Late Pleistocene fishes of the Tennessee River Basin: an analysis of a late Pleistocene freshwater fish fauna from Bell Cave (site ACb-2) in Colbert County, Alabama, USA

Bell cave site description

Bell Cave is located along the south bank of the Tennessee River in the northwest corner of Alabama in Colbert County (Fig. 2). Situated approximately 11 km west of the town of Tuscumbia, the cave is located near Tennessee River Mile 248.2 at 87°47′32″N, 34°43′46″W (Churcher et al., 1989; see also Holman, Bell & Lamb, 1990; Martin & Prince, 1990; Parmalee, 1992). The cave formed within the middle Mississippian-age Tuscumbia Limestone and the modern entrance to the cave overlooks the Tennessee River and sits approximately 15 m above the current mean water level (Martin & Prince, 1990; Parmalee, 1992). The cave consists of a 26.2 m horizontal crawlway that opens into a large 4.0 by 11.9 m oblong “pit-like” room with a ceiling that does not exceed 10.5 m in height (Fig. 3; Churcher et al., 1989; Holman, Bell & Lamb, 1990). This oblong room is oriented northeast to southwest and has become a settling basin for fissure-transported sediments (Churcher et al., 1989; Holman, Bell & Lamb, 1990; Parmalee, 1992). The southwest wall of the cave reveals a floor-to-ceiling fissure that likely served as the primary entrance during the period of bone accumulation (Parmalee, 1992). On the floor in the southwestern half of the cave, evidence of historic digging is present (Fig. 3), likely the result of 19th century saltpeter mining.

Museum excavations within Bell Cave revealed an accumulation of approximately 70 cm of sterile clay overlain by 40 cm of bone packed mud (Fig. 3; Holman, Bell & Lamb, 1990; Martin & Prince, 1990; Parmalee, 1992). Vertebrate remains from the site were recovered from four lithologically distinct, stratigraphically-controlled levels, referred to herein as “zones” (Fig. 3; Bell, 1985; see also Holman, Bell & Lamb, 1990; Martin & Prince, 1990; Parmalee, 1992). These bone-containing zones each represent unique depositional events and can literally be “peeled” from one another (Bell, 1985; Holman, Bell & Lamb, 1990; Martin & Prince, 1990; Parmalee, 1992). The lowest 70 cm of the site represents a sterile layer and is overlain by the lowermost bone-bearing stratum, Zone 4 (Bell, 1985; Martin & Prince, 1990). These two layers consist of yellow laminated clays that were subaqueously deposited at a time when the water level of the Tennessee River was much higher than that of present day (Bell, 1985; Churcher et al., 1989). Desiccation cracks and settling at the top of Zone 4 indicate a period of dewatering and consolidation of these two layers and the onset of travertine formation at the top of Zone 4 (Bell, 1985). These travertine deposits suggest that some time had elapsed between the dewatering and the deposition of the upper three layers, Zones 1 through 3.

Zones 1, 2, and 3 consist of bone-filled reddish textured clays that were deposited as independent mud flow events from fissures at the southwest end and in the ceiling of the main cave room (Martin & Prince, 1990). Zones 1 and 2 were combined and referred to as Zone 1–2 since they could not be distinguished in certain parts of the cave (Parmalee, 1992). Bell (1985) and Churcher et al. (1989) reported the following radiocarbon dates from bone samples taken from the four bone-bearing zones:

Zone 1–2 (0–15 cm)—11,820 + 480/−500 B.P. Dated from an Ursus americanus femur collected in the upper part of Zone 2 and the lower part of Zone 1. Date determined by Dicard Radioisotope Company (Norman, OK)—sample DIC 2929, 1(222-02-3).

Zone 3 (15–30 cm)—Not enough large bone was present for dating, but this zone has a similar sedimentary profile and is faunistically identical to Zone 1–2.

Zone 4 (30–40 cm)—26,500 + 870/−990 B.P. Dated from large longbone fragments collected at the base of Zone 4, just above the sterile layer. Date determined by Dicard Radioisotope Company (Norman, OK)—sample DIC 3117, 236-61. As a specific caveat to prior literature on the dating of these zones, in a study of the Bell Cave herpetofauna, Holman (1995) failed to report the presence of Zone 4 and incorrectly assigned the date of 26,500 B.P. to Zone 3.

Overall, Zones 1–2 and 3 represent secondary depositional events in which consolidation and mixing within each zone may have occurred (but not between them; Martin & Prince, 1990). Although the main sources of bone accumulation within the cave were the result of subaqueous deposition (Zone 4) and viscous mudflows (Zones 1–2 and 3), other methods of bone accumulation were also evident. Some bone material likely fell or “filtered” through a former fissure located above the cave as well as death during denning or hibernation; predation by predators also likely contributed to the accumulation (Churcher et al., 1989; Parmalee, 1992). The former fissure at the southwest end of the cave would have been large enough to serve as an entrance for large carnivores such as wolves and bears (which are among the mammalian remains recovered; see Ebersole & Ebersole, 2011) and the presence of raptor remains suggest the possibility of roosting within the cave (Parmalee, 1992). Other recovered specimens show excessive gnawing produced by rodents and other scavengers (Holman, Bell & Lamb, 1990), suggesting the possibility of rodent transport of some material. Human remains were recovered from clay and travertine cap deposits located above Zone 1–2. The discovery of no such remains from any of the underlying zones suggests the 40 cm of bone accumulation occurred before any significant human populations were in the area (Bell, 1985).

Systematic paleontology

Referred Material—see Appendix S1 for specific lot catalog numbers, type and quantity of specific elements, and stratigraphic zone information (arranged as Zones 1–2, 3, 4, and Mixed). Mixed areas are ‘disturbed areas’ and refer to any specimen in which a specific zone could not be determined, such as those from superficial caps, within fissures, resultant of historic diggings, or from rimstone areas within the cave deposits (Fig. 3; see Appendix S1 for specific position in the cave). All identified material referenced the University of Tennessee Zooarchaeology Collections (Knoxville, TN, USA) or comparative skeletal material from the personal collection of WC Dickinson. All taxonomy and references to biogeography, ecology, conservation status, and Tennessee River Basin occurrences referenced Etnier and Starnes’ The Fishes of Tennessee (1993), Boschung and Mayden’s Fishes of Alabama (2004), and Page and Burr’s Field Guide to Freshwater Fishes (2011). References to current mainstem taxa occurrences and abundances follow recent (1993–2014) Tennessee River mainstem electrofishing and gillnet surveys undertaken by Tennessee Valley Authority (TVA) Biologists (Kurt Lakin, TVA Biologist, personal communication). The systematic arrangement of higher taxonomic rankings follows that of Caroll (1988).

Material

Fin rays, vertebrae, dentaries, premaxillae, teeth, operculi, spines, unknown fragments.

Occurrences

Zones 1–2, 3, 4, Mixed.

Remarks

Although much of the Bell Cave material could be identified at least to the family level (450 unique identifications arranged in lots), 114 collections of material still remain problematic for definitive identification. The majority of this material is comprised of unidentifiable vertebrae and fin rays; however, several other fragments carry notes in Appendix S1 as to suspected identifications and potential significance. The majority of these unidentified fragments almost certainly come from the taxa outlined below—however, several fragments may amend new taxa to the list. For example, lot RMM 4544 contains the unidentified remains of a dentate fragment with several tiny recurved teeth in situ. This element could represent a possible Hiodontidae member. Additionally, lot RMM 4950 contains a few cranial elements that could represent a single Bowfin Amia calva (Amiidae) individual. Other unidentified fragments may actually be smaller non-fish vertebrate fragments such as mammal or bird remains mistakenly apportioned with fishes. Other fragments may be too eroded or amorphic for any identification attempts and remain cataloged as such. Regardless, additional matrix from future cave excavations may result in further taxonomic detail and resolution.

A caveat to any discussion of late Pleistocene fish remains is that these collections almost undoubtedly represent a subset of the actual living assemblage and that numerous taxa are likely missing due to some form of taphonomic bias occurring between decomposition, transport, deposition, preservation, and collection processing. For example, the complete absence of otoliths among the examined material or within the >100 kg of recovered bulk matrix from Bell Cave leads us to believe this absence is preservational. The chemical composition of nearly all Actinopterygiian otoliths is aragonite (as opposed to bone and teeth which are composed of apatite; Hall & Kennedy, 1967), a form of calcium carbonate that is often subject to erosion by synsedimentary or postsedimentary decalcification when in the presence of infiltrating acidic water (Nolf, 1985). The former presence of acidic water within Bell Cave is evident by not only the formation of the limestone cave itself, but by the geomorphological events that formed the various strata (subaqueous and mud-flow deposition) and cave formations (i.e., travertine and speleothems). Additionally, the likely underrepresentation of smaller taxa (e.g., Minnows and Darters) may represent a combination of factors ranging from deposition (lighter fragments may be less readily held in deposits), preservation (smaller bone fragments weather more rapidly than larger fragments), decomposition (may degrade more rapidly than larger fragments, particularly in digestive tracts of larger animals such as owls), and/or collection processing (screen mesh size may exclude the tiniest fragments) biases. These taphonomic biases are also evident within the nine other localities discussed in this study, as all report fish taxa that were not able to be identified at or below the order level.

Material

Fulcrum, scute, lateral plate, quadrate, unidentified skull fragments.

Occurrences

Zones 1–2, 3, 4, Mixed

Material

Fulcrum, scute, parasphenoid.

Occurrences

Zone 3.

Material

Scute.

Occurrences

Mixed.

Remarks

The presence of these remains confirms the native status of these taxa to the Tennessee River drainage. However, a combination of harvest and habitat alterations have resulted in the depletion of both of these taxa throughout the modern Tennessee River Basin (Etnier & Starnes, 1993; Boschung & Mayden, 2004; Appendix S2). Specifically, Boschung & Mayden (2004) list this family as entirely extirpated from the basin; however, recent TVA mainstem fisheries surveys did recover a small number of Lake Sturgeon (8 out of 687,334 total collected fish). The presence of Lake Sturgeon in recent surveys is unlikely an error in Boschung & Mayden (2004); rather, their presence is almost undoubtedly the result of recent stocking efforts to reestablish breeding populations in portions of the Lake Sturgeon’s native range. Scutes, plates, and fragmentary material were definitively identified on the basis of appearance and thickness. Reference to recent comparative material facilitated some differentiation of a few fragments given robustness and thickness differences as well as sheer size to distinguish between Acipenser and Scaphirhynchus material. In addition to the Bell Cave specimens, reports of the presence of Acipenseriformes remains have come from Dust Cave in Alabama (Walker, 1998), and more specifically, Lake Sturgeon, from Cheek Bend Cave in Tennessee (Dickinson, 1986). Interestingly, Bell Cave is the only locality within our study area to contain late Pleistocene remains of the Shovelnose Sturgeon (see Table 1).

Material

Teeth, scales, dentary, vertebrae, skull fragments.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Dentary, teeth.

Occurrences

Zones 1–2, 3.

Remarks

Four Lepisosteidae taxa are recognized as native to the Tennessee River drainage. Recent mainstem surveys have confirmed the presence of these taxa: Longnose Gar (Lepisosteus osseus), Spotted Gar (L. oculatus), and Shortnose Gar (L. platostomus), once considered extirpated from this section of the basin (Boschung & Mayden, 2004). However, modern surveys have not collected any Alligator Gar (Atractosteus spatula). Together, the three Lepisosteus taxa comprise 1,809 recorded individuals, with Spotted Gar representing the most common. In the Bell Cave samples, their characteristic ganoid scales coated with ganoin enamel were the most numerous elements identified from Gar taxa. The presence of fine needle-like striated teeth of varying position rooted with a circular base in a shallow socket were also used to delineate Lepisosteidae. Only one dentary element associated with the recovered material was diagnostic beyond the generic level. This particularly long and narrow dentary bone with teeth in situ strongly suggests the presence of Longnose Gar in the samples, while conversely, the presence of more compressed dentary bones with teeth in situ likely indicate the presence of Spotted and/or Shortnose Gar. Several extremely large Lepisosteidae fragments, including comparatively thick scales, may also indicate the presence of Alligator Gar in the Bell Cave samples. However, no Alligator Gar are known to have ever been collected from the mainstem Tennessee and there is not enough intact diagnostic material to confirm this taxon. In addition to the Bell Cave specimens, remains identified as Lepisosteus sp. have been identified from Baker Bluff (Guilday et al., 1978) and Cheek Bend (Dickinson, 1986) caves in Tennessee (see Table 1).

Material

Dentary, premaxillary.

Occurrences

Zone 3.

Remarks

The American Eel is an uncommon native catadromous taxon present in large rivers throughout the modern Mississippi River drainage, including the Tennessee River. Specific to the mainstem, only ten individuals were collected in recent TVA surveys that were composed of 687,334 total individuals. The American Eel depends on long migration runs between spawning areas in the north Atlantic and streams throughout the Mississippi River drainage to fulfill its life cycle, this highly migratory attribute has contributed to its decline in recent years as a result of impoundments and river fragmentation (Etnier & Starnes, 1993; Boschung & Mayden, 2004). Definitive remains in the Bell Cave samples were initially flagged as unique considering the elongated structure of the dentary and premaxillary bones while identification was confirmed through corresponding tooth socket shape and offset arrangement. None of the samples yielded any evidence of the tiny recurved teeth from these socket positions. No additional Anguilliformes have been reported from any other known late Pleistocene locality within our study area (see Table 1).

Material

Teeth.

Occurrences

Zones 3.

Material

Teeth, vomer, palatine, angular, dentary.

Occurrences

Zones 1–2, 3, 4.

Material

Teeth, vertebrae, angular, palatine, vomer, dentary.

Occurrences

Zones 1–2, 3, 4, Mixed.

Remarks

The presence of these remains indicate a range extension into Alabama for two of these taxa. Boschung & Mayden (2004) indicate that only two species of Esocidae are native to the Tennessee River drainage, the Chain Pickerel (Esox niger) and the Redfin Pickeral (E. americanus), while Etnier & Starnes (1993) add the Muskellunge to this list. Recent TVA fisheries surveys of the mainstem did result in the collection of Chain Pickerel, Redfin Pickerel, as well as Muskellunge. Interestingly, the presence of Northern Pike in Bell Cave samples coupled with the absence in modern fisheries surveys or listed as native in species accounts likely indicates a Pleistocene displacement event. Page & Burr (2011) indicate that the Northern Pike’s range encompasses only the northern portion of the Mississippi River, Great Lakes, Arctic, and Pacific basins while Muskellunge occupy a native range (irrespective of introduced populations) that also encompasses the northern portion of the Mississippi River Basin and the southern portion of the Great Lakes Basin.

Several teeth, vertebrae, skull fragments, and a single dentary were definitively identified on the basis of meristic and morphometric appearance and size. Crossman & Harrington (1970) point out that the teeth of Northern Pike exhibit compressed lateral edges on the larger teeth, a feature absent in Muskellunge. However, in this same publication, it is mentioned that these compressed edges are not always present, thus, the tentative assignment of the majority of the unassociated Bell Cave Esox tooth samples. The large size of the teeth were used to rule out either the ‘native’ Redfin or Chain Pickerel. One example of a large, intact palatine with tooth socket combinations was also used to definitively identify the presence of Muskellunge. When present, neural arches were also used to distinguish between Northern Pike and Muskellunge vertebrae. A near complete Muskellunge dentary was identified on the basis of submandibular pore counts (6–9 E. masquinongy, ∼4 E. lucius; Crossman & Harrington, 1970). In addition to the Bell Cave specimens, remains of Esox sp. have been reported from Baker Bluff Cave in Tennessee (Guilday, Hamilton & Parmalee, 1975), and remains of Muskellunge have been reported from Cheek Bend Cave in Tennessee (Dickinson, 1986). Bell Cave is the only locality within our study area to produce identifiable remains of the Northern Pike (see Table 1).

Material

Pharyngeal, pharyngeal teeth, vertebrae.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Pharyngeal.

Occurrences

Zones 1–2.

Material

Pharyngeal.

Occurrences

Zones 1–2.

Material

Pharyngeal.

Occurrences

Zones 1–2.

Remarks

Etnier & Starnes (1993) and Boschung & Mayden (2004) recognize 91 species of native Tennessee River Basin Cyprinidae, while recent surveys of the mainstem indicate the presence of 21 of these native taxa. The Largescale Stoneroller (Campostoma oligolepis), and Central Stoneroller (C. anomalum), were identified in recent mainstem surveys in Alabama; however, only the Central Stoneroller is considered native to the basin by Etnier & Starnes (1993). The Largescale Stoneroller could be a recent introduction or represent the recently described native Bluefin Stoneroller (C. pauciradii). However, the Bluefin Stoneroller is typically only associated with the Georgia portion of the Tennessee River Drainage Basin, well outside the vicinity of Bell Cave (Page & Burr, 2011). Regardless, the contemporary presence of these two Stoneroller taxa potentially narrow the identification of the Pleistocene remains. Specific to the River Chub group, Nocomis remains provide confirmation of a native taxon that has probably been extirpated from the mainstem. Two Nocomis taxa, the Bluehead Chub (N. leptocephalus) and River Chub, are noted to occur in the basin, although the former is not always noted as present in the drainage basin (Boschung & Mayden, 2004). Overall, mainstem extirpation or population changes are likely a function of habitat alterations in the past century as well as the potential for gear bias to under-represent this and other Cyprinidae groups. Cyprinidae material was identified using pharyngeal arch shape and tooth formula (Eastman & Underhill, 1973; Boschung & Mayden, 2004). Comparative material of several Minnow genera, including Erimystax and Cyprinella, closely matched some of the noted tooth configurations; however, too much degradation prevented anything more than tentative identification. The overall lack of Cyprinidae remains could represent a change in assemblage, preservation bias, or collection and processing bias.

Late Pleistocene specimens identified as Cyprinidae have also been identified from Appalachian Caverns (this study) and Baker Bluff (Guilday et al., 1978) in Tennessee, and Saltville in Virginia (Dickinson, 1986). In addition to the specimen identified from Bell Cave, Stoneroller remains identified as Campostoma sp. have also been reported from Cheek Bend Cave in Tennessee (Dickinson, 1986). Remains belonging to ‘Chubs’ Nocomis sp. have also been identified from Guy Wilson Cave in Tennessee (this study), and although remains belonging to the River Chub have been identified from both Bell Cave and Cheek Bend Cave, the latter locality also produced remains belonging to Hornyhead Chub N. biguttatus (Dickinson, 1986). Other Cyprinidae members not identified from the Bell Cave deposits include Common Shiner Luxilus cornutus from Saltville in Virginia (Dickinson, 1986); Bluntnose Minnow Pimephales notatus from Saltville, and Creek Chub Semotilus atromaculatus from Cheek Bend Cave (Dickinson, 1986).

Material

Epihyal, hypohyal, basipterygium, hyomandibular, basihyal, scapula, pharyngeal teeth, supracleithrum, fin ray, quadrate, cleithrum, vertebrae, operculum, maxilla, orbitosphenoid, parietal, basioccipital, pharyngeal, angular, ethmoid.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Dentary.

Occurrences

Zones 1–2, 3.

Material

Operculum, hyomandibular.

Occurrences

Zone 3.

Material

Angular, operculum, rostral, epihyal, angular, parasphenoid, pharyngeal.

Occurrences

Zones 1–2, 3.

Material

Pharyngeal teeth, rostral, cleithrum, epihyal, maxillary, dentary, neural process, basioccipital, parietal, orbitosphenoid, scapula, urohyal.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Premaxillary, palatine.

Occurrences

Zone 3.

Material

Dentary, maxillary.

Occurrences

Zones 1–2, 3.

Remarks

There are 22 species of North American Suckers native to the Tennessee River Basin (Etnier & Starnes, 1993; Boschung & Mayden, 2004), of which 14 are still collected in mainstem TVA surveys. The Bell Cave samples provide generic level evidence for all but the Spotted Sucker (Minytrema melanops), Northern Hogsucker (Hypentelium nigricans), and the Blue Sucker (Cycleptus elongatus). Genera represented include Moxostoma, Ictiobus, and Catostomus. All definitive species identified still occur in the mainstem, with one notable exception. Represented by only two cranial fragments, the remains of the extinct Harelip Sucker, are perhaps the most significant among the Bell Cave Catostomidae remains. Little is known about the Harelip Sucker’s niche, distribution, or eventual decline, except that the species is thought to have preferred medium to large warm-water lotic systems with low rates of sedimentation and that population declines occurred concurrently with these habitats becoming increasingly degraded as a result of agricultural sedimentation (Jenkins & Burkhead, 1994). The last living Harelip Sucker specimens were collected in 1893 in the Auglaize and Blanchard Rivers in Ohio, where previously collected individuals were distributed across portions of the Ohio River Basin and Great Lakes Basin (Jenkins & Burkhead, 1994). Manzano & Dickinson (1991) documented the first archaeological occurrences of the Harelip Sucker from several sites across Tennessee. Specimens from additional archaeological sites along the Tennessee River system in Tennessee and Alabama have been identified in recent years (WC Dickinson, 2015, unpublished data). The most diagnostic element found in the Bell Cave samples was the premaxillary bone, which can be described as having an obtuse angle approximately 140 degrees compared with more 90 degree premaxillary angles found in other Catostomidae (Manzano & Dickinson, 1991). The Harelip Sucker specimens identified from Bell Cave represent the first known Pleistocene occurrence for this taxon as well as the stratigraphically oldest occurrence recorded.

Other Catostomidae identifications were based on the morphology of dentaries, pharyngeal teeth, and operculi. Material from unidentified Redhorse is likely from one of the other identified Redhorse taxa. However, at least one example (RMM 5265) of a damaged or weathered cleithrum may be from a Silver Redhorse M. anisurum. Additionally, numerous damaged Catostomidae fragments may also be from one of three Carpsucker Carpoides taxa native to, and still collected in, the mainstem.

Within our study area, remains identified as belonging to Catostomidae have been identified from Dust Cave in Alabama (Walker, 1998), Appalachian Caverns (this study) and Guy Wilson Cave (this study) in Tennessee, and Saltville in Virginia (McDonald & Bartlett, 1983; Dickinson, 1986). The White Sucker has been reported from three localities aside from Bell Cave, including Baker Bluff (Guilday et al., 1978) and Cheek Bend caves in Tennessee (Dickinson, 1986), and Saltville in Virginia (Dickinson, 1986). Members of the Redhorse genus Moxostoma are widely represented within sites in our study area with specimens identified as Moxostoma sp. reported from Dust Cave, Appalachian Caverns, Baker Bluff Cave. Other Redhorse species found in Bell Cave have also been reported at other localities including River Redhorse from Appalachian Caverns, Baker Bluff Cave, and Cheek Bend Cave as well as Shorthead Redhorse from Cheek Bend Cave. Cheek Bend Cave has also produced three species of Redhorse not identified from Bell Cave, Silver Redhorse, Black Redhorse M. dusquesnei (also reported from Appalachian Caverns), and Golden Redhorse M. erythrurum (also reported from Dust Cave). Appalachian Caverns, Baker Bluff, Cheek Bend, and Guy Wilson caves also report the presence of Northern Hogsucker Hypentellum nigricans, another taxon not identified from Bell Cave. Of the localities within our study area, remains belonging to Buffalo are only known from Bell Cave (this study; see Table 1).

Material

Pectoral spine, cleithrum, maxillary, dorsal spine, quadrate, symplectic, ceratohyal, palatine, premaxillary, coracoid, lateral ethmoid, posttemporal.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Dentary, spine.

Occurrences

Zones 1–2.

Material

Dentary, pectoral spine, operculum, cleithrum, angular, premaxillary, quadrate, fin ray, hyomandibular, epihyal, ethmoid.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Pectoral spine, premaxilla, dentary, posttemporal, angular, hyomandibular, supraethmoid, cleithrum, maxillary.

Occurrences

Zones 1–2, 3, 4.

Material

Pectoral spine, dentary.

Occurrences

Zones 1–2, 3, Mixed.

Material

Pectoral spine, dentary, premaxillary.

Occurrences

Zones 1–2, 3, Mixed.

Material

Pectoral spine.

Occurrences

Zones 1–2, 4, Mixed.

Material

Pectoral spine.

Occurrences

Zone 3.

Material

Pectoral spine.

Occurrences

Zones 1/2.

Material

Pectoral spine.

Occurrences

Zones 1/2.

Material

Pectoral spine, articular, premaxilla, quadrate, ethmoid.

Occurrences

Zones 1–2, 3.

Remarks

There are 25 Ictaluridae considered native to the Tennessee River Basin (Etnier & Starnes, 1993; Boschung & Mayden, 2004), of which 6 were recently noted in TVA surveys of the mainstem. Overall, Ictaluridae are perhaps the most well-represented group in the Bell Cave assemblage. This is likely due to the preservation and identification bias of pectoral spines rather than indicating that this Pleistocene river system was actually dominated by Catfish. However, the comparatively large number of positively identified spine and cranial elements does convey their importance to the assemblage and to understanding this fish assemblage.

Ictaluridae elements were identified using comparative material and the descriptions and illustrations of individual elements from Taylor (1969), Lundberg (1982), Dickinson (1986), Thomas (2002) and Wright (2012). In particular, spines were identified by comparison of overall curvature, anterior and posterior serrae patterns, and insertion condyle shape relative to reference material. Of particular note regarding many of the described elements is the potential bias in identifications stemming from the effects of weathering and interspecific similarities among osteological patterns, such as similarities in spine serrae. These caveats likely suggest that our list is conservative compared to the true assemblage.

The most striking difference between modern mainstem diversity and Bell Cave diversity is in the Madtom group, which is entirely absent in recent standard mainstem fisheries surveys. The presence of the Madtom genus Noturus in Bell Cave collections provides evidence of a very different mainstem Pleistocene habitat compared with more recent habitats (currently modified into a chain of impoundments) as well as potential evidence for range expansions for several of these taxa. Madtom spines are often noted by their ornate serrae and curvature patterns coupled with their relatively small size. However, Stonecat Madtoms were distinguished primarily via a general lack of these ornamentations, including any anterior or posterior serrae, but with several posterior notches on the posterior distal end of the pectoral spine. Stonecat Madtom remains also included characteristic premaxillaries with elongated processes extending from functional tooth pads (Taylor, 1969; Lundberg, 1982; Dickinson, 1986). Conversely, Rabida-type spines were distinguished as having the typical elaborate or highly-ornamented posterior and anterior serrae patterns. Mountain Madtom was differentiated from Northern Madtom on the basis of less prominent anterior serrae (Thomas, 2002). Spines linked to the Northern Madtom are particularly unique as this taxon currently occurs in the Lake Erie and Ohio River drainage basins, well north of the Tennessee River Basin (Page & Burr, 2011). One caveat, however, is the historical lumping of Northern Madtom with the newly described Piebald Madtom (N. gladiatorThomas & Burr, 2004), which could have resulted in a bias of comparative collections and past descriptions. Checkered Madtoms were distinguished from other Madtoms (including the similar Brindled Madtom N. miurus) on the basis of their more prominent, recurved, and unique spacing of anterior serrae (Taylor, 1969). Interestingly, Checkered Madtom are considered an Ozark endemic today, yet given the Miocene divergence of this taxon (Hardman & Hardman, 2008) it seems that this contemporary pattern is best explained by the taxon’s extirpation from much of its former range. Dickinson’s (1986) identification of Checkered Madtom from Holocene deposits (not reported here) in Cheek Bend Cave support this hypothesis and indicate that the extirpation of this taxa occurred sometime after the Pleistocene.

Larger bodied taxa such as Channel Catfish, Flathead Catfish, and Bullhead Catfish were often more readily identifiable than Madtom due to their size and general lack of ornamentation compared to the smaller taxa. Channel Catfish were recognized from a variety of spine size classes using a general pattern of decreasing anterior serration size in more medial positions that tended to decrease in prominence in larger specimens (Lundberg, 1982). Bullhead Catfish were distinguished using serration groove features as well as noting similar-sized serrations along their posterior edge coupled with slight notches along the spine tip. Flathead Catfish were recognized primarily from spines exhibiting similar-sized and shaped serrae along both posterior and anterior spine edges (Lundberg, 1982).

Remains belonging to members of the Ictaluridae are diverse in Bell Cave, but have shown to be uncommon at other localities within our study area. Of the few reports, specimens identified as belonging to Ictaluridae have been recovered from site ACb-3 in Alabama (this study) and Cheek Bend Cave in Tennessee (Dickinson, 1986). Specimens identified as Ictalurus sp. have been identified at Baker Bluff (Guilday et al., 1978) and Beartown caves (Guilday, Hamilton & Parmalee, 1975) in Tennessee, and Channel Catfish has been identified from Bell Cave, Baker Bluff and Cheek Bend caves (Dickinson, 1986). Remains belonging to the Madtom group have been reported from sites ACb-3 and Baker Bluff, while remains of the Stonecat Madtom have been reported from only two localities, Bell Cave and Cheek Bend Cave. Bell Cave has also produced remains belonging to four additional Madtom species that have not been identified at any other locality within our study area, Mountain Madtom, Northern Madtom, Elegant Madtom, and Checkered Madtom. Bell Cave is also the only locality within our study area to produce late Pleistocene remains of Bullhead and Flathead Catfish (see Table 1).

Material

Premaxilla, dentary.

Occurrences

Zones 1–2, 3.

Remarks

Two Cottidae taxa are recognized natives of the Tennessee River Basin (Etnier & Starnes, 1993; Boschung & Mayden, 2004), the Mottled Sculpin (Cottus bairdi) and Banded Sculpin (Cottus carolinae). Both of these taxa are still recovered in mainstem fisheries surveys, albeit in relatively low numbers. The Sculpin fragments identified from Bell Cave lots were also some of the most uncommon of the taxa identified. Sculpin fragments were described based on characters outlined in Dickinson (1986) and Broughton (2000) which included recognition of lateral fossa, tooth arrangement, and generally narrow appearance along premaxilla and dentary pieces. Although the material is not definitive to the species level, we suggest these are the remains of the Banded Sculpin due to distribution and habitat preferences. Locating preopercles in the Bell Cave samples could potentially provide definitive identification to the species level (Broughton, 2000); unfortunately none were recovered.

Relative to the assemblage, the low abundances of Sculpin material might suggest something about the Pleistocene fish assemblage or it could reflect some sort of preservation bias. Today, Sculpin numbers are reduced in the mainstem as a function of habitat alterations occurring over the past century. Specifically, the reduction of riffle habitats in the modern Tennessee River mainstem has had implications for populations of these Sculpin taxa which key on gravel and rubble riffles (Boschung & Mayden, 2004) as an important part of their respective niches. In addition to the Cottus sp. specimens identified from Bell Cave, this taxon has also been identified from Appalachian Caverns in Tennessee (this study). Both Carolina Sculpin and Mottled Sculpin have been reported from Cheek Bend Cave in Tennessee (Dickinson, 1986; see Table 1).

Material

Dentary, premaxilla.

Occurrences

Zones 1–2, 3.

Material

Dentary, premaxillary, maxillary, quadrate, angular, supracleithrum, infrapharyngeal, gill raker, operculum, vomer, hypohyal.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Dentary, premaxilla, pharyngeal, operculum.

Occurrences

Zones 1–2, 3, 4, Mixed.

Material

Dentary.

Occurrences

Zones 3.

Remarks

Watershed records as well as recent mainstem survey efforts include all four (three unequivocal) of these taxa as common native members of the modern Tennessee River fish assemblage. Interestingly, the most common Centrarchidae in recent mainstem collections are Bluegill (Lepomis macrochirus) and Largemouth Bass (Micropterus salmoides), neither of which were definitely identified from Bell Cave samples. The absence of Largemouth Bass and Bluegill is not particularly surprising given their habitat affinities for the more recent impoundment conditions of the Tennessee River mainstem that were nonexistent in the Pleistocene, coupled with decades of introduction efforts throughout the watershed. However, the absence of any Sunfish Lepomis taxa is surprising and should be further explored in future studies as several of these (e.g., see Table 1 list for Dust Bend and Cheek Bend Caves) are known from the basin during this time period and would be expected to have occurred in the Bell Cave assemblage. Diagnostic characters used to identify Centrarchidae taxa were primarily dentary and premaxilla shape differences. Specifically, Black Bass taxa tend to exhibit more uniform tooth size and arrangement patterns on all surfaces compared with either Rock Bass or Sunfish taxa. Diagnostic differences in the ‘shelf’ that runs along the dentary as well as premaxillary were also used to supplement tooth arrangement pattern identifications to delineate Black Bass species where possible (Dickinson, 1986).

Relative to other Pleistocene locales, specimens identified as unknown Centrarchidae have been reported from Saltville in Virginia (Dickinson, 1986). More specifically, Rock Bass appear extremely widespread across the Tennessee River Basin in the late Pleistocene at this taxa has been reported from Bell Cave (this study); Appalachian Caverns (this study), Baker Bluff (Guilday et al., 1978), Cheek Bend (Dickinson, 1986), and Guy Wilson (this study) caves in Tennessee, and Saltville in Virginia (Dickinson, 1986). Remains belonging to unidentified Black Bass have been reported from site ACb-3 in Alabama and Baker Bluff, while in addition to the remains from Bell Cave, specimens identified as Smallmouth Bass have been reported from Baker Bluff and Spotted Bass from site ACb-3 in Alabama (this study). A specimen identified as an unidentified Sunfish taxa has been reported from Dust Cave in Alabama while Green Sunfish have specifically been reported from Cheek Bend Cave. Remains belonging to this genus have yet to be identified from Bell Cave.

Material

Teeth, dentary, premaxillary, angular, quadrate, maxilla, preoperculum, ceratohyal

Occurrences

Zones 1–2, 3, Mixed

Material

Teeth, dentary, articular, angular, premaxilla, epihyal, vomer, quadrate, palatine.

Occurrences

Zones 1–2, 3, Mixed.

Material

Preoperculum, premaxillary

Occurrences

Zones 1–2.

Remarks

Both of these taxa are native to the Tennessee River drainage and have been surveyed in recent mainstem samples. Relative to modern mainstem TVA collections, the Sauger is a more numerically dominant member of the assemblage than the Walleye. If this trend has historical continuity it could mean that the majority of the unidentified remains should be assigned to Sauger (Griffith & Smith, 1834). The primary identification metric used to separate out the Sander genus from other remains were the characteristic and often diagnostic canine teeth, both loose and in situ within their robust and deep dentary sockets. Additionally, the singular row-arrangement of teeth compared with cluster arrangements found in other Perciformes (e.g., Centrarchidae) was used as a diagnostic feature. Differentiation between the two species was often a function of size inference (e.g., Walleye attain larger sizes than Sauger; Page & Burr, 2011) using measurements noted in comparative material. Of particular interest in these samples is the lack of material that could be assigned to other members of the family Percidae. For example, the Tennessee River drainage is currently home to 80+ other Percid species in the genera: Etheostoma, Nothonotus, Percina, and Perca (Etnier & Starnes, 1993; Boschung & Mayden, 2004; Page & Burr, 2011), several of which (12 taxa representing all major genera) are still collected in mainstem surveys (although Etheostoma/Nothonotus are likely underrepresented due to gear bias in recent TVA efforts). Specific to the Pleistocene samples, the lack of any Darter remains is likely a function of preservation, size, collection, recovery and processing biases inherent in the Bell Cave samples.

Three members of the Percidae family, Greenside Darter E. blennioides, Yellow Perch Perca flacescens, and Logperch Percina caprodes have been identified among the remains at Cheek Bend Cave in Tennessee (Dickinson, 1986). Remains of these three taxa have not been identified at Bell Cave nor have they been reported from any other locality within our study area. Remains identified as belonging to Sander sp. have been reported from Dust Cave in Alabama (Walker, 1998) and Baker Bluff Cave (Guilday et al., 1978) in Tennessee, while in addition to the Bell Cave specimens, Sauger has been reported from Baker Bluff and Cheek Bend caves, and Walleye from Cheek Bend Cave (see Table 1).

Material

Pharyngeal, pharyngeal teeth, vertebrae, dentary, angular, spine, angular, parasphenoid, hypohyal, quadrate, unidentified skull fragments.

Occurrences

Zones 1–2, 3, 4, Mixed.

Remarks

The Freshwater Drum is the only Sciaenidae taxon found in freshwaters of North America and is extremely common in medium to large rivers throughout the Mississippi River drainage, including the Tennessee River. Recent TVA fisheries sampling efforts indicate that Freshwater Drum are one of the most abundant species (16/80) in the assemblage and likely higher in biomass than the majority of other taxa. This commentary on contemporary abundance appears to be consistent with the assemblages captured in the Bell Cave Pleistocene deposits as Drum samples were very well represented across all zones. Freshwater Drum remains, pharyngeal bones and teeth in particular, were identified from a variety of size classes and comprised the most prevalent materials in the Bell Cave samples. The pharyngeal bone was identified through robust thickness, triangular shape, and rounded tooth sockets of increasing circumference proximal to pharyngeal edges. Corresponding pharyngeal teeth are molariform in shape with rounded edges and a dome shaped crushing functional surface (i.e., bead or pearl shaped). Freshwater Drum pharyngeal bones and teeth are well-known in the Pleistocene fish literature; however, this represents the first associated museum records with collection information (but see Parmalee, 1992) for Alabama. In addition to the remains identified at Bell Cave, late Pleistocene remains of Freshwater Drum have been identified from site ACb-3 (this study) and Dust Cave (Walker, 1998) in Alabama, and Baker Bluff (Guilday et al., 1978) and Cheek Bend (Dickinson, 1986) caves in Tennessee (see Table 1).