INTRODUCTION

Modern scorpions are an abundant group generally occurring in tropical and warm temperate regions worldwide, although examples are known from cold high altitude environments in Patagonia. They are not known to occur naturally at latitudes higher than 50 degrees (Polis 1990). Scorpions are extremely conservative organisms that occur in the fossil record from the early Silurian period (Størmer 1977; Dunlop & Selden 2013) and are believed to have been an entirely terrestrial clade (e.g. Scholtz & Kamenz 2006; Dunlop et al. 2008; Kühl et al. 2012), with the exception of Devonian Waeringoscorpio Størmer, 1970 which may have been secondarily aquatic (Kühl et al. 2012).

Early scorpions were classically considered to have been aquatic and probably marine (e.g. Rolfe & Beckett 1984; Kjellesvig-Waering 1986; Polis 1990), all Silurian examples having been recovered from marginal marine lagoonal sediments (Jeram 1998). Terrestrialisation was believed to have occurred early in their history, probably during the Devonian period, with allegedly aquatic Proscorpiidae coexisting with terrestrial Mesoscorpionina and ‘palaeosterns’ until the early Carboniferous (Jeram 1998).

Evidence for terrestriality in mesoscorpions was provided by demonstration of booklungs in Pulmonoscorpius Jeram, 1994 from early Carboniferous strata of East Kirkton in Scotland (Jeram 1990, 1994). The presence of an oral tube (floored by the coxal apophyses) in structurally similar Pulmonoscorpius, Petaloscorpio Kjellesvig-Waering, 1986 and Hubeiscorpio Walossek, Li & Brauckmann, 1990 (Walossek et al. 1990) was considered indicative that the mesoscorpion clade that they comprise was terrestrial (Jeram 1998).

The view that basal scorpions were not terrestrial and that terrestrialisation occurred in scorpions independently of other Arachnida (Selden & Jeram 1989) carried the implication that the Araneae are not monophyletic (Scholtz & Kamenz 2006). This has been strongly refuted by Scholtz and Kamenz (2006) who argue that booklungs are an apomorphy of Araneae, that the Araneae are monophyletic and that they resulted from a single terrestrialisation event in their common stem lineage. This implies that all fossil scorpions were primarily terrestrial. Scholtz and Kamenz (2006) argue that several other arachnid characters such as trichobothria, slit-sense organs, endodermal malphigian tubules, and extra-intestinal digestion can be interpreted as apomorphic arachnid adaptations to land (Scholtz & Kamenz 2006).

Support for this view comes from the redescription of a number of apparently basal taxa. Dunlop et al. (2008) re-examined the oldest (Late Silurian) members of the Proscorpiidae, viz. Proscorpius Whitfield, 1885 (Whitfield 1885a , b ), Archaeophonus Kjellesvig-Waering, 1966 and Stoermeroscorpio Kjellesvig-Waering, 1986, and synonomised them under Proscorpius. They refuted evidence for gill openings and found no structural evidence to support an aquatic lifestyle. They assert that no definite evidence for internal gills or any unambiguously aquatic character has ever been unequivocally demonstrated in a fossil scorpion (Dunlop et al. 2008).

Palaeoscorpius devonicus Lehmann, 1944 from the Lower Devonian Hunsrück Slate Lagerstätte in Germany has also been re-investigated (Kühl et al. 2012). This key fossil was formerly interpreted as the most basal member of the Scorpiones and as one of the order's most likely candidates for an aquatic mode of life (Jeram 1998). New imagery was obtained using radiography, X-ray micro-tomography, and CT scanning techniques. Layered internal mesosomal organs were located and interpreted as probable book lungs (Kühl et al. 2012). Kühl et al. (2012) note that much of the argument for Palaeoscorpius Lehmann, 1944 being marine was based on the belief that the Hunsrück Slate was of deep marine origin. Conversely, recent studies of the palaeoenvironment suggest that the ‘Hunsrück Sea’ was part of an intrashelf basin close to the shore, into which terrestrial plants were periodically washed. Terrestrial material was likewise preserved in the lagoonal deposits that have yielded other early scorpions (Kühl et al. 2012). After reviewing all evidence Kühl et al. (2012) conclude that Palaeoscorpius was most likely to have been terrestrial. They also note that this provides strong evidence for arachnid monophyly and a single shared terrestrial arachnid ancestor. Highly unusual Lower Devonian Waeringoscorpio may have been secondarily aquatic as its unique filamentous external organs strongly resemble the tracheal gills of (secondarily) aquatic insects (Poschmann et al. 2008; Kühl et al. 2012).

Scorpion evolution is patchily represented in the fossil record due to the poor preservation potential of scorpions, particularly in terrestrial environments. Only 16 species, commonly represented by single specimens, of Silurian and Devonian scorpions were recognised in 15 genera by Jeram (1998). Of these, two Silurian genera and species have subsequently fallen away (Dunlop et al. 2008). All but one genus, Hubeiscorpio, have been recovered from European and North American strata (Sissom 1990), that at the time of their deposition formed a continuous land mass, Laurasia. Both Laurasia and China situated within tropical to subtropical latitudes during the Silurian and Devonian (Scotese & Mckerrow 1990; Mitchell et al. 2012).

Only two occurrences of scorpion fossils have been recorded from the Late Devonian (Jeram 1996, 1998). These are Petaloscorpio bureaui Kjellesvig-Waering, 1986 from the Frasnian Escuminac Formation of Canada (Jeram 1996) and Hubeiscorpio gracilitarsus Walossek, Li & Brauckmann, 1990 from the Frasnian of China (Walossek et al. 1990). Cladistic analysis of scorpions (Jeram 1998) places these two species as sister groups within the clade Mesoscorpionina, together with early Carboniferous Pulmonoscorpius (Acanthoscorpius Kjellesvig-Waering, 1986 having been subsequently removed from the Scorpiones (Legg et al. 2009)). The Mesoscorpionina form the sister group of late Carboniferous Palaeopisthacanthus Petrunkevitch, 1913, the earliest member of the crown group (Jeram 1998). Legg et al. (2009), however, suggest that mesoscorpions may be paraphyletic with regard to the crown group.

MATERIAL AND METHODS

New material consists of fragments recovered from the Witpoort Formation (Witteberg Group, Cape Supergroup) Waterloo Farm locality near Grahamstown, South Africa. This locality consists of black carbonaceous shale derived from anaerobic mud deposited in a back-barrier marginal marine lagoonal estuary (Gess & Hiller 1995a ). Organic remains accumulated in this setting include those of algae (Gess & Hiller 1995b ; Hiller & Gess 1996), terrestrial plants (Gess & Hiller 1995a; Anderson et al. 1995) and fish (Gess & Hiller 1995a ; Long et al. 1997; Gess 2001; Gess et al. 2006). Invertebrate remains include those of small smooth-shelled bivalves, ostracods, conchostacans, and a eurypterid (Gess & Hiller 1995a ). All fossil material consists of near two-dimensional compressions in which all original tissues have been replaced by secondary metamorphic mica. This has largely been altered to chlorite during uplift (Gess & Hiller 1995a ).

The Witpoort Formation strata at Waterloo Farm were deposited towards the end of the Famennian age of the Late Devonian period, approximately 360 million years ago. The Waterloo Farm estuary formed along the shoreline of the intracontinental Agulhas Sea, apparently within fifteen degrees of the south pole (Scotese & McKerrow 1990; Scotese & Barrett 1990; Mitchell et al. 2012) (Fig. 1).

Fig. 1.

Gondwanan reconstruction with position of the South Pole (star), main latitudes reconstructed for Late Devonian/Early Carboniferous (modified after Scotese & Barrett 1990), and the position of the Waterloo Farm (WF).

Scorpion fragments, including a pedipalp chela articulating with a patella (Fig. 2A, B) and a telson attached to a metasomal segment V (Fig. 2C, D) have been recovered. The telson is slightly damaged where it has been compressed together with small plant fragments.

TAXONOMY

Fig. 2.

Gondwanascorpio emzantsiensis gen. & sp. n.: (A, B) holotype AM5700, general appearance and details of pedipalp chela articulating with patella, (C, D) paratype AM5701, general appearance and details of telson and metasomal segment V. Abbreviations: ch — chela, fi.r. — fixed ramus, fr.r. — free ramus, ma — manus, pa — patella, m.V — metasomal segment V, p.f. — plant fragments, te — telson.

DISCUSSION

This represents the first record of a Palaeozoic scorpion from Gondwana. It is extremely unusual in that it has been recovered from rocks apparently deposited at a far higher latitude than that at which modern or fossil scorpions are known to have occurred. This may reflect amelioration of climatic gradients towards the end of the Devonian (Streel et al. 2000).

The presence of a scorpion in Gondwana, similar to Laurasian taxa, is consistent with growing evidence for globally comparable terrestrial ecosystems by the end of the Devonian. These were characterised by cosmopolitan plant genera such as the progymnosperm tree Archaeopteris, which has also been described from Waterloo Farm (Anderson et al. 1995). A breakdown in disparity between Gondwanan and Laurasian marginal marine ecosystems towards the end of the Devonian has previously been noted, possibly caused by increasing proximity of Laurasia to Gondwana (Young 1987).

Gondwanascorpio is the oldest known terrestrial animal from Gondwana. The only other putative terrestrial animal reported from the Devonian of Gondwana were tergites of Maldybulakia Tesakov & Alekseev, 1998 from Australia (Edgecombe 1998a , b ). Having at first been described from Kazakhstan as Lophodesmus, Maldybulakia was originally considered to most probably represent a type of myriapod (Tesakov & Alekseev 1992). It has subsequently been demonstrated that Maldybulakia most likely represents remains of an aquatic xiphosuran (Anderson et al. 1998; Edgecombe 2004).

The next oldest records of terrestrial animals from southern Africa are insects from the Early Permian Whitehill Formation (Ecca Group, Karoo Supergroup) (McLachlan & Anderson 1977) that was deposited approximately 90 million years later (Fildani et al. 2007). This remarkable gap in the regional fossil record is partially accounted for by a tectonically related 30 million year depositional hiatus, which began around 330 million years ago and was followed by a period of approximately 10 million years of deposition of the Dwyka Group (Karoo Supergroup) glacial diamictites. These resulted from southern Africa's passage over the South Pole during the Carboniferous and the consequent glaciation of most of Gondwana (Catuneanu et al. 2005).