Redescription of Marstonia comalensis (Pilsbry & Ferriss, 1906), a poorly known and possibly threatened freshwater gastropod from the Edwards Plateau region (Texas)

Abstract Marstonia comalensis, a poorly known nymphophiline gastropod (originally described from Comal Creek, Texas) that has often been confused with Cincinnatia integra, is re-described and the generic placement of this species, which was recently allocated to Marstonia based on unpublished evidence, is confirmed by anatomical study. Marstonia comalensis is a large congener having an ovate-conic, openly umbilicate shell and penis having a short filament and oblique, squarish lobe bearing a narrow gland along its distal edge. It is well differentiated morphologically from congeners having similar shells and penes and is also genetically divergent relative to those congeners that have been sequenced (mtCOI divergence 3.0–8.5%). A Bayesian analysis of a small COI dataset resolved Marstonia comalensis in a poorly supported sub-clade together with Marstonia hershleri, Marstonia lustrica and Marstonia pachyta. The predominantly new records presented herein indicate that Marstonia comalensis was historically distributed in the upper portions of the Brazos, Colorado, Guadalupe and Nueces River basins, south-central Texas. The species has been live collected at only 12 localities and only two of these have been re-visited since 1993. These data suggest that the conservation status of this snail, which has a critically imperiled (G1) NatureServe ranking and was recently proposed for federal listing, needs to be re-assessed.


Introduction
Th e freshwater gastropod genus Marstonia (Hydrobiidae: Nymphophilinae) is composed of 15 small (shell height < 5.0 mm), ovate-to elongate-shelled species that are distributed in springs, streams and lakes in eastern North America (Th ompson and Hershler 2002, Hershler et al. 2003a, Th ompson 2005. Most of these species have extremely narrow geographic ranges and consequently have become a focus of conservation activities; two are federally listed as endangered (USFWS 1994(USFWS , 2000 and others are variously listed by state wildlife agencies. Marstonia diff ers from the other eight North American nymphophiline genera in that the (female) oviduct and bursal duct join well in front of (instead of behind) the posterior wall of the pallial cavity (Th ompson and Hershler 2002); it has also been resolved as a well supported sub-clade within its subfamily based on mtDNA sequences (Hershler et al. 2003a). Although Marstonia has been reviewed three times since 1978 (Th ompson 1978, Hershler 1994, Th ompson and Hershler 2002, three of its congeners have been little studied beyond their original descriptions and their anatomy is unknown. Two of these -M. olivacea (Pilsbry), M. ozarkensis (Hinkley) -may be extinct (Th ompson 1978, NatureServe 2009) and thus will likely remain incertae sedis. Th e third, M. comalensis (Pilsbry and Ferriss) (originally Amnicola comalensis Pilsbry & Ferriss), which is extant, is the focus of this paper. Pilsbry and Ferriss (1906) described A. comalensis based on six shells from Comal Creek and the Guadalupe River near New Braunfels, south-central Texas. Th ey diff erentiated this species from A. limosa (Say) and two nomina (A. cincinnatiensis [Anthony], A. peracuta Pilsbry & Walker) that are currently recognized as synonyms of Cincinnatia integra (Say) (see Hershler and Th ompson 1996) by its much smaller size and noted that it further diff ered from the latter by its less shouldered whorls. Th e genus Amnicola was used at that time as "a catch-all for most American amnicoloid species that could not conveniently be placed elsewhere on the basis of their shells" (Th ompson 1968: 150). Amnicola comalensis was not further treated taxonomically until Taylor (1975) transferred it to Cincinnatia without comment in a bibliographic compilation; this allocation was widely followed in the subsequent literature (e.g., Burch andTottenham 1980, Turgeon et al. 1998). During the course of a revisionary study of C. integra, Hershler and Th ompson (1996) examined several alcohol preserved collections of a snail that they identifi ed as A. comalensis and noted that it closely resembled species of Marstonia (which were then placed in Pyrgulopsis); A. comalensis was subsequently transferred to Marstonia based on this unpublished work (Th ompson and Hershler 2002). Hershler et al. (2003a) recently published a molecular phylogenetic analysis of the North American nymphophilines that included a specimen of M. comalensis from Old Faithful Spring in Real County, Texas (ca. 180 km from the type locality), which was depicted as nested within the Marstonia clade. Th is is the only published record for M. comalensis subsequent to its original description.
We redescribe M. comalensis herein based on study of a large series of dry shell and alcohol-preserved material, most of which was collected by malacologists J.J. Landye and D.W. Taylor from 1971Taylor from -1993, and provide anatomical evidence supporting its current generic allocation. Th e new records detailed in this paper considerably expand the geographic range of M. comalensis, which lives in springs and fl uvial habitats spread among four river basins in south-central Texas. We also further analyze previously published molecular data (Hershler et al. 2003a) to evaluate the divergence and phylogenetic relationships of M. comalensis, whose geographic range is broadly disjunct relative to other members of the genus. Th e information presented in this paper may assist eff orts to protect this poorly known species, which was included in a recent federal listing petition (Rosmarino and Tutchton 2007) based on its critically imperiled (G1) NatureServe (2009) ranking, but found not to warrant listing owing to insuffi cient information (USFWS 2009).

Materials and methods
Anatomical study was based on specimens that were relaxed with menthol crystals and fi xed in dilute formalin. Types and other material of M. comalensis in the collections of the Academy of Natural Sciences of Philadelphia (ANSP); Florida Museum of Natural History (FMNH); National Museum of Natural History, Smithsonian Institution (USNM); and University of Minnesota Bell Museum of Natural History (UMBM-NH) were examined during the course of this study.
Variation in the number of cusps on the radular teeth was assessed using the method of Hershler et al. (2007). Other methods of morphological study and descriptive terminology are those used in recent taxonomic investigations of nymphophiline gastropods (Hershler 1998, Hershler et al. 2003b. Shell data were compiled using Systat for Windows 11.00.01 (SSI 2004).
Th e molecular phylogenetic analysis included single mtCOI sequences from M. comalensis, six other species of Marstonia and representatives of six other North American nymphophiline genera. Hydrobia acuta (Draparnaud) was used as the root. Sample information, GenBank accession numbers and publication references for the sequences are in Table 1. Sequence divergences (uncorrected p distance) were calculated using MEGA4 (Tamura et al. 2007). Phylogenetic relationships were inferred using Bayesian inference in MrBayes 3.12 (Ronquist and Huelsenbeck 2003). MrModeltest (Nylander 2004) selected the Hasegawa-Kishino-Yano model with some sites assumed to be invariable and with variable sites assumed to follow a discrete gamma distribution (HKY + I + G), which best fi t the data under the Akaike Information Criterion. In the initial Bayesian analysis the burn-in was set at 10% (10,000 generations) of the chain length (100,000 generations). Th ree runs were conducted in MrBayes using the HKY + I + G model and the default random tree option to determine when the loglikelihood sum reached a stable value (by plotting the log-likelihood scores of sample points against generation time). Th e ln likelihoods started around -4,300 and quickly converged upon a stable value of about -3,050 after 1,000 generations. For the fi nal run, Metropolis-coupled Markov chain Monte Carlo simulations were performed with four chains for 1,000,000 generations and Markov chains were sampled at intervals of 10 generations to obtain 100,000 sample points. Th e sampled trees with branch lengths were used to generate a 50% majority rule consensus tree with the fi rst 5000 trees (equal to 50,000 generations) removed to ensure that the chain sampled a stationary portion.    Revised diagnosis: Shell large for genus (maximum height, 4.6 mm), ovate-conic, openly umbilicate; penis with short fi lament and oblique, squarish lobe bearing a single terminal gland along its distal edge. Distinguished from congeners having closely similar shells and penes as follows: from M. gaddisorum Th ompson by its less convex shell whorls, distinctive pallial roof pigmentation, larger number of cusps on the inner side of the lateral teeth and on the outer marginal teeth, larger penial lobe, narrower terminal gland, and smaller overlap of the bursa copulatrix by the albumen gland; from M. lustrica by its smaller prostate gland, smaller penial lobe, narrower penial fi lament, straight anterior vas deferens, partly imbedded (in albumen gland) bursal duct, and larger seminal receptacle; and from M. ogmorhaphe Th ompson by its smaller size, broader shell, smaller prostate gland, straight anterior vas deferens, and smaller bursa copulatrix.
Cephalic tentacles pale except for black eyespots. Snout brown; distal lips pale; foot pale. Pallial roof having black pigment bands along edges of ctenidium and dorsal edge of genital duct (Fig. 1H); visceral coil pale except for black pigment on testis. Ctenidium positioned a little in front of pericardium; ctenidial fi laments 24-25 (n=5), broadly triangular, lateral surfaces ridged. Osphradium narrow, positioned slightly posterior to middle of ctenidium. Hypobranchial gland large, overlapping rectum and part of genital duct, thickened alongside kidney. Style sac longer than remaining portion of stomach, posterior stomach having small caecal appendix. Testis large (1.75 whorls), composed of compound lobes, broadly overlapping stomach anteriorly. Seminal vesicle opening near anterior edge of testis, composed of a few thickened coils, positioned along ventral side of anterior 33% of testis. Prostate gland small, pea-shaped, with about 50% of length in pallial roof. Anterior vas def- erens opening from antero-ventral edge of prostate gland, section of duct on columellar muscle straight. Penis large, base rectangular, inner edge without folds; fi lament short, narrow, tapering, oblique; lobe rather medium-sized, squarish, oblique (Fig. 3A). Terminal gland (Fig. 3A-D) narrow, usually transversely positioned along outer edge of lobe (58/86 specimens examined from three samples), less frequently horizontal (28/86), sometimes borne on short stalk. Penial duct narrow, near outer edge, almost straight. Penial fi lament having black internal pigment core along most of length. Ovary small (0.75 whorl), composed of simple, stalked lobes; slightly overlapping stomach anteriorly. Female glandular oviduct and associated structures shown in Figure 3E-F. Coiled oviduct narrow, vertical. Bursa copulatrix small, ovate, horizontal, about 50% overlapped by albumen gland. Bursal duct longer than bursa, narrow, opening from distal edge, partly embedded in albumen gland proximally, entirely embedded distally, junction with common duct well in front of posterior wall of pallial cavity. Seminal receptacle small, pouch-like, positioned near ventral edge of albumen gland slightly anterior to bursa copulatrix. Albumen gland largely visceral. Capsule gland composed of two distinct tissue sections. Genital aperture a terminal slit. Distribution and habitat: Marstonia comalensis is distributed in the upper portions of the Brazos, Colorado, Guadalupe and Nueces River basins, south-central Texas (Fig. 4); almost all of these localities are on the Edwards Plateau. We were unable to confi rm a previous report of this species from a drainage canal near Galveston Bay (Cable and Isserhoff 1969). Marstonia comalensis lives in cold water springs near their sources and slack water riverine habitats; it has been most commonly found on mud, aquatic vegetation and dead leaves.
Remarks: Th e material referred to M. comalensis herein, which includes specimens from the Guadalupe River above the Comal Creek confl uence, closely conforms to the types of this species both in size and shell shape ( Fig. 1A-C). Th is snail clearly belongs to Marstonia based on its strongly wrinkled shell protoconch, distally bifurcate penis ornamented with a gland along the edge of the lobe (terminal gland), and connection between the oviduct and bursal duct well in front of the posterior pallial wall (Th ompson and Hershler 2002). As noted above, this generic placement is also supported by molecular phylogenetic evidence (Hershler et al. 2003a).
Th e original collections of M. comalensis are worn shells having the appearance of drift material (Fig. 1A). We have not seen any live-collected specimens of this species in the numerous samples that we have examined from the type locality (Comal Springs) and other waters near New Braunfels. Th e various reports of living M. comalensis from this portion of the Guadalupe River basin (e.g., EHA 1975;Arsuffi 1994;Tolley-Jordan and Owen 2008) are probably of misidentifi ed C. integra, as evidenced by the illustrations of shells in several of these documents (Lindholm 1979, fi g. 4;Cauble 1998, fi g. 7). It is possible that M. comalensis became extinct at Comal Springs when this water body temporarily dried in 1964 (USFWS 1996); it is also possible that the shells of this species which have been found at this site were washed downfl ow from extant populations in the headwaters of the Guadalupe River. Taylor's (1975) allocation of A. comalensis to Cincinnatia appears to have been the result of a misidentifi cation as all of the material in his collection that he referred to this species (per the original labels), including several lots from the type locality, is C. integra (RH unpublished). Some of these records were detailed in an unpublished manuscript, "Freshwater molluscs from the Nueces River drainage, Texas" that Taylor circulated in the mid-1970's. Cincinnatia integra, which is widely distributed in Texas (Hershler and Th ompson 1996), has been frequently confused with M. comalensis in museum collections despite the obvious diff erences between their shells that were noted in the original description of the latter (Pilsbry and Ferriss 1906). Th ese two species also well diff erentiated anatomically (see Hershler and Th ompson 1996 for details of the former).

Discussion
Th ompson (1978) speculated that Marstonia is composed of two species lineages that are diff erentiated by the size and shape of the penial lobe and fi lament. Marstonia comalensis, which is distributed almost 800 km distant from its most proximal congener, conforms to the putative lineage characterized by a large, squarish lobe and small, slender fi lament. Th is group includes widely ranging M. lustrica and species distributed in the Tennessee (M. arga Th ompson, M. ogmorhaphe, M. pachyta), Alabama (M. hershleri; see Th ompson 1995) and Altamaha (M. gaddisorum; see Th ompson 2005) River basins. Th e molecular phylogenetic relationships of M. comalensis were previously delineated by Hershler et al. (2003a), who analyzed a mtCOI dataset using maximum parsimony and maximum likelihood methods. We re-analyzed the relevant portion of these data using a Bayesian algorithm Th e resulting topology (Fig. 5) is closely similar to those illustrated by Hershler et al. (2003a: fi gs. 2, 3) and has slightly better resolution. Marstonia is resolved as a well supported (100% posterior probability) clade. Th ree congeners living in Georgia coastal drainage, which Th ompson (1978) recognized as a distinct lineage based on their elongate penial lobe and robust fi lament, formed a well supported (95%) subclade. Th e other four congeners included in the analysis, which conform to the second putative lineage discussed above, formed a weakly supported subunit, within which M. comalensis was positioned as sister to a subclade containing M. lustrica and M. pachyta. Th e single sequenced specimen of M. comalensis diff ers from those of the six other congeners that have been similarly analyzed by 3.0-8.5%; it is most similar to M. pachyta.
Marstonia comalensis is ranked as critically imperiled (G1) by NatureServe (2009), but has a minimal conservation profi le otherwise. Following the negative fi nding per its proposed listing (USFWS 2009) it was categorized by the USFWS (2010) as "status undefi ned." It was misidentifi ed as a "mussel" in a ecological sustainability report for the Cibola National Forest Grasslands plan revision (USDAFS 2008) and is not mentioned on the Texas Parks and Wildlife's website (http://www.tpwd.state.tx.us/) or in the State Wildlife Action Plan (Texas Parks and Wildlife 2010). Given that M. comalensis has been live-collected at only 12 localities (Fig. 4) and only two of these sites (Leakey Springs, Old Faithful Spring) have been re-visited since 1993, it would seem prudent to add this species to the list of aquatic biota of the Edwards Plateau region meriting protection (Bowles and Arsuffi 1993) and assess its current conservation status. Depending on the extent of its possible decline since 1993, M. comalensis may merit addition to the IUCN Red List (e.g., if it consists of 10 or fewer extant populations; IUCN 2001) and other conservation watch lists.