Arthropleura, Jordan, 1854

Davies, Neil S., Garwood, Russell J., McMahon, William J., Schneider, Joerg W. & Shillito, Anthony P., 2021, The largest arthropod in Earth history: insights from newly discovered Arthropleura remains (Serpukhovian Stainmore Formation, Northumberland, England), Journal of the Geological Society (jgs 2021 - 115) 179 (3), pp. 1-18 : 6-14

Genus Arthropleura Jordan in Jordan and von Meyer, 1854

Type species Arthropleura armata Jordan, 1854 , plates 13–15, plate 2, figs 4–5 in Jordan and von Meyer, 1854, Arthropleura sp. Jordan in Jordan and von Meyer, 1854

( Figs 1 and 2 )

Material: CAMSM X.50355, partial remains comprising articulated anterior 12–14 tergites in two slabs.

Locality: Howick Bay (55° 27′ 19.2″ N, 01° 35′ 32.4″ W), Northumberland, England.

Age and formation: Early Serpukhovian, Stainmore Formation (Yoredale Group).

Description: specimen identified as the partial anterior dorsal exoskeleton of Arthropleura because of the trilobate tergites, coupled with the large dimensions. Remains consist of 12–14 tergites and paratergites, 76 cm in maximum length from the anterior to posterior, and 36 cm at the greatest width. Preserved as a three-dimensional cuticular infill by sand, with limited cuticular material. Ornamentation limited: some longitudinal striae are visible on the paratergites and there is a granular or verrucose texture on the anterior margins of the medial tergites. The specimen has an irregular morphology as a result of the taphonomy of a large three-dimensional exoskeleton interred within sand in a tectonically active setting.

Description of the specimen

The three-dimensional preservation of this large fossil is summarized in Figure 7 . The fossil is visible on a fracture surface within a block of cross-bedded fine-grained sandstone. The fracture splitting the well-indurated host lithology is recent and presumably formed when the host block fell from the cliff. The fossil is preserved on surfaces either side of this fracture. As the fracture may run through the middle of the three-dimensional fossil (see later discussion), it is inappropriate to refer to these as part and counterpart ( Fig. 1 ) and they are here referred to as slab Aand slab B. Slab Ais the lower stratum and hosts the bulk of the fossil. Slab Bis the upper stratum and preserves an impression that domes downward to a relief of c. 10 cm, creating a three-dimensional semi-cylindrical form.

The fossil consists of 12–14 sub-rectangular medial tergites, flanked on one side by right paratergites. The left paratergites are missing and the medial tergites terminate against a serrated edge. The right paratergites have frayed and irregular lateral margins and so are also imperfectly preserved ( Figs 2 and 8 ). The anterior five to six paratergites are increasingly recurved ( Figs 1 and 2 ).

Slab A broke into several pieces during extraction from the host block; these fragments reveal the three-dimensional form of the tergites. Each is filled with the host sediment, forming threedimensional imbricated pillows. The tergite sand infills are 4 mm thick in the medial tergites, thinning to 1 mm or less towards the paratergites ( Fig. 9 ).

The fossil is underlain in slab A, and overlain in slab B, by a carbonaceous smear that exactly mirrors the form of the frayed right paratergites ( Fig. 10 ). The offset between recognizable frayed paratergites and their displaced form in smears can be measured. The offset is consistent along the length of the fossil at 40 cm. Excluding the smear repetition, the total length of the fossil is 76 cm and its maximum preserved width from the right lateral paratergite to the termination of the left medial tergite is 36 cm.

In addition to this smear, remnants of carbonaceous material can be seen in patches on both slabs, most notably in the tergal margins in slab A ( Fig. 9 ), where the arthrodial membrane would have been present in life. The dorsal side of the tergites in slab Acan be seen, in broken fragments, to be carbon-rich and flecked with abundant mica, which possibly adhered to a sticky surface prior to burial ( Fig. 8 ). The majority of the fossil, however, has no organic material and its form is revealed by impression, or sand infill, alone.

As a result of the lack of well-preserved cuticle and the granular nature of the host sediment, ornamentation is not consistently visible on the specimen. However, the anterior margins of the medial tergites, where the segments would have been joined by a softer arthrodial membrane, have a rough, grainy appearance on the surfaces of both slabs. The absence of this texture across the remainder of the fossil and host sediment implies either that it is an original texture or that it is a taphonomic difference reflecting the contrasting nature of the exoskeleton between these regions ( Fig. 8 ). In addition, striae can be seen on some of the paratergites, some of which appear to form pronounced medial grooves that are parallel to the central axis of a paratergite ( Fig. 8 ).

This information can be combined to describe a stratigraphic transect downwards through the fossil as follows: (1) 10 cm of domed fine-grained sand (slab B); (2) a negligible thickness, offset carbonaceous smear (slab B); (3) an impression fossil of tergites/ paratergites, with grainy and striated surface textures (slab B); (4) remnant patches of organic material and the original form of the tergites, recording verrucose and striated surface textures (slab A); (5) 1–4 mm thickness of sand infill within three-dimensional tergites (slab A); (6) a highly micaceous and carbonaceous veneer on the surface of the tergites (slab A); (7) a negligible thickness, offset carbonaceous smear (slab A); and (8) underlying fine-grained sand (slab A).

Taphonomy of the specimen

The granular sandy host lithology of the specimen is remarkably coarse for preserved arthropleurid remains. All the other articulated remains are known from very fine-grained mudrocks, sandy siltstones or crystal tuffs ( Guthörl 1934, 1935; Hahn et al. 1986; Schneider and Barthel 1997; Schneider et al. 2010), although several isolated remains from the late Visean Hainichen basin in Saxony are well preserved in silty fine-grained sandstones to fine- to medium-grained sandstones ( Röβler and Schneider 1997). The wellsorted, granular nature of the host sediment created taphonomic conditions that were not conduciveto preserving chitinous cuticles in high fidelity ( Briggs et al. 1998) and the fossil is identifiable primarily because the cuticle was filled with sediment post-mortem. The only evidence of the original organic material is some carbonized material between the tergites in slab A ( Fig. 9a ) and the carbonaceous and micaceous material that appears offset and smeared across both the ventral and dorsal extremes of the fossil.

Missing body parts

The Howick specimen preserves only part of the dorsal exoskeleton of the organism, with no evidence of appendages. As is common to all other reports of giant Arthropleura , the head is also missing, but the lack of segmentation anterior to the first sizeable tergite suggests that the fossil may terminate where the head capsule was during life ( Fig. 2 ). No trace of appendage attachment points is present on either side of the specimen preserved in slab A, where both the ventral and dorsal surfaces of the dorsal exoskeleton can be observed. The most plausible explanation for these characteristics is that the specimen is an exuvium, potentially one in which the suture was located between the ventral edge of the paratergite and the body. This scenario would have resulted in a hollow mass of cuticle representing the dorsal and lateral exoskeleton, which was open to sediment infilling during an interval after moulting, but prior to ultimate internment in the sediment pile.

Despite missing key body parts, the remains are not fully disarticulated, which is unexpected given the sedimentological evidence for relatively high-energy deposition because arthropod exoskeletons rapidly disarticulate when tumbled in a fluid ( McCoy and Brandt 2009). Considered alongside the fact that the fossil is preserved in three dimensions, fully enveloped and partially coiled (longitudinal doming of the sediment in part B) within a finegrained cross-bedded sandstone, this suggests that the remains were instantaneously deposited with the host sediment. In a scenario where the fossil was parautochthonous, with the exuvium discarded and filled with sand away from the final resting location, this could feasibly have occurred as a pulse of bank margin debris (i.e. sand, exuvium and plant remains) that collapsed into a river channel and was subsequently sculpted by migrating bedforms.

Two further characteristics imply that the fossil represents an articulated exuvium that had already degraded prior to such collapse and interment: (1) the absence of any other fragments of arthropleurid material within the host bed (despite intensive searching), which implies that a complete organism was not disarticulated within the bedform in which it was ultimately preserved; and (2) the degradation recorded by the serrated leftlateral margin to the tergites and the irregular broken appearance of the majority of the right paratergites ( Figs 1 and 2 ).

Post-burial deformation

Two key characteristics of the fossil imply that it has been deformed post-burial. First, the sand-filled tergites in slab Acan be seen to buckle, suggesting that they experienced compressional stress within the sediment pile. Second, the repetition of form between the carbonaceous smears that sandwich the fossil and the sand infill implies that the internal cast has been squeezed out and offset from the dorsal and ventral cuticles after partial lithification. Both characteristics are unusual, but can be readily explained through the tectonic taphonomy of the host bed.

The fossil-bearing bed occurs within the hanging wall damage zone of the Howick Fault ( Fig. 4 ) and has previously been imaged, when still in situ in the cliff face, in earlier structural geology investigations of the locality (see De Paola et al. 2005, their fig. 8; Kjemperud 2011, their figs 14 and 17). The Stainmore Formation in the hanging wall damage zone contains several features – including small thrust faults, listric geometries, stratal thickening in the hanging wall, mudstone deformation and calcite veins – that together show that the master fault was syndepositional and occurred when the sediment was only partially lithified ( De Paola et al. 2005; Kjemperud 2011). Normal faulting in the Howick Fault Zone was initiated during thermal subsidence after the earliest Carboniferous cessation of rifting in the Northumberland Basin and was contemporaneous with the deposition of the Yoredale Group ( Kimbell et al. 1989; De Paola et al. 2005; Kjemperud 2011). Development continued when it was reactivated as a strike-slip fault during Variscan-induced shortening ( Leeder et al. 1989; Chadwick et al. 1995; Fraser and Gawthorpe 2003; De Paola et al. 2005), as well as during the Carboniferous–Permian emplacement of the adjacent Whin Sill dolerite ( De Paola et al. 2005; Kjemperud 2011).

Fault development at the precise fossil locality thus involved the near-continual deformation of the host sediment, prior to full lithification, throughout the Carboniferous and into the Permian. To accommodate the stress in the synsedimentary main fault, internal compressional strain in the fossil-hosting bed would have occurred. The arthropleurid fossil – as a significantly large material discontinuity within the unlithified to partly lithified sandstone bed – likely took up some of this strain, buckling the tergites and offsetting the internal sand moulds from the carbonized remains of the exuvium.

Interpretation of Arthropleurid identity

The partial preservation of cuticular material, the segmented nature of this fossil and the partial preservation of the lateral divisions of the segments into medial and paratergites strongly supports an arthropod identity for this fossil, even though no appendage is preserved. The morphology of the best-preserved paratergites (i.e. four to six; Fig. 2 ) indicates that the fossil records the anterior part of the animal.

The surface of slab Ais interpreted as recording the ventral surface of the dorsal exoskeleton, with the surface of slab B recording a three-dimensional counter-print of this, and the pillow forms in slab Abeing sand infills of the tergites. The dorsal surface of the dorsal exoskeleton is visible in some fragments that have broken off from slab A ( Fig. 8c ).

No other Carboniferous arthropod with this morphology, or of this size, is known. Therefore, based on these observations, we propose that this fossil represents a giant arthropleurid. Although it is possible that other – yet unreported – arthropod taxa reached this size during the Carboniferous, an identity as Arthropleura sp. remains the most parsimonious explanation. In addition, two characteristics of the specimen bear notable resemblances to other specimens of Arthropleura : (1) the recurvature of the anterior paratergites is similar to that seen in other specimens ( Hahn et al. 1986; Briggs and Almond 1994; Brauckmann et al. 1997; Kraus and Brauckmann 2003; Schneider and Werneburg 2010); and (2) the granular surface texture on the anterior border of the tergites bears a resemblance to that visible in partial specimens ( Brauckmann et al. 1997).

Five morphospecies of Arthropleura have previously been described, mainly based on cuticular ornamentation: A. armata and A. mammata are in common use (e.g. see Hahn et al. 1986) and there are less frequent or isolated reports of A. cristata ( Hannibal 1997) , A. fayoli ( Boule 1893) and A. maillieuxi ( Pruvost 1930) . Comparable large arthropleurids have been assigned to the species A. armata , but the lack of preserved appendages or detailed ornamentation in the Howick specimen precludes confident species-level diagnosis.

Original size

Assuming that the missing left paratergites were the same size as those preserved on the right, the original carcass must have been at least 55 cm in width and considerably more than the 76 cm length that is preserved. A number of alternative width to length ratios for Arthropleura have been posited, calculated on the basis of trackways and partial giant, or complete juvenile, specimens with appendages. Estimates range between 3.47 ( Martino and Greb 2009), 3.75 ( Ryan 1986), 3.6–4.4 ( Kraus 1993; Schneider and Werneburg 1998; Schneider et al. 2010) and 4.78 ( Hahn et al. 1986). The Howick specimen is the widest arthropleurid fossil thus far discovered. Based on these ratios, it would also represent the largest individual discovered to date – being between 190 and 263 cm in length ( Fig. 11 ). We contend that the true size is most likely to have been at the upper end of these estimates because the fossil has been tectonically compressed and only 12–14 tergites are preserved.

Estimates of the number of tergites in Arthropleura have improved with the discovery of new specimens, but all estimates are considerably greater than the 12–14 in the specimen described here. Early morphological details were based on the description of a c. 6.5 cm long, nearly complete, juvenile specimen ( Calman 1914) from below the Top Hard Coal in Derbyshire, England (late Bashkirian; Duckmantian; Sheppard 2005). That specimen has an indistinct head region, a nearly complete trunk in dorsal aspect and an indistinct terminal segment, permitting the recognition that Arthropleura had at least 28 tergites ( Calman 1914).

Several subsequent reconstructions (e.g. Rolfe and Ingham 1967, fig. 2; Briggs et al. 1984) were strongly influenced by the c. 90 cm long ‘Maybach specimen’ from the Moscovian Saarbrücker Schichten (Sulzbach Formation, Saarbrücken Subgroup) of the Saar Basin, Germany ( Guthörl 1935; first described and figured in detail by Hahn et al. 1986, fig. 1 and plates 1 and 2). That specimen shows 23 tergites from a dorsal aspect, but the head and tail regions are missing ( Hahn et al. 1986, plate 2). Hahn et al. (1986, fig. 2) suggested these represent the remains from a trunk of an estimated 30 tergites.

Other discoveries that inform on Arthropleura segment numbers include two associated remains of a distorted ventral exoskeleton from the Gzhelian–Asselian Döhlen Formation of Saxony, which preserve 25 articulated leg-bearing segments of an individual between 0.65 and 0.8 m in length ( Schneider and Barthel 1997, p. 195, plates 5–7). The most recent reconstructions of Arthropleura – a 2.20 m long three-dimensional reconstruction, figured by Schneider and Werneburg (2010, fig. 6C) – assume 32 tergites for adult giant arthropleurids. All of these reconstructions imply that the Howick specimen comprises less than half the length of the original organism.

The Howick specimen isthus analogous in size to the very largest Arthropleura previously interpreted from indirect evidence: the 51 cm wide organism interpreted from fragmentary preserved appendages in the Gzhelian–Asselian Manebach Formation, Germany ( Schneider and Werneburg 1998) and the organisms that left nearly 50 cm wide trackways in the Visean Strathclyde Group of Scotland ( Pearson 1992; Pearson and Gooday 2019) and the Gzhelian Cape John Formation of Nova Scotia ( Ryan 1986; Ryan and Boehner 1994).

Weights of 8–10 kg have previously been calculated for giant Arthropleura , estimated from interpretations of a fraction of a simplified cylindrical volume and a density equivalent to that of water ( Kraus and Brauckmann 2003). However, a cylinder is not representative of the true form of Arthropleura , which is better envisaged as a hemi-ellipsoid with a flat underside and raised topside tapering towards the lateral, anterior and posterior edges. In addition, the density of water (997 kg m − 3) is not representative of the densities of modern giant millipedes, which are typically 350– 550 kg m − 3 ( Bercovitz and Warburg 1985; Mwabvu et al. 2010; Horváthová et al. 2021).

We calculated two possible estimates for the weight of the Howick Arthropleura based on a reasonable estimate of 20 cm as the height and consequent dimensions of 20 cm × 55 cm × 263 cm. Calculating this volume as a hemi-ellipsoid (volume = 2/3π abc, where a, b and c are the half-height, width and length) equates to c. 158000 cm 3, suggesting substantial weights of c. 55–87 kg based on the densities of extant giant millipedes. This method provides a replicable estimate, but does not account for the true shape of the organism being a fraction of a complete hemi-ellipsoid. To account for the likely overestimate, we also purchased a commercially available three-dimensional mesh of a model Arthropleura from Turbosquid.com, loaded this into Blender ( Garwood and Dunlop 2014) and scaled it to the dimensions derived from this fossil (see Supplementary Information). The 3D Print Toolbox in Blender provided a volume measurement for an Arthropleura -shaped object with the specified dimensions of 91509 cm 3, equivalent to c. 32– 50 kg based on the densities of extant giant millipedes. This range of estimates converges at an approximate weight of c. 50 kg, which is substantially larger than previous estimates, but inevitable due to the extreme size of this specimen (applying the calculation method of Kraus and Brauckmann (2003) would lead to an implausible weight estimate of c. 205 kg).

With a surface area of c. 2.7 m 2, the Howick specimen is one of the largest individual arthropod fossils found to date globally, comparable with the largest specimen of the Ordovician trilobite, Isotelus rex ( Rudkin et al. 2003) . It may also record the largest known arthropod in Earth history. The upper size estimate of a 2.63 m length and c. 50 kg weight exceeds the 2.5 m length interpreted for Jaekelopterus rhenaniae, the Early Devonian eurypterid previously suggested to be the largest arthropod ever to have evolved ( Braddy et al. 2008).

Implications for the understanding of Arthropleura

The Howick specimen provides limited new information on Arthropleura Bauplan , being primarily the ventral surface of the dorsal exoskeleton. The specimen lacks the ornamentation of arthropleurid remains that are found in more taphonomically favourable settings (indeed, these would not be expected on the ventral surface of the dorsal exoskeleton). Despite this limited detail, the curvature of the specimen supports assertions of arthropleurid manoeuvrability and refutes the suggestion that Arthropleura may have had a weak, unmineralized cuticle and was stabilized by musculature and antagonistic hydraulics, as in caterpillars (e.g. Kraus and Brauckmann 2003; Kraus 2005; McGhee 2018). The fractured margins and sand infill of the tergites in the Howick specimen imply a sclerotized exoskeleton in life, as does the survival of an exuvium. Further supporting evidence is provided by arguments based on trackways, where leg stance ( Shear and Edgecombe 2010) and track depth ( Lucas et al. 2005; Schneider et al. 2010) suggest that skeletal support was offered by more than just haemolymph pressure, and the observation that arthropleurid remains are more recalcitrant than other arthropod fragments in depositional settings with significant transport histories ( Proctor 1998).

Arthropleurid habitat

The fossil-bearing bed was deposited in a minor fluvial distributary channel in direct proximity to the coast and the completeness of the arthropleurid fossil suggests that it has not been subject to a significant history of transport. Direct palaeobotanical evidence shows that the small river traversed a lower delta plain that was colonized by a mixed arborescent flora of lycopsids, medullosalean pteridosperms and cordaitaleans ( Fig. 6 ). The presence of only thin, discontinuous and infrequent coals implies that the vegetation was relatively open at the coast, rather than forming dense coal forests (at 326 Ma old, the fossil also predates the widespread dominance of equatorial wetland coal forests in Euramerica; Greb et al. 2006). Ichnological evidence shows that the lower delta plain also hosted communities of terrestrially adapted amphibians and small infaunal and surface-grazing invertebrates ( Fig. 6 ), whereas the adjacent marine waters were populated by a normal salinity community of vertical burrowers, bryozoans, brachiopods, crinoids and marine forams.

(continued)

This setting contrasts with the traditional view that arthropleurids predominantly occupied swampy environments (e.g. Donovan 2002; Kraus and Brauckmann 2003). Although even early investigations noted that Arthropleura was more common in fluvial sandstones that were intercalated with coals ( Guthörl 1940), the common perception of an association of Arthropleura with coal swamp environments appears to have arisen as an artefact of the earliest fossil discoveries of the organism being made in working coal mine settings and excavation dumps (e.g. Guthörl 1936). The interpretation of tightly vegetated coal-forming swamps as the preferred habitat of Arthropleura is not supported by finds of more or less allochthonous body remains and especially not by the absolutely autochthonous Arthropleura tracks ( Schneider et al. 2010). Awealth of more recent ichnological evidence is aligned to the setting recorded by the Howick example, namely sparsely wooded, alluvial and littoral environments (e.g. Pearson 1992; Lucas et al. 2005; Schneider et al. 2010; Getty et al. 2017; Pearson and Gooday 2019). In addition, the close proximity of the fossil to the trackway Baropezia ( Scarboro and Tucker 1995) provides direct evidence that confirms that arthropleurids shared an environmental niche with tetrapods, even by the end Mississippian ( Falcon-Lang et al. 2006; Martino and Greb 2009; Schneider et al. 2010; Minter et al. 2016; Getty et al. 2017; Dernov 2019), contrary to the traditional view that the latter would have outcompeted them (e.g. DiMichele et al. 1992).

Ichnological evidence that has been attributed to arthropleurid activity includes the large trackways Diplichnites cuithensis (e.g.

Briggs et al. 1979, 1984; Ryan 1986; Pearson 1992; Schneider et al. 2010; Moreau et al. 2019), possible large Beaconites aestivation burrows ( Falcon-Lang et al. 2006; Falcon-Lang and Miller 2007; Pearson and Gooday 2019) and rare coprolites ( Scott and Taylor 1983). The identification of an arthropleurid trace-maker for many of these ichnofossils is assumed primarily based on their size; sites that yield both trace and body fossils are thus far unknown ( Table 1 ). Trace fossil localities have a different bias to body fossil localities, requiring the presence of true substrates (bedding planes that have archived ancient air–substrate interfaces), which are most favourably exposed in areas of extensive rock outcrop rather than abundant spoil debris ( Davies and Shillito 2018, 2021; Shillito and Davies 2020).

Diplichnites cuithensis is thus far known from the Northumberland Basin, but the Stainmore Formation is directly contemporaneous in age, and comparable in facies, with the Upper Limestone Formation of the adjacent Midland Valley basin of southern Scotland. Despite different lithostratigraphic and basin names, the units were deposited within a linked deposystem during the Pendleian, connected by contiguous deltaic sedimentary environments in the present North Sea area, with upland and emergent areas in the region of the Southern Uplands Block ( Fig. 2 ) ( Kearsey et al. 2015, 2019). Diplichnites cuithensis trackways are common in the Upper Limestone Formation, recorded from both Glasgow city, 170 km WNW (Buckman 2021, pers. comm) and the Isle of Arran, 220 km WNW ( Fig. 12 ; the type locality of D. cuithensis; Briggs et al. 1979).

The marginally older Visean strata of the Midland Valley basin in Fife (the Anstruther and Pittenweem formations) also have abundant reported D. cuithensis ( Pearson 1992; Whyte 2018) and original fieldwork at these localities has yielded 26 individual instances of the track form, ranging in external width from 23 to 47 cm (mean 32 cm) and indented into sand to depths of up to 8 mm. These dimensions strongly suggest that the trackways were made by organisms of the same size and posited weight as recorded by the Howick body fossil. The trackways all occur in similar sedimentary facies to the Howick body fossil: delta-top alluvial and littoral sandstone facies with patchy standing tree fossils and abundant Stigmaria , but no evidence for extensive afforestation. They provide direct evidence for arthropleurid habitat preferences, with individual trackways traversing both submerged and emergent substrates ( Fig. 12 ). This ichnological evidence supports physiological evidence that arthropleurids were suited to both subaerial and very shallow water locomotion (e.g. Størmer 1976; Shear and Selden 1995; Schneider and Barthel 1997) and would have been well-suited to the patchily wet lower delta plain environment recorded in the Howick section.

Palaeogeographical and stratigraphic range

The full known stratigraphic and palaeogeographical range of arthropleurids is shown in Figure 13 and Table 1 . The organism is known from Visean to Sakmarian strata and has a tight palaeogeographical range in the narrow equatorial belt ( Schneider and Werneburg 2010).

The earliest fossil evidence is known from a handful of Mississippian sites in Britain and Germany (including this study) and becomes widespread across equatorial Euramerica later in the Carboniferous. The peak geographical distribution of unequivocal body and trace fossils is known from the Early to MidPennsylvanian. Post-Kasimovian body fossils are less common, with evidence primarily from a few sites in central and southern Europe ( Table 1 ), despite a notably abundant trackway record from this interval across North America ( Ryan 1986; Ryan and Boehner 1994; Mángano et al. 2002; Lucas et al. 2005; Martino and Greb 2009; Schneider et al. 2010; Chaney et al. 2013). Two secondary reports of fragmentary remains and trackways from Kazakhstan ( Novozhylov 1962; Nelikhov 2010), cited by Dernov (2019), represent a higher latitude palaeogeographical outlier, but these instances remain anecdotal in the absence of published illustration and the host strata are only coarsely dated to the Pennsylvanian.

The shifting distribution of both body and trace fossil evidence for Arthropleura ( Fig. 13 ) may imply that the palaeogeographical range of arthropleurids expanded from a localized subequatorial crucible in the Mid- to Late Mississippian. The fossil described here, in addition to the Scottish trackways, indicates Mississippian gigantism in this group before arthropleurid fossils become widespread in the late Carboniferous. By the Pennsylvanian, Arthropleura had an extensive west–east palaeogeographical range across the entire continent of Laurussia/Pangaea, but no verified evidence for the organism is present from palaeolatitudes higher than 10° Nor 10° Sand most known instances tightly follow the palaeoequator ( Schneider and Werneburg 2010, fig. 16). In the Late Pennsylvanian and early Permian, a transcontinental range was maintained, but almost all known arthropleurid and track instances remain within 10° of the palaeoequator. The Carboniferous– Permian northwards drift of Pangaea appears to be reflected by the increased abundance of younger Arthropleura remains from more southern modern latitudes because fossil evidence tracks the relative southwards migration of the palaeoequator during this interval.

The strong relationship between Arthropleura body fossils and the location of the palaeoequator could be counter-argued to be reflective of sampling biases, tracking the distribution of mined coal-bearing strata. However, the trend is also seen within the trace fossil record ( Fig. 13 ), which is subject to a different and mutually exclusive suite of biases (i.e. extensive bedding plane outcrop instead of excavated spoil tips). For example, in Britain, multiple outcrops of strata with bedding plane exposures persist through the latest Carboniferous and early Permian and have been investigated for (vertebrate) trackways (e.g. Sarjeant 1974; Hedge et al. 2019). However, despite these directed ichnological surveys and the correct outcrop type, the youngest British trace fossil evidence for D. cuithensis is Serpukhovian ( Briggs et al. 1979). By contrast, in Spain, the oldest worked coal measures are of Moscovian age ( Piedad-Sánchez et al. 2004) and spoil from these measures has been intensively interrogated for plant and other fossil remains (e.g. Wagner and Álvarez-Vázquez 2010). However, despite these directed palaeontological surveys and the correct outcrop type, the earliest Spanish body fossil evidence for Arthropleura is not known until the Kasimovian ( Castro 1997).

These examples illustrate that although different outcrop expressions can bias evidence for Arthropleura , the appearance and disappearance of suitable outcrop types is discordant with the appearance and disappearance of evidence for Arthropleura . The most parsimonious explanation for the southwards drift of evidence through the Carboniferous and Permian is that the affinity of Arthropleura for equatorial latitudes was robust and that the genus maintained its geographical range as the Carboniferous continents drifted northwards.

Figures

Fig. 1. Specimen of partial remains of a giant Arthropleura (anterior 12– 14 tergites) after excavation from the Serpukhovian Stainmore Formation, Howick Bay, Northumberland, England (CAMSM X.50355). Slab A and slab B are not true part and counterpart, but rather a split through the middle of a three- dimensional dorsal exoskeleton (see Fig. 7 and discussion in text).

Fig. 2. Interpreted sketch of the fossil, as preserved in slab A, while still intact in fallen block of host sandstone prior to excavation. The overlap pattern of the paratergites on the left indicates that this is the ventral site of the dorsal exoskeleton. The typical ornamentation of the dorsal site of Arthropleura tergites and paratergites is therefore not visible.

Fig. 8. Details of the Arthropleura specimen. (a) Alternation of granular (1) and smooth (2) texture on medial tergites. Granular texture interpreted to reflect degraded verrucose texture to anterior part of each tergite. View of underside of dorsal exoskeleton preserved on surface of slab A. Scale bar 1 cm, (b) Detail of right lateral paratergites seen on slab B showing striated patterning and frayed margins to the paratergites. Scale bar 1 cm. (c) Micaceous and carbonaceous smear on fragment of upper side of dorsal exoskeleton (slab A). Scale bar 1 cm. (d) Inflection of paratergites (1) relative to medial tergites (2) at anterior of specimen on the part. Head region seen at (3). Scale bar 1 cm. (e) Detail of invertebrate burrow (Planolites) seen in centre of slab B and likely exploiting remains shortly after interment. Scale bar 1 cm.

Fig. 9. Three-dimensional morphology of cuticular sand infill preserved in slab A. (a, b) Images of slab A before and after extraction of fossil. Part (a) shows a natural break in sandstone revealing the underside of the dorsal cuticle, notable for carbonaceous material (cb) between the paratergites and the presence of a granular texture. Part (b) shows slab A after extraction, revealing the cast of the dorsal exoskeleton, still yielding carbonaceous material, but with no granular texture. Scale bars 20 cm. (c, d) Images of partial fragment of cuticular sand infill between the underside and upper side of the dorsal cuticle, as extracted from slab A. Part (c) shows the exoskeleton underside with a granular texture and both images show cross-section through lenticular imbricated tergites, possibly compressed by synsedimentary tectonics. Scale bar 1 cm. An, anterior; Po, posterior.

Fig. 10. Details of (a, b) slab A and (c, d) slab B with interpreted sketches of margins showing offset of three-dimensional sand infill from carbonaceous smear. Same tergites on each slab are colour-coded (with reference to prominent tergite shown in white). Cuticular interior shown as full circles with correspondingly shaped carbonaceous smear shown as colour-coded partial circles, indicating consistent 40 cm squeezed offset of infill from cuticle in direction of red bidirectional arrow, associated with synsedimentary deformation. Scale bars 20 cm.

Fig. 4. Context of the specimen prior to extraction. (a) Slab A of the fossil (facing stratigraphically upwards). Note convex-down doming of the host sediment. Scale bar 20 cm. (b) Slab B of the fossil (facing stratigraphically downwards). (c) Fallen block in which the fossil was discovered in January 2018 (white circle), showing split in rock where fossil was exposed (white triangle). Black circle denotes position in cliff profile from where the block fell. Listric faults in hanging wall of Howick Fault are highlighted, showing hanging wall thickening of mudrock (HT), attesting to synsedimentary origin (red arrows indicate sense of fault movement). Scale bar 2 m. (d) View of the fallen block (white circle) and original position (black circle) in the context of the Howick Fault damage zone and master fault separating the Alston and Stainmore formations; HT is at same position as in part (c). Host bed is immediately adjacent to one of two synsedimentary (Mississippian) normal faults that were later reactivated as strike-slip faults during the latest Carboniferous to Permian emplacement of the Whin Sill dolerite intrusion (De Paola et al. 2005). This structural context attests to strain in the fossil-bearing sandstone bed prior to full lithification and may explain certain features of the fossil (see main text). Blue dashed line show approximate stratigraphic position of the Lickar Limestone, marking the onset of the late Serpukhovian (Cózar and Somerville 2021). Scale bar 5 m. Photograph in part (d) courtesy of Geospatial Research Limited.

Fig. 11. Reconstruction of the Howick Arthropleura. (a) Scale of the Howick Arthropleura relative to other articulated giant specimens (preserved remains highlighted pink) and the largest Diplichnites cuithensis trackways known from each Carboniferous-Permian stage. Note that the previously known partially complete body fossils were both markedly smaller than the dimensions of Arthropleura revealed by trace fossil evidence. For references and details of localities mentioned, see Table 1. (b) Reconstruction of the Howick Arthropleura within its habitat of a lower delta plain with open woodland.

Fig. 6. Ichnological and palaeobotanical features associated with the Arthropleura fossil. (a) Trackway of an anthracosaur tetrapod (Baropezia) (individual footprints arrowed). Located at 4.6 m on stratigraphic log in Figure 3d. Scale bar 10 cm. (b) Detail of Baropezia footprint (Ba) adjacent to Stigmaria root (black arrow). For details of vertebrate tracks prior to recent erosional degradation, see Scarboro and Tucker (1995). Located at 4.6 m on stratigraphic log in Figure 3d. Scale bar 20 cm. (c) Thinly bedded wave-rippled facies immediately below fossil horizon showing horizontal invertebrate burrows Planolites (Pl) and Archaeonassa (Ar). Located at 10.6 m on stratigraphic log in Figure 3d. Scale bar 10 cm. (d) Tightly curved invertebrate burrow identified as Taenidium on basis of meniscate fill and absence of lining. Fallen block of same lithology adjacent to fossil-bearing block. Scale bar 10 cm. (e) Ripplelaminated sandstone beds colonized with coalified, downwards-branching plant rootlets. Located at 9.0 m on stratigraphic log in Figure 3d. Scale bar 20 cm. (f) Plant remains in same bed as Arthropleura fossil, including fragment of branch or bark belonging to Lepidodendron sensu lato (black arrow) and thin axis that acutely bifurcates distally, possibly the proximal part of a medullosalean pteridosperm frond (white arrow). Located at 10.0 m on stratigraphic log in Figure 3d. Scale bar 20 cm. (g) Seed fossils, possibly Samaropsis, as produced by cordaitaleans. Note seeds occur immediately adjacent to the Arthropleura remains in the part, seen at top left of image. Located at 10.2 m on stratigraphic log in Figure 3d. Scale bar 5 cm. (h) Stigmaria on stratigraphic top of fallen block hosting the Arthropleura remains. Preserved fraction of main rhizomorph axis extends between white arrows and shows ornamentation in boxed area. Adjacent part of bedding plane contains traces of rootlets (r) branching from the main rhizomorph. Located at 9.8 m on stratigraphic log in Figure 3d Scale bar 1 m.

Fig. 12. Trace fossil evidence for Arthropleura from the trackway Diplichnites cuithensis, preserved in strata, near contemporaneous in age to the Howick specimen, from adjacent basins in northern Britain (see Fig. 3 for localities). (a) Trackway from which the D. cuithensis holotype was cast (Briggs et al. 1979) in the Serpukhovian Limestone Coal Formation of Laggan, Isle of Arran, Scotland. Stick is 1 m long, inner and outer width of trackway highlighted. (b) D. cuithensis trackway of near-analogous width to the Howick body fossil, Visean Pittenweem Formation, Crail, Fife, Scotland. (c) D. cuithensis trackway>10 m long (arrowed) in the Visean Pittenweem Formation, St Andrews, Fife, Scotland (see Whyte 2018). Geologist is 1.8 m tall. (d, e) Photograph and interpreted sketch of context of D. cuithensis on a true substrate (Davies and Shillito 2018, 2021) recording the passage of two animals between emergent and subaqueous substrates. Convex bedform, colonized with plants, provided small island surrounded by shallow water puddles in a delta-top setting. Trackways are well-defined on emergent substrate and less so on subaqueous parts of substrate. Visean Anstruther Formation, 3.5 km NW of Kingsbarns, Fife, Scotland. Measuring stick 60 cm long.

Fig. 13. Palaeogeographical range of Arthropleura body and trace fossils. (a) Known sites of evidence plotted on a palaeogeographical map of the late Carboniferous (c. 310 Ma; after Torsvik and Cocks 2016). Numbers refer to sites listed in Table 1. (b) Visean to Serpukhovian sites (latitudinal parallels shown for c. 320 Ma; from Torsvik and Cocks 2016). (c) Bashkirian to Moscovian sites (latitudinal parallels shown for c. 310 Ma; from Torsvik and Cocks 2016). (d) Kasimovian to Sakmarian sites (latitudinal parallels shown for c. 290 Ma; from Torsvik and Cocks 2016).