The fossil record of spiders as currently known is briefly reviewed, with special reference to Africa. The second specimen of Triassaraneus andersonorum Selden in Selden et al. 1999 is described from a different locality in the Triassic (Carnian, c. 225 Ma) Molteno Formation of South Africa.
Over 90 % of fossil spiders originate from ambers of Cainozoic age (less than 65 Ma old). Cainozoic ambers preserve representatives of the majority of the 109 spider families known today. Of greater interest for phylogeny, therefore, is the fossil record of Araneae from the Mesozoic and Palaeozoic eras, and what it can tell us about the origins and longevity of major clades. Fossil spiders have been known from the late Palaeozoic Coal Measures of Europe and North America for more than a century (although it must be mentioned that some of these specimens are certainly not spiders, Penney & Selden 2006), but the first genuine Mesozoic spider to be formally described was only in 1984 (Eskov 1984). Over the succeeding decades only a few new species have been reported. More recently, much new material is being discovered in, for example, lower Cretaceous deposits of north-east Brazil and China, which is being described at the moment.
It is because of their great rarity that almost every new specimen of spider from the Palaeozoic and Mesozoic eras is worthy of report, and each occurrence can drastically alter our perception of spider phylogeny. For example, the identification of a single record of a family earlier than previously known predicts the occurrence of its sister clade total group to the same date. Thus, Penney's (2002) description of Cretaceous Oonopidae predicted the occurrence of the family's sister group, Orsolobidae. As pointed out by Penney et al. (2003) and others, the true biodiversity at any time in the past is greater than estimated from raw fossil data because range extensions and ghost lineages need to be taken into account. Figure 1 illustrates the present state of knowledge of the fossil record of spiders.
For the last two decades, the oldest known spider in the fossil record was thought to be Attercopus fimbriunguis (Shear, Selden & Rolfe in Shear et al. 1987), from the Devonian of New York, USA. Attercopus was considered sister to all other spiders. Recently, however, Attercopus has been shown to belong to a new order of arachnids: the Uraraneida Selden & Shear in Selden et al. 2008, an order that also includes the Permian Permarachne Eskov & Selden, 2005, which was originally described as an unusual mesothele. Mesotheles are known from upper Carboniferous and Permian strata (Selden 1996a, b, 2000; Eskov & Selden 2005), but no opisthotheles until Triassic time; putative Palaeozoic opisthotheles have been shown to be other arachnids or unrecognizable (Penney & Selden 2006). The oldest opisthothele known is the mygalomorph Rosamygale grauvogeli Selden & Gall, 1992, from the Lower Triassic (Anisian) of France, which was the first spider to be described from rocks of this period. Selden in Selden et al. (1999) described the oldest araneomorphs, Triassaraneus andersonorum (Fig. 3) and Argyrarachne solitus, from the Carnian Molteno Formation of KwaZulu-Natal, South Africa, and the Carnian Cow Branch Formation of Virginia, USA, respectively; their presence in the Triassic Period having been predicted by the occurrence of their sister group in 1992 (Selden & Gall 1992). Single, monotypic specimens were described from each of these localities; additional, but poorly preserved, specimens were noted, but not described. In this paper an additional specimen of Triassaraneus, now confirmed as an araneomorph spider, is described from the Molteno Formation locality of Telemachus Spruit in Eastern Cape, South Africa.
Few spiders have been described from the Jurassic Period. The first to be described was Juraraneus Eskov, 1984, an orb-weaver from Transbaikalia, and the second, Jurarchaea, Eskov, 1987, is an archaeid (sensu lato) from Kazakhstan. One, from the Jurassic of Grimmen, Germany, has been figured but not yet formally described (Ansorge 2003; preliminary investigation by the senior author suggests this may be a palpimanoid), and additional specimens from Russian localities remain undescribed. More than 400 specimens recently become available for study from beds of Middle Jurassic age at Daohugou, Nincheng County, Inner Mongolia, China (Huang et al. 2006). Most belong to Uloboridae, and have yet to be studied, but some belonging to the superfamily Palpimanoidea (sensu Forster & Platnick 1984) were described recently (Selden et al. 2008).
Few Cretaceous spiders had been formally described and named until recently, but work is continuing apace to rectify this situation. Cretaceous records include: indeterminate Araneae from Koonwarra, South Australia (Jell & Duncan 1986); mygalomorphs from Transbaikalia and Mongolia (Eskov & Zonshtein 1990); orbicularian araneomorphs from the Sierra de Montsech and Las Hoyas localities, Spain (Selden 1989, 1990, 1991; Selden & Penney 2003); an unnamed lycosoid from Orapa, Botswana (Rayner & Dippenaar-Schoeman 1995); a poorly preserved specimen from Mexico (Feldmann et al. 1998); many new spider specimens reported from the Crato Formation of Brazil, of which just a few have been described so far (Mesquita 1996; Selden et al. 2002, 2006); and many also are now being collected from the Lower Cretaceous of China (Chang 2004; Cheng et al. 2008). Spiders are known from Cretaceous amber from Canada (McAlpine & Martin 1969; Penney & Selden 2006), the Caucasus and Siberia (Eskov & Wunderlich 1994; Zherikhin & Eskov 1999), France (Schlüter 1978; Saupe & Selden 2009), Lebanon (Penney 2003a; Penney & Selden 2002), Myanmar (Cockerell 1920; Rasnitsyn & Ross 2000; Zherikhin & Ross 2000; Penney 2003b; Wunderlich 2008), New Jersey (Grimaldi et al. 2000; Penney 2002), Álava, Spain (Alonso et al. 2000; Penney 2006), Asturias, Spain (Arbizu et al. 1999), Jordan (Kaddumi 2007; Wunderlich 2008) and the Isle of Wight, England (Selden 2002).
At the time of publication of Eskov and Zonshtein's (1990) paper on Cretaceous mygalomorphs, there were hardly any spiders known at all from the Mesozoic era. It appeared that mygalomorphs, perhaps already an archaic group, may have flourished at a time of biotic crisis, when araneomorphs became reduced in number (Eskov & Zonshtein 1990). Since then, however, many more Mesozoic spiders, and particularly araneomorphs, have been described. Figure 1 shows that by Cretaceous time, many of the modern families were in existence or predicted by sister-group relationships. By the time of formation of the mid-Cainozoic ambers, nearly all extant families have fossil representatives or are predicted to occur. Interestingly, there are very few families known only from fossils; some new families from Cretaceous ambers recently described by Wunderlich (2008) have yet to be critically assessed. Some families described initially from fossils were later found to be extant, e.g. Archaeidae were first described from Baltic amber and later found alive in Africa and Madagascar. The Cretaceous extinct family, Lagonomegopidae Eskov & Wunderlich, 1994, is interesting on account of (a) its enormous posterior median eyes, which suggest a possible relationship (homology or analogy) with Salticidae (the most speciose modern family, and unknown in preCainozoic strata), and (b) the characteristic peg-teeth, which place it within the Palpimanoidea sensu Forster and Platnick (1984). Few arachnologists concur with the composition of this superfamily; in particular Schütt (2002) has provided evidence, mainly from spinneret morphology, that the palpimanoid families formerly allied with Araneoidea should be returned to that superfamily. These include Mimetidae and the archaeid families, which are araneophages characterised by elongated chelicerae bearing long peg-teeth. Of the described Jurassic spiders, several (Jurarchaea, Patarchaea Selden, Huang & Ren, 2008, Metaxyostraca Selden, Huang & Ren, 2008, and an undescribed specimen from Germany) belong in the Palpimanoidea, and the other (Juraraneus) is an early orb-weaver. The earliest known araneomorphs, Triassaraneus and Argyrarachne, resemble araneoids more than any other group. Conversely, cursorial spiders such as the Dionycha, including such speciose families as Salticidae, Thomisidae, Gnaphosidae and the clubionoids, which represent some 35 % of genera and 37 % of species at the present day (data from Platnick 2009), are as yet unknown in pre-Cainozoic strata. The fossil record thus appears to support the hypothesis, well argued by Dippenaar-Schoeman and Jocqué (1997), that the more derived hunting spiders arose from ancestors which used webs for prey capture (Vollrath & Selden 2007).
A study by Penney et al. (2003) examined how spiders fared across well-known mass extinctions, specifically the mid-Cretaceous Cenomanian—Turonian (C/T, 93.5 Ma), and the better-known Cretaceous—Tertiary (K/T, 65 Ma) events. Using quantitative techniques combining taxonomic and numerical data on amber fossils only, these authors showed that spiders suffered no decline at the family level during these mass extinctions. On the contrary, they increased in relative numbers through the Cretaceous and beyond the K/T boundary. They suggested that this extinction resistance of spiders is due to the majority of them being generalist predators. Those feeding predominantly on plantspecific herbivores that became extinct could easily have switched to a new primary food source, such as many of the non-herbivorous insects or polyphagous herbivores, which appear to have been little affected by the K/T extinction (Labandeira & Sepkoski 1993; Labandeira et al. 2002). Thus, the mass extinction that killed off the dinosaurs seems to have had little effect on the Araneae, at least at family level.
African fossil spiders
Spiders are common in African copal, which is a kind of sub-fossil amber; it chiefly occurs on the coasts of Kenya, Tanzania and Mozambique, and on the north-west coast of Madagascar. Softer and paler yellow than true amber, it dates from about Miocene and younger in age, and so its inclusions are generally of less interest for phylogenetic studies. It can be of great interest, however, for research on palaeogeography and palaeoclimatology of more recent geological times. For example, interesting work has been done comparing the fossil and extant spider species in Dominican amber and alive on the island of Hispaniola respectively (Penney 2008). In spite of specimens having been available in gem shops relatively cheaply throughout the world, few descriptions of African copal spiders appeared until recently. There can be a problem, however, in that some amber dealers will heat copal to make it appear to be true amber and thus gain higher prices. For example, Wunderlich (2004) suggested that the specimen of Entomocephalus formicoides Holl, 1829 from Baltic amber was really a fake in Madagascan copal. Holl's one-sentence description and figure (pl. 8, fig. 68a) suggests it almost certainly belongs to an ant-mimicing salticid genus such as Myrmarachne (which occurs in Madagascan copal), although the figure and description mentioned only six eyes (Penney 2003b). It was placed (erroneously) in Archaeidae by Petrunkevitch (1958); the location of the holotype is unknown.
Lourenço (2000) described Archaea copalensis (Archaeidae), which was thought to be the first record of this genus from Madagascan copal; the species was later synonymized with the extant Archaea gracilicollis Millot, 1948 by Wunderlich (2004). Previously, Wunderlich (1998) had described Mysmena dominicana (Mysmenidae) and Grammonota deformans (Linyphiidae, later placed in Ceratinopsis by Wunderlich (2004)) from supposed Dominican amber. He later realized that the amber was actually Madagascan copal treated to make it appear to be older and from a different provenance (Wunderlich 2004). Similarly, Wunderlich's (1999) report of the family Archaeidae from Dominican Republic amber also proved to be a mistake for Madagascan copal. Species from the following spider families are currently known as sub-fossils in Madagascan copal: Araneidae, Archaeidae, Clubionidae, Corinnidae, Deinopidae, Dictynidae, Hahniidae, Hersiliidae, Linyphiidae, ?Migidae, ?Miturgidae, Mysmenidae, Oonopidae, Philodromidae, Pholcidae, Salticidae, Scytodidae, Segestriidae, Selenopidae, Tetragnathidae, Theridiidae, Thomisidae and Uloboridae (Wunderlich 2004, 2008).
Only two specimens of fossil spider have been described from the continent of Africa: Triassaraneus, as mentioned above, and a supposed lycosoid spider from the Orapa diamond mine in Botswana (Rayner & Dippenaar-Schoeman 1995). The Orapa mine exploits a kimberlite pipe in which a crater lake developed during the Cretaceous period; the matrix bearing the spider and abundant insect remains has been dated to 93 Ma ago (Late Cretaceous: Turonian) by radiometry (Rayner et al. 1997). The relatively soft volcanic breccias of kimberlite pipes commonly accommodate circular crater lakes (see Gernon et al. 2009 for a review of the sedimentology of the Orapa kimberlite). The Orapa lake was shallow and the fine lamination of the sediment suggests little disturbance, so it was probably anoxic bottom waters which prevented decay of the biota. Because there has been little appreciable erosion since Cretaceous times, the lake sediments are still preserved. More than half of the insects at Orapa are Coleoptera, followed by Blattodea (about 20 %), and less than 10 % each of Diptera, Hymenoptera, Hemiptera, Orthoptera, Dermaptera and the rest (Rayner et al. 1997). A diverse flora of mainly angiosperms (about 90 %) that was washed into the lake has been described by Bamford (1990); ferns account for the remainder of the flora. The only spider to have been described from Orapa is an unnamed specimen with a body length of approximately 25 mm. It was referred to superfamily Lycosoidea on the basis of its habitus being that of a hunting spider. This was the first fossil spider to be described from Africa.
The second, and most important, fossil spider to be described from Africa comes from the Molteno Formation of the Karoo (Fig. 2), which was deposited in an extensive, intracontinental foreland basin bounded by rising fold mountains to the south and traversed by a system of braided rivers (Cairncross et al. 1995). The Molteno Formation reaches a maximum thickness of about 600 m and the erosional remnant extends over an area roughly 400 km north to south and 200 km west to east. The age of the formation is not tightly established, but on the basis of global biostratigraphic correlations (Anderson & Anderson 1983, 1989) is considered to be Carnian (Late Triassic), 222– 229 Ma. No absolute radiometric ages are available. A 30-year collecting programme (Anderson 2001; Anderson & Anderson 1983, 1984, 1989; Anderson et al. 1998) has yielded 100 fossil plant assemblages from the Molteno Formation. The flora, the richest known globally from the Triassic, includes 56 genera with 204 vegetative species. It is particularly characterised by some 20 species of Dicroidium (Ginkgoopsida). There is a roughly equal diversity of gymnosperms, including conifers, cycads and ginkgos, along with several new orders, and ‘pteridophytes,’ primarily horsetails and ferns. Though rare, insects comprise by far the most frequently encountered element of the fauna. A remarkable diversity of 117 genera and 333 species in 18 orders is provisionally recognised in the over 2000 specimens from 43 of the 100 plant assemblages. Coleoptera, Blattodea and Hemiptera dominate. Conchostracan crustaceans, from 20 of the plant assemblages, are represented by some three genera and eight species. The remaining fauna is sparse: three species of fishes (impressions only) from three assemblages, two species of bivalve from one assemblage, and two spider specimens. Dinosaur trackways, but no skeletal remains, have been identified at a few (non-plant) sites. Anderson (2001) provided a review of the Molteno Formation flora and fauna.
At the Telemachus Spruit locality there is a 10 cm thick, buff mudstone that is interpreted as an abandoned channel-fill (Cairncross et al. 1995). The flora of 12 genera and 19 species is strongly dominated by the single coniferous species Heidiphyllum elongatum (Morris, 1845) Retallack, 1981. The assemblage most likely represents two distinct plant communities: a mono-dominant stand of reed-like conifers colonizing sand bars in the braided river and, from farther afield, a Dicroidium-dominated riparian forest occupying the river bank. The insect fauna at this site, is represented by only 17 fragmentary specimens, and is dominated, as at other Molteno localities, by Coleoptera, Blattodea and Hemiptera. Conchostracans have not been found. The rarity of the single spider specimen is emphasized by the fact that John and Heidi Anderson have spent 90 man-hours cleaving slabs at this site and that all 900 curated and catalogued slabs have been carefully scanned for insects or other faunal elements under the binocular microscope.
Infraorder Araneomorphae Smith, 1902
Superfamily ?Araneoidea Latreille, 1806
Triassaraneus Selden in Selden et al., 1999
Type and only species: Triassaraneus andersonorum Selden in Selden et al., 1999.
Triassaraneus andersonorum Selden in Selden et al., 1999
Material examined: SOUTH AFRICA: KwaZulu-Natal: Holotype PRE/F 18560a (part) and 18560b (counterpart), immature or mature female from Upper Umkomaas ‘Waterfall Locality’ (UMK III). Eastern Cape: PRE/F 17234 Tel 111, immature from Telemachus Spruit locality (Anderson & Anderson 1983, 1984). All material from from Member Z of the Late Triassic (Carnian) Molteno Formation and deposited in the palaeobotanical collection of the South African National Biodiversity Institute, Pretoria.
Description of new specimen: The specimen is preserved as brown, organic cuticle on a pale, grey-brown shale. The shale is splintery and pieces readily fall away. Scattered throughout the shale are abundant plant remains, including on the fossil slab a male cone of Ginkgoopsida. Parts of at least three legs are preserved, though it is not possible to determine which legs these are. No body parts are preserved. On the right in Figs 4, 5 is a nearly complete walking leg, showing most of the femur to the tarsus. On the left are some of the opposing walking legs, but less complete. Note that in Fig. 5A, there is a loose piece of matrix bearing parts of a leg; this became detached and is not present in Fig. 5B (under ethanol), which therefore reveals more of the tibia of on the left-hand walking legs. The walking legs bear setae on all podomeres, and larger, thin spines on all podomeres except the tarsus. These spines, for example on the right leg patella and tibia, are long, thin and erect. Although tarsal claws cannot be seen, it is certain from the shape of this podomere (only preserved on the right side) that its distal end is present, so the claws must be small. There is no evidence of tarsal scopulae or claw tufts. The legs appear to be rather short, although the only one which is nearly completely preserved, on the right, may be a third leg, which is commonly short in orbicularians (orb-web weavers).
The small claws and lack of tarsal scopulae and claw tufts indicate that this was a web-living rather than a ground spider. Short legs suggest that this is a juvenile spider; however, if the right leg is indeed a third leg of an orbicularian, then this may not be the case. If a third leg, then it is not a great deal smaller than that of T. andersonorum, so it is unlikely to be a juvenile of that species. The leg spination of the right leg matches that of leg 3 in T. andersonorum, with one distally on the femur, one on the patella, two in the proximal half and one in the distal half of the tibia (compare Fig. 4 with fig. 2 in Selden et al. 1999). Therefore, this fossil is considered to be a second specimen of T. andersonorum. Triassic spiders are extremely rare, and this is only the second spider to come out of the Molteno Formation, after many years of painstaking search. It comes from a different locality from T. andersonorum, but is of the same age and possibly occupied the same habitat (Dicroidium-dominated riparian forest).
A general impression gained from studying fossil arachnids is that one spider fossil occurs for approximately every 1000 insects in the Mesozoic lacustrine or lagoonal fossil insect localities, e.g. the Cretaceous Crato Formation of Brazil (Mesquita 1996; Selden et al. 2002), the Cretaceous of Spain (Selden 1989, 1990; Selden & Penney 2003) and the Jurassic and Cretaceous of China (e.g. Selden et al. 2008). The ratio in the Lower Cretaceous locality of Baissa, Transbaikalia is 1:1623 (24,366 arthropod remains) and that in the Middle-Upper Jurassic locality of Karatau, Kazakhstan it is 1:904 (19,909 arthropod remains) (M. Mostovski pers. comm., 2009). With some 2000 insects collected from the Molteno Formation, and a new specimen described herein, this ratio of spider to insect fossils is in agreement, and a similar ratio seems to occur in the Cow Branch Formation of Virginia. However, when comparing other Triassic and Permian localities the ratios are rather different. For example, there are no spiders among over 22,000 insect remains in the Ladinian/Carnian locality of Madygen, Kyrgyzstan (Shcherbakov 2008), and, similarly, no arachnids have ever been found among the tens of thousands of insect specimens from the Artinskian Wellington Formation of Kansas and Oklahoma (Beckemeyer & Hall 2007). The ratio is 1:2129 in the Kazanian locality of Verkhnii Kaltan, Siberia (M. Mostovski pers. comm., 2009), and only one spider specimen, one trigonotarbid and one uraraneid are known out of more than 8000 insects in the Kungurian Koshelevka Formation (including the Chekarda locality) of the Urals (Eskov & Selden 2005). A quantitative study of the ratios of insect remains to those of spiders and other arachnids in localities with abundant insect fossils (Rasnitsyn & Zherikhin 2002) would be extremely interesting.
The senior author thanks David Penney (Manchester) for discussion on Mesozoic spiders in amber, Roger Smith (Iziko South African Museum, Cape Town) for hospitality and a field visit to the Molteno Formation, Mike Mostovski (Natal Museum, Pietermaritzburg) for help with insect and spider abundances, and Jason Dunlop and Kirill Eskov for constructive reviews.