A Tournaisian (earliest Carboniferous) conglomerate-preserved non-marine faunal assemblage and its environmental and sedimentological context

Tetrapods: large elements Figs. 3–5

Braincase: Two braincase elements have been identified. The anteriormost part of a parasphenoid cultrifom process (2017.2.187 Fig. 3E) of a large tetrapod shows a weakly denticulate ventral surface. The dorsal surface, seen from micro-CT scans (Fig. 3E), is strongly concave, and the body is pierced by canals and other foramina of uncertain function (Fig. 3F). The element is similar in size to that of Crassigyrinus, although the cultriform process in that animal has only a small oval area of denticulation further posteriorly (Panchen, 1985).

A basal process (2011.9.12 Fig. 3A) from the right side of the braincase is preserved with part of the basisphenoid. Grooves for the carotid arteries are clearly shown, as are the two facets of the basipterygoid process and its lining of dermal bone from the parasphenoid. Dorsally, a cavity probably represents part of the dorsum sellae. The element is large, and would indicate an individual larger than known Crassigyrinus specimens.

Skull bones: A robust but highly incomplete example of a tabular (2011.9.5 Fig. 3C) has worn but rugose dermal ornament on the dorsal surface. The posterolateral margin carries a broken tabular horn, grooved dorsally. The posterior margin carries a thin, narrow flange along its ventral edge, which presumably contacted an extension from the postparietal. The posterior margin curves strongly toward its medial edge, and with the postparietal would have formed a so-called “widow’s peak” as described in some embolomeres, but is more exaggerated in form. The bone is much larger than the equivalent in the known specimens of Crassigyrinus.

Identified as a tetrapod postfrontal, 2017.2.545 (Fig. 3B) has typically tetrapod pit and ridge dermal ornament. It is rather strongly developed in the posterior part of the bone, but lighter anteriorly. The bone is complete, with the orbital margin well defined, and sutural surfaces on each of the other sides. It has a sutural surface anteriorly for the prefrontal, and there is an excavation on the medial side for contact with the nasal. Medially is a contact for the parietal, and posteriorly two contacts, one presumably for the supratemporal and one more laterally, suggesting the presence of an intertemporal or a dorsal process of a postorbital. The bone relationships more closely resemble those of Anthracosaurus (Panchen, 1977) than embolomeres (Panchen, 1972) or Eoherpeton (Smithson, 1985).

Tooth: An isolated tooth (2017.2.384 Fig. 3D) is a blunt cone. It has labyrinthine infolding at the base, and an acuminate tip. It does not resemble that of a rhizodont tooth and is interpreted as tetrapod.

Postcranial elements, axial skeleton: Centra, represented by two half-hoops (2017.2.374 Fig. 4D), show the typical morphology of gastrocentrous intercentra, similar in size to those known from Crassigyrinus (Panchen, 1985). The ventral and lateral surfaces are lined with smooth periosteal bone. Internally the surfaces are unfinished.

Neural arches are represented by one half of a neural arch (2017.2.189g Figs. 4B and 4C), probably an atlas arch somewhat similar to that of Greererpeton (Godfrey, 1989), and 2017.2.511, half of a neural arch (Fig. 4E). Specimen 2017.2.189g has little remaining evidence of zygapophyses, and the neural spine is thin with the distal portion somewhat compressed, leaving a transverse ridge across its base. Micro-CT scans show the mesial surface of unfinished bone where it adjoined its antimere and the underlying centra. An atlas arch of this size probably belonged to a moderately large individual, and it was found associated with the Crassigyrinus-like rib 2017.2.189a.

Several large ribs or parts of ribs most closely resemble those of Crassigyrinus. A large example (2017.2.189a Fig. 4A) has a striated surface. Its ends are somewhat crushed, but this seems to result purely from compression that did not exaggerate the expansion of both ends. In the figure, the probable proximal end is to the top. There is no apparent curvature, but that is usual for cervical ribs in early tetrapods including Crassigyrinus, as is the expanded distal end (Panchen, 1985). Rib 2017.2.49 (not figured) is of a similar size, but has lost both ends and is straight. The striation resembles that of 2017.2.189a, but identity as a lungfish rib is not ruled out.

Postcranial elements, appendicular skeleton: Several metapodials and phalanges including an ungual have been identified. A slightly asymmetrical autopodial bone (2017.2.388 Fig. 4J, left pair) is smooth and convex in dorsal view, and concave and grooved medially in ventral view. Others include 2017.2.380 (Fig. 4J, central image) exposed in ventral view only. It is more symmetrical but is also grooved medially. A shorter element, 2017.2.61, is probably a phalanx (Fig. 4J right pair). A probable ungual (2017.2.382) is exposed in dorsal view (Fig. 4I). It is flattened with an expanded proximal end and a rounded distal end. It most closely resembles the unguals of Greererpeton (Godfrey, 1989).

Narrow body scales with longitudinal grooves and ridges very similar to those of Crassigyrinus specimens (Panchen, 1985) have been found: 2017.2.107, 2017.2.175, and 2017.2.144 (Figs. 4F, 4G and 4H).

Tetrapods: small elements

Alongside the large elements are a few moderate sized and much smaller bones, showing the presence either of small-bodied individuals or juveniles. Delicate ribs, limb bones, and small phalanges have been found.

Skull and jaw bones: An almost complete jugal (2016.13 Clack et al., 2016, fig. 6c) is exposed in internal view. The postorbital region is relatively long, the orbit margin moderately deep, and the preorbital region expands slightly anteriorly. The internal view shows no evidence of a lateral line. In terms of its orbital component it falls somewhere between the shallow orbit of Ossirarus and the deep orbit with a narrow suborbital bar in Diploradus (Clack et al., 2016), and more closely resembles that of Whatcheeria in proportion (Lombard & Bolt, 1995; Daeschler, Clack & Shubin, 2009) although this jugal is from a much smaller individual. Whatcheeriid-like remains have been found elsewhere in the Burnmouth sequence (Clack et al., 2016; Otoo, 2015; Otoo et al., 2018).

Two small partial jaws are interpreted as tetrapod on the basis of tooth shape and surface ornament. Neither the smaller, 2017.2.117 (not figured) nor the larger, 2017.2.443 (Fig. 5G) is identifiable to a higher taxon, but with other smaller elements, they show the presence of small individuals.

Postcranial elements: A small tetrapod ulna 2017.2.51 (Figs. 5C and 5E) is preserved in two counterpart pieces that have been micro-CT scanned. It shows a moderate olecranon process and a thick cortex at mid-shaft.

Small, and some very small, ribs have been identified. A small almost parallel-sided rib 2017.2.476 (Fig. 5A) has an incipient uncinate process. Ribs 2017.2.149 (Fig. 5D) and 2017.2.170 are extremely slender and curved. 2017.2.170 (Fig. 5B) is more strongly curved at its thicker and presumed proximal end, which is also grooved and has a slight sigmoid bend at its distal end. The rib head is not preserved.

Lungfish Figs. 6–10

Skull bones: Conjoined J- and I- skull bones 2017.2.405 (Fig. 6A) have rugose ornament of pits and ridges externally, but the strongly adhering matrix does not allow the lateral lines to be traced. The I-bone has a posterior process, which would have articulated with an anocleithrum. Internally, the centers of growth can be seen from the direction of the striations. The sutures between the bones are almost fused and difficult to trace in places. These bones would represent the skull of a very large individual.

A large robust skull bone 2017.2.550 (Fig. 6H) is visible mainly in internal view. The bone is complete except for wedge-shaped gaps at two places in the margin. Regions of sutural overlap surface mark the edges indicating the contact with at least four different bones. The possible course of a lateral line is indicated at one end by a groove and for the rest of its path, a low smooth ridge. Its shape is comparable with the KL-bone of some Ctenodus specimens (Sharp & Clack, 2013). The small area of exposed external surface shows low irregular ornament.

A moderately large bone (2018.1.2 Fig. 6I) has an irregular hexagonal shape with sutural overlap areas on two sides, and an unsutured concave margin on another. It is identified as lungfish on the lack of dermal ornamentation. A probable lungfish skull bone 2017.2.722 (Fig. 6B) has a smooth surface and lateral line pits. It is a long oval with an overlap surface partly broken off. A rhizodont scale is adjacent. It is not possible to identify the skull bone precisely within the skull. Skull bone 2011.9.10 (Fig. 6E) is an irregular hexagon in internal view, with a more or less smooth surface, whereas skull bone 2017.2.189i (Fig. 6C) is an oval bone with a plane surface, no strong ornament, but marked with small pits. Numerous small skull bones are preserved, mainly hexagonal or somewhat elongated with smooth surfaces and simple straight or curved margins, exemplified by 2018.2.193 (Fig. 6F).

Bones 2011.9.7 (Fig. 6D), 2017.2.68, 2017.2.53, and 2017.2.547 (latter three not figured) are all probably circumorbital bones, the first three from a large individual. 2011.9.7 shows lateral line pores on the surface and has been micro-CT scanned (Fig. 6G) to reveal the course of the branches through the bone. All three bones are extremely robust. 2017.2.547 is more slender, elongated, and curved, showing two lateral line pores. Opercular series: Two very large opercula, 2017.2.188 (Fig. 7A) and 2017.2.667 (Fig. 7B) roughly similar in size, both appear to represent the same side of an animal, so represent at least two large individuals. The more complete 2017.2.188 is hexagonal in shape, with a small dorsally placed projection. Specimen 2017.2.667 was collected as an in situ block positioned further seaward in the bed than most of the other specimens, in a part of the bed that was subsequently completely removed by erosion. Its matrix is more coherent with fewer inclusions than seen in the more cliff-ward section, and may indicate one margin of the fossiliferous channel-fill. Potentially, the J + I-bones and the circumorbitals are from an individual of a similar size to those relating to the opercula. Smaller examples include 2018.1.5 (Fig. 7D) and 2011.9.6 (Fig.7C). The former is more or less circular, whereas the latter is an elongate hexagon in shape. Alternatively, that may be a suboperculum. A very small example 2017.2.720 (Fig. 7E) has a slightly striated surface.

A large suboperculum 2018.1.7 (Fig. 7F) is of a size commensurate with those of the large operculars and 2018.1.5 (Fig. 7G), and 2017.2.543 (not figured) are parts of bones that may be either subopercula or submandibulars.

Shoulder girdle: Several elements of lungfish shoulder girdles are present, although only the anocleithrum is well represented. The best preserved is 2017.2.57 (Fig. 8D), a large example. It is a flattish bone bearing two obvious processes, but is otherwise featureless. The broader of the two processes is assumed to be that which contacted the I-bone. The other has not been entirely exposed but is narrower and more tapered. The presumed external surface onto which the cleithrum would have fitted is not visible. It could have belonged to similarly-sized individuals as did the large opercula. Other smaller and incomplete examples are in the collection.

Partial clavicles and cleithra include 2017.2.70 (Fig. 8A), a partial clavicle and 2017.2.546 (not figured), the distal end of a clavicle. The clavicle is curved in cross section and probably represents the lower part of the bone, lacking the ventral and distal sections. Specimen 2017.2.546 (not figured) shows the rugose sutural surface for the cleithrum, with the external face showing low pustulate ornament. Alternatively, it may be the ventral end of the cleithrum. A small cleithrum is represented by 2017.2.494a (Fig. 8B).

A possible centrum element is 2017.2.398 (Fig. 8C). This bone consists of a shallow curved part that bears a grooved and ridged flange at right angles to it. A similar bone was identified by the late Mahala Andrews (T.R.S. Andrews, 1979, personal communication) from the Serpukhovian of the Dora Bone Bed near Cowdenbeath (see Smithson, 1985, for details of the excavation there). It does not resemble any element preserved on the almost complete postcranial skeleton of Ctenodus interruptus (Sharp & Clack, 2013), so its identity remains in doubt.

Tooth plates: Five lungfish tooth plates are present. These include two incomplete specimens representing tooth plates approximately 24 mm long and three much smaller tooth plates up to 3.5 mm long. One of the largest specimens, 2014.1.2 (Fig. 9E) was referred to Ballagadus rossi (Smithson, Richards & Clack, 2015). It consists of a complete tooth ridge 1 and partial tooth ridges 2 and 3 from a right pterygoid tooth plate. Ridge 1 is 24 mm long and bears eight laterally compressed teeth. The six oldest teeth are worn to a narrow blade. Teeth seven and eight show little wear, are curved slightly anteriorly and separated from the older teeth by a distinct gap. The teeth on ridges 2 and 3 are longitudinally compressed. The second of the larger incomplete specimens 2017.2.54 (Fig. 9A) consists of three incomplete rows of teeth representing parts of ridges 1, 2, and 3. The teeth are similar in size and shape to those on 2014.1.2 with anteriorly curved, laterally compressed teeth on ridge 1 and longitudinally compressed teeth on ridges 2 and 3. The teeth are relatively large and probably formed part of the labial edge of the plate.

The three smaller tooth plates (Figs. 9B, 9C and 9D) are of similar size and morphology. The most complete is 2017.2.513 (Fig. 9D). It is approximately triangular-shaped, 3.5 mm long and 2.5 mm wide with a length to width ratio of 1.4:1. There are four tooth ridges radiating from a point immediately behind the posterior end of tooth ridge 1. The tooth ridge angle is approximately 80°. Ridge 1 has four laterally compressed teeth, ridges 2 and 3 have four longitudinally compressed teeth and ridge 4 has three. 2107.2.67 (Fig. 9C) is missing many of the labial teeth but has five ridges and the tooth ridge angle is approximately 110°. These smaller tooth plates are similar to those of Ballagadus caustrimi (Smithson, Richards & Clack, 2015) which have been found in two older beds at Burnmouth (Bed 333 and Bed 340.5).

Rhizodonts Figs. 10–15

Skull roof: Postparietals include 2017.2.406 (Fig. 10B), which has part of its medial edge broken but is otherwise similar in shape to that of Strepsodus (Andrews, 1985; Jeffery, 1999). It has the typical pustulate ornament externally as found in Strepsodus in this and other beds at Burnmouth. Others include 2017.2.186 (Fig. 10A), a postparietal with overall fine pustulate texture externally, but with irregular grooved vermiform channels in which it is unlike that of 2017.2.406 or that of Strepsodus or Screbinodus. A probable partial postparietal, 2017.2.52 (Fig. 10E) has similar surface texture to 2017.2.186, particularly on the longer, left side. The internal surface shows a raised ridge. It is broader and shorter than that in 2017.2.406. A very small postparietal, 2017.2.643 (Fig. 10D), shows similar vermiform ornament invading a pustulate base, as 2017.2.186 and 2017.2.52, suggesting that this form of ornament is not the result of erosion or a feature of older individuals. It is narrow with a straight edge, presumed to be the medial edge, forming the midline. The lateral edge is somewhat broken, but may be a juvenile of that seen in Fig. 10D. An extrascapula, 2017.2.94 (Fig. 10C), has tuberculate ornament which is eroded away on the right-hand side.

The stout hyomandibular bone, 2017.2.338 (Fig. 10F), has a proximal articulation at an angle of about 45° to the long axis of the bone. It appears relatively short compared to that of a tristichopterid such as Eusthenopteron (Jarvik, 1980) and there is no sign of a foramen for the chorda tympani. Although compressed, the bone shows a prominent longitudinal ridge down one side, and a groove, not quite opposite it, on the other. It is most similar to that of GLAHM (Glasgow Hunterian Museum) V8038 illustrated by Jeffery (1999) except that in the latter the proximal end appears incomplete and its distal end wider.

Palate: A partial pterygoid, 2011.9.2 (Fig. 11), is one of the few known rhizodont pterygoids, which scarcely appear in the literature. Young, Long & Ritchie (1992, figs 30 and 31) described what they considered to be the palatoquadrate and part of the lower jaw of the rhizodont Notorhizodon but Johanson & Ahlberg (2001) concluded that these elements probably belonged to a tristichopterid. However, 2011.9.2 bears ornament strikingly similar to that carefully illustrated in Young, Long & Ritchie’s (1992, figure 31). It consists of fine denticulations that morph gradually into vermiform grooves. There is an abrupt transition into a smooth overlap surface, exactly as depicted by Young, Long & Ritchie (1992).

A problem arises in that tristichopterids are not known to occur in the Carboniferous, having become extinct at the end of the Devonian. Our specimen is thus unlikely to be a tristichopterid. An alternative explanation is that the awkward fit between parts of the jaw and palate described by Young, Long & Ritchie (1992), and that was noted by Johanson & Ahlberg (2001), result from the parts belonging to different individuals rather than different taxa.

The bone described here is possibly much distorted. One end curves out of the plane of the denticulated surface and is smooth underneath. It appears to represent an articular surface, presumably part of the jaw joint. The surface then curves away from an edge rimmed by small teeth (unfortunately damaged during preparation and restored photographically from before and after photographs in Fig. 11D). Along the side opposite to the denticulated surface, the bone is folded downward to a ledge, presumably also an overlap surface, and the row of small teeth runs along the edge of the fold. Comparison with the Young, Long & Ritchie’s (1992) figure 31 shows that contrary to that bone, the articular end here is continuous with the denticulated ornament, whereas in the Notorhizdon bone, it is separated from the ornamented area by the smooth overlap surface. The relative positions of the row of small teeth and the denticulated region in 2011.9.2 is more similar to that of the palatoquadrate of Notorhizodon, although the entire extent of the more medial parts of the bone are missing in that specimen. There would have been some contact with the parasphenoid mesially, possibly though a quite extensive smooth overlap area, and the ledge beneath the row of small teeth would be a contact for the marginal palatal bones.

Other cranial elements: Mandibular bones include 2017.2.477 (Fig. 12A), an elongate almost rectangular bone with pustular ornamentation best identified as an infradentary; 2017.2.404 (Fig. 11B) a tooth-bearing bone, probably a small, incomplete dentary with a symphysial tusk; and 2017.2.64 (Fig. 12E), an incomplete gular plate with pustular ornamentation.

Teeth are not usually well preserved in this bed, but a few isolated examples show characteristic features. Tooth 2017.2.99 (Fig. 12D) lacks its tip but shows the folded enamel at the base. The surface is damaged, but fine almost parallel striations are visible further away from the base. Tooth 2017.2.72b (Fig. 12C) has a complete tip, which shows no recurvature like that of Strepsodus sauroides (Jeffery, 2006). The striations more closely resemble those of Archichthys, but the tooth itself is more slender than those of the holotype series. Some of the external enamel is damaged, particularly along the convex (labial) surface, so that it is difficult to say that it lacked striations as in that genus. The concave (lingual) surface shows fine parallel striations similar to those of 2017.2.99 and Archichthys (Jeffery, 2006).

Shoulder girdle elements: Cleithra (Fig. 13) are among the elements more commonly recovered in the collection. A large example of a left cleithrum 2011.9.3 (Fig. 13C), is incomplete dorsally and ventrally. The external surface shows coarse meandering and anastomosing ridges on the lower blade and posterior edge, but is plain elsewhere. There is a deep groove along the posterior margin and up the middle of the mesial surface of the blade. The bone conforms to those of Strepsodus spp. and Strepsodus sauroides (Andrews, 1985; Jeffery, 1999; Parker, Warren & Johanson, 2005).

An exceptionally large example, 2017.2.716 (Fig. 13A), in which the matrix has not been completely removed from the external surface, where exposed shows similar ornament to 2011.9.3. A grooved and ridged cleithrum tip (not figured) collected at the same time and close by does not fit onto the main specimen but was likely part of it. The smallest cleithrum found in this bed is 2018.1.4 (Fig. 13B). It preserves only the upper part of the blade. Two incomplete cleithra, 2017.2.399 (Fig. 13D) and 2017.2.400 (see below), show evidence of erosion by rolling during transport. They are similar to 2011.9.3 but are smaller.

Anocleithra: Two specimens are identified as rhizodont anocleithra, 2017.2.385 (Fig. 14B) and 2017.2.386 (Fig. 14A), by comparison both with those illustrated by Parker, Warren & Johanson (2005) for Strepsodus from Ducabrook, Australia, and an unidentified rhizodont from the Dora Bone Bed, Cowdenbeath, Scotland. The two Burnmouth specimens differ in shape, although the cleithral flange in the smaller example 2017.2.386 and the dorsal flange in the larger example 2017.2.385 are both broken. Both have an anteroventral process, although that in 2017.2.386 is robust and rounded, whereas that in 2017.2.385 is pointed and more slender. Each has a clear overlap region where the cleithrum would fit, but in 2017.2.386 it is defined by a smooth rounded buttress, whereas in 2017.2.385 it is defined by a prominent ridge. Both are probably from the left side of the animal, but their cleithral flanges have distinct processes in differing positions. The lower margins of the cleithral flanges in the described examples are highly variable. It is unlikely that these two belong to the same taxon.

Lepidotrichia and scales: Both large and small lepidotrichia are common in the bed. Fin spine 2017.2.494b, with about 75 mm preserved, shows a longitudinal groove, and tapers toward its distal end (Fig. 15F). Fin spine 2017.2.568 (Fig. 15E) is circular in cross section along the approximately 80 mm of its preserved length. It does not taper, but bears faint longitudinal striations. Cross sections show a thick cortex with spongy medullary bone. It was originally identified as a plant stem. The best-preserved examples of fin spines are probably attributable to rhizodonts, but may also pertain to lungfish. Rhizodont scales of many sizes are present, identified by the characteristic boss near the center of ossification (Figs. 15A–15D) (Andrews, 1985).

Gyracanths and other chondrichthyan elements Figs. 16 and 17

Gyracanth spines are very common in this bed, although few are well preserved. Some eroded examples were to be seen in the sandstone wall above this conglomerate bed prior to the rock fall of 2018 (Figs. 16E and 16F). The most complete specimen collected is 2017.2.182, a pelvic spine measuring 40 cm. The striations are narrow, close together, and bear closely spaced tubercles. In other specimens such as 2017.2.185, an incomplete example, the ornamentation appears coarser than in 2017.2.182. There are several other examples like this. Large scapulocoracoid and procoracoid plates include 2011.9.11 and 2017.2.189b (Figs. 16A and 16B).

Agelodus teeth are occasionally found. The crown of Ageleodous tooth 2017.2.282 (Fig. 17A) originally had eight cusps, but one is missing. Cusps sit on a raised area, with the “root” depressed distally (see also the micropaleontology section for a different type of tooth).

Onychoselache-like spines are represented by 2017.2.397 (Fig. 17B), a small spine showing approximately longitudinal grooves and ridges, and 2017.2.51 (Fig. 17C) a smaller example.

Actinopterygian remains Fig. 18

Actinopterygian scales are not infrequent in this bed (see below), but are often fragmentary. More complete examples include 2017.2.200 and 2018.1.3 (Figs. 18A and 18B). Isolated bones, though rare, may include 2018.1.1 (Fig. 18C).

Coprolites Fig. 18

A few coprolites such as 2017.2.58 (Fig. 18D) have been found but cannot be associated with any particular taxon.

Elements of uncertain identity Fig. 19

A small tooth-shaped triangular element 2017.2.49 (Fig. 19A), probably flat on the unseen surface, has a rounded profile toward the tapered end, but the wider end is characterized by a raised horseshoe-shaped rim with fluting, grooves, and ridges. This bounds a depressed area. The wider end is broken, and this bone may be a small part of a larger and more complex one.

Element 2017.2.327 (Fig. 19B), appears to be from a rhizodont. It bears ornamentation similar to skull bones, including some vermiform grooves in a small area. One margin, the left side in the figure, is strongly curved downward. The other is broken. The ornamented area terminates in a distinct margin. Extending from this is an unornamented region with a smooth convex outline. It is marked on this surface by a deeply depressed area toward the side opposite the curved margin. On the underside, the whole surface is smooth, but there is a buttress beneath the curved margin that bounds a depressed and grooved area. There is no obvious distinction on the internal surface between where the ornamented segment of the external surface of the bone ends and the smooth area begins.

Suggestions for the identity of this bone include part of the shoulder girdle, a quadratojugal with attached quadrate, a surangular with attached articular, and a pterygoid with attached epipterygoid. We welcome other suggestions.

Abraded elements and plant macrofossils Fig. 20

Rolled bones such as 2017.2.214 of uncertain identity (Fig. 20B), appears smooth and featureless. It is convex in profile and slightly C-shaped. It may be part of a rib that has been rolled during transport. Other rolled and eroded bones include the two rhizodont cleithra listed above (Figs. 13D and 20A).

Plant macrofossils are common with some large elements represented (Fig. 20C), suggesting that large plants existed during that part of the Tournaisian. However, they are usually poorly preserved such that identification is not usually possible. Some appear to be lycopsid. They are often coated with iron oxide, as are some of the bones (see below).

Microfossil assemblage Figs. 21 and 22

In this study, we compared the microfossil assemblage to the macrofossil assemblage in Bed 383, confining the identity of specimens to major groups. The micropaleontological assemblage (SI1) is almost entirely composed of vertebrate material (5,660 out of 5,694 specimens), with some charcoal fragments (34 specimens). Microfossils are generally well preserved, although some abrasion, wear surfaces and cracks are observed. For vertebrates, 67.9% of specimens recovered have been identified. Rhizodont and actinopterygian material dominates the assemblage, with lesser occurrences of chondrichthyan and dipnoan microfossils (Fig. 21). Charcoal is identified by its black color, brittle texture, fibrous external structure and hollow internal structure of preserved cellular tissue (see below and Scott, 2010).

Rhizodont scales have an exterior surface which is cream colored, shiny and has a fibrous structure. When broken, the interior layers have a range of characteristic structural elements including sheets of either tubercules and pits or grooves and ridges that interlock together (Figs. 22A and 22B). These structures were identified by the examination of broken rhizodont scale macrofossils from this bed. A total of 13.9% of the rhizodont scales have a red staining on the interior surfaces, likely due to the presence of iron oxide. Other rhizodont elements include rhizodont dermal bone (Fig. 22C) with a distinctive ornament of tubercules or ridges (cf. Jeffery, 2012) and three indeterminate rhizodont teeth (Fig. 22G), which have a shape similar to that of Strepsodus, but an ornament of well-defined striae similar to that of Archichthys (Jeffery, 2006).

Actinopterygian scales have a rhombic shape with a smooth interior surface with keel, and a shiny, sculpted exterior surface (Fig. 22D) with transverse ridges or grooves (cf. Carpenter et al., 2014). Actinopterygian dermal bones and lepidotrichia (Fig. 22E) are also identified. Some of the dermal bone is similar to that of rhizodonts, with pustulate ornament on one side, but the fragments are small and they have a shiny, ganoine surface texture (Fig. 22F). The lepidotrichia are most common in the 250 μm size fraction, so are more likely to be actinopterygian than rhizodont, although there are no physical characteristics to distinguish them other than size. They have a range of surface textures, from smooth to longitudinal striations or ridges. The interior medullary bone of the fine spines is infilled with calcite in some specimens. Actinopterygian teeth are present in all apart from the 1,000 μm size fraction and are distinguished by their transparent apical cap (Fig. 22H) and shaft with cross-hatched surface ornamentation (Carpenter et al., 2011). Pharyngeal actinopterygian teeth are unornamented, curved, blunt at the tip, and occur in rows (Fig. 22I). In total 30 main teeth and 92 pharyngeal teeth are recorded.

Chondrichthyan teeth comprise three Ageleodus teeth with five to six tooth cusps (Fig. 22J) and one xenacanth (indeterminate) tooth (partly broken). Chondrichthyan material is dominated by denticles (140 specimens), which are common in the 1,000 and 425 μm fractions and are mostly present as fragments in the 250 and 125 μm fractions. Three types of scales have been identified; elasmobranch (n = 43), hybodont (n = 28) and ?ctenacanth (n = 5). Indeterminate elasmobranch scales are identified by a flat or concave base and a top of curved spines (Fig. 22K), which in dorsal view form clusters of irregular height, or individual spines (Burrow, Long & Trinajstic, 2009; Carpenter et al., 2011). Hybodont scales have a concave base, spinose top and distinctive grouping of spines (Fig. 22L) which forms a single flat star-shape, or multiple star-shaped clusters in dorsal view (Garvey & Turner, 2006; Yazdi & Turner, 2000). Scales which can be putatively assigned to a ctenacanth origin have a flat base, spinose top and similar appearance to contemporaneous specimens described by Ivanov (1996) and Yazdi & Turner (2000). The spines are strongly curved, numerous and are of irregular height (Fig. 22M).

Tetrapod microfossils have not been recorded from this bed and only one dipnoan element is identified (a toothplate fragment). This may result from problems in identification, given they are more common as macrofossils. Further direct comparative work on broken dipnoan and tetrapod macrofossils is necessary to determine their microfossil characteristics.

Charcoal description and identification Figs. 23–26

The charcoal fragments in Bed 383 range in size from a few millimeters to a centimeter in maximum dimensions (Fig. 23). All are woody, comprising both secondary xylem and ray cell tissue. All specimens show evidence of pitting in the tracheid cell walls that are often circular pits arranged in single or in several rows (Figs. 23B and 23D). The cells also show evidence of homogenization (Fig. 23B) (Scott, 2010). It is often considered that secondary xylem (wood), that is, dense or pycnoxylic is the most common feature of the arborescent pteridosperms (Galtier & Meyer-Berthaud, 2006) (Table 1). The tracheids are often less than 50 μm wide and have short uniseriate rays, but this is not easily visible. There is insufficient detail in the specimens found so far in Bed 383 to provide definitive identifications.

Charcoal fragments from Bed 362 (2018.4.13) are expected to preserve similar taxa to those of Bed 383, but they have a greater range of size, many specimens ranging from 10 to 20 mm in maximum length (Figs. 24 and 25). In addition, because of their larger size they show a wider range of characteristics including relatively small xylem cell sizes (Figs. 24B, 24D, 25C and 25D) but large numbers of wide rays (Fig. 24B). However, the specimens in Figs. 25A, 25B and 26C expose wide surfaces of longitudinal radial sections split at the level of the “soft” ray tissue, demonstrating that rays were very high (i.e., tens of cells high). They may exhibit several types of wall pitting in the xylem as well as in the ray cells (Figs. 25B and 25D). The pitting may comprise several rows of pits (Fig. 25D) and occasionally bordered pits (Fig. 26C). In one specimen (Fig. 24) there appears to be the presence of leaf traces (Fig. 24B) and also regular rays separating a few rows of tracheids (Figs. 24A and 24B). Cell wall homogenization is also seen (Fig. 26B).

Most of the charcoal appears to have come from small woody axes and possible larger wood from a range of arborescent pteridosperms (Galtier & Meyer-Berthaud, 2006). Such pteridosperms are widely reported throughout the Mississippian of Scotland (Scott, Galtier & Clayton, 1984; Scott & Galtier, 1988, 1996; Galtier & Meyer-Berthaud, 2006). Most common taxa that may be represented at Burnmouth include the large tree Pitus, that is, also common in the Borders (Long, 1979) and the smaller Eristophyton (Scott & Galtier, 1996; Galtier & Scott, 1990). Other taxa reported from Tournaisian rocks in the Midland Valley (Galtier & Meyer-Berthaud, 2006) include Bilignea, Endoxylon, and Buteoxylon, although the size of the charcoalified specimens do not allow the preservation of the combination of characters that will allow their certain identification.

Some of the specimens are of pycnoxylic wood, that is, compact with little parenchyma between the rows of tracheids (Fig. 26A). Dadoxylon wood may also be a candidate as they have uni- to biseriate rays. Calamopitys, in contrast, has been characterized as having manoxylic wood in the stems (i.e, abundant parenchyma cells between rows of tracheids) and a thick cortex with a range of cell sizes. The genus is more shrub-like with stems that are only two to three centimeters in diameter. A related form Stenomyelon, however, has been reported from Scotland (Scott & Galtier, 1988). All the specimens we have seen have a nonmanoxylic wood and cannot be referred to calamopityans. In Pitus (Pitys) the tracheids of the secondary wood typically have pits arranged in multiseriate rows (Gordon, 1935) and the trunks of this plant can be very large and may provide significant quantities of wood that are available to produce charcoal. Characters differ, however, between species of this genus (Galtier & Meyer-Berthaud, 2006).

In many cases small fragments of wood charcoal are difficult to identify as the tracheid dimensions and pitting types occur in many different taxa. What is necessary are the ray size measurements and ray organization (number of cells wide and number of cells high) and this is only possible if, in addition to a good transverse section, radial and tangential longitudinal sections are available. It is possible from the limited information available that the specimen in Fig. 24 has broad rays, but since there is no evidence that they are very high, they may be comparable to Eristophyton waltonii (cf. Galtier & Meyer-Berthaud, 2006, fig. 5E). Specimens in Figs. 25 and 26 (possibly 26 but without ray data) have very high rays but no evidence they were broad (i.e., more than three to four cells wide) and therefore may be comparable to Pitus antiqua (Galtier & Meyer-Berthaud, 2006, fig. 5H).

Macrofossil assemblage

The partial Crassigyrinus-like lower jaw, discovered in 2010 (UMZC 2011.9.1), from Bed 383, is about 20 Myr older than C. scoticus, and although attribution to the same genus is not unreasonable, it is unlikely to represent the same species (Clack, Porro & Bennett, 2018). Also found in 2010 were a partial parasphenoid and a basal process, not identified as Crassigyrinus-like at the time. These braincase elements indicate an animal at least 50% larger than C. scoticus.

Earlier-collected material includes central elements: one complete centrum, one half of a centrum, and one fragment. Phalanges or metapodials of a large tetrapod are also preserved, though their attribution to Crassigyrinus is less secure. Subsequent discoveries include ribs and body scales. Found in close proximity, probably in less than one square cubic meter of the 15–20 cm thick matrix, it is possible that many of these elements belonged to the same individual. The fact that few other elements have been found more recently suggests that most of the animal’s remains had been eroded away before collecting began.

Smaller elements among the macrofossils include a jugal (Clack et al., 2016, fig. 6), ribs with uncinate processes and an ulna. These much smaller and moderate sized bones show the presence either of small-bodied individuals or juveniles. Also included are delicate ribs, limb bones, and small phalanges. Their shapes are quite different from those in C. scoticus and furthermore, probably do not belong to a juvenile of the same taxon as the lower jaw.

As with the tetrapods at this bed, lungfish elements indicate a wide range of sizes of individuals, from tiny juveniles or small-bodied taxa up to large adults. At least two very large individuals are represented by opercula within this relatively small accumulation. To our knowledge, these represent the largest lungfishes found in the Carboniferous. A rough guide to the length of the fish can be obtained by a ratio of 15:1 from the diameter of the operculum (T.R. Smithson, 2017, personal observation). By this measure, the large opercula belonged to fish up to three meters in length. Those from the later Carboniferous found so far are no more than about half the size as judged by this ratio. This finding, as with some of the tetrapod, rhizodont, and gyracanth elements, is in contrast to the trends in vertebrate size documented by Sallan & Gallimberti (2015), who suggested a reduction in maximum sizes during the Tournaisian, followed by very slow recovery of larger forms later in the Carboniferous.

By contrast, no large tooth plates have been found, but several very small ones are present, including probable representatives of two species of Ballagadus (Smithson, Richards & Clack, 2015). Skull bones with varying surface ornamentation or none suggest that more than one taxon is present in the assemblage. Large elements with ornamented external surfaces and deeply interdigitated sutural surfaces contrast with small often oval or hexagonal elements with smooth or pitted surfaces and no obvious sutural overlap areas. Whether these differences could be accounted for by ontogeny is not clear.

Rhizodonts are represented mainly by bones from large or medium-sized individuals, although small scales, a small cleithrum and a small postparietal have been recovered. Although many specimens resemble bones of Strepsodus it is possible that two taxa of rhizodont were represented in Bed 383. The very large cleithrum suggests a fish of a size comparable with those of the large lungfish, again in contrast to Sallan & Gallimberti (2015).

There is some question of the taxonomic status of the rhizodonts from Burnmouth. Otoo (2015) and Otoo et al. (2018) suggest that two taxa might be present in a somewhat older assemblage from Burnmouth, possibly Strepsodus and Archichthys. This requires further investigation, but the anocleithra and the postparietals (Figs. 1A, 1B, 1E, 14A and 14B) do suggest the presence of two taxa in Bed 383. An isolated humerus from a lower bed (Bed 340.5 see also Otoo et al., 2018) under study by JAC and coauthors is quite different from those figured for Strepsodus.

Gyracanth spines are among the most common elements in this bed, though they are difficult to collect and prepare and are often broken. Some elements are exceptionally large, in keeping with the large elements of tetrapods, lungfishes, and rhizodonts. Similar spines, though not as long as 2017.2.182 were found in the base of the overlying sandstone (probably those noted in McAdam, Clarkson & Stone (1992), although they had been subject to erosion on the surface of the exposed wall. They were lost in the rock fall.

Rare chondrichthyan remains suggest their presence in the region at the time, and are better represented in the microfossil assemblage (see below).

One of the issues encountered in this investigation is the difficulty of identifying isolated elements. This is particularly the case with possible rhizodont elements that are known from very few other taxa. Where possible we therefore illustrate here both internal and external surfaces of individual bones. Internal surfaces of skull bones are often not accessible except by micro-CT scanning, so that it is important to illustrate both where circumstances allow. Our rhizodont skull bones also show unusual surface features of the ornament such as the vermiform grooves illustrated above. Similar grooving is known on some lungfish skulls, such as Ctenodus interruptus (Watson & Gill, 1923; Sharp & Clack, 2013) but it is not combined with an otherwise pustular surface. The ornament type in the postparietals illustrated here was confirmed as rhizodont by J. Jeffery (2018) from his personal observations, but has never been illustrated in the literature.

Unknown morphologies have also been uncovered. One of our specimens is a particular puzzle and does not fit easily into known categories, and a second, a rhizodont pterygoid, could challenge information from the published literature. These discoveries point out how little of the earliest Carboniferous is known in terms of vertebrate fossils, and isolated elements emphasize the level of our ignorance.

Microfossil assemblage

Given their limited representation in the macrofossil assemblage, an unexpected actinopterygian abundance is revealed in the micropaleontological sample. In total, actinopterygian microfossils comprise 55.3 specimens per gram, while rhizodont microfossils only 11.5 per gram (SI1). Actinopterygian microfossils are more common than rhizodont microfossils in the 250 and 125 μm size fractions (Fig. 21). This would imply that actinopterygians are more abundant than rhizodonts, surprising given their difference in size—rhizodonts are approximately 10 times larger than actinopterygians, and so could be expected to produce a lot more microfossil fragments. The greater fragility and breakage of actinopterygian scales compared to rhizodont scales counters this interpretation, along with the taphonomic concentration or winnowing of a certain size of material during transport. However, the teeth of actinopterygians and rhizodonts recovered in this study are of a similar size and there are approximately ten times the number of actinopterygian teeth (excluding pharyngeal teeth) than rhizodont teeth. This indicates that the relative abundance of actinopterygians was probably higher than that of rhizodonts.

Interpretation and environment of Bed 383

The six packages are interpreted as channelized river deposits, the channel forms being clear from the aerial views. Packages 1–3 are laterally adjacent channels and each one is successively younger, potentially representing shallow river channels laterally migrating through time (Fig. 2A). The dolostones overlying these three packages indicate a time during which a standing body of water developed, probably a lake (Bennett et al., 2016). Package 4 represents the return of a river system to this area, with the associated erosional relief suggesting a more significant system. Sediment deposited in this channel was partially eroded, but package 5 may represent a phase of channel abandonment and the deposition of fine-grained sediments. The youngest, package 6, has the greatest erosional relief. Only one side of the channel is exposed, therefore, the large-scale stratification could represent bar progradation or lateral migration of the channel itself.

The localized conglomerate probably represents deposition in the deepest part of this channel (package 6) through bedload processes. The sedimentology of the conglomerate lag (with clasts of red and gray siltstone) indicate that floodplain sediments were incorporated, including clasts of sandy siltstones (Bennett et al., 2016) and paleosols (Kearsey et al., 2016). Erosion by these river systems of the surrounding vegetated floodplains is clear from the erosional relief observed at this location. Several of the specimens show evidence of abrasion from rolling, and most are chaotically distributed in the matrix. The iron oxide staining of many rhizodont scales indicates that some rhizodont carcases may have been subject to subaerial exposure. Staining may have occurred during the desiccation of a floodplain lake and oxidation of fossil material, prior to the incorporation of these sediments into the river system.

The presence of ichnofauna, rare scolecodonts, orthocones, and evaporites (gypsum and anhydrite) within the Ballagan Formation provide evidence that the floodplain was subject to occasional marine water input (Bennett et al., 2016; Millward et al., 2018a, 2018b). The Chondrites in dolostones below Bed 383 has been interpreted as short-lived colonization of coastal lake sediments by opportunistic organisms that were transported from shallow marine environments during storm events (Bennett et al., 2016). These Chondrites-bioturbated dolostones suggest the development of a significant floodplain lake, prior to the fluvial systems (packages 4–6) being reestablished. This is consistent with the complex coastal to alluvial paleoenvironment that characterizes the Ballagan Formation. Dolostone and evaporite clasts have not been incorporated into the conglomerate. Xenacanths are often associated with freshwater deposits, and Carboniferous elasmobranchs, ctenacanths, and Ageleodus are also interpreted as having a euryhaline salinity tolerance and occur in a range of marine to continental environments (Carpenter et al., 2014). Elasmobranchs are more commonly associated with freshwater sedimentary deposits than contemporaneous holocephalan chondrichthyans (Friedman & Sallan, 2012).

The presence of plant and charcoal debris confirm that vegetated floodplains were eroded and were subject to fire. However, other evidence for subaerial floodplain ecosystems, for example, myriapods (Ross et al., 2018), are absent. The bedload transport process is likely to have broken and winnowed away arthropod cuticle and the fragile calcite shells of ostracods and bivalves, which are more common within floodplain facies (Bennett et al., 2016). The scarcity of tetrapod and dipnoan microfossil material may again reflect taphonomic processes related to the transport, concentration and deposition of fossil material. The presence of charcoal may indicate forest fires that increased the potential for erosion of the floodplain.

Burnmouth is a locality well known for the occurrence of anatomically preserved plants that were described in several papers by Albert Long (see Scott, Galtier & Clayton, 1984). The majority of the plants described were from dolostone facies of presumed late Tournaisian age obtained from loose pebbles. Scott, Galtier & Clayton (1984) reported in situ plants from nearby localities such as the shore section at Partanhall, Burnmouth (NT959614) and Scott & Rex (1987) illustrated a permineralized Stauropteris berwickensis in a silty-wackstone horizon that also contained “Lepidodendron” calamopsoides axes together with fish bones and ostracods. The flora described from loose blocks included a range of lycopsids, pteridosperms, and ferns. A similar flora was described in dolostone facies from the River Whiteadder, for example, at Edrom (Scott, Galtier & Clayton, 1984; Scott & Rex, 1987). In addition to permineralized plants, some preserved as charcoal were also recovered (Scott & Rex, 1987). Compression floras, together with permineralized plants and charcoal were also reported from Foulden, about four miles (6.4 km) from Burnmouth and also of Tournaisian age (Scott & Meyer-Berthaud, 1985; Scott & Rex, 1987; Scott & Galtier, 1996; Scott & Glasspool, 2006).

Fire has been an important influence on the Earth system throughout history (Scott, 2000, 2018). For there to be fires, there needs to be not only fuel to burn, and an ignition source (generally lightning) but also an atmospheric oxygen level of over 17% (Scott, 2010; Scott et al., 2014). The occurrence of charcoal in the sediments of this conglomerate allows two immediate conclusions. The first is that the oxygen level in the atmosphere was at least above 17% corroborating the hypothesis that the Tournaisian stage was not one subject to lower oxygen levels, and may even have been above the modern level of 21% (Glasspool et al., 2015; Rimmer et al., 2015). Secondly, there is evidence of wildfire that affected the floodplains. However, the occurrence of charcoal in the sediments offers a further possibility. If there are rainstorms following a wildfire then the area of the fire can experience postfire erosion (Moody & Martin, 2001; Cannon, 2001). This happens when the roots of the plants have been killed and the sudden influx of water leads to increased erosion and movement of both the sediment and plant material (as well as animal material on the soil surface). This slurry may move very quickly (Scott, 2010) and form almost instantaneous alluvial fans away from the place of the wildfire (Moody & Martin, 2001; Scott et al., 2014). Such sediment may be either be transported into a river or lake system directly (as following the frequent Yellowstone fires) (Meyer & Pierce, 2003; Pierce, Meyer & Jull, 2004) or else the sediments are quickly eroded by fluvial processes such as occurred in the Buffalo Creek fires of 1996 (Moody & Martin, 2001, 2009). Postfire erosion/depositional systems have only recently been recognized in the fossil record. Extensive sediment pulses with charcoal have been described from the Early Carboniferous (Mississippian) of Ireland where there has been a sudden influx of sediment into a marine near-shore environment with consequences for the fauna living there (Nichols & Jones, 1992; Falcon-Lang, 1998). The occurrence of vertebrates that have been caught up in such postfire flood events have also been described in the Triassic (Havlik et al., 2013) and Cretaceous by several authors (Brown, Collinson & Scott, 2013; Muir, Bordy & Prevec, 2015).

The vertebrate remains alongside the charcoal, and the greater depth of erosion observed associated with package 6 suggest that postfire erosion/deposition may have played a role in its formation. Charcoal often floats and takes time to settle in a water column, depending on size, type of organ, and temperature of formation and hence can travel considerable distances by water transport (Nichols et al., 2000; Scott, 2010). Charcoal is often deposited by currents of a certain velocity where fragments are also well sorted (Nichols et al., 2000). The occurrence of charcoal and uncharred plants with siltstone and sand-sized grains alongside larger pebbles (paleosols and sandy siltstone) could suggest a more chaotic flow and rapid deposition for at least this part of the bed. However, the stratification in the channel fill above the conglomerate indicates large-scale bed-forms and sustained period of sandstone deposition. The widespread occurrence of charcoal suggests that fire appears to have been a regular occurrence through the Mississippian of Berwickshire and across the south of Scotland and to have had an impact on both the vegetation and the landscape (Scott, 1988; Brown, Scott & Jones, 1994; Scott & Glasspool, 2006).