Dineley, D. & Metcalf, S. GCR Editor: D. Palmer. 1999. Fossil Fishes of Great Britain. Geological Conservation Review Series No. 16. JNCC, Peterborough, ISBN 1 86107 470 0.
The original source material for these web pages has been made available by the JNCC under the Open Government Licence 3.0. Full details in the JNCC Open Data Policy
The late Downton vertebrate faunas at this Shropshire site represent some of the earliest known freshwater vertebrates from Britain, and are locally succeeded by pteraspidids of the Dittonian stage. The site's importance is in showing the strength of the late Downtonian–early Dittonian vertebrate communities.
Devil's Hole is a stream section at the northern edge of the Anglo-Welsh Old Red Sandstone outcrop, where it cuts the 'Psammosteus' Limestone escarpment, and hence, the Downtonian–Dittonian junction. The GCR Unit undertook a series of excavations here between 1980 and 1982, enabling a detailed sedimentological and palaeontological analysis to be made of the sec tion by M.A. Rowlands and P. Tarrant.
The geology of the site has been referred to by Ball and Dineley (1961), Allen and Tarlo (1963), Banks (1980), Richardson et al. (1981), Halstead (1985), Allen (1985) and Blieck (1985), and the fish faunas by Wills (1948, 1950), White (1950b), Denison (1956), Robertson (1957), Dineley (1964), Turner (1973), Blieck (1981, 1984, 1985) and Tarrant (1991). The section was originally collected by Ball and Dineley (1961) and subsequently excavated by P Tarrant and M.A. Rowlands (Tarrant, 1991).
The sequence consists of 50 m of mudstone, sandstone and intraformational conglomerates containing fossils and calcretes
The four main fossil-bearing beds in Devil's Hole are called Lye Brook 1, Lye Brook 3 and Lye Brook 4, abbreviated to LB1, LB3 and LB4 (Ball and Dineley, 1961) and the 'Cephalaspis Sandstone', a unit named by WW King (1934), who believed it to be a laterally continuous sandstone recognizable throughout the Welsh Borders. It is not present at the Devil's Hole Stream section, but outcrops on Meadowley Bank to the north. All four beds can be recognized at several other localities locally.
AGNATHA
Heterostraci: Eriptychiformes: Tesserapididae
Tesseraspis tessellata Wills, 1935
Heterostraci: Phialaspidiformes: Poraspididae
Poraspis polaris Kiaer, 1930
Heterostraci: Phialaspidiformes: Traquairaspididae
Traquairaspis symondsi (Lankester, 1868)
Heterostraci: Pteraspidiformes: Pteraspidae
Protopteraspis gosseleti (Leriche, 1906)
( = Pteraspis leathensis White, 1948)
Osteostraci: Scolenaspidiformes: Scolenaspididae
'Scolenaspis' n. sp.
Pattenaspis whitei (Stensiö, 1932)
Osteostraci incertae sedis
'Cephalaspis' sp.
Thelodonti: Thelodontida: Turiniidae
Turinia pagei fauna' (see
GNATHOSTOMATA
Acanthodii incertae sedis
'Ischnacanthid and climatiid spines'
The largest accumulations of fossils, particularly the fish plates, occur near the base of the intraformational conglomerates, together with clasts of reworked mudstone and calcrete. The remains are disarticulated and have been water-sorted. Taken as a whole, acanthodian spines form the major percentage of fossils, followed by heterostracans. The material in LB4 is poorly preserved and often completely demineralized, even though some specimens are associated with well-preserved fragments of Protopteraspis. It is possible, therefore, that this material is derived.
There seem to be two quite distinct assemblages, one in the LB1 Bed and LB3 Bed, the other in the LB4 Bed and Cephalaspis Sandstone. These assemblages are listed separately below.
LB 1 Traquairaspis symondsi (Lankester, 1868), Tesseraspis tessellata Wills, 1936, 'Scolenaspis' n. sp., 'Cephalaspis' n. sp., Turinia pagei thelodont fauna, and ischnacanthid and climatiid acanthodian remains.
LB 3 Traquairaspis symondsi (Lankester, 1868), Tesseraspis tessellata Wills, 1936, 'Cephalaspis'spp., Turinia pagei thelodont fauna, and ischnacanthid and climatiid acanthodian remains.
LB 4 Protopteraspis gosseleti (Leriche, 1906) (syn. Pteraspis (Simopteraspis) leathensis White, 1935), Poraspis polaris Kiaer, 1930, 'Cephalaspis'sp., Turinia pagei thelodont fauna, and ischnacanthid and climatiid acanthodian remains.
Cephalaspis Sandstone Protopteraspis gosseleti (Leriche, 1906), cf. Pattenaspis whitei (Stensiö, 1932), 'Scolenaspis'n. sp. (Rowlands MS) and acanthodian scales and spines. Collections are housed in the NHM and SMLU.
The commonest heterostracan occurring below the 'Psammosteus' Limestone at Devil's Hole is Traquairaspis symondsi Lankester
Isolated tesserae of Tesseraspis tessellata are commonly found associated with T. symondsi in the Downtonian beds, where there appears to be a range of sizes of individuals at different stages of exoskeletal development. There are no records of T. tessellata from outside the Anglo-Welsh Old Red Sandstone.
Protopteraspis gosseleti is a common vertebrate macrofossil above the 'Psammosteus' Limestone in many Welsh Border localities
Five species of cornuate osteostracans have been discovered at Devils Hole, and a further three or four species are probably represented by indeterminate fragmentary material. Osteostracans occur in each of the main fossil beds, but are less common than acanthodians or heterostracans and usually occur as fragmentary pieces of head shield or indeterminate scales. Each fossiliferous horizon contains different osteostracan species, which is typical of their occurrence throughout the Welsh Borders.
Two species were found in the collections from LB1. Several head shields and fragments were of 'Scolenaspis'n. sp., a small scolenaspidiform with a dorsal spine. One head shield from LB1 is very different from other cornuates from the Welsh Borders, and has features of both scolenaspidiforms and thyestidians. The two species recovered from LB1 represent two of the earliest known cornuate cephalaspids in the Welsh Borders: most osteostracans from the Downtonian are non-cornuate, and cornuate forms become common only in the Dittonian.
Cornuate osteostracan material dominates in the collections from LB3, consisting of fragments, including many portions of robust tuberculated cornua, and pieces of thick tuberculated ventral rim. LB4 is a highly fossiliferous bed, containing many discs of Protopteraspis and acanthodian spines, but species diversity is lower than in LB1 or LB3. Osteostracan material is rare, and identifiable material probably represents only one species of a medium-sized cornuate.
Several good headshields of rare osteostracans have been recovered from the 'Cephalaspis'Sandstone. A single specimen of a headshield of cf. Pattenaspis whitei Stensiö was found in the basal intraclast layers of sandstone near the head of Devil's Hole section, and several other specimens of headshield of a second kind were recovered from loose blocks of the overlying sandstones from Meadowley Bank. The second species, 'Scolenaspis'n. sp., is a small cephalaspid with eyes well forward, and has similarities with Eucephalaspis agassizi. As in other instances throughout the Anglo-Welsh Lower Devonian, the osteostracans of this assemblage are rather uniform in size and composition, perhaps indicating that the osteostracans preferred habitat was separate from that of other fishes present.
Turner (1973) recorded denticles of Turinia pagei from the LB1 and LB3 Beds, and similar material has now been discovered in the LB4 Bed. Acanthodian specimens from Devil's Hole consist mainly of isolated fin spines representing a variety of species; some individuals reach 7 cm in length, indicating relatively large acanthodians.
The sequence at Devil's Hole represents interdistributary bay and ephemeral channel deposits formed in an alluvial plain. Below the 'Psammosteus' Limestone, the Downton sequence is dominated by overbank deposits with mud-cracks and calcretes. The overlying Ditton Group beds show the return of abundant fluvial sandstones, perhaps marking a wetter climate, with extensive river systems reappearing across the Anglo-Welsh floodplain. The fluvial channels were colonized by assemblages of fishes different from those that had been present during Downton Group time.
The excavations at Devil's Hole reinforced the earlier hypothesis (Ball and Dineley, 1961; Allen and Tarlo, 1963; Halstead, 1985) that the upper Downton Group and lower Ditton Group in the central Welsh Borders are entirely non-marine, although Allen (1985) and Barclay et al. (1994) have shown that the upper Downton Group was partly intertidal to the south. Blieck (1985) interpreted the Anglo-Welsh Old Red Sandstone faunas as occurring in marine intercalations in otherwise non-marine sequences, because similar faunas occur elsewhere associated with marine strata. However, neither the sedimentology nor the invertebrate palaeontology of the Devil's Hole sequence shows any sign of marine influence, since the deposits represent relatively small-scale high-energy regimes, with calcretes and desiccation-crack beds immediately below.
Most of the Downton Group fishes were seemingly non-marine in Britain. As is typical of the upper part of the Downton Group throughout the Welsh Borders, the faunas below the 'Psammosteus' Limestone at Devil's Hole are dominated by Traquairaspis symondsi, Tesseraspis tessellata and acanthodians, which occur only rarely in the Ditton Group. Protopteraspis gosseleti is the most common vertebrate fossil found immediately above the 'Psammosteus' Limestone in the Anglo-Welsh basin, which suggests the sudden replacement of the traquairaspid fauna at this level by new forms. The few Ditton Group species that occur below the 'Psammosteus' Limestone are mostly found in marine or brackish sequences. Cornuate osteostracans are well known in the Welsh Borders, but occur mainly in the Ditton Group, coinciding with the development of fluviatile deposits. This fauna is found in the LB4 Bed and, as elsewhere in the Anglo–Welsh region, is represented only by adults of a simple ontogenetic stage. The 'Psammosteus' Limestone is interpreted as representing a period of aridity lasting for up to 10 000 years (Allen, 1985), which would have had a drastic effect on any faunas restricted to freshwater conditions. The only fishes to occur locally both above and below the 'Psammosteus' Limestone are thelodonts, a group known to have been adapted to brackish and freshwater conditions (Turner, 1973), and would thus be able to escape from hostile arid climates.
Juvenile and adult forms of heterostracans occur together, which suggests that these fishes spent their entire life-history in freshwater environments. The extinction of some species at the 'Psammosteus' Limestone event suggests that they were unable to adapt to a changed habitat. On the other hand, the Ditton Group species appear to have been adapted to both marine and freshwater environments since they occur elsewhere in marine strata. Such adaptability to both marine and fresh water habitats at different stages in their life cycle is seen in modern fishes (Wills, 1950).
Reconstructions of Protopteraspis gosseleti and Traquairaspis symondsi (Tarrant, 1991) show them to have been opportunistic perhaps microphagous feeders that rooted in the substrate, consuming a variety of food. However, P. gosseleti has a more evenly vaulted cephalothoracic shape, and the less frequent occurrence of ventral abrasion suggests a more nectonic lifestyle (Blieck and Tarrant, pers. comm.). Only plates from adult animals occur at Devil's Hole, suggesting that its juvenile development occurred elsewhere. The migration of juvenile forms from marine to freshwater has been used to explain the succession of distinct species that is found in the Welsh Borders Old Red Sandstone, and also their distribution outside this province (Wills, 1950; Denison, 1956; Robertson, 1957; White, 1958a; Allen and Tarlo, 1963). The matter is, however, not resolved beyond doubt.
The exact chronostratigraphical position of these sediments is uncertain. Holland and Richardson (in Martinsson, 1977) have argued that, in the Welsh Borders, the Silurian–Devonian boundary should be provisionally placed at the 'Psammosteus' Limestone. However, a Turinia pagei thelodont fauna in LB1, the lowest fossil bed, might indicate that this section is entirely within the Lower Devonian (Karatajute-Talimaa, 1978), and that the boundary between the Přídolí Series and the Lochkovian Series must lie within the Downton Group in the Welsh Borders. A spore assemblage from the Protopteraspis Bed, just above the 'Psammosteus' Limestone, indicates an early Lochkovian age (Banks, 1980; Richardson et al., 1981). Current opinion, which places the boundary at the base of the lowest bed with the protopteraspid P. gosseleti, needs to be substantiated by a detailed account of a stratotype section for the base of the Devonian in this facies. The Devil's Hole section appears to be a strong candidate.
The site LB1 has a relatively similar relationship to the 'Psammosteus' Limestone to those of other Traquairaspis localities in the Clee Hills area. Little Oxenbold and Earnstrey Brook localities (Ball and Dineley, 1961) bear a similar fauna. Other localities around the Clee Hills and at approximately the same level below the 'Psammosteus' Limestone lack the distinctive Phialaspis symondsi element (e.g. Targrove Dingle 6; in Ball and Dineley, 1961). White (1946) and Tarrant list other sites of comparable position and fauna throughout the Welsh Borderland and Wales.
Blieck et al. (1995) have issued a revised Upper Silurian–Lower Devonian ichthyostratigraphy of northern France and Belgium, in which they list localities to be correlated to the 'Traquairaspis Zone'. Dineley (1964) found Traquairaspis cf. symondsi in a Canadian red-bed sequence that lacked any horizon to be correlated with the 'Psammosteus' Limestone, but the zone was subsequently identified in Spitsbergen (Blieck, 1984) and Arctic Canada (Dineley, 1990).
The recent finds in the Devil's Hole include enigmatic osteostracan specimens, and plentiful material of Traquairaspis symondsi, Protopteraspis leathensis ( = P. gosseleti)and P. aquilonia, which allow detailed morphological reconstructions (Blieck and Tarrant, pers. comm.). Since the varied forms of Protopteraspis gosseleti occur here in close association, together with a range of intermediates, it may be shown that the three previously described species are conspecific, and that this also occurs in the Psammites de Lievin of France and the Fraenkelryggen 'Division' (Red Bay Group) of Spitsbergen (Blieck, 1984). In Nova Scotia P. whitei occupies a similar stratigraphical position and is very like P. gosseleti (Denison, 1955).
The conservation value of this site arises from the importance of its late Downtonian to early Dittonian fish faunas. Furthermore, from the viewpoint of interregional correlation this important site has much potential for new discoveries, being an accessible and extensive stream section and adjacent scarp face. Also, the site is an unbroken section across the important Downtonian–Dittonian boundary.