X-ray microtomography of the late Carboniferous whip scorpions (Arachnida, Thelyphonida) Geralinura britannica and Proschizomus petrunkevitchi

Abstract

Whip scorpions (Thelyphonida) and schizomids (Schizomida) are closely related arachnid orders, whose low species diversity compared to other arachnid groups is reflected in a limited fossil record. Here we investigate two key fossil whip scorpions from the British Middle Coal Measures of Coseley, Staffordshire (late Carboniferous, c. 315 Ma), UK, using X-ray microtomography, and incorporate the taxa into an updated arachnid phylogeny. Geralinura brittanica Pocock, 1911 is an unequivocal whip scorpion, whose pedipalps are closer to the modern condition than previously assumed. The trochanter bears a dorsal flange with prodorsal teeth, and the whole pedipalp is rendered subchelate by the presence of at least a patellar apophysis. These results suggest a high degree of stasis in whip scorpions since the late Carboniferous, and the genus Geralinula Scudder, 1884 is within the crown-group; i.e. the extant family Thelyphonidae. Proschizomus petrunkevitchi Dunlop and Horrocks, 1996 lacks median eyes, a character shared with Schizomida, but unlike schizomids the pedipalp also has patellar and tibial apophyses; albeit with the appendage articulating at an angle of c. 45° rather than articulating horizontally as in living whip scorpions, or vertically as in schizomids. Our phylogeny refutes a previous hypothesis that P. petrunkevitchi is a stem-schizomid, and places it among the whip scorpions instead. However, the species may have branched early from the whip scorpion lineage, and thus its morphology might reflect a plesiomorphic arrangement in which the pedipalps were transitioning from a more leg-like to a fully raptorial and claw-like set of prey-catching appendages.

Keywords:

• Coal Measures
• fossil
• systematics
• Schizomida
• pedipalps

Introduction

The arachnid orders Thelyphonida (the whip scorpions, also known as vinegaroons) and Schizomida (schizomids, short-tailed whip scorpions or micro whip scorpions) are evidently closely related. For many years they were treated as a single order, usually under the name Uropygi, although some authors (e.g. Clouse et al., 2017; Harvey, 2003; Rowland & Cooke, 1973) continue to use Uropygi (or Uropygida) for whip scorpions only. The (Thelyphonida + Schizomida) clade is characterized by several synapomorphies (Shear et al., 1987; Shultz, 1990, 2007; Weygoldt & Paulus, 1979) such as fusion of the pedipalpal coxae into a single structure (the camerostome), the presence of repugnatorial glands opening close to the base of the telson, and similar mating behaviour. Summaries of whip scorpion and schizomid biology can be found in Haupt (2000) and Reddell and Cokendolpher (1995), respectively. The oldest whip scorpions come from the late Carboniferous Coal Measures of Germany (c. 319 Ma; Brauckmann & Koch, 1983). By contrast, the smaller and less heavily sclerotized schizomids are less likely to fossilize. As a result, schizomids have a limited fossil record: they are first recorded from several extinct genera in mid-Cretaceous (c. 99 Ma) Burmese amber (Müller et al., 2019; De Francesco Magnussen et al., 2022).

While there seems little doubt that whip scorpions and schizomids are sister-taxa, the ground pattern of their common ancestor is less obvious. Both lineages belong to the wider Pantetrapulmonata clade (Shultz, 2007) which also includes whip spiders (Amblypygi), spiders (Araneae), and the extinct groups Trigonotarbida, Haptopoda and the spider-like Uraraneida. Although often resolved as derived pantetrapulmonates (Shear et al., 1987; Shultz, 2007; Wang et al., 2018) on the basis of characters such as their raptorial pedipalps, both whip scorpions and schizomids also retain some putatively plesiomorphic characters. These include the postanal telson and the possible retention of a basal suture on the chelicerae of whip scorpions (Haupt, 2009a; Kästner, 1949), which may reflect the original three-articled condition of the chelicerae in the arachnid ground pattern. This suture is also seen in schizomids (Krüger, 2011). Thelyphonida are usually fairly large (c. 25–74 mm body length), which raises the question whether the smaller (c. 3–15 mm) Schizomida are essentially derived whip scorpions in miniature. As discussed by Weygoldt and Paulus (1979), schizomids have a divided prosomal dorsal shield which lacks median eyes; both of which have been assumed to be derived character states within arachnids.

Two fossil species which may shed light on this debate are Geralinura britannica Pocock, 1911 and Proschizomus petrunkevitchi Dunlop and Horrocks, 1996, both from the c. 315 Ma British Middle Coal Measures of the English West Midlands. Both are preserved three-dimensionally within ironstone (siderite) concretions. This three-dimensionality means that some details, such as the exact shape, orientation and ornamentation of the limbs, remain hidden within the matrix and could not be extracted using traditional study methods based on photographs and drawings (Dunlop & Horrocks, 1996; Petrunkevitch, 1949; Pocock, 1911). X-ray microtomography (μCT) has proven to be a powerful tool that can overcome these limitations, which has been successfully applied to a range of Coal Measures fossils preserved in concretions (e.g. Garwood et al., 2009; Garwood & Sutton, 2010, 2012; Jones et al., 2014; Selden et al., 2016a). ‘Virtual fossils’ are reconstructed by scanning the void left by the animal within its encasing concretion; for an overview see Sutton et al. (2014). Under ideal circumstances, μCT can recover novel, phylogenetically informative characters as well as giving a better feel for the habitus of an animal in life.

Here, we use tomography to re-study the two fossils assigned to P. petrunkevitchi, a Carboniferous species interpreted as potentially being more closely related to schizomids than thelyphonids (Dunlop & Horrocks, 1996). Our re-investigation has revealed that one of the fossils was incorrectly identified (see also Material and methods) and actually represents G. britannica as per its original assignment by Pocock (1911). The present study also received impetus from the molecular phylogeny of Clouse et al. (2017). Their analyses supported a sister-group relationship between whip scorpions and schizomids, and inferred that they both probably originated during the late Carboniferous (c. 299–323 Ma); a date consistent with the oldest fossil whip scorpion at 319 Ma. By reconciling fossil data with modern distributions, the place of origin for these animals was suggested to be tropical Pangea, a region corresponding to modern day North America, Europe, Africa and South America (Clouse et al., 2017, fig. 4). This is again consistent with Carboniferous whip scorpions, which have been recovered from North America and several countries in Europe (see below). Here, we use the novel characters recovered by tomography to integrate these two Coal Measures whip scorpions into a phylogenetic dataset derived from Garwood and Dunlop (2014), and thus test the position of the fossils in relation to their modern counterparts.

Historical background

Early work on Carboniferous whip scorpions was summarized by Dunlop and Horrocks (1996) and Tetlie and Dunlop (2008). The genus Geralinura Scudder, 1884 was raised for a North American species from the Coal Measures of Mazon Creek. Geralinura britannica Pocock, 1911 was subsequently described from the British West Midlands. Four specimens were available to Pocock. Two of them, one of which is obviously the holotype, were figured by Pocock (1911, pls I, II). A third specimen was included as a text-figure (Pocock, 1911, text-fig. 9). This has a more pointed prosomal dorsal shield, which Pocock thought had broken anteriorly. All four specimens were assumed to be conspecific, a conclusion supported by Petrunkevitch (1949) who re-examined the material and offered new descriptions of both the holotype and the specimen in Pocock’s text-figure. Subsequently, Petrunkevitch (1953) transferred the British species to a different genus, Prothelyphonus Frič, 1904, originally created for some spectacular fossil whip scorpions from the Carboniferous of Bohemia in the Czech Republic. This transfer was justified on the supposition that both the Bohemian and British species retained eyes. Geralinura was redefined as whip scorpions without eyes, and restricted to a suite of rather ill-defined American species from Mazon Creek.

Dunlop and Horrocks (1996) rejected Petrunkevitch’s transfer of the British species to Prothelyphonus. They also recognized that the specimen in Pocock’s (1911) text-figure with a pointed dorsal shield is not damaged, but is actually quite well preserved. The pointed tip and the absence of median eyes on the dorsal shield appear to be genuine features which differentiate this from the holotype of Geralinura britannica. A new genus and species, Proschizomus petrunkevitchi, was proposed using the specimen in Pocock’s text-figure as the holotype. A second specimen was also assigned to P. petrunkevitchi as it appeared to lack median eyes too; although the present study now reveals this was a misinterpretation (see below). As its name implies, the authors suggested that Proschizomus petrunkevitchi could be a stem-schizomid with loss of the median eyes proposed as a potential synapomorphy of (P. petrunkevitchi + Schizomida). In this scenario Schizomida would be defined by a divided prosomal dorsal shield as per Dunlop and Horrocks (1996, fig. 6).

Tetlie and Dunlop (2008) retained Geralinura for both Pocock’s British and Scudder’s Mazon Creek fossils. Prothelyphonus was restricted to the large and slender Bohemian material. A new genus, Parageralinura Tetlie and Dunlop, 2008, was proposed for two further fossil whip scorpions described previously from the Netherlands (Laurentiaux-Vieira & Laurentiaux, 1961) and Germany (Brauckmann & Koch, 1983). Proschizomus Dunlop and Horrocks, 1996 was retained for the British species without median eyes, but bearing a pointed tip to the prosomal shield. Tetlie and Dunlop (2008) also suggested that in all well-preserved Coal Measures whip scorpions the pedipalps do not appear to be subchelate, and apparently lack the distinct inwards-facing projections (apophyses) seen on the patella and tibia in modern species. For this reason, the Coal Measures taxa were placed as plesion genera, outside the family Thelyphonidae, with this family rediagnosed as comprising whip scorpions with patellar and tibial apophyses.

Since the publication of Tetlie and Dunlop (2008), whip scorpions unequivocally belonging to the crown-group family Thelyphonidae have been found in Cretaceous Burmese amber (Cai & Huang, 2017; Wunderlich, 2015), and two further Coal Measures whip scorpions have been described. Selden et al. (2014) reported a fossil assigned to Thelyphonida sp. from the Carboniferous of the Ukraine. It is only known from the dorsal shield, although this is quite modern-looking. Selden et al. (2016b) added a new species of Parageralinura from the Carnic Alps of Italy.

Material and methods

The specimens scanned herein were obtained from their repository in the Natural History Museum, London (NHMUK). The holotype of Proschizomus petrunkevitchi (NHMUK PI In 7912, ex William Madeley collection) is a former paratype of G. britannica (see Morris, 1980). We also scanned the second specimen (NHMUK PI In 31265, ex Walter Egginton collection) which Dunlop and Horrocks (1996) also assigned to their new species. The CT-scans reveal characters, such as the eyes and the structure and orientation of the pedipalps, which indicate that these fossils actually represent two different species. NHMUK PI In 31265, which is also one of the original paratypes of G. britannica, was misidentified by Dunlop and Horrocks (1996) and should again be treated as a paratype example of G. britannica. The CT scans in turn facilitated a comparative study of the two British Carboniferous whip scorpion species. Both fossils investigated here originate from Coseley, Staffordshire, UK. They are preserved as voids within ironstone (siderite; FeCO₃) concretions, and belong stratigraphically to the British Middle Coal Measures. These are conventionally dated to the Duckmantian stage of the late Carboniferous (the Westphalian B of older terminologies), which translates to an absolute age of approximately 315 Ma (Pointon et al., 2012).

There are relatively few detailed accounts of whip scorpion and schizomid morphology; see also comments in Reddell and Cokendolpher (1995). The two classic studies are those of Börner (1904) and Hansen and Sørensen (1905), but see also Werner (1935). These are supplemented by more recent overviews in van der Hammen (1989; and references therein) and Shultz’s (1993) detailed study of skeletomuscular anatomy in whip scorpions. The morphological terminology employed herein also draws on Barrales-Alcalá et al. (2018), and the references on the systematics of modern species contained within that work. Comparative modern whip scorpions and schizomids were studied from the collections of the Museum für Naturkunde, Berlin. Live schizomids collected by Karl-Hinrich Kielhorn (Berlin) – primarily from the reptile house of the Berlin Tierpark – were also available for study. Although these animals are not native to Europe, they are often encountered in botanical gardens, reptile houses and similar artificial environments (e.g. Armas & Rehfeld, 2015).

Imaging

We scanned both fossils at the NHMUK using a Nikon HMX-ST 225 scanner and a tungsten reflection target with a 0.1 mm copper filter. For both scans, 3142 projections were collected, and then reconstructed in CTPro V2.1. NHMUK PI In 7912 was scanned with a source current/voltage of 205μA/170kV and the resulting volume has a 17.6 μm voxel size. For NHMUK PI In 31265 we used a source current/voltage of 205 μA/175kV, and the dataset comprises 19.2 μm voxels. Following the methods of Garwood et al. (2012), we created a digital visualization using the SPIERS software suite (Sutton et al., 2012). For publication images we raytraced isosurfaces in Blender (Garwood & Dunlop, 2014), and created missing elements of the limbs using that software’s mesh creation tools. In the figures and Supplemental animations these are rendered partially transparent. Photographs show the specimens whitened using ammonium chloride, as per Dunlop and Horrocks (1996).

Phylogeny

Proschizomus petrunkevitchi and Geralinura brittanica were scored into a dataset modified from that of Garwood and Dunlop (2014), and developed through the course of a number of further publications (Garwood et al., 2016, 2017; Wang et al., 2018). To better sample the breadth of morphology within the schizomids and thelyphonids we added a number of terminals: the extant Schizomida genera Schizomus, Agastoschizomus, Protoschizomus, and the Burmese amber genus Mesozomus, and for Thelyphonida Typopeltis, Hypoctonus and Mayacentrum. Within the family Hubbardiidae, we changed the terminal representing the genus Trithyreus to the genus Mayazomus, which has a better-reported morphology; see e.g. Monjaraz-Ruedas and Francke (2016).

In order to clarify the relationships within the thelyphonids and schizomids in the phylogeny, we added the following characters (full character descriptions for the current analysis are included in the Supplemental material): seta at base of anterior process of propeltidium (character 14), present in family Protoschizomidae (Cokendolpher & Reddell, 1992; Monjaraz-Ruedas et al., 2017); number of annuli of the female flagellum (character 37), which provides some resolution within the Schizomida (Cokendolpher & Reddell, 1992); dorsal depressions on male flagellum, which is present in some members of the schizomid family Hubbardiidae (Monjaraz-Ruedas & Francke, 2016; character 36). Coding was also updated for the Coal Measures taxa within based on these models – in particular for characters 15 (prosomal dorsal shield with keels), 60 (palpal apophyses) and 61 (male patellar apophysis modified). Furthermore, as a result of ongoing development of this matrix, several new characters have been added which do not impact upon the taxa reported herein. We outline these additional characters in the Supplementary material.

We analysed the matrix (included in the Supplemental material in formats ready for TNT and MrBayes) using both Bayesian and maximum parsimony approaches. Bayesian analysis was conducted in MrBayes v.3.2.6 (Ronquist et al., 2012) using the Lewis (2001) discrete (morphology) model, ascertainment (coding) bias variable and a gamma-distributed rate model. This was implemented using two runs, each comprising one cold and three heated chains. The analysis sampled every 500 generations, and ran for 5,000,000 generations, the first 25% burn in of which was discarded. Figure 3 shows the 50% majority rule consensus output by MrBayes. This was imported into R (R Core Team, 2019), and the package STRAP (Bell & Lloyd, 2015) was used to plot the topology against a geological timescale: branch lengths are derived from the fossil ages (also in the Supplemental material) and the equal method in STRAP. Supports are posterior probabilities. Convergence was assessed for both runs in MrBayes (PRSF 1.0) and using Tracer 1.6 (Rambaut et al., 2018; overlapping traces, effective sample size >1200).

Parsimony analysis was conducted using TNT v.1.5 (Goloboff & Catalano, 2016; made available with the sponsorship of the Willi Hennig Society). The characters are unordered where multistate, and we conducted a tree bisection-reconnection (TBR) search (100 trees saved per cycle for 1000 replicates). We present topologies using both equal and implied weighting schemes (Supplemental material Fig. 2 shows k = 3; the results presented are unaffected across a range of k values between 0.25 and 10). Supports, calculated in TNT, represent absolute frequencies, and for equal weights comprise jackknife (33% removal probability, 10,000 replicates; Farris et al., 1996) then bootstrap (10,000 replicates; Felsenstein, 1985) values. Implied weight supports employ symmetric resampling (change probability 33%, 10,000 replicates, Goloboff et al., 2003).

Strict consensus trees were created in TNT and exported as a Nexus file, then imported into FigTree (Rambaut, 2020) to create the Supplemental figures. The clade inset on Supplemental Fig. 2 was created via the R workflow used to plot the Bayesian figures. Figures were prepared for publication in Inkscape and GIMP.

Morphological interpretation – Geralinura britannica

Prosoma

The prosomal dorsal shield, or carapace in traditional terminologies, is a single unitary plate in G. britannica (Fig. 1A). The dorsal shield is longer than wide, the anterior margin is slightly procurved and the posterior margin is straight. The lateral margins curve outwards slightly, giving the whole structure an elongate, roughly oval appearance. The shield shows faint sulci, similar to modern species, defining an anterior region, and these sulci come together near the middle of the shield to form a faint depression which probably acted as a muscle insertion site (or fovea). The CT scan confirms the presence of both median and lateral eye tubercles, as in modern whip scorpion species. The median eyes consist of two lenses (Fig. 1B), although we note these are better resolved in other specimens described by Dunlop and Horrocks (1996). The number of lateral eye lenses is equivocal. Ventral prosomal structures were not well resolved, although a large sternum in the shape of an inverted triangle – the anterior sternum sensu Shultz (1993) – can be seen, and we refer here to the hand specimen description by Dunlop and Horrocks (1996) for other features.

Figure 1. A, reconstruction of a paratype of Geralinura brittanica (NHMUK PI In 31265) in dorsal view. B, expanded view of the dorsal anterior of G. brittanica with the median eyes outlined. C, photograph of G. brittanica paratype, part. D, photograph of G. brittanica paratype, counterpart. E, right chelicera of G. brittanica. F, left pedipalp of G. brittanica. G, G. brittanica in frontal view. Abbreviations: 1–4, legs 1–4; Ap, apophysis; Fe, femur; Fl, flange; LK, lateral keel; ME, median eyes; Pa, patella; Pp, pedipalps; Ta, tarsus; Ti, tibia; Tr, trochanter. Scale bars: A–D = 5 mm; E, F = 1 mm; G = 2 mm.

Chelicerae

The chelicerae in G. britannica were only hinted at previously (e.g. Dunlop & Horrocks, 1996, fig. 23). The CT scan now demarcates them quite clearly (Fig. 1E) and reveals that they project some distance in front of the prosomal dorsal shield. However, this may not be their life position as the chelicerae in whip scorpions can be voluntarily extended forward when, for example, masticating food. Thus, if the chelicerae were pushed out during slight compression of the dead animal in the sediment, their position here could be a post-mortem effect. Something like this is seen in, for example, the holotype of G. carbonaria Scudder, 1884 from Mazon Creek (Tetlie & Dunlop, 2008, fig. 3). Our CT scan resolves the chelicerae of G. britannica as ‘clasp-knife’ structures, as is typical for members of the Pantetrapulmonata, in which a basal element articulates against a slightly curving fang. There is evidence for at least one large tooth on the proximal article in the left chelicera and the overall morphology is similar to that of living whip scorpion species. Although two cheliceral articles are obvious, the scans do not allow us to test for the presence of a third basal element sensu Haupt (2009a).

Figure 3. Results of the cladistic analysis presented herein under equal weights parsimony and Bayesian inference. Top: topology within the pantetrapulmonates, in particular between the Haptopoda, Amblypygi, Thelyphopnida and Schizomida, in both parsimony and Bayesian analyses. Bottom: the relationships recovered for all arachnid and chelicerate orders using the topology from the Bayesian analysis (the arachnid-wide parsimony topology is included in the Supplemental material). Support values are bootstrap/jackknife (parsimony) or posterior probabilities (Bayesian); plotted against geological time using equal branch lengths between fossil taxa (see Methods). Taxon images either drawn for this publication, or from Lozano-Fernandez et al. (2019).

Pedipalps

The CT scans offer particularly valuable insights into the structure of the pedipalp. As in living whip scorpions the principal plane of movement in G. britannica is horizontal; i.e. a side-to-side movement (Fig. 1G). The expected limb articles of trochanter, femur, patella, tibia and tarsus can be resolved (Fig. 1F). The prodorsal margin of the trochanter, which is always developed into a series of spines in modern species, also protrudes markedly in G. britannica and forms an approximately diamond-shaped flange. In detail, there are hints of four small prodorsal teeth. As in modern forms, this structure essentially creates dentition immediately in front of the chelicerae and is presumably something against which captured prey items could have been pulled in order to help kill and masticate them. Another significant and novel discovery from the CT data (Fig. 1) is the presence of a patellar apophysis; contra the interpretation of Tetlie and Dunlop (2008) based on hand specimens and published illustrations only. This apophysis renders the pedipalp of G. britannica subchelate (i.e. allows the distal podomeres to articulate against this apophysis) and means that it is very similar to the pedipalp of modern whip scorpion species. The only possible difference compared to modern species is the absence of a similar tibial apophysis, which in modern species yields a small chelate tip to the pedipalp. However, the preservation of the pedipalps lowers in quality distally, and as such the presence of a tibial apophysis is equivocal rather than being absent.

Legs

The CT scans clarify the limb morphology of G. britannica, at least down to the patella of leg I and the tibiae/metatarsi of legs II–IV. They confirm that leg I was long and slender, as in living species, where this elongate first leg is used more as a tactile appendage to search for prey. Whip scorpions (and schizomids) thus walk hexapodally on the remaining three leg pairs. In G. britannicus legs II–IV are preserved with the prolateral surface uppermost. This could be close to their life position as the same habitus is often seen in living whip scorpions where it may facilitate crawling into burrows, or other narrow spaces. Legs II and III in the fossils are shorter and less gracile than leg I. As in living species, the patella is invariably shorter than the adjacent femur and tibia. None of the scans reveal the full length of the tarsi, but as in living species the tarsi of legs II–IV are likely to have been fairly short and divided into three tarsomeres. Leg IV in G. britannicus is noticeably longer and more robust than the other walking legs. The relative size of legs II–IV in extant whip scorpion taxa varies between subfamilies. In some species legs II–IV are of a similar size, and leg IV is not markedly longer or larger; see e.g. habitus photographs of Hypoctoninae in Huff and Prendini (2009) and of Mastigoproctinae in Barrales-Alcalá et al. (2018). In other subfamilies leg IV does appear to be a little longer, i.e. Thelyphoninae and Typopeltinae; see e.g. figures in Seraphim et al. (2019).

Opisthosoma

The opisthosoma of G. britannicus is preserved partially crushed. It comprises 12 segments, although the smaller first tergite cannot be clearly resolved. The opisthosoma has an elongate–oval shape in outline. It narrows anteriorly towards the prosoma and posteriorly towards the terminal segments forming the postabdomen. Both tergites and sternites have straight posterior margins where preserved and lack obvious ornament (e.g. tuberculation, etc.). The postabdomen is composed of three ring-like elements (segments 10–12) which increase in length from anterior to posterior. This postabdomen is a prominent feature in G. britannica and Tetlie and Dunlop (2008) rediagnosed the extinct genus Geralinura as Carboniferous whip scorpions with a large postabdomen in which the terminal segment is longest. The whip scorpion telson is a long, annulated flagellum: the ‘whip’ in their common name. In those whip scorpion fossils where a telson is preserved the structure is also whip-like (e.g. Pocock, 1911). Our specimen of G. britannicus has very little of this preserved: there is a small slender element behind the postabdomen also picked up by Dunlop and Horrocks (1996, fig. 19). This probably represents the telson. Because the opisthosoma terminates near the edge of the nodule, neither the hand specimen nor the CT scan reveal any more of this structure.

Morphological interpretation – Proschizomus petrunkevitchi

Prosoma

The prosomal dorsal shield of P. petrunkevitchi is a single unitary plate, and is not divided into a pro-, meso- and metapeltidium as in schizomids (Fig. 2). The small anterior spine at the front of the dorsal shield is better seen in the hand specimen (Dunlop & Horrocks, 1996, fig. 17) as compared to the CT scan. The scan does, however, clarify the topology of the dorsal shield. It is longer than wide, and dominated by a raised ‘V’-shaped region, quite different from the morphology seen in any known whip scorpions (or schizomids), and thus a good diagnostic character for the genus Proschizomus. A slit-like depression, probably a muscle apodeme, towards the middle of the dorsal shield slots in at the base of the ‘V’. Anteriorly, there is no evidence in hand specimen or CT scan for median eyes or a median eye tubercle, and we interpret these eyes as being wholly absent. However, at the margins of the raised ‘V’ there is a pair of raised tubercles which are consistent with being lateral eye tubercles as they are in the same position as the (rare) examples of eyes seen in some living schizomids (e.g. Cokendolpher et al., 1988; Sissom, 1980) and in several of the Burmese amber fossil schizomids (De Francesco Magnussen et al., 2022; Müller et al., 2019). Any individual lenses on these tubercles in P. petrunkevitchi could not be resolved. The chelicerae could not be resolved in the scan either, and are also not obvious from the hand specimen.

Figure 2. A, reconstruction of the holotype of Proschizomus petrunkevitchi (NHMUK PI In 7912) in dorsal aspect. B, photograph of the P. petrunkevitchi holotype, part. C, photograph of the P. petrunkevitchi holotype, counterpart. D, left pedipalp of P. petrunkevitchi. E, P. petrunkevitchi in frontal view. Translucent elements are inferred through symmetry or comparison with extant taxa. Abbreviations: 1–4, legs 1–4; Ap, apophysis; Fe, femur; Pa, patella; Pp, pedipalps; Ta, tarsus; Ti, tibia; Tr, trochanter. Scale bars: A–C = 5 mm; D = 1 mm; E = 2 mm.

Ventrally, the camerostome – the large fused coxae of the pedipalps – is well resolved, each coxa bearing a slight protrusion on the mesal side close to the midline. This feature is typical for both living whip scorpions and schizomids. There is a large roughly diamond-shaped sclerite between the leg II coxae; the anterior sternum sensu Shultz (1993). A posterior sternum between the leg III coxae cannot be resolved, although this is unsurprising as the element is tiny in the living animals. Again following the interpretation of Shultz (1993), the sub-triangular sclerite between the leg IV coxae represents the sternite of opisthosomal segment 1.

Pedipalps

An interesting aspect of P. petrunkevitchi is the orientation of the pedipalps. The hand specimen appears to show pedipalps primarily articulating up-and-down, more like a schizomid, as opposed to side-to-side as in modern whip scorpions (W. Shear, pers. comm. to JAD). We set out to test this important character further, and the scans now reveal that the orientation in NHMUK PI In 7912 is essentially intermediate between the whip scorpion and the schizomid condition. This is best seen in a frontal view of the 3D image (Fig. 2G), which reveals pedipalps articulating at a c. 45° angle to the horizontal. The palpal trochanter uniquely bears a single ventral spine towards the anterior end. Another important character is the degree of pedipalp ornamentation. As noted above, whip scorpions (now including G. britannica) have subchelate pedipalps with a patellar apophysis, while schizomids have functionally raptorial but essentially smooth pedipalps. The scans of P. petrunkevitchi now reveal that a small patellar apophysis was present in this Carboniferous species too (Fig. 2D), and an even smaller tibial apophysis was probably present as well. In this sense the pedipalps are more like those of a whip scorpion than a schizomid, albeit with less pronounced apophyses in the fossil. A further feature of interest is the tarsus. In whip scorpions there is no differentiation into a tarsus and apotele (or claw), but this differentiation is retained in schizomids. The scans are largely equivocal on this feature, but a faint line provides hints that a separate apotele was present in P. petrunkevitchi too.

Legs

The legs of P. petrunkevitchi are similar in shape and build to those of G. britannica. All legs are preserved at least down to the tibia and leg IV seems to be complete in the P. petrunkevitchi scans. Leg I is again slender, and was presumably tactile in function. As with G. britannica (see above), leg IV is noticeably larger than the preceding legs II and III. Among schizomids, a larger leg IV is more typical of the family Hubbardidae, while in Protoschizomidae legs II–IV are less differentiated and more uniform (e.g. Monjaraz-Ruedas et al., 2016, fig. 1).

Opisthosoma

The opisthosoma in P. petrunkevitchi is lozenge-shaped, slightly wider towards the back, but not as oval as in G. britannica. The opisthosomal tergites and sternites have straight posterior margins and the postabdomen of P. petrunkevitchi is smaller and more compact than in G. britannica. Unfortunately, the opisthosoma ends towards the edge of the nodule and the telson is not visible, neither in the hand specimen nor in the CT scan. Schizomids characteristically have either a truncated flagellum of only a few segments (in females) or a telson modified into a club-or plate like flagellum (in males).

Phylogenetic results

Equal and implied weights parsimony, and the Bayesian analysis (Fig. 3; Supplemental Figs 1, 2) recovered Thelyphonida and Schizomida as monophyletic. This is uncontroversial and, as expected, the two orders resolve as sister-taxa forming the clade referred to by several authors (e.g. Shultz, 2007) as Uropygi. In the Bayesian analysis the sister-group of Uropygi was Amblypygi (whip spiders). This (Amblypygi + Uropygi) clade has been recovered in several studies, summarized by Giribet (2018), and is usually referred to as Pedipalpi. It is supported morphologically by a shift towards raptorial pedipalps (hence the name) and a more slender and tactile first pair of legs. The extinct order Haptopoda, represented by a single genus Plesiosiro Pocock, 1911, also has tactile, albeit still quite robust first legs, but retains unmodified and leg-like pedipalps. Haptopoda is the sister-group of Pedipalpi in the present Baysian analysis (Fig. 3). This position is consistent with previous phylogenies to include fossils (e.g. Garwood & Dunlop, 2014; Shultz, 2007). Parsimony analyses differed slightly in recovering (Amblypygi (Haptopoda + Uropygi)) instead. We note, however, that Haptopoda is equivocal for a number of phylogenetically informative characters, e.g. cheliceral morphology, and this could underlie its lability between analyses.

Significantly, both the parsimony and the Bayesian analyses (Fig. 3, Supplemental Figs 1, 2) recovered Geralinura britannica and Proschizomus petrunkevitchi within Thelyphonida. In other words, G. britannica is confirmed as a bone fide whip scorpion, as would be expected given its morphology, but the hypothesis that P. petrunkevitchi is a stem-schizomid (cf. Dunlop & Horrocks, 1996) is not corroborated by the new features identified, and phylogenetic analyses reported here. The principal characters supporting thelyphonid affinities for P. petrunkevitchi are the presence of apophyses on the pedipalp (Fig. 2). In detail, in both the parsimony and the Bayesian trees (Fig. 3), the extinct P. petrunkevitchi and the extant genus Hypoctonus Thorell, 1889 form a polytomy with a group including the other whip scorpions being tested. Geralinura britannica then resolves in a polytomy with the two extant genera included here, Mastigoproctus Pocock, 1894 and Typopeltis Pocock, 1894. Characters in support of this placement for G. britannica are the presence of keels between the median and lateral eyes and the absence of a pointed anterior margin of the prosomal dorsal shield. We caution that while this anterior projection of the dorsal shield is observed in some hypoctonines (e.g. Rowland & Cooke, 1973, fig. 2), including that coded here, it is absent in some taxa currently also placed in this subfamily (e.g. Huff & Prendini, 2009, figs 1, 2). Our taxon sampling does not allow us to resolve the relationships of whip scorpions at the subfamily level, and even the molecular tree(s) of Clouse et al. (2017, figs 2, 3) have not provided clear resolution among the four currently recognized whip scorpion subgroups.

Within Schizomida, the parsimony and Bayesian analyses (Fig. 3) both recovered the basic division into two families: Protoschizomidae and Hubbardiidae. We should caution that protoschizomids may not be monophyletic as they currently stand; see e.g. discussion in Monjaraz-Ruedas et al. (2016). Equal weights parsimony did not offer any resolution within Hubbardiidae, but the Bayesian analysis (Fig. 3, Supplemental material Fig. 1) grouped the Burmese amber genus Mesozomus Müller et al., 2019 with the extant Schizomus Cook, 1899 and Stenochrus Chamberlin, 1922. Implied weights parsimony (Supplemental material Fig. 2) also recovered a Schizomus/Stenochrus clade, but within a polytomy with Mayazomus and Mesozomus. We are cautious of overinterpreting this result as the taxon sampling for Schizomida here is poor and Schizomus in particular has been treated as something of a ‘bucket genus’ in the past. For example, it includes species from several biogeographical regions. We refer to the analysis of Clouse et al. (2017) for a more detailed account of schizomids’ internal relationships: the extinct genus Mesozomus could fit both geographically and stratigraphically into their inferred Asian–Pacific radiation of schizomids during the Cretaceous.

Systematic palaeontology

Order Thelyphonida Latreille, 1804

Plesion (Genus) Proschizomus Dunlop & Horrocks, 1996

Type and only species

Proschizomus petrunkevitchi Dunlop & Horrocks, 1996, by original designation.

Emended diagnosis

Medium-sized, fairly gracile Coal Measures whip scorpions with a prosomal dorsal shield uniquely drawn into a blunt anterior projection; prosomal shield not divided and lacking median eyes, with a ‘V’-shaped raised topology; pedipalps articulating at an angle of c. 45° to the vertical plane and both patella and tibia with small apophyses. (Emended from Tetlie & Dunlop, 2008.)

Proschizomus petrunkevitchi Dunlop and Horrocks, 1996

(Fig. 1B, E)

1911 Geralinura britannica Pocock (misidentification in part): 29–30, text-fig. 9.

1913 Geralinura britannica Pocock (misidentification in part); Petrunkevitch: 59.

1949 Geralinura britannica Pocock (misidentification in part); Petrunkevitch: 264–266, figs 157, 265.

1953 Prothelyphonus britannicus (Pocock) (misidentification in part); Petrunkevitch: 97–98.

1955 Prothelyphonus britannicus (Pocock) (misidentification in part); Petrunkevitch: 120, fig. 5.

1961 Prothelyphonus britannicus (Pocock) (misidentification in part); Laurentiaux-Viera & Laurentiaux: 25.

1962 Prothelyphonus britannicus (Pocock) (misidentification in part); Dubinin, fig. 1263.

1980 Prothelyphonus britannicus (Pocock) (misidentification in part); Morris: 45.

1983 Prothelyphonus britannicus (Pocock) (misidentification in part); Brauckmann and Koch: 65, 68.

1985 Prothelyphonus britannicus (Pocock) (misidentification in part); Brauckmann, Koch and Kemper, 22–23.

1991 Prothelyphonus britannicus (Pocock) (misidentification in part); Brauckmann: 43–44.

1996 Proschizomus petrunkevitchi Dunlop and Horrocks: 304–305, figs 2, 3, 5, 17–24.

2003 Proschizomus petrunkevitchi Dunlop and Horrocks; Harvey: 80.

2007 Proschizomus petrunkevitchi Dunlop and Horrocks; Dunlop et al.: 167.

2008 Proschizomus petrunkevitchi Dunlop and Horrocks; Tetlie & Dunlop: 311, fig. 6.6.

2016b Proschizomus petrunkevitchi Dunlop and Horrocks; Selden et al.: 7.

2019 Proschizomus petrunkevitchi Dunlop and Horrocks; Müller et al.: 8.

Diagnosis

As for the genus.

Holotype

NHMUK PI In 7912 (holotype), from the British Middle Coal Measures of Coseley near Dudley, Staffordshire, UK. Late Carboniferous (Duckmantian).

Description

Detailed descriptions can be found in Petrunkevitch (1949) and Dunlop and Horrocks (1996); supplemented here by novel appendage data recovered through tomography. Total body length (excluding telson) 16 mm. Prosomal dorsal shield undivided, length, 5.8 mm, maximum width 3.8 mm. Pedipalps subchelate, length 6.7 mm. Leg I slender, total preserved length 13.2 mm. Leg II, total preserved length 11.5 mm. Leg III, total preserved length 9.9 mm. Leg IV, total preserved length 14.0 mm. Opisthsoma elongate; length c. 12 mm, maximum width 4.2 mm. Terminal three segments form a postabdomen, length, 2 mm. Telson equivocal.

Family ?Thelyphonidae Lucas, 1835

Geralinura britannica Pocock, 1911

1911 Geralinura britannica Pocock (misidentification in part): 29–30, pl. 1, fig. 3, pl. 2, fig. 3.

1913 Geralinura britannica Pocock (misidentification in part); Petrunkevitch: 59.

1935 Geralinura britannica Pocock (misidentification in part); Werner: 462, fig. 164.

1949 Geralinura britannica Pocock (misidentification in part); Petrunkevitch: 264–266.

1953 Prothelyphonus britannicus (Pocock) (misidentification in part); Petrunkevitch: 97–98.

1955 Prothelyphonus britannicus (Pocock) (misidentification in part); Petrunkevitch: 120.

1961 Prothelyphonus britannicus (Pocock) (misidentification in part); Laurentiaux-Viera and Laurentiaux: 25.

1962 Prothelyphonus britannicus (Pocock) (misidentification in part); Dubinin, fig. 1263.

1980 Prothelyphonus britannicus (Pocock) (misidentification in part); Morris: 45.

1983 Prothelyphonus britannicus (Pocock) (part); Brauckmann and Koch: 65, 68.

1985 Prothelyphonus britannicus (Pocock) (part); Brauckmann, Koch and Kemper, 22–23.

1991 Prothelyphonus britannicus (Pocock) (part); Brauckmann: 43–44.

1996 Geralinura britannica Pocock; Dunlop and Horrocks: 301–302, figs 4, 7–16.

1997 Geralinura britannica Pocock; Anderson et al.: 206, fig. 4e.

2003 Proschizomus petrunkevitchi Dunlop and Horrocks; Harvey: 79.

2008 Geralinura britannica Pocock; Tetlie and Dunlop: 310, fig. 6.4.

2016b Geralinura britannica Dunlop and Horrocks; Selden et al.: 8.

Emended diagnosis

Geralinura with a more oval, anteriorly narrowed prosomal shield than G. carbonaria; opisthosoma with well-developed muscle apodemes, but lacking tubercles and/or median divisions; patella of fourth leg without ornament; palpal trochanter bearing prominent prodorsal flange. (Emended from Tetlie & Dunlop, 2008.)

Material

NHMUK PI In 31265 (paratype) from the British Middle Coal Measures of Coseley near Dudley, Staffordshire, UK. Late Carboniferous (Duckmantian).

Description

Detailed descriptions can be found in Petrunkevitch (1949) and Dunlop and Horrocks (1996); supplemented here by novel appendage data recovered through tomography. Total body length (excluding telson) 18 mm. Prosomal dorsal shield undivided, length 7.1 mm, maximum width 4.3 mm. Pedipalps subchelate, length 7.4 mm. Leg I slender, total preserved length 15.1 mm. Leg II, total preserved length 13.3 mm. Leg III, total preserved length 13.0mm. Leg IV, total preserved length 16.4 mm. Opisthsoma elongate; length 13.1 mm, maximum width 5.3 mm. Terminal three segments form a postabdomen, length, 4.3 mm. Telson equivocal.

Discussion

The key results of the present study are: (1) the recognition that one of the fossils assigned to Proschizomus petrunkevitchi actually belongs to the species Geralinura britannica; (2) that G. britannica has pedipalps with at least patellar apophyses rendering the limb subchelate, and very similar to the pedipalps of living species; (3) that both fossils have pedipalpal apophyses, a consequence of which is that P. petrunkevitchi resolves among the Thelyphonida and is not recovered as a member of the Schizomida stem-lineage; and (4) that P. petrunkevitchi has pedipalps which articulate at a c. 45° angle, which is essentially intermediate between the whip scorpion and schizomid condition. All of these results would be difficult to obtain using traditional study methods (cf. Dunlop & Horrocks, 1996) and reiterate how useful computed tomography can be for recovering morphological characters in three-dimensionally preserved Coal Measures fossils hosted in ironstone concretions.

Affinities of Geralinura britannica

Modern whip scorpions are a fairly homogeneous group, reflected in the fact that all 124 living species (Seraphim et al., 2019) are accommodated within a single family; indeed, Haupt (2009b, p. 18) referred to the “monotony of the Thelyphonida”. Previous attempts to recognize the hypoctonids as a separate family (e.g. Rowland & Cooke, 1973) were not supported by subsequent authors. The Carboniferous species Geralinura britannica is here shown to be anatomically very similar to living whip scorpions; specifically the recognition of apophyses on at least the patella of the pedipalp contradicts Tetlie and Dunlop’s (2008) proposal that this character is absent in all Palaeozoic taxa. Unfortunately, the other Coal Measures taxa are not always preserved well enough, or with the pedipalp in an amenable unfolded position, to test whether the absence of patellar and/or tibial apophyses is genuine or taphonomic. We also note, for example, that the habitus images of Huff and Prendini (2008) reveal that in some modern species the degree to which patellar apophysis is expressed is sexually dimorphic; in Etienneus africanus (Hentschel, 1899) it is barely visible in males. These caveats aside, our phylogenetic analyses (Fig. 3) recover G. britannica nested within Thelyphonida, which can inform a taxonomic decision to place this species within the single crown group family Thelyphonidae too. Specifically, the presence of keels between the eyes resolves G. britannica in a more derived position than the hypoctonids, i.e. within the crown-group, rather than as a stem-thelyphonid.

So should Geralinura be referred to Thelyphonidae, as in some previous studies (e.g. Petrunkevitch, 1949, 1955)? The type species of this genus comes from Mazon Creek in the USA (Scudder, 1884), but does not show an obvious patellar apophysis as described here in the British species; see also the re-description by Tetlie and Dunlop (2008). Given that the American and British species are otherwise very similar, we cannot rule out the possibility that the apophysis was present, but cannot be clearly resolved in the Mazon Creek material. A fossil family name Geralinuridae proposed by Scudder (1885) is available, but it is difficult to identify unequivocal apomorphies in the Geralinura fossils which would justify (and define) an extinct family. As noted above, the postabdomen in this Carboniferous genus is quite large and the fourth leg is perhaps a little longer than in living species, but neither of these are fundamentally different to the characters present in extant whip scorpions. Another point of interest is that the pedipalps in G. britannica are about as long as the prosomal dorsal shield. In some modern species the pedipalps are noticeably longer than the dorsal shield – see e.g. habitus images of Mastigoproctus in Barrales-Alcalá et al. (2018) – although this is not true of all living whip scorpions. In genera like Thelyphonus and Typopeltis, the pedipalp is similar in length to the dorsal shield; see e.g. habitus images in Javed et al. (2009).

On balance, given that our phylogenetic results (Fig. 3) suggest that at least Geralinura, from among the Carboniferous genera, can be tentatively accepted as a member of the crown-group, we suggest this species is placed in the extant family Thelyphonidae. The other Carboniferous genera are currently difficult to place because the pedipalp is not known in the same level of detail, and should perhaps be retained as plesion genera of Thelyphonida sensu Tetlie and Dunlop (2008). Given the morphology of Proschizomus (see below) we should also entertain the possibility that the Carboniferous hosted a mixture of stem- and crown-group whip scorpions.

Affinities of Proschizomus petrunkevitchi

The new characters recovered here, and our phylogenetic analyses incorporating these, do not support the hypothesis that P. petrunkevitchi is a stem-schizomid. Instead, they place the fossil in an unresolved position within the Thelyphonida. As noted above, the main features supporting this are the apophyses on the patella and tibia of the pedipalp which are wholly absent in schizomids. This implies that the loss of the median eyes could be a homoplastic feature, i.e. it could have occurred independently in this Carboniferous species and in schizomids. Alternatively, it could be that median eyes were lost on the Uropygid stem, then reacquired within crown thelyphonids. The orientation of the pedipalps in P. petrunketichi is also of note. The c. 45° angle of the plane of motion of the pedipalp is quite different from the horizontal orientation in modern whip scorpions, and exclusion from the modern family Thelyphonidae based on this character seems justified. We thus follow Tetlie and Dunlop (2008) in treating Proschizomus as a plesion genus within Thelyphonida.

Based on these results, we can hypothesize that the common ancestor of whip scorpions and schizomids had a unitary prosomal dorsal shield, with relatively unmodified pedipalps articulating up and down like the legs, and a long, whip-like flagellum. In this scenario, within the Thelyphonida lineage the pedipalps acquired patella and tibial apophyses to render them subchelate, and presumably enhance their prey-capturing efficiency. However, we caution that sexual selection may also have modified the pedipalps in some taxa. Modern genera like Typopeltis have sexually dimorphic pedipalps (e.g. Haupt, 2009b; Seraphim et al., 2019) and the males’ pedipalps are actively used to wrestle with rivals during competition for females (Watari & Komine, 2016; Weygoldt, 1988). This is also true in the outgroup of the Uropygid (+/– Haptopoda) clade (McLean et al., 2020). More generally, the plane of movement in the whip scorpion pedipalps rotated from primarily vertical to a more horizontal one, perhaps also allowing them to better grasp at prey items in front of the animal. Proschizomus petrunkevitchi may represent an intermediate stage of this process in which the pedipalps have partially rotated and are beginning to develop their apophyses. A comparable transformation appears to have taken place in whip spiders (Amblypygi), in which the earliest branching taxa had pedipalps articulating up and down, while more derived groups shifted to motion in the horizontal plane (Dunlop et al., 2007; Garwood et al., 2017).

In schizomids the raptorial, but otherwise unmodified, pedipalps were retained and the main additional transformations within this lineage were the partial (and eventually total) loss of eyes, the division of the dorsal shield into separate plates and the reduction of the flagellum to a rather short telson. This may reflect Weygoldt and Paulus’ (1979) suggestion that the schizomid body is adapted for movement in confined and cryptic spaces (e.g. leaf litter, rotting logs) where eyes are less useful and a flexible prosoma and less cumbersome telson may be advantageous. In any case, the fossil record (Fig. 3) imposes a minimum age constraint of 318 Ma for the origins of both Thelyphonida and Schizomida. The two fossils restudied here also demonstrate that there were whip scorpions with both modern and extinct character suites living alongside each other in the Carboniferous coal forests. A similar situation is also seen among the Coal Measures harvestmen (Opiliones) (e.g. Garwood et al., 2011, 2014; Selden et al., 2016a, and references therein), Araneae (Garwood et al., 2016) and to some extent in scorpions (Jeram, 1994).

Associate Editor: Greg Edgecombe

Acknowledgements

We thank Claire Mellish (NHMUK) for access to material in her care and William Shear (Hampden-Sydney) and the late Joachim Haupt (Berlin) for helpful comments prior to commencing this study. We are also grateful to Karl-Hinrich Kielhorn (Berlin) for providing modern schizomid material, Carsten Kamenz (formerly Berlin) for providing photographs of schizomids and Jessica Krüger (formerly Berlin) for allowing us to use material from her Master’s thesis. RJG was supported by NERC award NE/T000813/1. This research received support from the SYNTHESYS Project (http://www.synthesys.info/), which is financed by European Community Research Infrastructure Action under the FP7 ‘Capacities’ Program. We are grateful for the comments of an anonymous reviewer and Carolin Haug, and Associate Editor Greg Edgecombe, all of whose suggestions improved the manuscript.

Supplemental material

Supplemental material for this article can be accessed here: https://doi.org/10.1080/14772019.2023.2180450.

Additional supplemental material for this paper has been uploaded to the Zenodo repository. This can be accessed here: https://doi.org/10.5281/zenodo.7070927