The late Ordovician Soom Shale Lagerstätte: an extraordinary post-glacial fossil and sedimentary record Open Access

The late Ordovician Soom Shale Lagerstätte, Western Cape Province, South Africa has produced an array of marine fossils, many of which preserve spectacular details of their soft (i.e. nonbiomineralized) tissues. Anatomical features usually lost rapidly after death, through scavenging or decomposition, such as eyes, gills, muscles and livers, are preserved. Some of the fossil assemblage is typical of a shallow marine Ordovician setting, such as brachiopods, trilobites and orthoconic nautiloids. However, other taxa, such as the naraoiid Soomaspis (Fortey & Theron 1995), a marine lobopodian (Whittle et al. 2009) and undescribed taxa including a naked agnathan (jawless fish) and a multi-segmented enigmatic fossil would not seem out of place in a Cambrian Lagerstätte (Selden & Nudds 2004). Thus, the Soom Shale appears to chronicle surviving taxa from the Cambrian radiation together with more conventional Ordovician taxa, with aspects of their soft part anatomy preserved.

The Soom Shale Lagerstätte is distinctive in several ways: the host shale itself is conspicuous as the only fine-grained lithology in a 3000 m thick sequence of dominantly arenaceous rocks; the fossils record life in a cold-water, glacially influenced marine setting in the immediate aftermath of a mass extinction and major glaciation; the mode of preservation is the inverse of the ‘normal’ fossil record: in the Soom Shale the hard, biomineralized components of animals are demineralized, and are often comparatively poorly preserved, whereas nonbiomineralized decay-prone anatomy is frequently preserved in exquisite detail. This uniqueness, however, does not diminish the significance and importance of the fossil assemblage from this deposit. As well as recording the earliest and latest known examples of several taxa, thus extending the known temporal ranges of several organisms, the Soom Shale also preserves several taxa unrecorded elsewhere, adding to our knowledge of end-Ordovician biodiversity and ecology. Finally, as well as exceptional fossil preservation, the lithology itself constitutes an undisturbed and probably continuous archive of depositional processes and climate change during and immediately after the dramatic collapse of the Hirnantian (latest Ordovician) glaciation.

The first fossil discovery in the Soom Shale was in 1981, when a team of geologists from the Geological Survey of South Africa (now Council for Geoscience), guided by Johannes Theron, were searching for potential biostratigraphical markers. The team focused on the Cedarberg Formation, the most prominent argillaceous unit in a thick sequence of Palaeozoic sandstones. In the basal member of the Cedarberg Formation, the Soom Shale Member, they discovered several centimetre-sized spiky elements, which occurred together in loose bundles. These fossils were first published as the earliest record of vascular land plants or their ancestors (Kovács-Endrödy 1986). However, this interpretation was overturned and the fossils were convincingly shown to be bedding plane assemblages of conodonts (Theron et al. 1990). Subsequently, the search for additional fossil material has been spearheaded by Johannes Theron and Dick Aldridge (to whom this paper is dedicated) and this yielded a variety of taxa, many with soft tissues preserved.

The Soom Shale Member and overlying Disa Siltstone Member together form the Cedarberg Formation of the Table Mountain Group, part of the Cape Supergroup (Figs 1 and 2). To date, fossils in the Soom Shale Member have been found across a geographically wide area: an age-diagnostic mucronaspidine trilobite in the Hex River Mountains (Cocks & Fortey 1986) as well as the typical Hirnantian brachiopod fauna from Wellington to Porterville (Fig. 1). The principal fossil site for exceptional preservation in the Soom Shale Member is at Keurbos farm.

The Cedarberg Formation forms a conspicuous recessive, laterally persistent, marker horizon (about 120 m thick), among predominantly arenitic units of the Table Mountain Group (Fig. 1b). The Soom Shale Member is about 10 – 15 m thick and is underlain by the glacially derived Pakhuis Formation (Aldridge et al. 2001). Upwards it merges with the thinly bedded Disa Siltstone Member composed of alternating mudstone and siltstone.

The Cedarberg Formation contains the only datable fossils in the entire Table Mountain Group, a succession of latest Cambrian to Early Devonian age (Rust, 1967). The highest resolution dating of the Soom Shale Member (and also the Cedarberg Formation) is provided by chitinozoans (Spinachitina and its morphotypes). The presence of S. oulebsiri restricts the age to a 1 myr interval of late Hirnantian to Rhuddanian time (Vandenbroucke et al. 2009). The absence of S. fragilis chitinozoans suggests that it underlies Silurian (Rhuddanian) strata. Thus, a late Hirnantian age is most likely (Vandenbroucke et al. 2009).

During the Hirnantian an extensive, but short-lived, glaciation, centred on the South Pole, is widely considered to be the leading cause of the second largest extinction event in the Phanerozoic (Sepkoski 1996). The precise mechanisms that caused up to 85% of marine species to become extinct are still debated (see Harper et al. 2014) but it is clear that the event had two phases (e.g. Sheehan 2001; Sutcliffe et al. 2000). The first is associated with the onset of glaciation (Delabroye & Vecoli 2010); the second was prompted by the melting of the Gondwanan ice sheet, which led to glacioeustatic rise (Berry et al. 1995; Brenchley et al. 1995; Sheehan 2001) and extensive black shale deposition (Armstrong & Coe 1997; Delabroye & Vecoli 2010). A large continental ice sheet extended outwards from central Africa (Beuf et al. 1971; Deynoux 1985; Sutcliffe et al. 2000), possibly as far south as the Western Cape Province, which was situated c. 30 – 45°S at the time (Sutcliffe et al. 2000; Torsvik & Cocks 2009; Fig. 1). Although the late Ordovician was certainly a cold environment in South Africa, there is debate on the extent to which ice sheets influenced the deposition of the Pakhuis Formation and associated Peninsula Formation ‘Fold Zone’ (Backeberg & Rowe 2009; Blignault & Theron 2010, 2011; Rowe & Backeberg 2011). The distinctive lithologies of the Pakhuis Formation comprise sandy diamictites, sandstones and rarely varved mudrocks with dropstones. Piedmont glaciers, resulting from late Ordovician orogeny, are also implicated in Pakhuis Formation deposition (Young et al. 2004). The Soom Shale directly overlies the Pakhuis Formation with a gradational contact; indeed, within the lowermost metre of the Soom Shale diamict-like layers occur sporadically. The main part of the Soom Shale Member is composed of alternating laminae of mud and silt lacking evidence of bioturbation, which were deposited during post-glacial transgression (Gabbott 1998; Aldridge et al. 2001). High molybdenum abundance and carbon, sulphur and iron geochemistry point to a dominantly anoxic and at times euxinic basin (Fig. 2; Gabbott 1998). Thus, together the Pakhuis and Cedarberg Formations provide a continuous lithological record of the rapid collapse of the ice sheet in South Africa, and subsequent climatic amelioration. However, the Soom Shale sediments suggest a lingering glacial influence (see Box 1).

The Soom Shale preserves some taxa of the distinctive Hirnantian brachiopod fauna (Bassett et al. 2009). Elsewhere, this fauna is known to have survived the first extinction phase across high to moderate latitudes but the fauna disappeared almost entirely during the second extinction phase, except in a couple of refugia (Rong & Harper 1988; Harper et al. 2014). It is possible that the Soom Shale represents such a refugium, but in a localized postglacial cold-water setting (Sheehan 2001).

The Soom Shale is perhaps most famous for conodonts. The remains of this taxon exemplify the quality of preservation, the unusual composition of the fossils and that Soom taxa are often anatomically peculiar members of their kind. Conodonts are mostly represented in the fossil record by millimetre-sized phosphatic elements recovered after acid preparation of rocks from Cambrian to Late Triassic age; they are the teeth of otherwise nonbiomineralized jawless fish (Aldridge et al. 1986). Rarely, these phosphatic elements occur as natural assemblages reflecting the morphology and topology of the conodont tooth-set, albeit collapsed into two dimensions on the bedding plane. The Soom Shale preserves hundreds of bedding plane assemblages of conodonts, which can be seen on the shale surface with the naked eye. All are from the priniodontid order; those of the taxon Promissum pulchrum form a significant proportion of the fossils collected from the Lagerstätte (Aldridge et al. 2013). They are both the most complex conodont apparatus known (Aldridge et al. 1995) and the largest, with single elements up to 25 mm long (Fig. 4c). A second taxon, Notiodella keblon, was the first 17-element apparatus discovered from the fossil record (Theron et al. 1990; Aldridge et al. 2013). Interestingly, both taxa preserve complete tooth-sets more frequently than single elements, which is rare in the fossil record. This suggests that once the carcasses were deposited on the seafloor they lay undisturbed by physical or biological disturbance. Many of these bedding plane assemblages also preserve the eyes of the animal as either dark, carbonaceous rings or circles, or as silvery films of clay minerals (Aldridge & Theron 1993; Gabbott 1998).

As well as eyes, a single specimen of Promissum pulchrum also preserves aspects of the body (Fig. 4d). The preserved part of the Soom Shale conodont is 109 mm long, although the posterior portion is missing from the edge of the slab; scaling using complete animals from Scotland (see below) suggests that Promissum pulchrum would have been c. 40 cm in length (compared with 4 cm for the Scottish specimens).

Only four localities worldwide record conodont soft tissues. In the Silurian, one locality preserves a remnant of the trunk (Smith et al. 1987), another locality the eyes (von Bitter et al. 2007), whereas in the Carboniferous Granton Shrimp Bed of Scotland (Briggs et al. 1983; Aldridge et al. 1986, 1993) and the Soom Shale more details of the animal head and trunk are preserved. There are several specimens of conodonts in the Granton Shrimp Bed showing a long slender body with eyes, trunk musculature and a notochord, along with other features including a ray-supported caudal fin (Briggs et al. 1983; Aldridge et al. 1986, 1993). The Soom Shale conodont preserves eyes, V-shaped muscle blocks and a ventrally positioned dark patch of material interpreted to be the liver (Gabbott et al. 1995). Scanning electron microscopy of small samples of the eye and V-shaped muscle blocks showed them to comprise fibres of c. 5 μm diameter, which were interpreted to represent, in the eyes, extrinsic eye musculature, and in the trunk, muscle fibres (Gabbott et al. 1995). However, subsequent analyses of these fibres across the entire animal show that they are all similarly oriented, which may point towards an alternative explanation: fibrous clay minerals may have precipitated on the position of the eyes and muscle blocks in a manner similar to that postulated for late diagenetic–metamorphic clays on Burgess Shale fossils (see Butterfield et al. 2007; Page et al. 2008). Regardless of how muscle blocks and other soft tissue anatomy were preserved in the Soom Shale, specimen anatomy from both the Granton Shrimp Bed and the Soom Shale conodonts demonstrates a vertebrate affinity. They are shown cladistically to be more derived than either hagfishes or lampreys (both nonbiomineralized taxa) because they possess a mineralized dermal skeleton (e.g. Donoghue et al. 2000; Sansom et al. 2010).

The only other possible vertebrates in the Soom Shale are the remains of naked and completely unbiomineralized agnatha, the largest of which is about 11 cm in length (Aldridge et al. 2001) showing excellent preservation of the eyes, branchial structures and possibly the liver. This taxon is being described.

Arthropods are the most diverse group represented in the Soom Shale; trilobites, naraoiids, eurypterids, ostracods and caryocarids have been identified (Braddy et al. 1995, 1999; Aldridge et al. 2001; Gabbott et al. 2003; Whittle et al. 2007), along with a selection of unidentified specimens (Whittle 2007). Muscles, gut traces and lamellate book gills are preserved in eurypterid specimens (Fig. 4b; Onychopterella augusti), the presence of four pairs of book-gills supporting a sister group relationship between eurypterids and scorpions (Braddy et al. 1995, 1999). Whittle et al. (2009) identified a lobopodian from the Soom Shale providing a link between Cambrian lobopodians from sites such as the Burgess Shale and Chengjiang and younger forms from the Silurian and Carboniferous. A fossil belonging to a group more typically associated with Cambrian strata is the uncalcified arthropod Soomaspsis splendidaFortey & Theron 1995. This taxon has a cephalic shield and pygidium of roughly equal size between which are three thoracic segments (with possibly a fourth concealed beneath the cephalic shield); like other primitive, uncalcified trilobites it is in the Naraoiidae family. Myodocope ostracods (Myodoprimigenia fistuca) are found in the Soom Shale in exclusive association with orthoconic cephalopods, suggesting that they fed on cephalopod carrion as epibenthic scavengers (Gabbott et al. 2003). The morphology, including a weakly mineralized, thin flexible shell and pronounced rostrum and rostal incisure, most strongly suggest an ostracod that lived off the seafloor, perhaps being nektobenthic. The caryocarid Soomicaris cedarbergensis is an extremely rare component of the Soom Shale fauna but it is important as it extends the geographical, ecological and temporal range of the family and provides the only evidence of a zooplanktonic constituent to the Soom Shale biota, as no graptolites have been found (Whittle et al. 2009; Racheboeuf & Crasquin 2010).

The most common molluscs recovered from the Soom Shale are orthoconic nautiloids, the shells of which occur as moulds with some relief, after being completely demineralized. The specimens preserve a possible stomach and an unambiguous radula, which consists of up to 26 transverse rows with five simple tooth-like elements in each row (Gabbott 1999). Radulae are a key synapomorphy of the Mollusca and provide important phylogenetic and ecological information, but they are rarely preserved. For example, from the entire Lower Palaeozoic succession, radulae occur in the early Cambrian Mahto Formation, Alberta, Canada (Butterfield 2008) and are interpreted in both Wiwaxia and Odontogriphus from the Burgess Shale (Caron et al. 2006; but see Butterfield 2006, for an alternative view). In the Silurian there are two occurrences (Mehl 1984; Sutton et al. 2006).

Other molluscs include incomplete specimens of a bellerophontid monoplacophoran belonging to the family Sinuitidae, tentatively identified as Bucanella sp., and specimens of infaunal deposit feeding nuculanacean bivalves, attributed to Concavodonta (Cocks & Fortey 1986). All are extremely rare.

Scolecodonts are the small (usually less than 1 mm) complex jaw apparatuses of polychaete annelid worms, which are usually recovered as single elements after microfossil processing. In the Soom Shale scolecondonts occur as complete sets of elements that comprise jaw apparatuses on the bedding planes, a rarity in the fossil record. What makes them more important, however, is that the record of Ordovician scolecondonts from Gondwana is poor, being mostly based on either single-element taxa or records of ‘scolecodonts’ with no accompanying taxonomic treatment (Hints et al. 2015). Scolecondonts described from the Soom Shale were low in diversity and were identified as probable mochtyellids and xanioprionids (Synaptogenys and Xanioprion?) (Whittle et al. 2008). No associated soft tissue remains have been found.

The four brachiopod genera (three Linguliformea and one Rhynchonelliformea) described from the Soom Shale have been identified as members of the distinctive world-wide late Ordovician Hirnantia Fauna (Bassett et al. 2009). Rarely soft tissue preservation occurs. For example, in K. cedarbergensis 3D preservation of the pedicle represents the earliest documented, and possibly the only record of this kind for Palaeozoic discinoids; its structure is similar to that of Recent lingulids. Trematis taljaardi preserves bands of periostracum on the flanks of the pedicle notch.

The most frequently found fossil in the Soom Shale is its most enigmatic. Carbonaceous compressions of small (typically 5 – 15 mm in length and 1 – 2 mm in maximum width) pointed axes occur on bedding planes singly, as a few specimens together and as dense masses (Fig. 4f,g). They do not show any consistency in orientation, but are commonly found in association with conodont assemblages. The axes are mostly straight, although flexed and folded, and even sinuous specimens are known. Parallel to their long axes are a series of longitudinal lines (usually four or five) with relief, and small sets of depressions running between them. Another morphotype of the same fossil has small, paired projections orthogonal to the main axes. They were originally described by Kovács-Endrödy (1986) as the remains of a non-vascular land plant and given the designation Eohostimella parva. Subsequently, Chesselet (1992) reinterpreted these structures as possible disarticulated remains of a problematic Ordovician metazoan and renamed them as Siphonacis parva. However, there is little evidence to constrain what these common fossils are.

The largest fossil in the Soom Shale is of a multisegmented (c. 45 segments) animal some 380 mm long and 140 mm wide (Aldridge et al. 2001). The animal is a bilaterian as evinced by its obvious cardinal axes and bilateral symmetry. There are two very well-preserved specimens with complex, well-defined anatomical features, which overlap in the main body. The body margin is characterized by imbricating flaps, one per segment, which are similar, although less robust in appearance, to those in anomalocariids (e.g. Daley 2013). This fossil awaits a full description.

There are also many other soft-bodied fossils with unknown affinities, which have yet to be identified, and others that are represented by a single occurrence where a tentative assignment awaits confirmation through discovery of other specimens (Aldridge et al. 2001; Whittle 2007). One such occurrence is an unidentified arthropod with inwardly directed appendages of two distinct types: one is multiarticulate and biramous; the other is paddle-shaped with setiferous terminations (Fig. 4e).

Although not constituting ‘exceptional preservation’, algae and phytoplankton form an important component of the Soom Shale, being at the base of the food chain (Fig. 4a). Almost all of the bedding surfaces of the Soom Shale are covered with algae, which probably constituted the greatest biomass of the biota (Gabbott 1998). It is not known whether the algae lived on the seafloor forming a mat, or in the water column, but their intimate association with large quartz grains may indicate the latter (see Box 1). Phytoplankton have been recognized in the form of leiospheres, sphaeromorphs and spherical bodies attributed to the Prasinophyta, a group of green algae (Tappan 1980; Gray et al. 1986). Acritarchs (Gray et al. 1986; Aldridge et al. 2001) and small spore tetrads (Cocks & Fortey 1986; Gray et al. 1986) also occur.

Chitinozoans are recorded after standard palynological preparation (Cramer et al. 1974; Vandenbroucke et al. 2009) but can also be seen on bedding planes with a hand lens. Here they occur as scattered individuals, in linked chains, in aggregated masses and with associated organic envelopes interpreted to be cocoons (Gabbott et al. 1998).

The parallel-laminated Soom Shale Member sediment lacks clear bioturbation, although extremely rarely there may be some surficial disturbance and there are possible pelletal textures. Unsurprisingly then, the fauna of the Soom Shale preserves no infauna, except for a handful of extremely small (<5 mm long) lingulid brachiopods. These are bedding-parallel and may have been transported in, or more probably they colonized during brief periods of bottom and pore water oxygenation, dying as juveniles when anoxia was re-established. There is a paucity of sessile epibenthos, accounting to some extent for the low-diversity fauna when compared with other marine, shallow-water Ordovician communities such as those in north Wales and Scotland (Selden & Nudds 2012). Occasional single, but more commonly clusters of brachiopods occur on bedding planes, unassociated with orthoconic nautiloids, demonstrating occasional bottom water oxygenation, but such occurrences are rare. Gregarious nektobenthos include lobopods, eurypterids, naraoiids, ostracods and trilobites. Evidently, bottom waters were oxygenated for periods of time.

The fully motile nekton niche is represented by orthoconic nautiloids (including their pseudoplanktic brachiopods), conodonts and agnathans. As in many Lagerstätten the record of zooplankton is impoverished but a caryocaridid represents a mesopelagic taxon that was probably an active swimmer (Vannier et al. 2003; Whittle et al. 2007).

There are several taxa associated with the orthoconic nautiloids. Brachiopods and cornulitids are found on the orthocones; the aligned, adaperturally opening cornulitids suggest that attachment and growth occurred during the life of the orthocone, and size distribution analysis of the brachiopods also indicated colonization within the lifetime of the orthocone (Gabbott 1999). The presence of ostracods in the body chambers of orthocones suggests a benthic, or possibly nektobenthic scavenging mode of life, feeding on cephalopod carrion (Gabbott et al. 2003).

There are taxa that are conspicuously missing from the fauna and their absence provides some indication of the environment. For example, that there are no echinoderms, gastropods and bivalves is best explained through a persistently anoxic or euxinic seafloor, with brief interludes of oxygenation; there is no evidence of a soupy substrate prohibiting benthic colonization. Graptolites are another fossil group notable for their absence, particularly because, in all other records worldwide of carycarids (present in the Soom Shale) they co-occur with graptolites, both being cosmopolitan constituents of Ordovician plankton (Whittle et al. 2007). Therefore the absence of graptolites in the Soom Shale is puzzling. Non-preservation is unlikely; perhaps a cold-water setting precluded them.

Five categories of bromalite have been identified from the deposit based on shape, content and internal structure (Aldridge et al. 2006). The presence of crushed discinoid valves in several bromalite occurrences demonstrates that an effective durophagous predator was present in the Soom Shale community. Furthermore, the presence of broken conodont elements in other specimens provides evidence that there was some tiering within the predatory trophic level (Aldridge et al. 2006).

The fossils of the Soom Shale show virtually no evidence of transport and the spine-like Siphonacis parva, which would have been very susceptible to even weak currents, demonstrates alignment only in rare instances. The disarticulation and scavenging of remains is also not apparent. Carcasses were delivered rapidly to a relatively cold and often anoxic (and intermittently euxinic) seafloor, where they may have been rapidly buried by either distal turbidites or algal material (Gabbott et al. 2010).

In the Soom Shale original biominerals such as aragonite (orthoconic nautiloids), calcite (trilobites) and even apatite (conodont elements and brachiopods) are lost. Morphology is either recorded as a mould or replicated by clays, alunite group minerals and rarely quartz (Gabbott 1998). The nonbiomineralized tissues occur either as clays (usually illites) or as carbonaceous films, with more labile tissues exhibiting the former and more recalcitrant tissues the latter preservation modes. This unusual mode of preservation where soft tissues have apparently undergone clay mineralization (Gabbott 1998; Gabbott et al. 2001) has since been recorded in other deposits (e.g. Zhu et al. 2005; Callow & Brasier 2009; Anderson et al. 2011; Cai et al. 2012; Wacey et al. 2014). However, although rapid stabilization of decaying tissues and their replacement by aluminosilicates was postulated by Gabbott (1998), the complex diagenetic and anchizone metamorphism experienced by the Soom Shale makes it challenging to disentangle the role, if any, of clays in preservation. Whether they undertook an active role in decay retardation, and directly replaced the tissues, or whether they are diagenetic or metamorphic artefacts is unknown. In the Burgess Shale aluminosilicates are found on many taxa and are thought to have been the result of volatilization of carbonaceous material during metamorphism, which allowed clay mineral growth in voids (Butterfield et al. 2007; Page et al. 2008). However, convincing instances of aluminosilicification of organic matter, and fossils, are known (Wacey et al. 2014) as are records of bacterial mediation of clay mineralization (e.g. Urrutia & Beveridge 1994; Fiore et al. 2011). Furthermore, clay minerals are seen to attach to lobster eggs in decay experiments (Martin et al. 2004). More work is required to determine precisely the nature of the clays in the Soom Shale.

A range of geochemical redox proxies, including Mo abundance and Fe, S and C relationships, indicate that the Soom Shale was a dominantly anoxic and occasionally euxinic environment (Gabbott 1998). Loss of carbonate biominerals is attributed to low-pH waters produced by the decomposition of copious organic matter. However, extremely low pH values are required for dissolution of apatite and this may have been facilitated by Fe becoming limiting to pyrite formation, allowing a build-up of H₂S in pore and bottom waters. The occasional oxidation of such water may have produced acidity sufficient to demineralize apatite (Gabbott 1998). In a core of black, unweathered Soom Shale, a few conodont elements have been recovered and these demonstrate demineralization, so at least in these cases recent weathering cannot be invoked to account for mineral loss. However, the extent to which this is the case across the deposit is not tested.

Fossils are rare and difficult to find in the Soom Shale but the endeavour is worth while because the fauna provides us with an unmatched view of Late Ordovician ecology in a unique environment. The remains of hundreds of conodont bedding plane assemblages, many with associated eyes, and a single specimen with the trunk preserved, first brought attention to the Soom Shale. However, it is likely to be the description of enigmatic taxa that will ensure that this Lagerstätte continues to add significantly to palaeobiological reconstructions of the biosphere of the end-Ordovician, a time of dramatic biotic upheaval.

We dedicate this paper to Dick Aldridge, who, alongside Johannes Theron, recognized the significance and importance of the Soom Shale and did much to promote its research over the last 25 years. This work was supported by: National Geographic Global Exploration Fund–Northern Europe Award (GEFNE90-13); Geological Society of London Gloyne Outdoor Geological Research Fund; Royal Society Equipment Grant (all to S.G.). C. Nedza is thanked for photography and T. Clements and J. Zalasiewicz for proofreading. J. Zalasiewicz and M. Norry are thanked for many helpful insights and discussion on aspects of the Soom Shale. Some logistic support in the field and in the laboratory was provided by the Council for Geoscience, South Africa, which also made the specimens available for study. The South African Heritage Resources Agency provided permission for us to collect and temporarily export the fossils. Mr and Mrs J. Nieuwoudt (Keurbos), Mr and Mrs R. Nieuwoudt (Kromrivier) and Mr and Mrs J. Coetze (Sandfontein) kindly allowed us to collect on their land, and provided friendly hospitality. The Palaeontological Association is thanked for permission to reproduce Figure 4b.