during recent surveys in 2019–2020 demonstrates that the species is not yet extinct, as has been hypothesized by past authors (Buhay and Crandall 2005; Elliott 2005; Niemiller and Taylor 2019). However, the species remains critically imperiled and on the brink of extinction, as the population appears to be extremely small in size given only three individuals have been documented since the 1970s and overall cave crayfish abundance since 1985 at Shelta Cave is just ~2% of counts during Cooper’s (1975) dissertation work in the late 1960s into the early 1970s. The rediscovery of O. sheltae also offers hope that other imperiled stygobiotic species thought to be extirpated may be rediscovered at Shelta Cave in the future. The federally endangered Palaemonias alabamae was last observed on 24 November 1973 at Shelta Cave (Cooper 1975; Cooper and Cooper 2011) but has been documented at two new sites in the last 20 years (USFWS 2016; Niemiller et al. 2019). The salamander Gyinophilus palleucus was never abundant at Shelta Cave or other caves in and around the metropolitan Huntsville area (Cooper 1968, 1975; Cooper and Cooper 1968, 2011; Niemiller and Niemiller 2020). This top aquatic predator has not been observed since 1968 at Shelta Cave (Cooper 1975). Orconectes sheltae co-occurs with two other stygobiotic crayfishes at Shelta Cave. While recent surveys have confirmed the continued persistence of O. australis, Cambarus jonesi has not been documented since 1988 (Hobbs and Bagley 1989).

Population declines in O. sheltae and other stygobionts at Shelta Cave have been linked to impaired water quality and reduction in energy input into the aquatic ecosystem (Hobbs and Bagley 1989; Moser and Rheams 1992; Wilson and Robison 1993; McGregor et al. 1997; Elliott 2000, 2012). Elevated levels of cadmium and the pesticides heptachlor and dieldrin have been detected in the groundwater of Shelta Cave, the latter likely associated with increased urbanization (Hobbs and Bagley 1989; Moser and Rheams 1992; McGregor et al. 1997). The pesticides likely entered via rapid infiltration through epikarst from residential areas located within the recharge basin of the cave system. The aquatic ecosystem likely also was negatively impacted by the loss of a Myotis grisescens maternity colony (Elliott 2000, 2012), which was an important source of organic input through deposition of guano and dead individuals. After purchasing the cave in 1967, the National Speleological Society constructed a bat-unfriendly gate in 1968. The colony was estimated at 54,000 bats in 1969, and bats were occasionally observed during the early 1970s (Cooper 1975; Cooper and Cooper 2011); however, the colony had completely disappeared by the late 1970s (Hobbs and Bagley 1989). The gate was replaced with a more appropriate design in 1981 (Hobbs and Bagley 1989) and completely removed and replaced with a high fence around the entrance in the early 2000s (Elliott 2012), but the bat colony has yet to return, and the aquatic ecosystem has yet to recover at Shelta Cave.

The taxonomy of crayfishes in the family Cambaridae has been based historically on morphology; however, phylogenetic relationships and evolutionary histories may be obfuscated by convergent evolution (Crandall and Fitzpatrick 1996; Taylor and Knouft 2006; Breinholt et al. 2012), which may be of particular concern when elucidating phylogenetic relationships in cave-obligate crayfishes (Sinclair et al. 2004; Buhay and Crandall 2009) and cave organisms in general (Christiansen 1961; Culver et al. 1995; Wiens et al. 2003). The taxonomy of Cambaridae continues to be in a state of flux. Several studies have uncovered lack of support for monophyly of several genera (Taylor and Knouft 2006; Johnson et al. 2011; Breinholt et al. 2012; Crandall and De Grave 2017; Stern et al. 2017; Glon et al. 2018), including Orconectes (Crandall and Fitzpatrick 1996; Fetzner 1996; Sinclair et al. 2004; Buhay and Crandall 2005, 2008; Owen et al. 2015; Stern et al. 2017). Recently, Crandall and De Grave (2017) presented a phylogeny based on a subset of molecular data generated in Stern et al. (2017) which showed that cave Orconectes form a distinct clade more closely related to members of Cambarus, while surface Orconectes are more closely related to Barbicambarus, Creaserinus, and other members of Cambarus. Consequently, Crandall and De Grave (2017) restricted Orconectes to the cave taxa, as the type species of the genus is Orconectes inermis, and resurrected the genus Faxonius Ortmann, 1905 for the surface-dwelling group. Phylogenetic relationships of cave-obligate cambarid crayfishes based on 16S and CO1 datasets estimated in this study are similar to relationships estimated based on a maximum-likelihood analysis of partial mitochondrial (12S, 16S, and CO1) and nuclear (28S) sequence data by Carroll et al. (2021) and a six-locus dataset (12S, 16S, CO1, 18S, 28S, and histone H3) by Stern et al. (2017).

Our phylogenetic analyses revealed that the genus Orconectes as currently recognized is in need of additional taxonomic refinement, as we did not find support for inclusion of O. sheltae within Orconectes. Cooper and Cooper (1997) placed the Shelta Cave Crayfish in the genus Orconectes based primarily on gonopod morphology, and this classification has been followed since its description (e.g., Crandall and de Grave 2017). However, Cooper and Cooper (1997) noted that O. sheltae is “quite different from the other troglobitic members of the genus.” Specifically, O. sheltae i) lacks the first pleopods in females; ii) possesses a broad median trough of the annulus; ii) possesses elongate, narrow chelae, with a long palm and subvertical orientation; iv) has longer terminal elements of the form I male gonopod, with a greater degree of curvature and cephalocaudal flattening of the central projection; v) possesses a great depth of the cephalocaudal axis of the shaft of the gonopod proximal to the base of the central projection; vi) lacks prominent spines on the mesial margin of the carpus; and vii) is small in size, with a maximum carapace length of 19.7 mm. Our molecular results demonstrate that O. sheltae is a member of a clade that contains other geographically proximate cave-obligate species in northern Alabama in the genus Cambarus. Therefore, we advocate for recognition of this species as Cambarus sheltae to more accurately reflect evolutionary relationships in this taxon.

Discordance between gonopodal morphology and genetics in cave crayfishes of northern Alabama is not without precedence. Buhay and Crandall (2009) discovered that Procambarus pecki was in fact also a member of Cambarus based on molecular analyses and recognized the species as Cambarus pecki. Like O. sheltae, P. pecki was noted previously to be “disjunct with other members of the genus” and once belonged to its own monotypic subgenus Remoticambarus (Hobbs 1972). It is becoming increasingly clear that relying solely on morphological characters is inadequate to infer species’ boundaries and taxonomic relationships in cave-obligate crayfishes because of convergent evolution on troglomorphic characters, such as reduction in eye structures, loss of pigmentation, and attenuation of antennae and limb, as well as confounding gonopodal structures. Rather, genetic data paired with geographic information, other ecological data, and diagnosable characters (outlined in Buhay and Crandall 2009) appear to be sufficient for species identification in these stygobiotic crayfishes. It is important to note that past authors have cautioned against relying solely on mitochondrial loci for inferring phylogenetic relationships, as this has the potential to yield inaccurate hypotheses in arthropods (Fontaine et al. 2007; Song et al. 2008; Leite 2012), including crayfishes (Song et al. 2008; Buhay 2009; Schubart 2009), due to factors such as paternal leakage into the mitogenome (e.g., Fontaine et al. 2007; Mastrantonio et al. 2019) and the presence of mitochondrial pseudogenes (Song et al. 2008; Buhay 2009). We followed recommendations by Buhay (2009) and found no evidence of pseudogenes (i.e., premature stop codons and presence of indels) in our O. sheltae CO1 sequences. We did detect the presence of putative pseudogenes in some O. australis sequences, however, and these sequences were excluded from analyses.