Morphology and systematics of anthracomartidae (Arachnida: Trigonotarbida)

Authors


Abstract

Abstract:  X-ray microtomography (XMT) was applied to three species of the extinct arachnid order Trigonotarbida, hosted in siderite nodules from the late Carboniferous British Middle Coal Measures. All three, Cryptomartus hindi (Pocock, 1911), Cleptomartus plautus (Petrunkevitch, 1949) and Maiocercus celticus (Pocock, 1902), belong to the family Anthracomartidae. As well as providing interactive three-dimensional visualisations of their likely appearance in life, XMT study has resolved new morphological detail, yielding key data about the likely stance and habitus of these early arachnids. Similarities in the form of the carapace, eyes and coxosternal region between anthracomartids and the Devonian genus PalaeocharinusHirst, 1923 are highlighted. Analogies in limb disposition are drawn between anthracomartids and modern crab spiders (Araneae: Thomisidae), which exhibit sit-and-wait style predatory behaviour. Anthracomartids are relatively common fossils at many Coal Measures localities. Although they appear to exhibit limited morphological variation, numerous genera and species have been proposed in the literature, often distinguished from each other on rather trivial characters. From our reconstruction of well-preserved examples, we have resolved a number of common features likely to be present in a typical anthracomartid. Comparisons to this model suggest that the characters on which many anthracomartid genera are distinguished may be artefacts of preservation. We hence treat PromygaleFrič, 1901, BrachylycosaFrič, 1904, PerneriaFrič, 1904, CoryphomartusPetrunkevitch, 1945, CryptomartusPetrunkevitch, 1945, PleomartusPetrunkevitch, 1945, CleptomartusPetrunkevitch, 1949 and OomartusPetrunkevitch, 1953 as junior synonyms of the type genus of the family, AnthracomartusKarsch, 1882. Only the genera BrachypygeWoodward, 1878 and MaiocercusPocock, 1911 are additionally retained. Within Anthracomartus, Cleptomartus planusPetrunkevitch, 1949, Cryptomartus meyeriGuthörl, 1964, Cleptomartus hangardiGuthörl, 1965 and Cryptomartus rebskeiBrauckmann, 1984 are treated as junior synonyms of A. hindiPocock, 1911 (all syn. nov.). Cleptomartus plautusPetrunkevitch, 1949 and Anthracomartus denuitiPruvost, 1922 are treated as junior synonyms of A. priestiPocock, 1911 (both syn. nov.).

Trigonotarbids are an extinct order of arachnids which superficially resemble spiders, but lacked silk-producing spinnerets. About seventy species are currently known, and as a group, they ranged from the late Silurian to the early Permian. The majority – and the greatest diversity of families, genera and species – are found in the late Carboniferous Coal Measures of Europe and North America. Trigonotarbids can easily be recognised by the dorsal division of the opisthosoma into median and lateral plates. In most cases, the tergites are divided into three distinct sclerites, but in the family Anthracomartidae there are five, a convincing synapomorphy of the numerous anthracomartid genera. Indeed, anthracomartids are one of the most widespread trigonotarbid clades and are found at many of the classic Coal Measures localities e.g. Coseley, UK (Pocock 1911) and Nýřany and Rakovník in the Czech Republic (Frič 1901, 1904), where they can be locally quite abundant.

The best preserved anthracomartids come from siderite (ironstone) concretions, such as those found at Coseley in the English West Midlands, and other contemporary localities in the British Middle Coal Measures. Here, the fossils are preserved as external moulds or casts, and these impressions in the matrix can, under ideal conditions, preserve a high degree of three-dimensionality and fidelity of detail. Traditional methods of study (Pocock 1911; Petrunkevitch 1913, 1949) were nevertheless limited to breaking open nodules and drawing or photographing the part and counterpart, respectively.

In recent years, the development of tomographic methods applicable to small fossils (see Sutton 2008 for a review) has revolutionised our understanding of early fossil arthropods. In a pilot study, Garwood et al. (2009) demonstrated that X-ray microtomography (XMT) has great potential for recovering new data about arthropods – and other fossils – in siderite concretions (see also Selden et al. 2008). The technique differentiates the void or mineral infill representing the fossil from the host siderite, allowing its reconstruction as a three-dimensional computer model, which can be manipulated at will. These models represent accurate reconstructions of the animal’s morphology and include fine details significant for understanding both relationships and mode of life. In this study, they have allowed us to build up a picture of a typical anthracomartid and compare it with the diagnostic characters proposed for genera and species within the family. As suspected by Dunlop and Rössler (2002), we argue that many anthracomartid genera, particularly those described from shales, are based on inappropriate interpretations of morphology. We argue here that previous diagnoses of taxa were largely derived from taphonomic effects such as compression and/or truncation of the carapace and the consequent loss of key anatomical features like eyes or the anteriorly projecting clypeus. Based on these observations, a number of genera are synonymized with the type genus of the family, AnthracomartusKarsch, 1882. Within this genus, a number of species-level synonymies can also now be recognised.

Materials and methods

Anthracomartid fossils currently assigned to CryptomartusPetrunkevitch, 1945 and CleptomartusPetrunkevitch, 1949 were obtained from their repository at the Natural History Museum London (NHM). The specimens originate from the British Middle Coal Measures of Coseley near Dudley, Staffordshire, UK and can be dated to the Late Carboniferous (Duckmantian, Westphalian B in traditional terminology, Ogg et al. 2008), c. 311 Ma. Particularly well-preserved and complete examples of Cryptomarus hindi (Pocock, 1911) (NHM In 22841) and Cleptomartus plautus (Petrunkevitch, 1949) (NHM I. 15857) were selected for further investigation and scanned as detailed below. Additionally, a specimen of Maiocercus celticus (Pocock, 1902), which is the type and only known species of the genus MaiocercusPocock, 1911, was obtained from the private collection of Mr Lee Cherry and scanned using the same procedure. This very well-preserved specimen was collected at the Crock Hey open cast pit near Manchester, UK, which at uppermost Langsettian (=Westphalian A, c. 313 Ma, Ogg et al. 2008) in age is marginally older than the Coseley site. All three scanned specimens are preserved in siderite concretions. Braznell (2006) offered a taphonomic model for this form of preservation: decaying tissues initiated the development of framboidal plus microbial or crystalline siderite; the latter forming the encasing concretion. Tissue decay left an external mould, later part filled with kaolinite (±secondary siderite and sulphides). This form of preservation is typical for a number of Coal Measures Lagerstätten. Significantly, it involves early mineralisation (i.e. prior to compaction) and hence yields three-dimensional fossils, which can now be visualised with the aid of tomography.

Comparative anthracomartid material was examined by JAD. This included Czech Republic anthracomatids held in the National Museum Prague (NMP), mostly preserved compressed in shales from the Coal Measures of Bohemia. Further specimens in siderite nodules collected by Carl Horrocks from Lancashire, UK are held in the Manchester Museum (MM) (see Dunlop and Horrocks 1996). The holotype of Maiocercus celticus (Pocock, 1902) from the South Wales coalfield is held in the National Museum of Wales (NMW). The holotype of its junior synonym Maiocercus orbicularisGill, 1911 was recently rediscovered in the Bolton Museum, UK (see Craven and Dunlop 2008). Both were examined within the framework of earlier studies. It was not possible to locate the type material of the anthracomartids Brachypyge carbonisWoodward, 1878 and Cleptomartus denuiti (Pruvost, 1922), which should have been in Belgium.

Tomography

As noted above, siderite-hosted fossils are often preserved in three-dimensions, yet complete data recovery has always been problematic. Traditional study relies on splitting the concretion and examining the resulting surface(s). This approach leaves morphology not revealed on the split-surface largely unresolved. Latex casts have previously been used in an attempt to retrieve such information (e.g. Petrunkevitch 1949, pl. 62), but these rarely recover fine details and risk damaging the fossil. These problems can be overcome by tomography (reviewed by Sutton 2008), i.e. the slice-based investigation of a fossil. Data for tomographic reconstruction can be recovered using physical-optical tomographic methods, such as serial grinding or sawing coupled with photography (e.g. Sutton et al. 2002). However, these techniques are laborious and also destructive, rendering them inappropriate for existing type and figured material. For this study, a nondestructive approach i.e. high-resolution X-ray computed microtomography (XMT) was chosen. This method generates closely separated parallel slice images of a fossil, derived computationally from a large number of x-rays taken of a rotating sample (Ketcham and Carlson 2001). All models presented here were scanned on the Metris X-Tek HMX-ST of the Natural History Museum, London, with a current of 200 mA, a voltage of 225 kV and exposure times of between 0.25 and 2 s. A tungsten reflection target and 1-mm copper filter were used. The machine is fitted with a 2000 × 2000 Perkin Elmer detector panel, giving a resolution in the order of 20 μm for these specimens.

The resulting datasets provided the basis for the three-dimensional, virtual models of the fossils within their nodules. These models were created using the custom SPIERS software suite, which implements the methods of Sutton et al. (2001, 2002). Inverted linear threshold images were created, i.e. all pixels darker than a certain grey level were assumed to belong to the fossil. The data were then manually cleaned to ensure that spurious pixels were assigned correctly to either fossil or matrix. Finally, the images were masked; selected anatomical features were manually assigned to different (coloured) structures on a slice-by-slice basis, allowing removal of cracks from the model. The part and counterpart of the Cleptomartus plautus specimen (NHM I. 15857) were scanned separately to test the potential of this approach, which yields higher resolution scans. A negative consequence of this was that manually removing the crack between the nodule halves in the final image proved unfeasible, thus in the other scans part and counterpart were scanned together. The resulting 3D models were rendered as a collection of isosurfaces, and through interactive visualisations, iterative improvements were made to the masks and editing to produce clean and accurate models of the fossils. Final images (Text-figs 1–3) were ray traced using the open source application Blender (blender.org).

Figure TEXT‐FIG. 1..

 Reconstruction of the anthracomartid Anthracomartus hindi (formerly Cryptomartus) from the Coseley Largerstätte, UK (NHM In 22841). A, dorsal view. B, ventral view; of note are the coxal endites on the inside of each walking limb and the prominent ventral sacs. C, anterior view showing the lateral and median eye tubercles, clypeus and laterigrade stance. Text-figure key: 1–12, segment numbers; CE1–4, coxal endites 1–4; CH, Chelicerae; CL, clypeus; LE, lateral eye tubercle; L1–4, legs 1–4; LN, lateral notch; ME, median eye tubercle; PG, pygidium; PP, pedipalp; VS, ventral sacs. Scale bar represents 5 mm.

Figure TEXT‐FIG. 2..

 A reconstruction of the anthracomartid Anthracomartus priesti (formerly Cleptomartus plautus) from the Coseley Largerstätte, UK (NHM I. 15857). The crack through the nodule used to study the fossil has not been removed from this model. A, dorsal view. B, ventral view. C, anterior view. Text-figure key: 1–12, segment numbers; CE1–4, coxal endites 1–4; CH, Chelicerae; CL, clypeus; LE, lateral eye tubercle; L1–4, legs 1–4; ME, median eye tubercle; PG, pygidium; PP, pedipalp; VS, ventral sacs. Scale bar represents 5 mm.

Figure TEXT‐FIG. 3..

 A reconstruction of the anthracomartid Maiocercus celticus from the Crock Hey Largerstätte, UK (from the private collection of Mr Lee Cherry). A, dorsal view. B, ventral view. C, anterior view. Text-figure key: 1–12, segment numbers; CE1–4, coxal endites 1–4; CH, Chelicerae; CL, clypeus; LE, lateral eye tubercle; L1–4, legs 1–4; PG, pygidium; PP, pedipalp. Scale bar represents 5 mm.

Systematic palaeontology

Order TRIGONOTARBIDA Petrunkevitch, 1949

  • 1882 Anthracomarti Karsch, p. 560.

  • 1885 Meridogastra Thorell and Lindström, p. 31.

  • 1895 Eurymarti Matthew, p. 277.

  • 1949 Trigonotarbi Petrunkevitch, pp. 235–236.

Remarks. Petrunkevitch (1949) divided the anthracomartid fossils into two quite separate and supposedly unrelated orders: Anthracomarti Karsch, 1882 and Trigonotarbi Petrunkevitch, 1949. The endings of these ordinal names were subsequently modified by Petrunkevitch (1955) to the ‘-ida’ form to fit systematic conventions at that time. Anthracomartida, sensu Petrunkevitch was restricted to those taxa with abdominal tergites divided into five distinct plates and thus corresponds with the modern concept of Anthracomartidae. Formalised in the Treatise on Invertebrate Palaeontology (Petrunkevitch 1955), this division into two distinct orders was maintained by later workers (Guthörl 1964, 1965; Brauckmann 1984; Opluštil 1985, 1986).

In their important revision of trigonotarbid morphology and phylogeny, Shear et al. (1987) questioned the validity of this anthracomartid–trigonotarbid division sensu Petrunkevitch. Subsequently, Dunlop (1996a) further argued that characters which supposedly distinguished anthracomartids from trigonotarbids – such as the orientation of the mouthparts and three pairs of lungs with longitudinal spiracles – were largely based on misinterpretations of the fossils. Similarities between Anthracomartidae and the Devonian trigonotarbid family Palaeocharinidae were highlighted, and these similarities have been largely supported (carapace shape, opisthosomal segmentation) and augmented (coxal endites; pedipalpal claw) by this study. Anthracomartids and trigonotarbids were thus recombined as a single ordinal taxon (Dunlop 1996a). Although younger, Petrunkevitch’s name was adopted because Trigonotarbida now encompassed more of the species and through Shear et al.’s work had become more widespread and better defined phylogenetically in the literature. Two further, poorly known and redundant names for trigonotarbids can be found in earlier publications (Thorell and Lindström 1885; Matthew 1895), see also Dunlop and Miller (2007), and are listed here for completeness.

Family ANTHRACOMARTIDAE Haase, 1890

  • 1890 Anthracomartidae Haase, pp. 650–651.

  • 1903 Promygalidae Frič, p. 865 [nomen nudum].

  • 1904 Promygalidae Frič, p. 19.

  • 1911 Brachypygidae Pocock, pp. 58–59.

  • 1945 Coryphomartidae Petrunkevitch, p. 50.

  • 1945 Pleomartidae Petrunkevitch, p. 49.

Type genus. AnthracomartusKarsch, 1882.

Included genera. BrachypygeWoodward, 1878, MaiocercusPocock, 1911.

Emended diagnosis.  Trigonotarbids with tergites 2–9 divided by longitudinal sutures to form rows of five plates across the dorsal opisthosoma; the outer suture line continuing round in parallel with the outline of the opisthosoma to subdivide the median plate of tergite 9 into an anterior and posterior element. Tergite 1 retained as a locking ridge. Carapace subquadrate, somewhat box-like. Median and lateral eye tubercles retained and anterior margin of the carapace pronounced into a short, steeply descending projection or clypeus. Endite-like elements retained on the mesal part of the leg coxae (emended from Dunlop and Horrocks 1996, p. 29).

Remarks.  Anthracomartids are one of the most instantly recognisable trigonotarbid groups by virtue of their five plates across the dorsal opisthosoma, as opposed to the three seen in all other members of the group. Study of some known and some new Early Devonian fossils from Alken and der Mosel and related localities in Germany now suggests that there is a bridging taxon, whose morphology is consistent with being intermediate between the palaeocharinid and anthracomartid grades of organisation (Poschmann and Dunlop, in press). Anthracomartidae have been recorded widely from numerous Late Carboniferous Coal Measures localities across Europe and North America. The family currently comprises 23 valid species in 10 genera (JAD, unpublished data). As suggested by Dunlop and Rössler (2002), many of these genera seem to be based on preservational features rather than explicit apomorphies, and their apparent palaeodiversity is almost certainly an overestimate.

Frič (1904), Pocock (1911) and Petrunkevitch (1945), all proposed separate families for particular anthracomartid genera, but none of their family groups have become established in the literature. As its name implies, Promygalidae was actually created as a family of Araneae, as its type genus was mistaken for a spider. Pocock (1910, pp. 505–507) effectively referred Promygalidae to Anthracomartidae through his synonymy of the respective genera (see PromygaleFrič, 1901 below). Pocock (1911) created Brachypygidae to accommodate those anthracomartid genera (Brachypyge and Maiocercus) with a scalloped opisthosomal margin. Petrunkevitch (1913, p. 94) formally referred Pocock’s Brachypygidae to Anthracomartidae. Petrunkevitch (1945) introduced two monotypic families, Coryphomartidae and Pleomartidae, for their respective genera based on minor differences, in which segments had lateral plates (his marginal fields). These differences seem to have been based on misinterpretations and Petrunkevitch (1949, p. 208) himself listed Coryphomartidae together with Promygalidae and Brachypygidae as synonyms of Anthracomartidae. Finally, Petrunkevitch (1953, p. 59) formally added Pleomartidae to the Anthracomartidae synonymy list.

Note that Frič (1903) published a summary of his forthcoming monograph which, unfortunately, listed all his new taxa, but without diagnoses or indications. His 1904 family, genus and species names can thus be found as nomina nuda in the 1903 publication. It should also be noted that he typically published under the germanised form of his name, Fritsch.

Genus ANTHRACOMARTUS Karsch, 1882

  • 1882 Anthracomartus Karsch, p. 560.

  • 1901 Promygale Frič, p. 58. [syn. by Pocock (1910)]

  • 1903 Perneria Frič, p. 866 [nomen nudum].

  • 1904 Perneria Frič, p. 22. [syn. with Brachylycosa by Petrunkevitch (1953)]

  • 1903 Brachylycosa Frič, p. 866 [nomen nudum].

  • 1904 Brachylycosa Frič, p. 24. syn. nov.

  • 1945 Coryphomartus Petrunkevitch, p. 50. syn. nov.

  • 1945 Cryptomartus Petrunkevitch, p. 49. syn. nov.

  • 1945 Pleomartus Petrunkevitch, p. 49. syn. nov.

  • 1949 Cleptomartus Petrunkevitch, p. 211. syn. nov.

  • 1953 Oomartus Petrunkevitch, p. 66. syn. nov.

Diagnosis.  Anthracomartids with a smooth opisthosomal margin, lacking the scalloping seen in Brachypyge and Maiocercus (after Dunlop and Rössler 2002.)

Type species. Anthracomartus voelkelianusKarsch, 1882.

Included species. A. bohemica (Frič, 1901), A. carcinoides (Frič, 1901), A. elegans (Frič, 1901), A. granulatusFrič, 1904, A. hindiPocock, 1911, A. kustae (Petrunkevitch, 1953) (comb. nov.), A. janae (Opluštil, 1986) (comb. nov.), A. minorKušta, 1884, A. palatinusAmmon, 1901, A. planus (Petrunkevitch, 1949) (comb. nov.), A. plautus (Petrunkevitch, 1949), A. priestiPocock, 1911, A. nyranensis (Petrunkevitch, 1953) (comb. nov.), A. radvanicensis (Opluštil, 1985) (comb. nov.), A. triangularisPetrunkevitch, 1913, A. trilobitusScudder, 1884.

Remarks.  Despite their apparent diversity in the literature (Petrunkevitch 1955), anthracomartids seem to be, anatomically, a fairly homogeneous group. There are, however, considerable differences in their mode of preservation. Material from nodules (e.g. this study) tends to be better preserved and more three-dimensional, compared to material from shales which has usually been compressed. As well as lacking external relief, shale fossils can also be deformed by shearing or stretching. Alexander Petrunkevitch, in particular, seems to have had a poor appreciation for taphonomic processes and the way in which they can influence the final appearance of a fossil. This is important given the fairly box-like construction of the anthracomartid carapace (Dunlop 1996a; Dunlop and Horrocks 1996; Garwood et al. 2009; see also below) and the fact that numerous genera were diagnosed on carapace shape: e.g. high versus flat, with or without a clypeus (Petrunkevitch’s median crest), or a rectangular versus a rounded outline. We argue here that our 3D models of complete and well-preserved anthracomartids from the British Middle Coal Measures offer a good approximation of the appearance of a typical anthracomartid in life (Text-figs 1–3). Indeed, a similar gross morphology was reconstructed in Dunlop and Horrocks’ (1996, fig. 6) study of Maiocercus celticusPocock, 1902. We suggest that most of the previously proposed diagnostic characters for other genera, which differ from this groundplan are likely to be taphonomic artefacts and thus poor grounds for maintaining separate taxa.

A further problem (a discussion of which can be found in Dunlop and Rössler 2002) was that Petrunkevitch was unable to study the Anthracomartus genotype, A. voelkelianusKarsch, 1882, from its repository in (East) Berlin. He was thus reluctant to compare other anthracomartids with the holotype of the oldest available name. He even (Petrunkevitch 1953) went so far as to treat Anthracomartus as an incertae sedis genus. Dunlop and Rössler (2002) redescribed A. voelkelianus, from the Langsettian (=Westphalian A) Coal Measures of Silesia in Poland, and we are now able to compare other taxa directly with this species. Eight previously proposed genera are treated here as synonyms of Anthracomartus, and the fate of individual taxa is discussed in detail below. Two species described by Goldenberg (1873) – which were subsequently referred to Anthracomartus– are based on unidentifiable material and have already been effectively treated as nomina dubia (Petrunkevitch 1953; Guthörl 1965, Dunlop and Rössler 2002).

Promygale.  This genus was introduced by Frič (1901) in an important paper on the ‘Fauna der Gaskohle und der Kalksteine der Permformation Böhmens’. Note that there is some confusion about the date of publication of this work which is sometimes cited as 1899 or 1902, but was cited by Frič himself in subsequent work as 1901, see also comments in Harvey and Selden (1995). Promygale was established for three species from Nýřany in the Czech Republic; a locality which is now dated at late Carboniferous (Westphalian D) (cf. Opluštil 1986), rather than Permian as originally presumed. Frič interpreted Promygale as true spiders (Araneae); Mygale being an older genus name for tarantulas (Theraphosidae). Curiously, Frič believed he could see comb-like organs, similar to the pectines of scorpions, on the underside of the opisthosoma and used these to diagnose the genus. These features could not be confirmed by Petrunkevitch (1953). In his subsequent monograph on ‘Palaeozoische Arachniden’, Frič (1904) formally raised a suborder Pleuraraneae for spiders, possessing divided opisthosomal tergites (or pleurae). Promygale, and some other trigonotarbid genera, was erroneously included here as putative spiders. Pocock (1910) argued convincingly that Promygale was a trigonotarbid and not a spider, and he explicitly (p. 507) synonymized Promygale with Anthracomartus.

Petrunkevitch (1953) resurrected Promygale for some of the Nýřany anthracomartids and transferred a further species described by Kušta (1884) from the slightly younger Rakovník locality in the Czech Republic to this genus too. Opluštil (1986) more recently added another species, differentiated from a previous taxon on the most trivial of characters. Promygale was redefined by Petrunkevitch as anthracomartids with a flat carapace, longer than wide, and no clypeus. The flatness of the carapace is quite simply because of the fact that these fossils are compressed in shales, and probably also the reason why the clypeus is not clearly preserved. In fact, careful examination of his figures (Petrunkevitch 1953, figs 67, 153) together with the original material (JAD, pers. obs.) suggests that in some Nýřany fossils the carapace does (in outline) come to a point anteriorly, which is entirely consistent with the clypeal region in our XMT models (e.g. Text-fig. 1). The front of the carapace is probably missing in the genotype, A. voelkelianus, which leaves it without the clypeus and a preserved carapace only about as long as wide. In our more complete A. hindi reconstruction, the carapace is slightly longer than wide, as per the diagnosis of Promygale, and for this reason, we follow Pocock (1910) and accept the synonymy of this genus with Anthracomartus.

Brachylycosa.  This genus was introduced by Frič (1904) for a single species from Nýřany, B. carcinoides (Frič 1901) and interpreted as a ‘spider of uncertain position’. The genus name implies a truncated type of wolf spider (Lycosidae), but Frič’s (1904) reconstruction does not inspire confidence, being based on ‘an imperfectly preserved example’. Four eyes were recognised and used to diagnose the genus. In fact, this fossil shows the typical outline of a fairly stocky anthracomartid, but without the characteristic opisthosomal segmentation preserved. Petrunkevitch (1913) correctly listed Brachylycosa as a trigonotarbid, albeit under the family Eophrynidae. However, he incorrectly assigned the genus name to Frič’s 1901 paper rather than the 1904 monograph.

In 1953, Petrunkevitch assigned Brachylycosa to Anthracomartidae and added a second species from Rakovník. Brachylycosa was redefined on having a rounded, rather disc-like carapace. We find this character unconvincing and the roundness in the line drawings –Petrunkevitch (1953, fig. 64) essentially just sketched a circle – is much too strongly emphasised when compared to both the photographs and the original material (JAD, pers. obs.). Even in our A. hindi model, the postero-lateral corners of the carapace curve slightly inwards and a poorly preserved fossil under compression could easily yield the impression of a more rounded structure. In the absence of other characters differentiating this genus from Anthracomartus, we consider Brachylycosa to be a junior synonym.

Perneria.  This genus was introduced by Frič (1904) for another single species from Nýřany, P. salticoides (Frič 1901), and again interpreted as a ‘spider of uncertain position’. The species name is clearly implicit of the jumping spider family Salticidae, and this is reflected in the drawing of a rather squat, short-legged arachnid. Salticids are not known prior to the early Cainozoic, however. The P. salticoides holotype is quite small (5 mm) when compared to typical anthracomartids (15–25 mm) and thus possibly immature. Petrunkevitch (1953) examined the type and recognised it as a synonym of the anthracomartid species Brachylycosa carcinoides. Perneria was explicitly mentioned as a synonym of Brachylycosa (Petrunkevitch 1953, p. 63) and can thus now be treated as a synonym of Anthracomartus.

Coryphomartus.  This genus was introduced by Petrunkevitch (1945) for a species previously described (Scudder 1884) under Anthracomartus from the Joggins Mines of Nova Scotia in Canada. Petrunkevitch (1945, 1953, 1955) interpreted both Braychpyge and Coryphomartus as having a distinctly triangular carapace. In the case of Brachypyge, earlier studies (Woodward 1878; Pruvost 1922, 1930) found no evidence for a carapace, but Petrunkevitch claimed to have prepared the fossil and revealed a subtriangular structure. His photographs of this (Petrunkevitch 1953, fig. 145) are unconvincing. The opisthosoma is very well preserved, but the alleged carapace is, by contrast, at best a vague outline, which could conceivably be owing to fortuitous planes of fracture between the part and counterpart.

The situation for Coryphomartus is similar. We have not had the opportunity to examine the C. triangularis (Petrunkevitch, 1913) type, but the original photographs are relatively clear (Petrunkevitch, 1913, fig. 61; Petrunkevitch 1949, fig. 191) and reveal a typical Anthracomartus opisthosoma associated with a badly deformed carapace region crossed diagonally by a large fracture plane through the nodule. These, and other lines within the nodule, again admittedly leave the impression of a subtriangular area in front of the opisthosoma, but in this state of preservation, we would caution against reading too much into this observation. Petrunkevitch (1953) differentiated Coryphomartus from Brachypyge on the ratios of some of the opisthosomal sclerites and Brachypyge’s marginal scalloping. Given the smooth margin of the C. triangularis opisthosoma and the unreliability of the triangular carapace, we regard Coryphomartus as a junior synonym of Anthracomartus.

Cryptomartus.  This genus was introduced by Petrunkevitch (1945) for the two British species of Anthracomartus from the Duckmantian (= Westphalian B) of Coseley described by Pocock (1911). Further species were added by Guthörl (1964) and Brauckmann (1984) from Germany and by Opluštil (1985) from the Czech Republic, respectively. The diagnosis of Cryptomartus was expanded by Petrunkevitch (1949, 1953, 1955) and related to the carapace being high and steep-sided with the anterior region pronounced into a median crest (our clypeus). These features are all correct and can clearly be seen in our 3D model (Text-fig. 1). They are, however, not seen in anthracomartids like Karsch’s Anthracomartus voelkelianus, which is the oldest available genus and species name, or in many of the flattened specimens from Nýřany, Rakovník and other localities in the Coal Measures of Bohemia. That said, we suspect that Petrunkevitch failed to appreciate that the exact plane at which a rock splits directly affects the appearance of three-dimensional structures. Thus, the flatness of the carapace and the absence of the clypeus in Karsch’s essentially less well-preserved fossil are more likely to be artefacts of taphonomy rather than useful taxonomic features. Fossils assigned to Cryptomartus are probably the most complete and least distorted example of anthracomartids available. We suspect that the diagnostic features of carapace shape proposed for this genus are probably part of the ground pattern for Anthracomartidae in general. In the absence of any convincing biological differences, we regard Cryptomartus as a junior synonym of Anthracomartus.

Pleomartus.  This genus was introduced by Petrunkevitch (1945) for a species originally assigned to Anthracomartus from the Coal Measures of Arkansas, USA (Scudder 1884). A further species from Germany (Ammon 1901) was subsequently referred to Pleomartus by Petrunkevitch (1949). Pleomartus was defined primarily on a flattened, subrectangular carapace, wider than long (Petrunkevitch 1949, 1953, 1955). Problems in using the degree of flatness of the carapace and in the presence/absence of a clypeus have been noted above. The proportionally wider carapace may be a genuine feature, although the photograph of one specimen (not the type) of P. trilobitus (Scudder, 1884) offered by Petrunkevitch (1913) suggests that the width has been exaggerated in his drawing of the same specimen. The German P. palatinus (Ammon, 1901) does look quite broad from published illustrations (Ammon 1901; Petrunkevitch 1953; Guthörl 1965), but restudy would be welcomed. Pending formal revision, we feel that a wide carapace, in isolation, does not necessarily justify a separate genus and prefer to treat Pleomartus for now as a junior synonym of Anthracomartus.

Cleptomartus.  This genus was introduced by Petrunkevitch (1949) for two new Coseley species. A Belgian species (Pruvost 1922) previously described as Anthracomartus was also referred to Cleptomartus and Guthörl (1965) later added another species from the German Saar region. Cleptomartus was diagnosed by Petrunkevitch (1949, 1953) on a flattened carapace, about as wide as long, and with a rounded anterior margin. Some of the material contributing to C. plautus was originally part of the paratype series for Pocock’s A. priesti and Petrunkevitch’s (1949, p. 211) remark that ‘The abdomen does not present distinctive characters.’ is illuminating. A Cryptomartus style carapace could be easily turned into a Cleptomartus one simply by truncating the anterior margin and removing the clypeus, eyes, etc. Exactly, this seems to have happened in one of the otherwise very well-preserved C. plautus examples (NHM In 15896; Petrunkevitch 1949, fig. 217). As with the Anthracomartus/Cryptomartus comparison, the flatness of the carapace is dependent on taphonomic processes, even in nodules. In this case, it is not the whole carapace, which has been compressed, as happens at Nýřany for example. Instead, a nodule which splits near the top (dorsal) surface of the carapace will reveal only this uppermost part and would superficially appear to lack depth (see In 15896 again and especially Petrunkevitch 1949, fig. 66). A split further down, near where the legs emerge, would reveal the carapace at its full thickness and probably includes most or all of the clypeus too. It would be closer to the 3D reconstructions presented in this paper. As the defining characters of Cleptomartus are likely to be preservational artefacts, we regard Cleptomartus as a junior synonym of Anthracomartus too.

Oomartus.  This genus was introduced for a single species, O. nyranyensisPetrunkevitch, 1953, from Nýřany and defined on an essentially egg-shaped body with no obvious constriction between the pro- and opisthosoma. As with Brachylycosa, the sketch reconstruction overemphasises features seen in compressed, and in this case slightly sheared, material (cf. Petrunkevitch 1953, figs 70, 158, 176). This, coupled with poor preservation of the specimens, leaves little justification for maintaining a separate genus, and we refer Oomartus to Anthracomartus.

Anthracomartus hindiPocock, 1911
Text-figure 1

  • 1911 Anthracomartus hindi Pocock, pp. 64–67,text-figs 30–32, pl. 3, fig. 3.

  • 1913 Anthracomartus hindi Pocock; Petrunkevitch, p. 95.

  • 1919 Anthracomartus cf. hindi Pocock; Pruvost, pp. 355–357, text-fig. 43, pl. 23, figs 4, 4a.

  • 1930 Anthracomartus hindi Pocock; Pruvost,pp. 214–215.

  • 1945 Cryptomartus hindi (Pocock); Petrunkevitch, p. 49.

  • 1949 Anthracomartus hindi Pocock; Millot, p. 759, fig. 552.

  • 1949 Anthracomartus hindi Pocock; Waterlot, p. 904, fig. 685.

  • 1949 Cryptomartus hindi (Pocock); Petrunkevitch,pp. 223–227, figs 34, 36–37, 42, 64–71, 195–199.

  • 1949 Cleptomartus planus Petrunkevitch, pp. 220–222, figs 90–93, 218–220.

  • 1953 Cryptomartus hindi (Pocock); Petrunkevitch; p. 67, fig. 155.

  • 1953 Cleptomartus planus Petrunkevitch; Petrunkevitch, p. 66.

  • 1955 Cryptomartus hindi (Pocock); Petrunkevitch, p. 105, figs 66(3), 68(3).

  • 1955 Cleptomartus planus Petrunkevitch; Petrunkevitch, p. 107, fig. 67(4).

  • 1964 Cryptomartus hindi (Pocock); Guthörl, p. 101.

  • 1964 Cryptomartus meyeri Guthörl, pp. 98–101, text-fig. 1, pls 13, fig. 1A–B, pl. 14, figs 1A–B.

  • 1965 Cleptomartus hangardi Güthorl, pp. 15–17,text-fig. 4, pl. 2, figs 2a–b.

  • 1984 Cryptomartus hindi (Pocock); Brauckmann,pp. 97–99, fig. 4.

  • 1984 Cryptomartus meyeri Guthörl; Brauckmann,p. 96–99, fig. 3.

  • 1965 Cleptomartus hangardi Güthorl; Brauckmann, p. 96.

  • 1984 Cryptomartus rebskei Brauckmann, pp. 97–99, figs 1a–b, 2.

  • 1985 Cryptomartus hindi (Pocock); Opluštil, p. 42.

  • 1985 Cleptomartus planus Petrunkevitch; Opluštil, p. 41.

Holotype.  GSM 60173 (ex Kidston collection).

Type locality and horizon.  Coseley near Dudley, Staffordshire, UK. Late Carboniferous, Duckmantian (= Westphalian B in more traditional terminology).

Additional material.  NHM I. 7918 (former holotype of Cleptomartus planus); NHM In 22841 (the scanned specimen); Geological–Palaeontological Institute of the Technical Hochschule Aachen, Germany (Nr. 415) (holotype of Cryptomartus meyeri; not seen); Bergschule Saarbrücken, Germany (Nr. E/883) (holotype of Cleptomartus hangardi; not seen); Rebske collection, Bergisch Gladbach, Germany (Nr. Ca 1293) (holotype of Cryptomartus rebskei; not seen). Other material listed in Petrunkevitch (1949).

Distribution.  British Middle Coal Measures. Also recorded from the Coal Measures of Anzin in north-eastern France (Pruvost 1919) and the Aachen and Saarland regions of Germany. Stratigraphic distribution thus Langsettian–Bolsovian.

Diagnosis. Anthracomartus with a relatively narrow median tergal region, the lateral margins of tergites 4–8 tapering only slightly posteriorly. Chevron-shaped ventral sternites forming an angle of ca. 115 degrees where they come to a point anteriorly.

Description.  Scanned specimen of A. hindi, NHM In 22841 (Text-fig. 1), typical of species; 23 mm long, 8 mm wide across prosoma, 14 mm across opisthosoma. Carapace well-resolved, undivided and box-like, a little longer than wide, 4.1 mm deep. Anterior displays ventrally directed clypeus (Text-fig. 1A), with small median eye tubercle towards carapace front, immediately posterior to clypeus. Larger lateral eye tubercles on antero-lateral margins of carapace (small raised bumps forming the corners; Text-fig. 1C). Immediately anterior to tubercles are two shallow, semi-circular notches into dorsal margin of the carapace. Two further, evenly spaced notches on each side posterior to lateral eye tubercles. Slight median depression running length of carapace, meeting transverse linear depression with raised edges marking posterior of prosoma. Carapace with granular texture comprising microornament of sub-mm tubercles.

Ventral prosomal region (Text-fig. 1B) comprises recessed sternum accommodating prominent coxal endites of walking legs. Latter well-developed, anteriorly flattened in cross-section, more cylindrical posteriorly. Positioned closer to midline posteriorly; leg 4 endites almost touching. Chelicerae immediately posterior to clypeus comprise two elements (basal pautron and moveable fang), and of clasp-knife form (cf. Shear et al. 1987). Chelicerae posteriorly directed in palaeognath orientation (hanging down in parallel under the animal, facilitating a primarily backwards bite). Other appendages fully resolved. Pedipalps consist of six podomeres: long tarsus; tibia and patella about half this length; femur and trochanter, two-thirds tarsal length; basal coxa, similar size to tibia. Pedipalp terminates in chelate tarsal claw. Anterior walking limbs with prolateral side facing upward, allowing forward-directed appendages which can be held aloft (see Discussion). Walking legs comprise seven podomeres: tarsus; metatarsus (a third tarsal length); tarsal-sized tibia and patella; longest element the femur (twice the length of tarsus); and smaller trochanter and coxa. Legs robust, increasingly triangular in cross-section towards tarsus.

Opisthosoma suboval to pear-shaped, wider posteriorly (Text-fig. 1A). First tergite raised above those that follow (a locking ridge; see Remarks). Other tergites ex-sagittally divided into median, two lateral and two marginal sections, creating five dorsal sclerites per segment. Posterior-most (ninth) tergite further divided longitudinally. Segments two and three fused to form diplotergite posterior to locking ridge. Chevron-shaped sternites at c. 115 degree angle. Terminal (tenth) sternite consequently almost triangular, bearing prominent pygidium, comprising opisthosomal segments 11 and 12. Two strong ridges run outwards and anteriorly from sternite eight towards base of walking leg 4. Anterior opisthosoma bears two prominent lobed structures anterior to gently curving transverse ridge. The opisthosoma strongly dorsoventrally compressed, 0.23 mm thick towards posterior. Opisthosomal thickness at diplotergite two/three 1 mm. Detailed examination reveals a row of sub-mm tubercles bordering dorsal opisthosomal margin and subdivision of sternites forming a broad marginal rim ventrally.

Remarks.  The palaeognath chelicerae seen here are similar to those observed in the Rhynie chert palaeocharinids, where it has been suggested (Dunlop 1994) that they could have been partially withdrawn into the clypeal region. As in other trigonotarbids, tergite one of A. hindi forms a locking ridge (Dunlop 1996b) whereby the modified first tergite apparently slots into a corresponding fold under the posterior margin of the carapace, securing the two halves of the body together. This feature is reduced in some Carboniferous trigonotarbids, but its presence here demonstrates that this is not the case with the anthracomartids. Pocock (1911) originally described two Anthracomartus species from Coseley as A. hindi and A. priesti. They were redescribed by Petrunkevitch (1949), who reported body lengths for A. hindi of up to 25 mm. The most obvious difference between these two species (cf. Pocock 1911, figs 30, 33; Petrunkevitch 1949, p. 223) is the relative width of the median tergal plates on the opisthosoma. In A. hindi, these form a comparatively narrow band which tapers only slightly towards the posterior end (see Diagnosis). In A. priesti, by contrast, they are wider, at least anteriorly, and thus the median tergal region tapers more distinctly. These straightforward criteria appear to neatly distinguish two Coseley [morpho]species and can be further applied to some stratigraphically coeval (i.e. Westphalian) material from continental Europe. Hence, we suggest that some further species may be junior synonyms of either A. hindi or A. priesti, respectively.

In general, the systematics of the anthracomartids is complicated by Petrunkevitch’s application of the names Crypto- and Cleptomartus for different modes of preservation. Individual cases for species-level synonymies are outlined in detail below. Shale-preserved material from the Czech Republic has not been integrated into this revision; many specimens have been deformed by shearing, and we suggest that a retro-deformation approach will be required to analyse them effectively. The Bohemian fossils are typically less well-preserved than the western European nodular material, but as Frič’s nomenclature predates Pocock’s, we cannot rule out the possibility that some the former names will eventually have to take precedence.

Cleptomartus planusPetrunkevitch, 1949 was raised for a single Coseley specimen (NHM I. 7918) with a body length of only 9 mm. Its supposed generic position was evidently based on preservational factors: a flat carapace without eyes or a clypeus (see generic discussion above). NHM I. 7918 appears to us to show no further unique characters that justify a separate species, and as it expresses the relatively narrow median tergites characteristic of Anthracomartus hindi, we regard it as a junior synonym of this species, probably an immature specimen.

Cryptomartus meyeriGuthörl, 1964 is held in the geological–palaeontological institute of the Technische Hochschule in Aachen and originates from the Langsettian (= Westphalian A) of Palenberg near Aachen, Germany. Guthörl’s (1964) diagnosis is essentially a description, offering little in the way of unique, apomorphic characters and discussing instead various ratios of body measurements. We believe that such ratio-based differences derived from single specimens are highly problematic; each fossil might yield a unique ratio combination depending on its ontogenetic stage and/or the way it has been preserved and that the application of such ratios has been one of the main contributory factors towards the current inflation of diversity among Coal Measures anthracomartids. In the case of C. meyeri, the opisthosomal tergites match those of A. hindi rather well, with the diagnostic narrow median band. Apart from a slight difference in age, we see little morphologically to differentiate these species and treat C. meyeri as a junior synonym of A. hindi.

Cleptomartus hangardiGuthörl, 1965 is held in the Bergschule Saarbrücken and originates from the Westphalian D of Neunkirchen in the Saar region of Germany. The holotype, and only known specimen, consists of nothing more than an isolated dorsal opisthosoma. Guthörl attempted to diagnose this species on its somewhat narrow form, although this could be influenced by taphonomy and in the absence of any other useful characters there would be grounds for treating the name as nomen dubium. As the median tergal region again appears to be rather narrow, we tentatively refer this species to Anthracomartus hindi as a putative junior synonym.

Cryptomartus rebskeiBrauckmann, 1984 was described from the private Rebske collection (Nr. Ca 1293) and originates from the Bolsovian (= Wesphalian C) of Luisenthal in the Saar region of Germany. Brauckmann (1984) diagnosed this species based again on various ratios (prosoma:opisthosoma length; opisthosoma length:width) and a slightly more angular prosoma. The absence of wide intersegmental membranes between the tergal plates was also noted; although as mentioned below (see Discussion), this may have more to do with the degree of expansion of the opisthosoma and is a poor taxonomic feature. This feature is not consistent even within Coseley material assigned to A. hindi; compare NHM I. 13955 (with thick membranes: also the source of Brauckmann (1984, fig. 4)) and In 31250 (with little visible membrane). Overall, C. rebskei expresses a narrow tergal field, rather than a wide and tapering one. In the absence of convincing autapomorphies, we regard C. rebskei as a junior synonym of A. hindi.

Anthracomartus priestiPocock, 1911
Text-figure 2

  • 1911 Anthracomartus priesti Pocock, pp. 67–68,text-figs 33–34.

  • 1913 Anthracomartus priesti Pocock; Petrunkevitch, p. 95.

  • 1922 Anthracomartus Denuiti Pruvost, pp. 353–354, fig. 2.

  • 1930 Anthracomartus Denuiti Pruvost; Pruvost,pp. 214–215, text-fig. 9, pl. 11, fig. 4.

  • 1945 Cryptomartus priesti (Pocock); Petrunkevitch, p. 49.

  • 1949 Cryptomartus priesti (Pocock); Petrunkevitch, pp. 227–232, figs 26, 29, 35, 41, 53–54, 66, 72–80, 200–210, 213–215, 225–226.

  • 1949 Cleptomartus plautus Petrunkevitch, pp. 212–220, figs 30, 47–48, 81–89, 211, 216–217, 221–224, 227.

  • 1930 ?Cleptomartus denuiti Pruvost; Petrunkevitch, p. 211.

  • 1953 Cryptomartus priesti (Pocock); Petrunkevitch, pp. 67–68, fig. 156.

  • 1953 Cryptomartus priesti (Pocock); Waterlot, p. 571, figs 26–27.

  • 1953 Cleptomartus plautus Petrunkevitch; Petrunkevitch, pp. 65–66.

  • 1953 Cleptomartus denuiti Pruvost; Petrunkevitch p. 66, figs 69, 154.

  • 1955 Cryptomartus priesti (Pocock); Petrunkevitch, p. 105, figs 64B, 66(2), 68(5).

  • 1955 Cleptomartus plautus Petrunkevitch; Petrunkevitch, pp. 105–107, figs 66(1), 68(2), 69.

  • 1965 Cryptomartus priesti (Pocock); Guthörl, p. 101.

  • 1965 Cleptomartus plautus Petrunkevitch; Guthörl, pp. 15–16.

  • 1984 Cryptomartus priesti (Pocock); Brauckmann, pp. 97–99, fig. 5.

  • 1985 Cryptomartus priesti (Pocock); Opluštil, p. 42.

Holotype.  NHM In 18333.

Type locality and horizon.  Coseley near Dudley, Staffordshire, UK. Late Carboniferous, Duckmantian (= Westphalian B in more traditional terminology).

Additional material.  NHM I. 15857 (the scanned specimen); NHM I. 15896 (holotype of C. plautus), J. G. 8938 (holotype of C. denuiti), apparently in the Royal Belgian Institute of Natural Sciences; not seen. Other material listed in Petrunkevitch (1949).

Diagnosis. Anthracomartus with a relatively broad median tergal region, the lateral margins of tergites 4–8 tapering quite distinctly. Chevron-shaped ventral sternites forming an angle of ca. 135 degrees where they come to a point anteriorly.

Description.  Scanned specimen, NHM I. 15857, Cleptomartus plautus of Petrunkevitch (1949) slightly smaller than other members of the species, length 14, 4.5 mm wide across carapace, 6 mm across opisthosoma. Carapace well-resolved, box-like in form, with anterior ventrally directed clypeus, median and lateral eye tubercles, and longitudinal median depression posterior to median eye tubercles leading into a transverse linear depression with raised edges (Text-fig. 2A). Ventral prosoma (Text-fig. 2B) comprises coxal endties of limbs 2–4, flat in cross-section, decreasing in size anteriorly and coming closer to the midline posteriorly. Chelicerae present, two-segmented, of clasp-knife form and palaeognath in orientation. Limbs largely resolved, apart from leg 3 and proximal pedipalps, both obscured by cracks. Pedipalps robust and circular in cross-section; preserved extended in front of the body. Because of crack, only three terminal podomeres visible; all similar in size. Legs better resolved – front pair orientated with prolateral side facing upward, appendages forward-facing and held slightly aloft. All elements visible, metatarsus and trochanter significantly smaller, femur notably longer than otherwise similarly proportioned podomeres. Limbs triangular in cross-section distally.

Opisthosoma is suboval in outline, wider posteriorly. Tergite one locking ridge, all other tergites divided to create five dorsal sclerites per segment. Ninth tergite divided longitudinally, segments two and three form diplotergite. Sternites are chevron-shaped and meet medially at c. 135 degrees. Tenth sternite triangular and bears pygidium (segments 11 and 12). Prominent ventral sacs present, forming two lobes between a posterior gently curving transverse ridge and an anterior small, straight transverse ridge. Opisthosoma dorsoventrally compressed, but scanning method prevents measurement of this thickness. Subdivision of sternites create a broad marginal rim visible in reconstruction.

Remarks.  The morphology of A. priesti is largely congruent with that of A. hindi. Minor differences include a carapace which lacks well-developed lateral notches, and a lateral ridge at the back of the carapace (Text-fig. 2A), sternites which are comparatively narrower, most markedly in segments eight, nine and ten, and all are shorter in length. The angle between the sides of the chevron-shaped sternites is about 135 degrees, whereas this angle in A. hindi is closer to 115 degrees (Text-fig. 2B). Coxal endites two, three and four are present, but less well developed and more flattened than those of A. hindi, while the endite on leg 1 is either absent or not resolved. The dorsal opisthosoma also has subtle differences in the suture line that marks the border of the median set of plates. In A. hindi, these are largely parallel between segments four and eight, whereas here the margins of the median region converge and taper slightly between segments four and six, and more strongly posterior to this. The scan also reveals a small possible transverse ridge anterior to the ventral sacs that is less well developed in A. hindi.

Pocock (1911) and Petrunkevitch (1949) both previously described this species and noted body lengths of up to 13 mm, i.e. animals somewhat smaller than A. hindi. The species Cleptomartus plautus was largely derived from the original paratypes of Anthracomartus priesti, which Petrunkevitch (1949) separated out based primarily on the supposedly flattened carapace characterising his new genus (discussed above). Representatives of both A. preisti and C. plautus have the wider and more tapering median tergal region, and as differences in carapace shape are interpreted here as poor taxonomic characters, we treat these taxa as synonyms. Within the Coseley material, there are further minor differences among the fossils currently referred to A. priesti. Some have a slightly more rounded opisthosoma, such as I. 15896 – the designated holotype of C. plautus– and Petrunkevitch (1949) attempted to designate andro- and gynotypes according to his interpretation of their likely gender. The functional significance of any differences in these ventral sac regions remains equivocal; thus, complicating their interpretation as definitive secondary sexual characters. However, we note that minor differences in opisthosoma shape between males and females often occur among living arachnids, hence Petrunkevitch’s interpretation is not implausible.

Anthracomartus denuitiPruvost, 1922 is held in the Belgian Museum of Natural History and originates from the Langsettian (= Westphalian A) of Mariemon-Bascoup in Belgium. The carapace is only preserved in outline, but has the same medial and posterior depressions (Petrunkevitch 1953, fig. 154) as those resolved in our XMT studies of the Coseley material. It was subsequently referred to Cleptomartus on account of its flat carapace (Petrunkevitch 1949, 1953), and in the latter monograph, the species was defined on a median tergal area considerably wider than long. As Cleptomartus is no longer considered valid here, this previous diagnostic character of a wider median region accords well with the A. priesti opisthosomal pattern. In the absence of further diagnostic characters, we regard A. denuiti as a junior synonym of A. priesti.

Genus BRACHYPYGE Woodward, 1878

  • 1878 Brachypyge Woodward, p. 436.

Type and only species. Brachypyge carbonisWoodward, 1878; see Petrunkevitch (1953) for a synonymy list.

Emended diagnosis.  Anthracomartids with a scalloped opisthosomal margin; opisthosoma longer than wide and with median plates of tergites 4–8 narrow and barely tapering posteriorly.

Remarks.  This genus is based on an isolated opisthosoma from the Coal Measures of the Belle-et-Bonne colliery near Mons in Belgium. Woodward (1878) originally misinterpreted it as the abdomen of a crab, before Scudder (1885) correctly recognised it as an arachnid and transferred the species to Anthracomartus. Also now accepting its arachnid affinities, Woodward (1887, 1896) preferred (in asides or footnotes to papers on different subjects) to refer the Belgian fossil to his own genus EophrynusWoodward, 1871. Eophrynus has posterior spines on the opisthosoma which vaguely resemble the marginal scalloping in the Belgian fossil, but Woodward’s transfer would later cause confusion about the correct name for the type species of the genus Maiocerus (see comments in Dunlop and Horrocks 1996). Pocock (1902) returned to the original genus name Brachypyge and added a second species from south Wales, again based on an isolated opisthosoma. Brachypyge was redefined by Pocock on the scalloped margin of the opisthosoma which differentiated it from the smooth-margined Anthracomartus.

In his later monograph, Pocock (1911) proposed using the scalloping to raise a separate family, Brachypygidae. This family name was accepted by some subsequent workers (e.g. Pruvost 1930; Waterlot 1953), but has since been abandoned. We see no particular reason to resurrect Brachypygidae here, as it would leave Anthracomartus in a monogeneric Anthracomartidae family. Pocock (1911) also established a new genus, Maiocercus, for the south Wales specimen; whereby Brachypyge was diagnosed on an opisthosoma longer than wide and Maiocercus on an opisthosoma wider than long. This essentially reflects the current status of the genera. Both may well be closely related and the scalloped opisthosoma is a good potential synapomorphy. There may even be grounds for considering these genera as synonyms, as indeed Pruvost (1930) did, or even synonymizing their species, whereby the thin Brachypyge and fat Maiocercus could be seen as extremes along a taphonomic gradient.

We here prefer to retain Brachypyge and Maiocercus as separate taxa pending further investigation; we were unable to gain access to the B. carbonis type specimen or to modern photographs of the original material. Despite the lack of a carapace, published illustrations of the monotype of Brachypyge carbonis suggest it to be well-preserved fossil, with some surface relief and little obvious indication of postmortem deformation. Significantly, the median band of tergites is quite narrow and of almost constant width, i.e. the lateral margins of tergites 4–8 essentially form parallel lines. By contrast, Maiocercus is not only broader, but this median band of tergites visibly taper from anterior to posterior.

Genus MAIOCERCUS Pocock, 1911

  • 1911 Maiocercus Pocock, p. 60.

Type and only species. Maiocercus celticus (Pocock, 1902); see Dunlop and Horrocks (1996) for a synonymy list.

Emended diagnosis.  Anthracomartids with a scalloped opisthosomal margin; opisthosoma wider than long and median tergites 4–8 broad and distinctly tapering posteriorly.

Description.  Scanned specimen 28 mm in length, 8 mm across carapace and 18 mm across opisthosoma at its widest point. Carapace box-shaped with slight anterior taper, probably quite deep in life. Reconstructed specimen suffered from postmortem dorsoventral flattening; only an anterior peak represents the prosoma’s true depth (Text-fig. 3A,C). Posterior to this peak, internal (sclerotised) features of the prosoma are resolved rather than the true dorsal surface; small bi-lobed (1–3) and single-lobed (4) outgrowths associated with the emerging legs 1–4 are probably internal and related to the coxae (see Remarks). Scan reveals projecting clypeus, lateral eye tubercles (Text-fig. 3A,C) and posterior transverse depression in contact with a specialised segment one locking ridge. Ventral prosoma relatively poorly preserved. Pedipalps taper, almost triangular in cross-section distally. One curled beneath the body; femur, patella, tibia and tarsus discernable. Other disarticulated, comprising patella, tibia and tarsus only (Text-fig. 3B). Chelicerae resolved off-centre; palaeognath in orientation, and reconstruction likely represents fangs (see Remarks). Walking legs well-resolved seven podomeres slightly dorsoventrally flattened, proportions similar to A. hindi. Weakly developed coxal endites present, front two limbs similar in orientation to A. hindi and A. priesti. Granular tuberculation of cuticle has previously been reported (Dunlop and Horrocks 1996), but was not adequately recovered in these scans.

Opisthosoma well resolved dorsally and ventrally. Outline subcircular, slightly wider than long. Segment one locking ridge, all posterior segments divided ex-sagittally into five plates. Segments two and three fused into diplotergite. Tergite nine divided longitudinally, resulting suture line follows opisthosomal margin fairly closely. Margin scalloped, each tergite corresponds to an embayment in opisthosomal boundary. Ventral opisthosoma features marginal suture line, chevron-shaped sternites meeting medially in a small longitudinal ridge and triangular sternite of segment ten, bearing prominent pygidium. Immediately posterior of coxae four are possible ventral sacs – two small transverse ridges, poorly resolved.

Remarks.  The most complete example of Maiocercus celticus previously reported (Dunlop and Horrocks 1996) displayed the carapace, dorsal and ventral opisthosoma and fragments of a single leg. The new example of this species (in the private collection of Mr Lee Cherry) is scanned here and described above; it is complete yielding the entire prosoma and opisthosoma as well as all the appendages (Text-fig. 3). As mentioned above, the carapace displays a degree of postmortem dorsoventral flattening, and the resulting outgrowths associated with the emerging legs are very similar to the internal views of the coxae found in spider exuviae (JAD, pers. obs.). In these, the coxosternal morphology becomes disarticulated from the carapace during moulting. The groove between these features is thus more likely to be related to the sternum than to the dorsal surface of the carapace (in contrast to the median depression seen in A. hindi). This raises the possibility that this specimen represents the moult of a trigonotarbid, rather than a mortality, a suggestion supported by the poor preservation of the ventral prosoma, which would be expected in exuviae. The chelicerae, as seen, are interpreted as showing the fangs, rather than being drawn upwards beneath the clypeus (as they are in the other reconstructions). In this specimen, the cheliceral fangs appear to be folded out and lie more or less in parallel along the midline of the body, appearing somewhat larger as a result. The origin of the opisthosomal scalloping is unclear; Dunlop and Horrocks (1996) suggested that it may relate to the presence of spines on sternites, but our data suggest that the involvement of the tergites in this marginal ornament cannot be excluded.

As noted above, this genus was raised by Pocock (1911) for a species from the South Wales coalfield. Gill (1911) added a second species from Lancashire in the north-west of England. Gill’s name was synonymized with the south Wales species by Pruvost (1930); a synonymy confirmed by Craven and Dunlop (2008) who recently rediscovered and refigured Gill’s holotype. M. celticus is quite widespread across Europe and has now been recorded from the Coal Measures of England, France, Belgium, the Netherlands and Germany (see e.g. Essen et al. 1999).

Discussion

The major advantage of the 3D models generated here (Text-figs 1–3) is their ability to offer high-quality spatial reconstructions of Coal Measures arthropods preserved in nodules. XMT offers a unique window into the appearance of these extinct animals in life and can reveal well-preserved specimens with greater fidelity that was possible using traditional methods of study. Here, XMT has allowed us to verify, and in many cases augment, previous descriptions – and thus to build up what we believe to be an accurate picture of the typical anthracomartid body. It has also allowed increasingly accurate idealised reconstructions to be created (Text-figs 4–5). We can confirm here the presence of a carapace bearing both median and lateral eye tubercles, and a projecting anterior clypeus. The opisthosoma typically has five sclerites per tergite, but also modification of tergite 1 into a locking ridge and fusion of tergites 2 + 3 into a diplotergite. On the underside, a ventral opisthosomal suture line and chevron-shaped sternites are present. Additionally, our studies have revealed a number of novel traits significant for understanding the relationships and mode of life in these animals. As some are present in at least two genera, we suggest these features may be typical for the Anthracomartidae in general.

Figure TEXT‐FIG. 4..

 Reconstruction of Anthracomartus hindi in dorsal view (reproduced with permission from Garwood et al. 2009). Scale bar represents 5 mm.

Figure TEXT‐FIG. 5..

 Reconstruction of Maiocercus celticus in dorsal view. Scale bar represents 5 mm.

Coxal endites

All Anthracomartidae scanned here possessed coxal endites on each walking leg. These have not been formally described in anthracomartids before and were probably overlooked using the techniques then available. They have, however, at least been illustrated in exceptionally preserved members of the family Palaeocharinidae from the Rhynie chert (cf. Hirst 1923, fig. 5, pl. 14b). The function of these endites is unclear, although they could have played a role in the immobilisation and retention of prey while the animal was feeding. Trigonotarbid endites largely correspond in position to the gnathobases seen on the inner (mesal) side of the coxae in arachnid outgroups such as Xiphosura and Eurypterida, where the dentate gnathobases flank a food groove and are obviously used to actively masticate food preorally. It is possible therefore that palaeocharinid–anthracomartid endites are gnathobasic elements retained plesiomorphically in an early grade of arachnid; the alternative scenario would be to view them as an apomorphic character supporting (Palaeocharinidae + Anthracomartidae). A formal assessment of the significance of this character would require a cladistic study of in-group trigonotarbid relationships, which is beyond the scope of this study.

Chelate pedipalps

The claw-like tip of the anthracomartid pedipalp is a particularly interesting find, having only recently been identified in Rhynie chert palaeocharinids (Dunlop et al. 2009). The present study thus confirms its presence in at least two trigonotarbid families. Although it could be seen as another potential synapomorphy of (Anthracomartidae + Palaeocharinidae), comparisons with arachnids in general reveal a very similar distal end of the pedipalp in the enigmatic Ricinulei (see Talarico et al. 2008 for a recent detailed account). Ricinuleids usually resolve in cladistic analyses as the sister group of mites (Acari), but Dunlop et al. (2009) suggested that the chelate pedipalp joins divided tergites and a locking ridge between the prosoma and opisthosoma as a suite of characters supporting an alternative (Trigonotarbida + Ricinulei) relationship (see also Dunlop 1996b). Chelate pedipalps could thus also be interpreted as a plesiomorphy with respect to anthracomartids and palaeocharinids; in this scenario palpal claws would be part of the trigonotarbid groundplan.

Laterigrade stance

An intriguing feature revealed by the scans is the laterigrade stance of anthracomartids, with the first two pairs of legs in particular having their prolateral side facing uppermost, with the limbs directed forwards and held slightly aloft (Text-figs 1C, 2C, 3C). Pocock’s (1911, text-fig. 32) original reconstruction of the Coseley fossils hinted at this too. If the preserved morphology really does reflect the stance of these animals in life, it would have brought the anterior legs into an outstretched, forward position, while transferring their bodyweight more onto the posterior limbs. Such a stance is comparable to that used by modern crab spiders (Thomisidae), which are sit-and-wait/ambush predators, who use their outstretched forelimbs to grasp at prey items which come within their reach (Morse 2007). It is conceivable that anthracomartids may have hunted in a similar fashion, grabbing and overpowering prey with their forelimbs. Other taxa may have hunted in a similar fashion, for example members of the family Anthracosironidae appear to have semi-raptorial forelimbs and may show similar hunting adaptations, but a full investigation will require comparable three-dimensional reconstructions of additional Coal Measures trigonotarbids.

Flattened opisthosoma

A further feature common to all the anthracomartid reconstructions is a rather flattened, almost disc-like opisthosoma. This was also noted by Pocock (1911, text-fig. 32) and makes the animals appear superficially like ticks (Ixodida) in some orientations. A putative relationship between anthracomartids and ticks even has historical precedence (Schulze 1932), but can be easily discounted by the structure of the trigonotarbid mouthparts and the fact that they had book lungs rather than tracheae. Any similarities in habitus compared to ticks would thus be better treated as convergent. Sections through the opisthosoma of Rhynie Chert palaeocharinids (e.g. Shear et al. 1987, fig. 5) suggest that the opisthosoma in these animals was not so thin as that reconstructed here for anthracomartids. What is not clear is whether a collapse in opisthosomal thickness had occurred in the Carboniferous forms prior to preservation. Several specimens in the NHM collections preserve small gaps between the sclerites, suggesting that when well fed (or perhaps gravid), the opisthosoma could expand, as is often seen in extant arachnids. This would require flexible membranes between the opisthosomal plates, and as such this region would provide less resistance to flattening during burial. Although these models show a marked decrease in thickness posteriorly, this could be a taphonomic effect. Nevertheless, given that other aspects of the body are well preserved, we suspect that the opisthosoma in life was indeed dorsoventrally flattened, albeit perhaps of a more constant thickness than that reconstructed here, conceivably allowing the animals to conceal themselves in narrow spaces.

Acknowledgements.  We thank Claire Mellish (NHM), John Nudds (formerly MM) and Vojtech Turek (NMP) for access to material in their care. We are also greatly indebted to Andrew Tenny, Lee Cherry, Steve Keen, and Babara Mercer for giving access to fossils in their collections. The manuscript benefitted from comments by Carsten Brauckmann and an anonymous reviewer. RG thanks Mark Sutton for comments and advice, and Richie Abel and Stig Walsh for training in CT techniques. RG’s work is funded by a NERC PhD scholarship.

Editor. Lyall Anderson

Ancillary