Introduction

Trigonotarbida is an extinct arachnid order known from the late Silurian to the early Permian (ca. 419–290 Ma). They resolve at the base of the Pantetrapulmonata clade (Shultz 2007), i.e., spiders and their closest relatives, and lack any indication of silk-producing spinnerets. Sixty-seven species are currently recognised in the literature, although some of these are probably based on trivial differences; or even the way the fossils have been preserved. Revisions have invariably reduced their overall alpha diversity (Dunlop 1994; Rössler 1998; Garwood and Dunlop 2011). Trigonotarbid fossils have been recovered most frequently from the Late Carboniferous Coal Measures of Europe and North America (e.g., Frič 1904; Pocock 1911; Petrunkevitch 1913, 1949, 1953; Heide 1951; Brauckmann et al. 1985), together with a further Carboniferous record from Argentina. Eight of the nine described families occur in the late Carboniferous, and the fauna includes some highly ornamented arachnids with bodies up to a fewcentimetres long (cf. Rössler and Dunlop 1997).

One of the best known trigonotarbids is the large, spiny and tuberculate species Eophrynus prestvicii (Buckland, 1837) from the Coal Measures of the English West Midlands. Frequently mentioned in the literature (see synonomy list) or figured as a general example of a fossil arachnid in both historical (Roemer 1876; Scudder 1885) and modern (e.g., Black 1988) palaeontological textbooks, it is a key species for understanding both the systematics and paleobiology of the group. Historically, E. prestvicii was the first trigonotarbid ever to be described—albeit originally as a beetle—and it is thus the oldest available name. Furthermore, a second specimen (Figs. 1–3) added by Woodward (1871) is exquisitely preserved and offers an unparalleled opportunity to reconstruct the appearance of one of the more anatomically interesting examples of these animals in life. Indeed casts of this important fossil are fairly common in museum displays and at one stage were even commercially available; often referred to as a fossil “spider”. The Eophrynidae family to which Eophrynus prestvicii belongs is probably among the more derived trigonotarbid lineages (cf. Dunlop and Brauckmann 2006). Thus new morphological data from this largely complete fossil should help to clarify both the ground pattern and affinities of the family.

Our chosen study method. X-ray micro-tomography (XMT), has recently emerged as a powerful tool for studying Coal Measures fossils hosted in siderite concretions (e.g., Garwood and Sutton 2010; Garwood et al. 2011). Essentially, the three-dimensional void left by the original organism within the encasing ironstone (siderite) concretion is scanned and reconstructed (Fig. 3). A pilot study, including E. prestvicii, was carried out by Garwood et al. (2009). Here, we aim to follow up the results of this work in more detail, complementing Garwood and Dunlop (2011) on another trigononotarbid family Anthracomartidae. Here we offer an assessment of the wider significance of the new morphological data recovered for E. prestvicii, supplemented by a detailed historical account, synonymy list and a novel reconstruction (Fig. 4). We also discuss the implications of our results for other eophrynid genera and species (Fig. 5) and—picking up a theme from Loman (1900)—we consider whether the often similarly or-namented laniatorid harvestmen (Opiliones: Laniatores) (Fig. 6) offer a reasonable analogue for the eophrynid mode of life.

Institutional abbreviations.—BU, Lapworth Museum of Birmingham University, Birmingham, UK; NHM, Natural History Museum, London, UK.

Other abbreviations.—XMT, X-ray micro-tomography.

Historical background

Prestwich (1834) mentioned a putative spider [as “Aranea”] from a concretion in the collection of Mr Antice [sic] from Madeley. This fossil was subsequently described by Buckland (1837) as Curculioides prestvicii in one of his Bridgewater Treatises “On the Power Wisdom and Goodness of God as Manifested in the Creation Volume IV: Geology and mineralogy considered with reference to natural theology”. The holotype originates from the Carboniferous ironstone of Coalbrookdale, Shropshire, UK and was named in honour of Joseph Prestwich (1812–-1896); the local geological historian who first noted the specimen (see above), and who later became a Fellow of the Royal Society. The owner of the holotype was presumably the same William Anstice, who inherited the Madeley Wood Ironworks, and several other mining and ironworking enterprises in the Coalbrookdale area. Although initially noted as an arachnid, this fossil—together with another now recognised as the first ever (fossil or living) ricinuleid (see Seiden 1992)—was interpreted by Buckland (1837) as a beetle belonging to the weevil family Curculionidae (Coleoptera). From the annotations on his original figure it is evident that Buckland (1837) interpreted both the right pedipalp and the right first leg as antennae. This left three limb pairs on the right side contributed towards the misinterpretation as an insect. The projecting anterior part of the carapace was regarded as a proboscis; i.e. the weevil rostrum. Extensive tuberculation on the upper surface of the body—which in this fossil is partly overlain by the smooth sternites—was also noted.

Hollier's specimen.—Woodward (1871) described a second fossil from the Coal Measures of Coseley, Staffordshire, UK; this time coming from a series of nodules supplied by E. Hollier of Dudley. This may well have been Elliott Hollier (1813–1905,  http://www.hollyer.info/elliott.php), a chemist based in the Market Place of Dudley who was much involved in local civic affairs and even became the town's mayor. As of 1862, an E. Hollier from the Market Pace is listed both as Honorary Secretary and Curator of the Dudley and Midland Geological and Scientific Society and Field Club. For a historical account of this organisation see Cutler (1981). A society museum was apparently established as of 1863 but it is unclear whether members' fossils were habitually deposited there, or remained in private hands (see below).

Woodward (1871: 1) referred to Hollier's specimen as a “very perfect Arachnide”. The exact circumstances of its discovery are not recorded, but it remains to this day one of the most impressive Carboniferous arthropods (Fig. 2). This specimen fully reveals both the dorsal and ventral surfaces. Woodward (1871) recognised that it was conspecific with Buckland's (1837) fossil and now correctly identified them as arachnids. His genus name, Eophrynus Woodward, 1871, is clearly derived from the modern whip spider (Amblypygi) genus Phrynus Lamarck, 1801, to which the fossils were compared. It was, however, explicitly treated by Woodward (1871) as a pseudoscorpion (Arachnida: Pseudoscorpiones). Although these Coal Measures fossils are an order of magnitude larger than all known living pseudoscorpions, Woodward's (1871) interpretation is easier to understand when the habitus of modern genera like Che iridium Menge, 1855 is taken into account (e.g., Weygoldt 1969: fig. 100). These extant pseudoscorpions also have a triangular prosoma and a rounded, and dorsally highly tuberculate, opisthosoma. Woodward (1871) correctly described numerous morphological details for E. prestvicii, including lobes on the carapace and the anteriorly projecting rostrum, as well as the tuberculation pattern on the opisthosoma and the four spines around its posterior margin. Nine opisthosomal tergites, but only seven sternites, were recognised, the sternites supposedly bearing pairs of putative tracheal openings.

Geinitz (1882) described a new (large) species of trigono-tarbid from Germany and again interpreted both his fossil and E. prestvicii as pseudoscorpions. Karsch (1882) rejected any similarities between Eophrynus and whip spiders and in the same paper raised a new, extinct order Anthracomarti. This was divided into two families (spellings as in the original). Architarboidae included his new trigonotarbid genus, Anthracomartus Karsch, 1882, together with fossils belonging to what would later be recognized as another extinct order, Phalangiotarbida. The second family, Eophrynoidae, was restricted to Eophrynus and diagnosed on having more tergites than sternites and a tuberculate dorsal surface. Haase (1890) adopted the modern spelling Eophrynidae and redefined the family primarily on the supposedly segmented nature of the carapace, the presence of epimera—i.e., tergites with median and lateral plates—and, oddly, the absence of trochanters. Loman (1900) proposed that trigonotarbids were closely related to various modern groups of harvestmen (Opiliones). Eophrynus was explicitly compared to the suborder Laniatores and a drawing of E. prestvicii was juxtaposed against the modern Sumatran harvestman Gnomulus segnipes (Loman, 1893) which does indeed have a similar overall appearance.

Lost and found.—Pocock (1902) offered the first formal redescription of E. prestvicii, although by this time Hollier's fossil was no longer available and Pocock (1902) had to rely on casts and drawings of the original. Pocock (1902) demonstrated that the carapace was lobed rather than truly segmented (contra Haase 1890), but was unable to identify any eyes. Ventrally, a sternum was recognised along with two putative cheliceral articles. The pedipalps were interpreted as long and slender with a short patella barely distinguishable from the tibia. The larger fourth pair of legs was noted, and the legs were described as having a pitted surface and longitudinal grooves, and in one case as supposedly ending in a single claw. The opisthosoma was described as having nine segments—the first short without division into median and lateral plates—with a discussion of possible segmental fusion (see also Discussion). Ventrally the first sternite was thought to be absent, but the tergites and sternites were largely matched to each other, with a discussion of the segmental interpretation of the “anal plate” (the pygidium in modern terminology) and its surrounding sclerites. “Elevations” were described on the posterior border of the third sternite and rows of tubercles (Woodward's [1871] putative tracheal openings) were also highlighted. Eophrynidae was retained in the extinct order Anthracomarti and redefined on, among other things, the presence of spines on the posterior margin of the opisthosoma.

Frič (1904) also received a cast of E. prestvicii and offered his own description and illustration. His interpretations largely matched those of Pocock (1902), but Frič (1904) also figured a supposed single tarsal claw of the legs and novel interpretations of the putative genital region on the ventral opisthosoma. Pocock (1911) reproduced his earlier drawings, but could add little to the previous description. However, centrally located eyes were now cited as present and used in the diagnosis of Eophrynus. The monographs of Petrunkevitch (1913, 1949) briefly mentioned E. prestvicii, but the latter work is of more significance for dividing the old order Anthracomarti into two groups (see Systematic palaeontology). The family Anthracomartidae were retained under the older name, while the other families, including Eophrynidae, were transferred to a new, supposedly unrelated. order Trigonotarbida. A further putative example of E. prestvicii was described by van der Heide (1951) from the Coal Measures of Limbourg in Belgium. Since it is preserved only in ventral view it is difficult to determine its generic affinities unequivocally. Further material from the same locality was assigned to “Eophrynus spec, indet”.

Petrunkevitch (1953) incorrectly reported Buckland's (1837) holotype as lost. It is actually in London (see Material and methods). However, Petrunkevitch (1953) did rediscover Woodward's (1871) specimen which, according to his notes, was in private hands until 1945 when it was presented to the University of Birmingham, UK by the Misses Tilley. Note, however, that Strachen (1979)—who was based at Birmingham University—dated this transfer to 1936 and gave the name as “Titley”. How the Titley's (or Tilley's) acquired Hollier's fossil is unfortunately not recorded. Woodward's (1871) specimen was designated the “lectatype” [sic], by Petrunkevitch (1953). In fact he should have treated it as a neotype (Woodward's [1871] specimen was not part of the original type series), but since the holotype still exists this neotype designation is now superfluous. Morphologically, Petrunkevitch (1953) further prepared Hollier's specimen and was able to reveal novel features such as the long anterior spine on the carapace, which Pocock (1902) misinterpreted as a chelicera. Petrunkevitch (1953) confirmed the presence of a pair of median eyes and described pitting on the both lateral lobes of the carapace and, like Pocock (1902), on the limbs. Nine opisthosomal segments were again recognised, but ventrally the raised structures on the third sternites were thought to be associated with the genital opening and were thus assigned to the second opisthosomal segment. While numerous authors have since mentioned E. prestvicii (see synonymy list) there have been no new studies of the original material until the provisional XMT results of Garwood et al. (2009); which we expand upon in detail here.

Material and methods

The holotype of E. prestvicii, from Coalbrookdale, Shropshire, UK (NHM In 49322) is not particularly well preserved, and given that dorsal and ventral features are partly superimposed it was not considered appropriate for tomography. Much better (Figs. 1–3) is Hollier's more complete specimen BU 699 (cf. Woodward 1871) from the nearby locality of Coseley near Dudley, Staffordshire, UK. Both these examples of E. prestvicii come from the British Middle Coal Measures. This can be dated to the Duckmantian substage of the late Carboniferous (ca. 311 Ma); equivalent to the Westphalian B of more traditional stratigraphic terminology. Heide's (1951) specimen is marginally older (Langsettian), but was not examined. As noted above, it is only known in ventral view making its assignment questionable. With respect to related trigonotarbids, the holotype of Eophrynus udus Brauckmann, Koch, and Kemper, 1985 is in the private collection of K.H. Hellwig in Hagen, Germany, but a cast and photographs were kindly provided by Carsten Brauckmann. The original comes from the Hagen-Vorhalle brick pit (Ziegelgrube), North Rhein-Westphalia, Germany and is dated at Marsdenian (= Namurian B; ca. 319 Ma). Comparative material assigned to the eophrynid genera Pleophrynus Petrunkevitch, 1945 and Stenotrogulus Frič, 1904 was studied by Dunlop (1994, 1995a) respectively. These papers contain further details of localities, stratigraphy and repositories.

Older descriptions of BU 699 can be found in Pocock (1902) and Petrunkevitch (1953). These are expanded here on the basis of the XMT scans (Fig. 3) and examination of the hand specimens. Comparisons with exceptionally preserved Devonian fossils from the Rhynie and Windyfield cherts of Scotland and the Gilboa mudstone of New York (Hirst 1923; Shear et al. 1987; Fayers et al. 2005) yield data on the morphology of presumably basal trigonotarbids and facilitate the interpretation of the larger and more derived Coal Measures species. Comparisons were also drawn with our recent XMT study (Garwood and Dunlop 2011) of the family Anthracomartidae and an overview of trigononotarbid morphology can also be found in Garwood and Dunlop (2010).

X-ray micro-tomography.—BU 699 was scanned on a Nikon HMX-ST (Natural History Museum, London). A tungsten reflection target with 200 mA current and 225 kV voltage was used, and 3142 projections taken with a 1.4 second exposure, and a 1 mm copper filter. A 2000 × 2000 detector panel provided a voxel size of 45 microns. Computer reconstructions were created with the custom software suite SPIERS (Sutton et al. 2012). Images were thresholded, and thresholds manually cleaned, prior to the assigning “masks”. These allowed different morphological features to be rendered as individual isosurfaces, and facilitated the removal from the model of the crack along which the nodule was split. Visualisation and iterative improvement allowed the creation of accurate models which were exported to the open source raytracing application Blender ( http://www.blender.org/) to create high resolution images (Fig. 3) and films (Supplementary Online Material, SOM at  http://app.pan.pl/SOM/app59-Dunlop_Garwood_SOM.pdf). The supplementary data also includes a ZIP archive containing the tomographic model of E. prestvicii described herein, using the VAXML interchange format (Sutton et al. 2012). This has a reduced triangle count, and should run on most computers. Free viewing software is available at  http://spiers-software.org/.

Reconstruction.—In addition to XMT a new, principally hand-drawn, idealised reconstruction is included (Fig. 4). This was constructed in traditional fashion with pen and ink, based on both the XMT model and the hand specimen. Colouration was added digitally—for a justification of the colours see Discussion—using the open source raster graphics editor GIMP. The scanned figure was first quarted to provide manageable image sizes, and then thresholded. The white pixels in the threshold, within the reconstruction were then selected, and coloured with the colouring tool. Textures were achieved by adding a textured layer above the reconstruction layer and combined using the overlay mode.

Systematic palaeontology

Fig. 1.

Historical images of Hollier's specimen of Eophrynus prestvicii (Buckland, 1837) from the British Middle Coal Measures (Late Carboniferous, Duckmantian; ca. 311 Ma) of Coseley near Dudley, Staffordshire, UK. Dorsal (A₁–C₁ and ventral (A₂–C₂) views. A, after Woodward (1871: pl. 11); B, after Pocock (1902: fig. 1); C, after Petrunkevitch (1953: textfigs. 82, 83).

Fig. 2.

Photograph of Hollier's specimen (BU 699) of triehotarbid arachnid Eophrynus prestvicii (Buckland, 1837), whitened with ammonium chloride to improve contrast. A. Dorsal view. B. Ventral view.

Fig. 3.

XMT-based reconstruction trigonotarbid arachnid Eophrynus prestvicii (Buckland, 1837), from scans of BU 699. A. Dorsal view. B. Posterior view, legs removed, to show heavy opisthosomal ornamentation and spines (maximum width 15 mm). C. Fourth walking limb with podomeres labelled. D. Pedipalps with podomeres labelled. E. Ventral view. 1–12, segment numbers.

Discussion

Chelicerae.—Pocock's (1902) observation of chelicerae in Eophrynus prestvicii was challenged by Petrunkevitch (1953) as a misidentification of the clypeus (his "spike of the carapace"). This is corroborated by the XMT results, which show that the true chelicerae are positioned—as might be expected—between the pedipalp coxae, and do not extend in front of the carapace as shown in Pocock's (1902) figures. The chelicerae have two articles, a basal pautron and a backwards-directed fang. The chelicerae in E. prestvicii are smaller than those seen in the contemporaneous anthracomartid trigonotarbids and their sinistral position may be a result of post-mortem displacement.

Pedipalps.—Another point of debate was the boundary in the pedipalp between the femur, patella and tibia. Woodward (1871) figured a patella about as long as the femur— implying rather elongate pedipalps verging towards being antenniform structures—while other authors figured a more "typical" short patella (Pocock 1911; Petrunkevitch 1953). The XMT model suggests the latter view is correct and while the resolution of the XMT scan is not optimal as a result of the large nodule in which the specimen is hosted, there does seem to be a bend and a slight swelling at the “knee”, which is more consistent with a short patella.

Legs.—On the walking legs Frič (1904) identified single terminal claw. This could not be confirmed in the XMT scans. A single tarsal claw is typical for certain subgroups of harvestmen (Opiliones), and may relate to Frič's (1904) interpretation of trigonotarbids as fossil harvestmen suborder. In fact most well-preserved trigonotarbids express a pair of large claws or ungues (e.g., Garwood et al. 2009) at the ends of the legs; sometimes with a smaller medial or empodial clawbetween the ungues.

Fig. 4.

Anew reconstruction of trigonotarbid arachnid Eophrynus prestvicii (Buckland, 1837), with colouration based on modern laniatorid harvestmen (Opiliones: Laniatores) (see also Fig. 5). Not to scale.

Segmentation.—Trigonotarbid opisthosomal segmentation has traditionally proven problematic. For example Petrunkevitch (1955) argued that, across the group as a whole, the total number of opisthosomal segments varies between eight and eleven. The implications of such an inconsistent character state for trigonotarbid monophyly were critically discussed by Shear et al. (1987). Interpretations are complicated by the fact that in many trigonotarbid families the first tergite appears to be reduced in size and/or partially hidden, tucked under the posterior margin of the carapace as a functional "locking ridge" holding the prosoma and opisthosoma together. Dunlop (1994, 1995b) speculated that the locking ridge may be secondarily absent in the Eophrynidae and that a plesiomorphic fused diplotergite (2+3) may have undergone a reversal back to two separate sclerites. The present study demonstrates that this view is incorrect. Tergite one is retained as a very narrow and reduced dorsal element, which has apparently lost the locking function which may be part of the trigonotarbid ground pattern. Similarly there is no evidence of a diplotergite.

Fig. 5.

Comparative images of modern South American laniatorids as potential ecological analogues for Carboniferous eophrynids. Although these harvestmen are not particularly closely related to trigonotarbids, a number of them also express a tuberculate dorsal body surface and marginal spination of the opisthosoma; both of which presumably deter predation by increasing handling time. All images courtesy of Ricardo Pinto da Rocha (São Paulo).

As noted above, trigonotarbids are currently placed in the Pantetrapulmonata group which have a ground pattern character of two pairs of book lungs; later modified in some ingroups such as the more derived spiders. The Devonian Rhynie Chert trigonotarbids in the family Palaeocharinidae clearly preserve the ground pattern condition, with two pairs of book lungs opening on the second and third opisthosomal segments respectively (e.g., Kamenz et al. 2008). For this reason Woodward's (1871) putative tracheal openings on sternites 5–9 of E. prestvicii—while poorly resolved in the scans—seem likely to be a misidentification.

Respiratory organs and ventral sacs.—Another controversial feature is a pair of structures on the anterior underside of the opisthosoma, towards the anterior end. First recognised by Pocock (1902), he described them as “arcuate crests” and speculated that they could be associated with respiratory openings. Frič (1904) interpreted them—somewhat speculatively—as openings leading into some sort of a sperm-storage device: a receptacula seminis. Petrunkevitch (1953) preferred to interpret them as part of the second opisthosomal somite and thus explicitly associated with a putative genital opening. Similar raised lobes are seen in the Rhynie chert palaeocharinids (Fayers et al. 2005: fig. 5), and here they clearly occur on the third (i.e., postgenital) segment. These authors suggested that the trigonotarbid structures could be ventral sacs; enigmatic, eversible structures which occur in a topographically homologous position in many whip spiders (Amblypygi) (e.g., Weygoldt 2000: fig. 5). Ventral sacs are conceivably highly modified appendage remnants and are also seen in palpigrades (Palpigradi), albeit here on multiple segments. In whip spiders the ventral sacs are thought to play a role in water balance, but since their physiology is uncertain even among living arachnids further interpretations of the fossil structures would be speculative. Irrespective of their function, our computer model resolves them as a pair of small lobes either side of the midline of the third sternite (or posterior operculum); just anterior to sternite four. Confirmation that they belong to the third opisthosomal segment also argues against them being part of the primary genitalia given that the arachnid gonopore consistently opens on segment two. Unlike Anthracomartidae (images in Garwood and Dunlop 2011), there is no anterior curved transverse ridge associated with these ventral sac structures.

Pygidium.—The almost-triangular posteriormost sternite hosts the pygidium: the anal plate of Pocock (1902) or the anal operculum of Petrunkevitch (1953). The trigonotarbid pygidium (or postabdomen in some terminologies) is not, in fact, plate-like, but consists of two ring-like segments: topographically numbers 11 and 12. This has been adequately demonstrated in other material, including the Rhynie chert palaeocharinids (Hirst 1923; Fayers et al. 2005), and the same ground pattern morphology of a twelve-segmented opisthosoma can be confirmed here for eophrynids too.

Mode of life.—Trigonotarbids and harvestmen clearly resolve at quite different places on the arachnid tree (Shultz 2007), but taking up the theme of Loman (1900) some laniatorid harvestmen appear to offer good modern analogues for eophrynids. Numerous laniatorid species also bear dorso-marginal spines and prominent dorsal tubercles. Some laniatores are also brightly patterned, while others are more conservative in their coloration. It is thus interesting to speculate whether eophrynids employed aposematism (i.e.; warning colouration) or whether they were more cryptic animals, camouflaged against the substrate. For our new reconstruction (Fig. 4) a major consideration was whether Carboniferous predators were largely visual, to appreciate (or avoid) deimatic/aposematic coloration, or whether they used other means or locating prey. The evidence here is inconclusive. It is likely that early tetrapods possessed chromic vision (Bailes et al. 2007; Hart et al. 2008), and there is evidence of other forms of aposematism is some Carboniferous arthropod taxa, such as the ozopores (and thus chemical defences) in euphoberiid millipedes (Shear and Edgecombe 2010). However, there is no evidence for chemical, or any other form of aposematism in any known trigonotarbid, and mimicry is common in Carboniferous arthropod taxa (Scott and Taylor 1983). Furthermore, trigonotarbids were probably habitual predators based on their somewhat spider-like mouthpart structure (Garwood and Dunlop 2010), and excessively bright coloration could have been detrimental to their ability to catch prey. Thus we chose a more conservative model—in keeping with the likely background colours of a Carboniferous forest floors, but with highlighted colour to provide an element of disruptive crypsis (Jarzembowski 2005). As a basis we used a range of colours (yellows and browns) seen in laniatored harvestmen such as Acutisoma longipes (Gonyleptidae, Gonyleptoidea) (e.g., Giribet et al. 2010), with pale appendage membranes in keeping with many modern arachnids.

Anatomical similarities between laniatorid harvestmen and the heavily ornamented eophrynid trigonotarbids may also imply that these extinct arachnids had similar mode of life on the forest floor. In general, the opisthosoma of E. prestvicii resolves under XMT as relatively flat and disc-like, similar to the situation in Anthracomartidae (Garwood and Dunlop 2011). While there may have been a certain degree of post-mortem compression we still feel that the XMT results largely reflect the appearance of the animal in life, and that a flattened body may have been advantageous for crawling into narrow spaces and/or living within the detritus. The length of the limbs and lack of apparent raptorial adaptation (such as that seen in the Anthracomartidae) suggests E. prestvicii could have been a cursorial predator.

Eophrynus prestvicii is amongst the most heavily ornamented trigonotarbids. The prosomal transverse ridges of cuticle are likely to be an adaptation for strength, but the robust ornamentation of E. prestvicii—and other eophrynids—is clearly a defensive adaptation which would have increased the handling time for predators and made these trigonotarbids generally less palatable. This is further supported by their high-risk mode of life on the forest floor, and the likely inertial feeding mechanism and lack of differentiated teeth of their likely predators (Reilly and Lauder 1990). The Eophrynus ornamentation would also have provided protection against attacks from above and helped prevent crushing. This adaptation is not seen in the Anthracomartidae, which may instead have been ambush predators (Garwood and Dunlop 2011)—a comparatively safe mode of life. Both these adaptations (i.e., defensive ornamentation and ambush predation) could be a response to increasing predation at a time when tetrapod predators were becoming more numerous (Shear and Kukalová-Peck 1990). Ambush predation would allow anthracomartids to spend much of their time in the relative safety of hides, waiting for prey—in contrast to being in a vulnerable position in the leaf litter where Eophrynus, with no such behavioural specialisations, would be found. Thus it seems that the Eophrynidae and associated taxa could have responded instead by increasing their defences.