Geology of the Trout Rock caves (Hamilton Cave, Trout Cave, New Trout Cave) in Pendleton County, West Virginia (USA), and implications regarding the origin of maze caves Open Access

The Trout Rock caves are located in the John Guilday Caves Nature Preserve (JGCNP), which is situated in a hill named Cave Knob overlooking the South Branch of the Potomac River, ~7.2 km or 4.5 mi southwest of the town of Franklin in Pendleton County, West Virginia, USA (Fig. 1). The JGCNP is owned by the National Speleological Society (NSS) and was established in 1983 in order to protect the caves and their resources and to ensure access for cavers. The property contains three relatively large caves: Hamilton Cave, Trout Cave, and New Trout Cave. The property also contains six small caves: NSS Flood Cave, Film Can Cave, Trammelton Cave, Cathy’s Crack, Boulder Crawl Cave, and Spider Cave (West, 2001a,b,c,d,e). In addition, the property contains two very small “For the Record Only” (FRO) caves named Trout Rock FRO #1 and Trout Rock FRO #2 (Ganter, 1984). Most caves on the property are open to general visitation by cavers. Trout Cave, however, was gated in 2008, and for many years it was closed throughout the year to reduce disturbance to the federally endangered Indiana bats that roost within the cave. In 2019, however, Trout Cave was reopened for limited visitation under strict guidelines during a few weeks of April and May.

The caves of the JGCNP are located at various elevations above the South Branch of the Potomac River (Fig. 2). In this area, the river is at an elevation of ~546 m or 1791 ft above sea level. According to Dasher (2001), the Hamilton Cave entrance is located at an elevation of 640 m (2099 ft), the Trout Cave entrance is located at an elevation of 622 m (2040 ft), and the New Trout Cave entrance is located at an elevation of 585 m (1919 ft). The entrances to Film Can Cave, Trammelton Cave, and Cathy’s Crack are located between Hamilton Cave and Trout Cave at elevations of 617 m (2024 ft), 634 m (2080 ft), and 622 m (2040 ft), respectively. The entrances to Trout Rock FRO#1 and Trout Rock FRO#2 are located between Trout Cave and New Trout Cave at elevations of 579 m (1899 ft) and 582 m (1909 ft), respectively. The entrances to Boulder Crawl Cave and Spider Cave are located to the east of New Trout Cave at elevations of 597 m (1959 ft) and 610 m (2001 ft), respectively. The entrance to NSS Flood Cave is located at an elevation of 561 m (1840 ft).

The elevations of the entrances to some of the smaller caves cluster around the elevations of the entrances to the larger caves. For example, the entrance of Cathy’s Crack is at approximately the same elevation as the entrance of Trout Cave. Likewise, the entrances of Trout Rock FRO#1 and Trout Rock FRO#2 are at approximately the same elevation as the entrance of New Trout Cave. This clustering suggests that these caves with entrances at similar elevations may have developed during approximately the same time, when the river was at a certain elevation (base level) for a long enough duration for large caves to develop.

The Trout Rock caves are located within limestone of the Silurian–Devonian Helderberg Group, which extends from New York to Virginia (Fig. 3). At Cave Knob, the Helderberg Group is a >90-m- (295-ft-) thick unit of limestone, argillaceous limestone, cherty limestone, and chert that is overlain by a >7-m- (>23-ft-) thick unit of sandstone that is mapped as the Devonian Oriskany Sandstone. At this location, the Helderberg Group is divided from base to top into the following four formations: (1) the Silurian–Devonian Keyser Limestone; (2) the Devonian New Creek Limestone (previously called the “Coeymans Limestone” in some early publications, but this name has been discontinued in central Pennsylvania and further south, and the name “New Creek Limestone” is used instead; Bowen, 1967); (3) the Devonian Corriganville Limestone (previously called the “New Scotland Limestone” in some early publications, but this name has been discontinued in central Pennsylvania and further south, and the name “Corriganville Limestone” is used instead; Head, 1972); and (4) the Devonian Shriver Chert. More detailed descriptions of the Helderberg Group in eastern West Virginia may be found in Dorobek and Read (1986), Báez Rodríguez (2005), and Ryder et al. (2008).

Most of the strata exposed along the paths to New Trout Cave, Trout Cave, and Hamilton Cave are within the Keyser Limestone, and consist of (1) an ~2-m- (7-ft-) thick unit of limestone and argillaceous limestone displaying laminations, cross-bedding, and small thrust faults; overlain by (2) an ~8-m- (26-ft-) thick unit of limestone consisting of numerous relatively homogenous 0.1–0.3-m- (0.3–1.0-ft-) thick beds of wackestone-packstone. At Cave Knob, the Keyser Limestone is overlain by an ~9-m- (30-ft-) thick unit of gray crinoid-bearing limestone that is mapped as the New Creek Limestone. The New Creek Limestone is a packstone to grainstone (according to the limestone classification of Dunham, 1962) that contains abundant fossil debris. Crinoid columnals (“stems”) are particularly common, and horn corals and stromatoporoid fragments are also present. Cross-bedding is visible in a few places. Other features of the New Creek Limestone include stylolites, which are irregular suture-like contacts that form via dissolution under conditions of deep burial and high pressure. The New Creek Limestone is overlain by an ~7-m- (23-ft-) thick unit of gray limestone (packstone to grainstone) that is mapped as the Corriganville Limestone. The Corriganville Limestone contains lenses and nodules of light-gray chert, as well as abundant brachiopod shells and occasional horn corals and gastropods. Many of the brachiopod fossils are Spirifer macropleurus, which is characteristic of the Corriganville Limestone (Butts, 1940). The Corriganville Limestone is overlain by an ~54-m- (177-ft-) thick unit of dark-gray to black chert and cherty limestone (cherty carbonate mudstone, according to the limestone classification of Dunham, 1962) that is mapped as the Shriver Chert. Beds range from nodular to irregular with laminations. The unit is not exposed very well in outcrop, but is well exposed in the Square Room of Trout Cave (Fig. 1). The Shriver Chert is capped by an unconformity, above which lies an >7-m- (>23-ft-) thick unit of yellow to red-brown sandstone that is mapped as the Oriskany Sandstone (Butts, 1940; Woodward, 1943; Dennison, 1985). The sandstone is a quartzarenite composed of coarse to very fine quartz sand. Fossils include common molds of the brachiopod Spirifer arenosus. At Cave Knob, the Oriskany Sandstone is exposed only at the top of the hill above the caves.

The major structural feature of Cave Knob is a northeast-trending anticline named Cave Knob anticline (Fig. 1). As with most other anticlines in the Appalachian region, Cave Knob anticline is very steep on its west flank and less steep on its east flank. The axis of the anticline transects Hamilton Cave, and the strata dip at very low angles near the anticline axis. Some secondary anticlines, which also trend northeast, are present on the southeast limb of Cave Knob anticline. One secondary anticline is visible in the Triangle Room of Trout Cave (Figs. 1 and 4) and another is visible at Trout Rock FRO #2 (Fig. 1).

Joints are another prominent structural feature of Cave Knob. Many of the cave passages follow primary joint sets that trend N50E or N40E, and some passages follow secondary joint sets that trend N40W. These joint sets are particularly well exposed in the maze (front) portion of Hamilton Cave, where the cave passages have developed along vertical or near-vertical joints (Fig. 1).

Both Hamilton Cave and Trout Cave are infected by a terrestrial saprophyte (fungus) named Pseudogymnoascus destructans that causes white-nose syndrome (WNS) in bats. At present, the disease is not known to affect humans. In order to minimize the risk of spreading WNS to other karst regions, field-trip participants are asked to remove sediment/dirt from their clothes and gear upon exiting a cave. After this field trip, participants are advised to wash their clothes and gear in hot water and to avoid taking the clothing and gear used on this trip into other karst regions. Additional details about WNS decontamination protocols are available at www.whitenosesyndrome.org/topics/decontamination.

The WNS-causing fungus came to North America from Europe; it first appeared in North America in New York during 2006, and since then it has been spreading across North America and killing millions of bats (Gargas et al., 2009; Lorch et al., 2011; Swezey and Garrity, 2011a, 2011b; Minnis and Lindner, 2013; Reeder and Moore, 2013; Forsythe et al., 2018; Verant et al., 2018). White-nose syndrome was first detected in Pendleton County during January 2009 at both Hamilton Cave and Trout Cave (U.S. Geological Survey National Wildlife Health Center, 2009; Anonymous, 2010a, 2010b; Dasher, 2009; Hoke, 2009a, 2009b; Stihler, 2009; Turner et al., 2011). Interestingly, WNS has not been reported from New Trout Cave (as of December 2019), despite proximity to Hamilton Cave and Trout Cave. The absence of WNS in New Trout Cave might be attributed to differences in cave meteorology and/or cave soil chemistry. Reliable data are sparse, but the relative humidity is notably lower in New Trout Cave compared to Hamilton Cave and Trout Cave. Anecdotal field evidence suggests that the fungus prefers sites of high humidity (68–100%; Swezey and Garrity, 2011a, 2011b), although laboratory studies on this topic have not been published.

Hamilton Cave, Trout Cave, and New Trout Cave were inhabited by several species of bats prior to the arrival of WNS (Fowler, 1941, 1942, 1943; Garton et al., 1993; Grady, 1993, 1994; Hoke, 1991, 2003). Hamilton Cave was inhabited by the tri-colored bat (Perimyotis subflavus) [formerly the eastern pipistrelle], the little brown bat (Myotis lucifugus), and the northern long-eared bat (Myotis septentrionalis). Trout Cave was inhabited by the big brown bat (Eptesicus fuscus), the tri-colored bat, the Indiana bat (Myotis sodalis), the little brown bat, the northern long-eared bat, and the Virginia big-eared bat (Corynorhinus townsendii virginianus). New Trout Cave was inhabited by the big brown bat, the tri-colored bat, the eastern small-footed bat (Myotis leibii), and the little brown bat. According to Hoke (2010, 2011), Owens (2012), and Owens and Hoke (2013), the bat population in Hamilton Cave plummeted from a pre-WNS average of 450 to only 5 in 2013, and the bat population in Trout Cave decreased from a pre-WNS average of 676 to only 135 in 2013. In contrast, the bats in New Trout Cave do not appear to be afflicted by WNS, and their annual population has fluctuated between mostly 30 and 50 (data from 1986 to 2013). Population counts during March 2017 reported 9 bats in Hamilton Cave, 90 bats in Trout Cave, and 30 bats in New Trout Cave (Owens, 2017). The bat species in West Virginia that are most affected by WNS are the little brown bat, Indiana bat, northern long-eared bat, and the tri-colored bat (Stihler, 2017).

The caves of the JGCNP contain a variety of other fauna, in addition to bats. In Hamilton Cave, Garton et al. (1993) identified the raccoon (Procyon lotor), Allegheny woodrat (Neotoma magister), slimy salamander (Plethodon glutinosus), and phoebe (cf. Sayornis phoebe). In Trout Cave, Loomis (1939) identified the Luray Caverns blind cave millipede (Zygonopus whitei), Holsinger et al. (1976) identified two species of millipede (Pseudotremia simulans and Trichopetalum whitei) and one species of spider (Islandiana speophila), and Fong et al. (2007) identified Hoffman’s springtail (Sinella hofmani) and the Luray Caverns blind cave millipede. In addition, Landolt et al. (1992) identified the slime molds Dictyostelium sphaerocephalum and D. rosarium. In New Trout Cave, Garton et al. (1993) identified the raccoon, and Fong et al. (2007) identified Hoffman’s springtail. In NSS Flood Cave, Garton et al. (1993) identified the Allegheny woodrat. In Film Can Cave, West (2001b, p. 142) identified a “cave rat that collects old film cans.” In Spider Cave, West (2001d) identified many large, black spiders.

The following three stops are associated with this field trip: (1) entrance to New Trout Cave; (2) entrance to Trout Cave; and (3) tour of Hamilton Cave. The coordinates of the cave entrances are not given in this field guide in order to protect the caves from possible vandalism. Nevertheless, the three stops of this field trip are located on the property of the JGCNP, which is owned by the NSS. The NSS requests that each caver on the property use a helmet with chin strap and carry at least three sources of light. In addition, caving groups should consist of at least three people, and there should be someone outside the caves who knows when to call for rescue.

As stated previously, the Trout Rock caves have developed within limestone of the Silurian–Devonian Helderberg Group (Tilton et al., 1927; Haas, 1961; Davies, 1965; Palmer, 1975; Dyas, 1977; Medville, 2000a, 2000c; Dasher, 2001). More precisely, most passages of Hamilton Cave, Trout Cave, and New Trout Cave are located along the contact between the New Creek Limestone and the overlying Corriganville Limestone (Swezey, 2014). Many other caves in Pendleton County have also developed along the New Creek–Corriganville contact, including the Sinnett-Thorn Mountain Cave System (Swezey et al., 2004a), Cave Mountain Cave (Swezey and Dulong, 2010), and Short Cave (Dove, 2012). This stratigraphic contact appears to have been a preferential zone of fluid movement and/or a preferential zone of soluble minerals throughout Pendleton County. As such, this contact may be a karst “inception horizon,” which is a feature thought to develop most commonly under phreatic conditions (e.g., Lowe and Gunn, 1997; Filipponi et al., 2009).

The New Creek Limestone and the overlying Corriganville Limestone are well exposed at the entrance to New Trout Cave (Fig. 5). New Trout Cave has been known since at least the time of the American Civil War of 1861–1865 (Faust, 1964; Davies, 1965). The first published mention of the name “New Trout Cave” in the scientific/caving literature, however, appears to have been in a newsletter of the Washington, D.C., chapter of the NSS (Emmons, 1951). The cave was subsequently described in the book Caverns of West Virginia published by the West Virginia Geological and Economic Survey (Davies, 1958). A modern (1993) map of the cave may be found in the book The Caves and Karst of Pendleton County published by the West Virginia Speleological Survey (Dasher, 2001).

The front part of New Trout Cave (near the entrance) consists of a long passage in strata that dip to the southeast at an angle of 35°. This passage is followed by a series of three rooms, beyond which is a complex and multilevel maze of passages that is ~183 m (600 ft) long. Beyond the maze, there is a long and low passage that trends northeast along a bedding plane that dips southeast at an angle of ~30°. This passage leads to additional northeast-trending passages. Beyond a location named Thrush Hall (Fig. 1), the northeast-trending passages trend slightly more toward the east than in other areas of the cave.

Most of New Trout Cave has developed along the contact between the New Creek Limestone and the overlying Corriganville Limestone (Davies, 1965), and most of the passages follow joints that trend N50E or N40E. The longer passages in the cave are approximately horizontal and extend northeast along strike.

Meteorological data have not been reported from New Trout Cave. Much of the cave, however, is persistently very dry and dusty, and the cave does not appear to receive appreciable moisture either from groundwater or from air circulation. Furthermore, the fact that white-nose syndrome has not been detected in New Trout Cave, whereas it was detected in nearby Trout Cave and Hamilton Cave during January 2009, suggests that New Trout Cave would be an excellent location to study the influences of cave air humidity on the distribution of WNS.

Sediments in New Trout Cave include sand and mud on the cave floor, and in some places this sediment is >2.2 m (7.2 ft) thick (Holman and Grady, 1987; Grady and Garton, 2000). In the front (southwest) part of the cave, large boulders are common (classified as “breakdown blocks” and “breakdown slabs,” using terminology of White and White, 1969). The northeast end of the cave “ends in a series of mud chokes” (Grady, 2001a, p. 253).

Gypsum crusts are present in New Trout Cave in the maze section ~213–305 m (~700–1000 ft) from the entrance. An analysis by Swezey et al. (2002) of two gypsum samples from the cave revealed sulfur isotope (δ³⁴S) values of –18.4 and –7.6 per mil (parts per thousand). The negative δ³⁴S values and their large range suggest that the sulfur in these gypsum samples was not derived from the simple dissolution and reprecipitation of marine evaporite sulfate within the stratigraphic section. Instead, the sulfate-sulfur was probably derived from the oxidation of diagenetic sulfide minerals (such as pyrite) and/or from organically bound sulfur in nearby strata.

During the American Civil War, sediment in New Trout Cave was mined for nitrate (Faust, 1964; Davies, 1965; Anonymous, 1970; Powers, 1981; Garton and Garton, 2001; Taylor, 2001). A black substance, generally attributed to smoke from the miners’ torches, is present in some places on the cave ceiling. Secretan (1952) identified iron, manganese, aluminum, calcium, and silica (and possibly titanium, zirconium, and chromium) in one sample of dark powder from the roof of New Trout Cave.

Excavations at various locations in New Trout Cave have produced numerous vertebrate fossils, including shrew, mole, bat, ground sloth, pika, rabbit, chipmunk, woodchuck, squirrel, rat, mouse, vole, lemming, muskrat, gopher, beaver, porcupine, wolf, coyote, fox, bear, raccoon, marten, fisher, skunk, weasel, mink, badger, peccary, caribou, deer, musk ox, and horse, as well as various birds, fish, hellbender, newt, salamander, frog, toad, turtle, lizard, skink, and snake (Grady and Garton, 1980, 1981, 1998, 2000, 2001; Grady, 1982, 1983, 1984a, 1984b, 1987b, 1991a, 2001b, 2001c; Holman and Grady, 1987; Mead and Grady, 1996; Holman, 1999; Grady and Baker, 2007). Fossils of invertebrate species include crayfish, terrestrial snails, aquatic snails, and bivalves (Bogan and Grady, 1991). At one site within the cave, Grady (1987b) described a 2.2-m- (7.2-ft-) thick interval of fossil-bearing sediment, overlying clay without fossils. He divided the fossil-bearing interval into eight units, which he designated from top to bottom as Level A through Level H. Each of these units was ~0.3 m (1.0 ft) thick except for Level H, which was 10 cm (3.9 in) thick. According to Grady and Garton (1980), Grady (1987b), and Holman and Grady (1987), bone collagen from Level A yielded an age of 17,060 ± 220 in radiocarbon (¹⁴C) years before present (B.P.) (Smithsonian Institution radiocarbon number 4102), bone collagen from Level B yielded an age of 28,250 ± 850 ¹⁴C yr B.P. (Smithsonian Institution radiocarbon number 4103), and bone collagen from Level C yielded an age of 29,400 ± 1,700 ¹⁴C yr B.P. (Smithsonian Institution radiocarbon number 4104). In addition, Grady (1987b) reported that a seed from Level C yielded an age of 23,270 ± 270 ¹⁴C yr B.P., and a seed from Level D yielded an age of 25,730 ± 380 ¹⁴C yr B.P. Lower beds of cave sediment proved to be >30,000 years old, and thus were too old to date by the radiocarbon methods that were available at that time (Grady, 1986b; Holman and Grady, 1987). A subsequent study by Semken et al. (2010) reported that one accelerator mass spectrometry (AMS) date from the mandible of a taiga vole in New Trout Cave yielded an age of 13,360 ± 60 ¹⁴C yr B.P. (University of Georgia, Center for Accelerator Mass Spectrometry [CAMS], AMS Number CAMS-20595). They also reported that two AMS dates from mandibles of the eastern woodrat in New Trout Cave yielded ages of 47,200 and 50,300 ¹⁴C yr B.P. (AMS Number CAMS-20593 and AMS Number CAMS-20592). These two older ages approach the limit for radiocarbon dating, and thus may be minimum ages for the samples. All of these ages are reported in radiocarbon (¹⁴C) yr B.P. with a 1 sigma standard deviation of the age uncertainty, using the Libby half-life of 5568 yr and with 0 B.P. being equivalent to AD 1950. On the basis of comparison with fossils from other sites, the faunal assemblage recovered from sediment in New Trout Cave is thought to be of Rancholabrean age, ~300–10 thousand years ago (ka). Grady and Baker (2007), however, indicated that the presence of the teeth of a small muskrat (Ondatra hiatidens) suggests a much older age of ca. 500 ka.

Along the northwesterly path from New Trout Cave to Trout Cave, the following two units may be seen within the Keyser Limestone (Figs. 6 and 7): (1) an ~2-m- (7-ft-) thick unit of limestone and argillaceous limestone displaying laminations, cross-bedding, and small thrust faults; overlain by (2) an ~8-m- (26-ft-) thick unit of limestone consisting of numerous relatively homogenous 0.1–0.3-m- (0.3–1.0-ft-) thick beds of wackestone-packstone. These two units have been repeated by faulting and thrusting. At a site where the path changes from relatively horizontal to going steeply up the hill, there is some folded and vertical bedding where a thrust fault may be located (Fig. 8). Above the zone of folded and vertical bedding, the New Creek Limestone and the overlying Corriganville Limestone are well exposed at the entrance to Trout Cave (Fig. 9), which is located ~100 m (328 ft) northwest of the New Trout Cave entrance.

Like New Trout Cave, Trout Cave has been known since at least the time of the American Civil War (Faust, 1964; Davies, 1965). The first published mention of the name “Trout Cave” in the scientific/caving literature, however, appears to have been in the book Pendleton County published by the West Virginia Geological Survey (Tilton et al., 1927). In a study of Devonian strata of West Virginia, Woodward (1943) noted that the roof of Trout Cave contains excellent specimens of the brachiopod Eospirifer. The cave was subsequently described in the book Caverns of West Virginia published by the West Virginia Geological and Economic Survey (Davies, 1958). A modern (1993) map of the cave may be found in the book The Caves and Karst of Pendleton County published by the West Virginia Speleological Survey (Dasher, 2001).

Trout Cave is characterized by many northeast-trending passages, although a small area of network maze passages is present in the westernmost portion of Trout Cave that is closest to Hamilton Cave. These maze passages lie along the axis of a secondary northeast-trending anticline on the east flank of the main anticline (Fig. 1). The “Triangle Room” also lies along the axis of this secondary anticline (Fig. 4). As with Hamilton Cave, most of Trout Cave has developed along the contact between the New Creek Limestone and the overlying Corriganville Limestone (Haas, 1961; Davies, 1965; Palmer, 1975; Medville, 2000b; Dasher, 2001). One location in Trout Cave (the “Square Room”), however, extends from the Corriganville Limestone into the overlying Shriver Chert (Swezey, 2003). The very back (northeast) end of Trout Cave is filled with large boulders of sandstone that are described as breakdown slabs (West, 2003). These sandstone boulders may be derived from the Oriskany Sandstone. The longer passages in Trout Cave are approximately horizontal and extend northeast along strike. Most of the passages in the cave have developed along joints that trend N50E, N40E, or N40W. In the front (southwest) part of the cave, strata dip 12° to the east.

Meteorological data from Trout Cave are sparse, consisting mostly of air temperature measurements. Davies (1965) reported air temperatures of ~12 °C (54 °F); Hoke (2001) reported air temperatures ranging from 6.7 to 8.9 °C (44–48.1 °F); and Swezey et al. (2004b) reported air temperatures ranging from 43 to 55 °F (6–13 °C) and relative humidity ranging from 81 to 92%. In addition, Dyroff (1977) published an account of a one-day study (10 April 1977), which found that during the morning when the outside air temperature was 0 °C (32 °F), air flowed into the cave along the bottom of the cave passage and air flowed out of the cave along the top of the cave passage. In that study, the mean air temperature was 9 °C (48 °F) 152.4 m (~500 ft) inside the cave, although the air temperature near the ceiling was several degrees warmer than the air temperature near the floor. During the afternoon when the outside air temperature was 13 °C (55 °F), the air currents reversed so that warm air flowed in along the top of the cave passage and cooler air flowed out along the bottom of the cave passage. Despite this reversal in flow directions, the mean air temperature inside the cave remained at 9 °C (48 °F).

Much of Trout Cave has a floor of dry to slightly damp mud, and near the entrance this sediment is at least 4 m (13 ft) thick (Guilday, 1967). In the front (southwest) part of the cave, large boulders are common (classified as breakdown blocks and breakdown slabs, using terminology of White and White, 1969). Northeast of the Square Room in an area known as Crisco Madness (Fig. 1), the passage floor is covered by a very sticky and wet mud. According to Landolt et al. (1992), two sediment samples collected from somewhere in the cave had the following chemical parameters: calcium (Ca) = >1500 ppm; magnesium (Mg) = 594 ppm; phosphorus (P) = 125 ppm; potassium (K) = 508 ppm; pH = 7.4. The significance of these chemical parameters is not certain because there are very few published studies on the composition and chemistry of cave sediments. Such data, however, may be important for determining controls on the distribution of white-nose syndrome.

During the American Civil War, sediment in Trout Cave was mined for nitrate (Faust, 1964; Davies, 1965; Anonymous, 1970; Powers, 1981; Garton and Garton, 2001; Taylor, 2001). Analysis of one sediment sample from Trout Cave revealed a nitrate concentration of ~2100 ppm, which is similar to the concentrations of nitrate in the sediment of other West Virginia caves that were mined for nitrate (Swezey et al., 2004b).

Excavations at various locations in Trout Cave have produced numerous vertebrate fossils, including shrew, mole, bat, pika, rabbit or hare, woodchuck, chipmunk, squirrel, rat, mouse, vole, lemming, muskrat, gopher, beaver, porcupine, bear, coyote, raccoon, bobcat, weasel, skunk, horse, peccary, deer, pigeon, hellbender, and catfish, as well as passenger pigeon, salamander, frog, toad, lizard, and snake (Guilday, 1967, 1971, 1979; Zakrzewski, 1975; Holman, 1982, 1999; Grady, 1981d, 1984b, 1985, 2001b, 2001c; Pfaff, 1990; Winkler and Grady, 1990; Grady and Garton, 1998, 2000, 2001). Fossils of invertebrate species include both terrestrial and aquatic snails (Guilday, 1967; Bogan and Grady, 1991). In addition, fossil Celtis (hackberry) seeds have also been recovered from sediment in Trout Cave (Holman, 1982). The faunal assemblages recovered from sediment in the upper levels of Trout Cave are thought to be of Rancholabrean age (ca. 300–10 ka), whereas faunal assemblages recovered from sediment in the lower levels of the cave are thought to be of Irvingtonian age, ca. 1.81 million years ago (Ma)–300 ka.

Along the northwesterly path from Trout Cave to Hamilton Cave, the same following two units may be seen within the Keyser Limestone (from base to top): (1) an ~2-m- (7-ft-) thick unit of limestone and argillaceous limestone displaying laminations, cross-bedding, and small thrust faults; overlain by (2) an ~8-m- (26-ft-) thick unit of limestone consisting of numerous relatively homogenous 0.1–0.3-m- (0.3–1.0-ft-) thick beds of wackestone-packstone. The entrance of Hamilton Cave is located along the axis of Cave Knob anticline, ~165 m (541 ft) northwest of the Trout Cave entrance, at the northwest end of the path that runs along the base of the outcrops at the JGCNP. Near the entrance to Hamilton Cave, where the path changes from relatively horizontal to going steeply up the hill, there are excellent exposures of the New Creek Limestone and the overlying Corriganville Limestone. At the outcrop just before the path goes steeply uphill, the New Creek Limestone contains cross-bedding, abundant crinoid columnals (“stems”), some stromatoporoid fragments, and stylolites (Fig. 10). The contact between the New Creek Limestone and the overlying Corriganville Limestone is located ~1–2 m (3.3–6.6 ft) below the entrance to Hamilton Cave (Fig. 11), and at this location the Corriganville Limestone is characterized by brachiopod shells as well as chert nodules and lenses.

Like New Trout Cave and Trout Cave, Hamilton Cave has been known for a long time, but there is no evidence that sediment in the cave was mined for nitrate during the American Civil War. The first published mention of the name “Hamilton Cave” in the scientific/caving literature appears to have been in the Bulletin of the National Speleological Society (Stephenson, 1942). The cave was subsequently described in the book Caverns of West Virginia published by the West Virginia Geological and Economic Survey (Davies, 1958). A modern (1988) map of the cave may be found in the book The Caves and Karst of Pendleton County published by the West Virginia Speleological Survey (Dasher, 2001).

The front (upper) part of Hamilton Cave has a classic network maze pattern that is an angular grid of passages, most of which follow vertical or near-vertical joints that intersect at right angles. This part of the cave lies along the axis of Cave Knob anticline (Fig. 1), and most of the passages are located along the contact between the New Creek Limestone and the overlying Corriganville Limestone (Figs. 12 and 13) along joints that trend N50E or N40W. In many of the maze passages, the New Creek Limestone is obscured by sediment (predominantly mud) that has accumulated on the passage floors, and the visible cave walls are mostly the Corriganville Limestone, which contains prominent lenses of chert that protrude into the cave passages (Fig. 14).

In contrast with the front (maze) section of Hamilton Cave, the passages in the back (lower) part of the cave (beyond the Slab Room) lie along the west flank of Cave Knob anticline, where strata dip ~30° to the northwest. Many of these passages are also located along the New Creek–Corriganville contact, but the passages have formed more within the New Creek Limestone (limestone without chert lenses) such that much more of the New Creek Limestone is exposed along the passage walls. In this area, many of the passage walls have a sinuous (meandering) morphology (Fig. 15), and the cave lacks well-developed network maze morphologies. The passages in the back (lower) part of Hamilton Cave beyond the Slab Room may be divided into the following two categories: (1) northeast-trending passages that follow the strike of the beds; and (2) northwest-trending passages that follow the dip of the beds. The northeast-trending passages that follow the strike of the beds are relatively longer and approximately horizontal (e.g., passage at the entrance to the Airblower; Figs. 1 and 16). In contrast, the northwest-trending passages that follow the dip of the beds are relatively short and descend steeply to the northwest (e.g., the Rosslyn Escalator; Figs. 1 and 17).

Meteorological data from Hamilton Cave are sparse, consisting mostly of air temperature measurements. Dyroff’s (1977) one-day study (10 April 1977) found that the air flowed consistently out of the cave during the morning and afternoon, and that the average air temperature inside the cave was 12 °C (53 °F). Hoke (2009a, 2009b) reported that air always flows out of the cave during the winter, and that data loggers just inside the cave entrance recorded air temperatures that ranged from 13 °C (55 °F) at the beginning of winter to 10.8 °C (51.5 °F) at the end of winter.

Much of Hamilton Cave has a floor of dry to damp mud. In some parts of the cave (e.g., Slab Room, room before Rosslyn Escalator), large boulders are common (classified as breakdown blocks and breakdown slabs, using terminology of White and White, 1969). In parts of the cave west of the Airblower, water drips from stalactites and wet mud is present on the cave floor. Columns and flowstone are also present in the Old Room and in the room before the Old Room.

Gypsum crusts are present within Hamilton Cave along the passage walls of the New Creek Limestone from the Slab Room to the Airblower (Fig. 18). Swezey et al. (2002) reported that the analysis of four gypsum samples collected in the passage immediately southwest of the Airblower revealed sulfur isotope (δ³⁴S) values that range from –11.8 to –10.3 per mil (‱). Additional analyses of two of these samples revealed oxygen isotope (δ¹⁸O) values of –11.8 and –11.2 ‱. The negative δ³⁴S values and their large range suggest that the sulfur in these gypsum samples was not derived from the simple dissolution and reprecipitation of marine evaporite sulfate within the stratigraphic section. Instead, the sulfate-sulfur was probably derived from the oxidation of diagenetic sulfide minerals (such as pyrite) and/or from organically bound sulfur in nearby strata. The δ¹⁸O values are consistent with values expected for sulfate derived from the weathering of sulfide minerals where the dominant oxidizing agent is oxygen dissolved in local meteoric water (Swezey et al., 2002).

Excavations in the front (maze) part of Hamilton Cave have produced numerous vertebrate fossils, including bat, shrew, mole, pika, rabbit, woodchuck, squirrel, rat, mouse, vole, lemming, muskrat, gopher, beaver, porcupine, coyote, dog, raccoon, bear, weasel, skunk, otter, badger, saber-tooth cat, jaguar, cheetah, bobcat, peccary, deer, tapir, and horse, as well as snake, lizard, turtle, toad, frog, and salamander (Grady, 1981a,b,c; 1983, 1985, 1986a, 1987a, 1988, 1991a,b, 1992, 2001b,c, 2005; Repenning and Grady, 1988; Holman and Grady, 1989; Van Valkenburgh et al., 1990; Winkler and Grady, 1990; Mead and Grady, 1996; Holman, 1999; Grady and Garton, 1998, 2000, 2001, 2005; Martin et al., 2003, 2009). Fossils of invertebrate species include crayfish, millipede, terrestrial snails, aquatic snails, and bivalves (Bogan and Grady, 1991). The vertebrate faunal assemblage recovered from sediment in Hamilton Cave is thought to be of Irvingtonian age (ca. 1.81 Ma–300 ka). In most of the publications by Grady and colleagues (listed above), it is estimated that this faunal assemblage is ~800 thousand years old, on the basis of the assemblage of rodent and pika fossils. Martin et al. (2003, 2009), however, estimated that the vertebrate faunal assemblage in Hamilton Cave is ~1.6–1.3 million years (m.y.) old. In addition, Semken et al. (2010) reported that one accelerator mass spectrometry (AMS) date from the mandible of a taiga vole in Hamilton Cave yielded an age of 12,730 ± 60 ¹⁴C yr B.P. (AMS Number CAMS-20589). This age is reported in radiocarbon yr before present (B.P.) with a 1 sigma standard deviation of the age uncertainty, using the Libby half-life of 5568 yr and with 0 B.P. being equivalent to AD 1950.

The Trout Rock caves are excellent sites for examining the geologic controls on cave morphology and cave location. The recognition that individual caves have both network maze passages and non-maze passages yields important information on the variables required for the development of maze caves. In addition, the variations in passage morphology and passage location with respect to strike and dip may be related to positions of the water table during cave development. Finally, the relative elevations of the caves with respect to the modern river provide a framework for speculating about the ages of the Trout Rock caves, a framework for proposing a chronology of water-table behavior at this site during the past few million years, and a framework for possible regional correlations with cave levels elsewhere in the Appalachian region.

Hamilton Cave is an instructive place to investigate the controls on network maze caves. In a classic study, Palmer (1975) attributed the maze character of Hamilton Cave to the combination of the following three factors:

1. Tension along the axis of the anticline creating prominent joints;

2. The thickness of the overlying bedrock (Oriskany Sandstone) being less over Hamilton Cave (compared to Trout Cave and New Trout Cave), resulting in greater joint enlargement by erosional unloading;

3. Groundwater recharge to Hamilton Cave both as diffuse recharge from the surface and as backflooding from the South Branch of the Potomac River.

In the publication by Palmer (1975), however, the map of Hamilton Cave shows only the front (maze) portion of the cave. With the subsequent availability of more extensive and more detailed maps (Fig. 1), it is apparent that classic maze pattern is restricted to only the front (upper) part of the cave. Hamilton Cave is much more extensive than was mapped originally (Davies, 1965), and some parts of the cave have maze characteristics, whereas other parts of the cave do not. Therefore, with the availability of new data, it now appears that the network maze portion of Hamilton Cave developed where the following conditions were present:

1. Location on or near the axis of an anticline (the location of the greatest amount of flexure);

2. Abundant vertical or near vertical joints, favored by location in area of greatest amount of flexure and by a lithologic unit (limestone with chert lenses) that is more likely to experience brittle deformation than ductile deformation;

3. Widening of joints to enhance ease of water infiltration, favored by location in area of greatest amount of flexure (a correlation with the thickness of the Oriskany Sandstone may or may not be important here);

4. Dissolution along nearly all major joints to produce cave passages of approximately the same size (which would most likely occur via water without a free surface in contact with air—in other words, cave development at or near the top of the water table or under deep phreatic conditions, not vadose conditions). The interpretation that the network maze portion of Hamilton Cave developed at or near the top of the water table or under deeper phreatic conditions is supported by the observation that most of the cave is located along the same stratigraphic contact, which is interpreted as a karst “inception horizon” (thought to develop most commonly under phreatic conditions; Lowe and Gunn, 1997; Filipponi et al., 2009). In addition, the major cave passages of Hamilton Cave are oriented nearly parallel to strike, and the cave passages terminate abruptly in both updip and downdip directions. In areas of dipping strata, such characteristics suggest that a cave developed near the top of the water table/phreatic zone (Palmer, 1991, 2005a, 2005b; Audra and Palmer, 2011).

In contrast, the back (lower) level of Hamilton Cave is interpreted as having formed later as the water table fell, and passage enlargement occurred preferentially within the New Creek Limestone by flowing water with a free surface in contact with air (vadose conditions). A phreatic to vadose transition is suggested by the presence from the Slab Room to the top of the Rosslyn Escalator of large boulders, classified as breakdown blocks and breakdown slabs, which according to Palmer (2007) are most likely to form when a passage is first drained of water and buoyant support of the ceiling is removed. In addition, the wavy and scalloped morphology of the cave walls in Hamilton Cave between the Slab Room and the Rosslyn Escalator (Fig. 15) strongly suggests that passage morphology in this area was shaped by flowing water with a free surface in contact with air (vadose conditions). Vadose conditions are also suggested by the northwest-trending passages of relatively short length that follow the dip of the beds and descend steeply to the northwest in the back (lower) part of Hamilton Cave beyond the Slab Room (e.g., the Rosslyn Escalator; Fig. 19).

Despite being located along the axis of an anticline, a maze pattern did not develop within the back (lower) part of Hamilton Cave. The absence of a network maze pattern may be attributed to the combination of the following features: (1) a less dense network of joints, possibly related to the lower portion of the cave being located predominantly within the New Creek Limestone where the lack of chert lenses resulted in slightly more ductile deformation; and (2) flowing water with a free surface in contact with air (vadose conditions), which prevented slow and simultaneous dissolution along all possible conduits.

Published details of other nearby maze caves are relatively sparse, but it seems likely that at least some other maze caves may share similar geologic and geomorphological characteristics. For example, both Withero’s Cave (Bath County, Virginia) and Paxton’s Cave (Alleghany County, Virginia) have network maze passages and are located in Silurian–Devonian limestone on or near the axes of anticlines (Stephenson, 1945; Baroody, 1966–1967). Likewise, Helictite Cave and Owl Cave (both in Highland County, Virginia) also have network maze passages, and are located in Silurian–Devonian limestone on or near the axes of anticlines (Swezey et al., 2017; Brent et al., 2019). Furthermore, both Helictite Cave and Owl Cave have the following geomorphological characteristics, which are similar to those of Hamilton Cave (Swezey et al., 2017; Brent et al., 2019):

1. An upper level of horizontal passages that follow intersecting joints at right angles, thought to have developed under conditions where the water did not have a free surface in contact with air (either at shallow depths beneath the water table or under deeper phreatic conditions);

2. A lower level of passages with wavy (meandering) passage wall morphology, thought to have developed subsequently under vadose conditions where the water had a free surface in contact with air.

In contrast, nearby caves that are located along the axes of synclines tend to have a major trunk channel that follows the syncline axis, but they lack well-developed network maze patterns. Examples from Bath County (Virginia) include Sand Canyon in Butler Cave and the Burns Streamway in the greater Chestnut Ridge Cave System (White, 2015; Swezey et al., 2017). Thus, the location of the greatest amount of flexure (anticline or syncline axis) is not the sole criterion necessary for the development of network maze caves.

Some early publications suggested that caves develop deep below the water table where groundwater flow paths are likely to be stable for a long duration (e.g., Davis, 1930; Bretz, 1942), whereas other publications suggested that caves are most likely to develop at and/or just below the water table (especially the zone of water-table fluctuation), where the groundwater flow is most vigorous (e.g., Swinnerton, 1932). Subsequent publications concluded that caves may develop both above the water table (in the vadose zone), below the water table (in the phreatic zone), and in the zone of transition (epiphreatic or epigenic zone) where water-table fluctuations occur (Ford and Ewers, 1978; Palmer, 1991, 2005a, 2005b). On a worldwide basis, however, at least 80%–90% of known caves are thought to have developed in epigenetic settings (Audra and Palmer, 2011).

Some clues about cave development with respect to the water table may be determined by answering certain questions regarding cave location and passage morphology. For example:

1. Are the major cave passages located along strike, or along dip, or do they cut across strike or dip?

2. Are the slopes of the major cave passages gentle or steep?

3. Do the major cave passages change size in the direction of the nearest river valley?

4. What are the shapes in profile of the major cave passages, and how are these profile shapes related to stratigraphic bedding or structural features?

5. Do the elevations of major cave passages correlate with the elevations of nearby river terraces?

With respect to strike and dip, cave passages that follow the dip of bedding are generally considered to have developed above the water table (under vadose conditions), whereas cave passages that follow the strike of bedding are generally considered to have developed under the influence of a water table or under deeper phreatic conditions (Palmer, 1991, 2005a; Audra and Palmer, 2011). Under vadose conditions, the water has a free surface in contact with air, and water flow is governed primarily by gravitational processes. Therefore, in areas of dipping strata, most cave passages that form under vadose conditions are relatively short and they descend abruptly along dip (unless prominent fractures impose other trends). Excellent examples include Breathing Cave and Butler Cave in Bath County, Virginia (Deike, 1960; Swezey et al., 2017). In contrast, under phreatic conditions (either at shallow depths beneath the water table or under deeper phreatic conditions), the water does not have a free surface in contact with air, and water flow tends to follow the hydraulic gradient at gentle slopes. Therefore, in areas of dipping strata, most cave passages that form under phreatic conditions (either at shallow depths beneath the water table, or under deeper phreatic conditions) do not follow the dip of the strata. Furthermore, in areas of dipping strata, elongate cave passages that extend along the strike of bedding, and that terminate abruptly in both updip and downdip directions, are thought to have developed at or near the top of the water table/phreatic zone (Palmer, 1991, 2005a, 2005b; Audra and Palmer, 2011). Excellent examples from Pendleton County include Cave Mountain Cave (Swezey and Dulong, 2010) and the longer northeast-trending passages of the Trout Rock caves (Hamilton Cave, Trout Cave, New Trout Cave), where the major cave passages are oriented nearly parallel to strike.

Cave passage length and slope can provide some clues to water-table influences on cave development (Palmer, 1991, 2005a; Audra and Palmer, 2011). As stated above, in areas of dipping strata, most cave passages that are relatively short and that descend abruptly are thought to have developed under vadose conditions. In contrast, cave passages that are relatively long and that have a relatively gentle slope are thought to have developed at or near the top of the water table/phreatic zone.

Changes in the size of major cave passages may provide additional clues to water-table influences on cave development. For example, Davies (1960) noted that many caves in the Appalachian region have passages that are large in the part of the cave closest to a nearby river valley, whereas cave passages are progressively narrow and lower in height with increasing distance from the nearby river valley. He implied that this feature was indicative of a water-table control on cave development, and he cited both Trout Cave and New Trout Cave as examples.

The shape in profile of a cave passage may provide useful information on cave development (Deike, 1989; White and Deike, 1989; Palmer, 1991, 2005a; Audra and Palmer, 2011). For example, unless prominent fractures impose other trends, dissolution by groundwater at or below the water table (in the phreatic zone) tends to form tubular passages with lenticular, elliptical, or nearly circular cross sections. If the water table drops and phreatic passages are abandoned, then the water will have a free surface in contact with air (vadose conditions), and the water may cut a canyon in the cave floor to form a keyhole cross section. Within a given cave, some passages have profiles that are clearly controlled by joints, whereas other passages have profiles that are clearly controlled by bedding planes (e.g., Butler Cave; Swezey et al., 2017). “Joint-controlled passages” usually have greater height to width ratios than “bedding-plane controlled passages.” In the maze (front) portion of Hamilton Cave and in the northwest portion of Trout Cave, the cave passages are clearly joint controlled. Palmer (1991) attributed the development of most maze caves to “simultaneous enlargement of many competing openings.” In contrast, other portions of Trout Cave and some portions of New Trout Cave have passages with trapeziform cross sections (Davies, 1960), which are governed primarily by bedding planes of dipping strata. Yet other passages have bell-shaped profiles where the passage morphology is controlled by a fold in the strata (e.g., Triangle Room in Trout Cave; Fig. 4).

Finally, if the elevations of major cave passages correlate with the elevations of nearby river terraces, then it is likely that the development of these cave passages occurred at or near the elevation of a water table (Davies, 1960; Palmer, 1987, 1991). Furthermore, in a cave with multiple levels, the larger cave passages are those that remained active for a longer duration, and the development of these larger passages usually correlates with times of nearly static base level and with times of greater development of river floodplains (Palmer, 1987, 1991). At Cave Knob, for example, each of the three major caves (Hamilton, Trout, New Trout) has passages at discrete levels or elevations (Fig. 2), and Davies (1960) thought that these different cave-level elevations correspond closely with the elevations of nearby river terraces at the town of Franklin and at the mouth of nearby Thorn Creek.

In summary, the assumption that each of the three large Trout Rock caves (Hamilton Cave, Trout Cave, New Trout Cave) formed initially at or near the water table is supported by the following observations:

1. The major cave passages are elongate along the strike of bedding, and they terminate abruptly in both updip and downdip directions.

2. The major cave passages are inclined at gentle gradients.

3. The major cave passages of Trout Cave and New Trout Cave are larger where they are closest to the nearby river valley, and they are progressively narrow and lower in height with increasing distance from the river valley.

4. The three caves have developed along the same stratigraphic contact, which is interpreted as a karst “inception horizon” (thought to develop most commonly under phreatic conditions; Lowe and Gunn, 1997; Filipponi et al., 2009).

5. The elevations of the major cave passages correlate with the elevations of nearby river terraces (postulated by Davies, 1960).

Likewise, the assumption that each of the three large Trout Rock caves experienced a subsequent phreatic to vadose transition is supported by the following observations:

1. The wavy and scalloped morphology of the cave walls in the New Creek Limestone in the lower (sub-maze) portion of Hamilton Cave from the Slab Room to the top of the Rosslyn Escalator;

2. The northwest-trending passages of relatively short length that follow the dip of the beds and descend steeply to the northwest in the back (lower) part of Hamilton Cave beyond the Slab Room (e.g., the Rosslyn Escalator);

3. The presence throughout the caves of large boulders that are classified as breakdown blocks and breakdown slabs, which according to Palmer (2007) are most likely to form when a passage is first drained of water and buoyant support of the ceiling is removed.

Reliable estimates for the age of a cave are difficult to obtain, although an approximate age may be estimated by using the following assumptions and observations: (1) assuming that the cave formed at and/or not far below the top of the water table; (2) noting the elevation of the cave above the elevation of the nearby river; and (3) assuming an average rate of fluvial incision and erosion during the past several million years.

There are several means of estimating an average rate of fluvial incision and erosion. One estimate is provided by Springer et al. (1997), who used cave magneto-stratigraphy and elevation data to estimate an incision rate of ~61 m (200 ft) per m.y., or 30.5 m (100 ft) of incision every 0.5 m.y., for the Cheat River Basin in Monongalia and Preston Counties, West Virginia. Their incision rate would suggest that Hamilton Cave formed ca. 1.5 Ma, Trout Cave formed ca. 1.2 Ma, New Trout Cave formed ca. 0.6 Ma, and NSS Flood Cave formed ca. 0.25 Ma.

Another estimate is provided by Matmon et al. (2003), who used ¹⁰Be techniques in the Great Smoky Mountain National Park (North Carolina, Tennessee) to estimate an erosion rate of ~27 m (89 ft) per m.y. for the Mesozoic and Cenozoic (i.e., during the past 252 m.y. according to the geologic time scale of Gradstein et al., 2012). This erosion rate is equivalent to ~30.5 m (100 ft) of erosion every 1.1 m.y., and would suggest that Hamilton Cave formed ca. 3.5 Ma, Trout Cave formed ca. 2.8 Ma, New Trout Cave formed ca. 1.4 Ma, and NSS Flood Cave formed ca. 0.6 Ma.

In summary, as a very general first-order approximation, for every 30 m (~100 ft) of elevation that a cave is located above the modern river, one might estimate ~1 m.y. (±50%) of time for incision and erosion. Interestingly, a similar order of magnitude was presented by Dole and Stabler (1909), who calculated that the surface of the United States is being removed at the rate of 2.54 cm (1 in) per 760 yr, or 30 m (~100 ft) per ~912,000 yr.

Despite approximately similar orders of magnitude, the erosion rate estimates should be viewed very cautiously. At Dolly Sods Wilderness, which is located ~48 km (30 miles) due north of the Trout Rock caves, Hancock and Kirwan (2007) calculated a mean bare-bedrock erosion rate of 5.7 m (18.7 ft) per m.y., which is much lower than typical rates of fluvial incision in the central Appalachian region. Furthermore, Mills (2000) demonstrated that incision rates in the Appalachian region have not been constant, and that changes in climate may accelerate or decelerate rates of fluvial incision. Nevertheless, the rates listed above yield a general estimate that Hamilton Cave may have started to develop ca. 3.5–1.5 Ma, Trout Cave may have started to develop ca. 3–1 Ma, New Trout Cave may have started to develop ca. 1.5–0.5 Ma and NSS Flood Cave may have started to develop ca. 600–250 ka. The paleo-fauna (fossils) found among sediment in a cave must be younger than the age of the cave, and thus the age estimate of 3.5–1.5 m.y. for Hamilton Cave is consistent with the estimate of 0.80 m.y. for the fauna found in the cave. Likewise, the age estimate of 3–1 m.y. for Trout Cave is consistent with the estimate of 1.8–0.3 m.y. for the fauna from the lower levels of the cave, and the age estimate of 1.5–0.5 m.y. for New Trout Cave is consistent with the estimate of 0.5 m.y. for the older fauna from the cave.

If the Trout Rock caves are a few million years old, and if the along-strike northeast-trending passages developed under the influence of a water table that was at a stable position for a long duration, then it is tempting to correlate portions of each of the three caves with distinct episodes of water-table stillstand during the past few million years. The biggest drop in water table during the past few million years was probably coincident with the sudden increase in ice sheet extent in the Arctic that occurred ca. 3.1–2.5 Ma (e.g., Shackleton et al., 1984; Maslin et al., 1998; Haug et al., 1999; Prueher and Rea, 2001). With subsequent behavior of the northern hemisphere ice sheet and corresponding influences on sea level, episodes of ice sheet stability might correlate with times of water table at stable positions (approximately horizontal cave passages developing along strike), and episodes of ice sheet growth might correlate with times of water-table falls (steeply dipping cave passages developing along the dip of bedding).

With these assumptions in mind, a proposed chronology of events is outlined as follows:

1. ca. 3.5–3.1 million years ago (Ma): Water table stillstand; upper (maze) part of Hamilton Cave at ~631–641 m elevation (85–95 m above modern river) developed in phreatic zone (below water table) along New Creek–Corriganville contact;

2. ca. 3.1–2.5 Ma: Water table dropped coincident with the onset of glaciation in the northern hemisphere; in some upper parts of Hamilton Cave (e.g., Slab Room, Room before Rosslyn Escalator), northwest-trending passages that follow the dip of the bedding were modified in the vadose zone or epiphreatic zone;

3. ca. 2.5–1.5 Ma: Several water-table falls and stillstands; Development of much of Trout Cave and lower portions of Hamilton Cave, specifically passages at 601.7–623.6 m elevations (or 55.7–77.6 m above modern river);

4. ca. 1.5–0.5 Ma: Another water table stillstand; most of New Trout Cave (~574.6–587.7 m elevation, or 28.6–41.7 m above modern river) developed in phreatic zone (below water table) along New Creek–Corriganville contact;

5. After 0.5 Ma: Water table dropped again to near present position.

Using the chronology proposed above for the Trout Rock caves, it is tempting to make comparisons with ages proposed by Granger et al. (2001) and Anthony and Granger (2007) for different cave levels at Mammoth Cave in Kentucky and in the Cumberland River Basin in Tennessee and Kentucky. For example:

1. The upper (maze) portion of Hamilton Cave (~623.6–641.6 m elevation, or 77.6–95.6 m above modern river) might be comparable with Cave Level A at Mammoth Cave (>80 m above the Green River; age range estimated at 4.12–3.12 Ma) and with the Highland Rim upper cave level of the Cumberland River Basin (66–91 m above modern river; age range estimated at 6.77–3.10 Ma).

2. Much of Trout Cave (~601.7–623.6 m elevation, or 55.7–77.6 m above modern river) might be comparable with Cave Level B at Mammoth Cave (50–80 m above the Green River; age range estimated at 2.39–1.91 Ma) and with the middle cave level/Parker strath of the Cumberland River Basin (40–54 m above modern river; age range estimated at 2.62–0.68 Ma).

3. Most of New Trout Cave (~574.6–587.7 m elevation, or 28.6–41.7 m above modern river) might be comparable with Cave Level D at Mammoth Cave (30 m above the Green River; age range estimated at 1.57–1.33 Ma) and with the lower cave level/first terrace below Parker strath of the Cumberland River Basin (28–38 m above modern river; age range estimated at 1.24–0.48 Ma).

Although the chronology and correlations proposed above are quite speculative, as more data become available on the elevations, morphologies, geologic settings, and ages of caves, it might be possible to determine whether the different cave levels at Cave Knob have developed in response to local changes or in response to regional/global changes in water table, climate, and tectonic activity. In other words, it might be possible to determine whether or not the different cave levels of the Trout Rock caves are part of a larger assemblage of cave levels in the eastern United States that developed synchronously in response to regional forcing mechanisms.

The authors thank the Potomac Speleological Club, Bob Hoke, Dave West, and numerous other cavers who have accompanied us on various trips to the Trout Rock caves during the past several years. Grant Colip, Andrew Everett, Karlee Prince, and Katherine Worms provided additional field assistance in preparation for this GSA field trip. The manuscript was improved by suggestions from reviewers John Repetski (U.S. Geological Survey) and John Haynes (James Madison University). Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.