The main extinction zone is approximately 40 m thick at both the Bethulie and the Graaff-Reinet sections and is stratigraphically positioned in the lower strata of the Palingkloof Member at the top of the Balfour Formation (Fig. 1). Detailed sedimentological logs compiled from several sections at each locality show that the facies sequences in the boundary beds are remarkably similar (Fig. 2). The overall lithological transitions are from drab greenish grey and blue-grey thickly-bedded and pedogenically modified massive siltstone beds with interbedded, laterally-accreted fine grained sandstone bodies grading upwards into progressively more maroon coloured siltstone beds interbedded with vertically accreted sandstones with distinctive gullied basal scours and include a regionally extensive interval of maroon coloured thinly bedded laminites. The PTB sequence is capped by a thick conglomeratic sandstone-dominated succession – the Katberg Formation.

The facies sequence through the PTB is interpreted as a relatively rapid change in fluvial landscape from an alluvial plain traversed by a few large, highly meandering rivers with expansive lowland floodplains (massive dark grey mudrock) through a transitional stage when the rivers straightened and branched into a distributary channel network that scoured the now abandoned floodplains (massive red siltstone) [52]. The field evidence of more flashy discharge in the Triassic is the appearance of distinctive gullies at the base of the channel sandstones filled with reworked pedogenic nodules and large mudrock rip up clasts. The rapid switching of the talweg within a wide, low sinuosity channel is indicated by the numerous vertically-stacked sheet sandstones with glaebule conglomerate-lined disconformities. It has been proposed that this switch in fluvial style was triggered primarily by the die-off of bank strengthening vegetation [55]. As the run-off charged sediment load increased, these channels widened and in-channel bars eventually separated the flow into a braided pattern of interconnected sand-dominated ephemeral channels (conglomeratic sandstone). During deposition of the massive red siltstone facies, there was an apparently synchronous depositional event that immediately followed the extinction of Dicynodon – the last of the Permian dicynodonts to disappear from the Karoo Basin. This resulted in the accumulation of up to 5 m of finely laminated mudrocks (maroon laminites) that show evidence of shallow standing water with periodic sub-aerial exposure and desiccation. Thin sections of the laminae show light dark alternations that may be interpreted as varves and suggest accumulation within a thermally stratified standing water body which in this climatic setting would be termed a playa lake [48]. Periodic flooding deposited sand/mud couplets that show little post-depositional colonization by either animal or plant life except minor root traces of a hydrophilic quillwort [48], and burrows of a possible crustacean. The conclusion drawn from this evidence is that for a short period following the disappearance of Dicynodon, the central Karoo Basin floodplains were almost devoid of vegetation, and were seasonally occupied by widespread playa lakes. Using the average floodplain accretion rate of 1.5 mm calculated by Retallack et al. [48] from the weak palaeosol development in the laminites, applying a 60% compaction ratio [1] and adding two short periods of pedogenesis, gives an estimated time interval for the 5m thick laminites as approximately 10 000 years.

Above the laminites, the floodplain accretion rates decrease and pedogenic overprinting is again evident, except the palaeosol character has changed. The massive red siltstone beds are interpreted as having been initially deposited by overbank flood events but subsequently re-worked by wind action as loess [52]. The palaeosols, although distinctively red in colour, have a slightly greater depth to calcic horizon than those of the Latest Permian. Retallack et al. [48] interpret this as a modest increase in seasonal precipitation from an estimated 346 mm to 732 mm across the boundary, combined with an increase in mean annual temperature and a change in vegetation type from an arid Glossopteris dominated flora to a semi-arid equisetalian flora. Perhaps a more plausible interpretation of the observed changes in fluvial style and pedogenesis is that they were brought about by an increase in mean annual temperature combined with the onset of a monsoonal rainfall regime [48]. Within a completely continental setting such as the Karoo Basin, this resulted in more highly seasonal rainfall and increased storm intensity, but with lower reliability.

The fossils show a pattern of disappearance and a taphonomic mode that suggests that drought may have been a contributing factor to the extinctions. At the beginning of the main period of extinction a new dicynodont genus, Lystrosaurus, first appeared in the basin as rare large isolated skulls and skeletons of the long-nosed Lystrosaurus maccaigi. This new arrival co-existed with the gradually dwindling Dicynodon fauna for an estimated 250–425 kyr (40 m of floodplain strata corrected for 60% compaction and accumulated at an estimated 0.15–0.25 mm yr^–1 [48]) before Dicynodon, the last remnant of the Permian fauna, finally went extinct. By the time the last of the Dicynodon disappeared Lystrosaurus had already become the most common large vertebrate in the basin, yet it was not until between 7.5 and 22.5 kyr later (7 m above the PTB with 60% compaction ratio and accumulation rates of 0.5–1.5 mm yr^–1 [48]) that the rest of the Lystrosaurus Assemblage Zone recovery fauna became abundant enough to be preserved in the rock record. The field evidence shows that the small dicynodonts disappeared before the medium and large herbivores, which suggests that their food source, presumably the undergrowth of lush ferns and clubmosses, disappeared before the Glossopteris shrubs and trees. This is in keeping with the sedimentological interpretation of the lowering of water tables and the onset of drought conditions. The fact that the medium- and large-sized Lystrosaurus seemingly flourished as the Dicynodon fauna began to fade, indicates that it was somehow pre-adapted to survive the worsening conditions. This is discussed more fully further in this report.

Several ‘bonebeds’ comprising numerous (±10) jumbled-up disarticulated sub-adult Lystrosaurus skeletons occur in the maroon floodplain mudrocks of the Lower Triassic Katberg Formation (Fig. 3). They are interpreted as drought induced aggregations where animals died en masse on the periphery of a shrinking waterhole. Subsequent, possibly monsoonal, downpours resulted in short-lived sheet washes that transported the desiccated bones into shallow depressions along with reworked cohesive mud clasts and nodules [53] and [54].

The changes in fluvial style in the Karoo Basin at the end of the Permian were initially thought to be triggered by a pulse of thrusting in the southerly source area, which brought about rapid progradation of a large sandy braided fan system (the Katberg Fm.) into the central parts of the basin [21] and [52]. Ward et al. [55] suggested an alternative that better fits the observed palaeoclimatic changes. They proposed that the switch in fluvial regime was a consequence of the extinction event rather than the cause. The global aridification and concomitant de-vegetation of the continental interiors caused unstable channel banks and increased run-off that is reflected by the onset of gullying of the floodplains and the switch from high to low sinuosity channel pattern.

Stable isotopic analyses of pedogenic nodules and teeth taken from embedded fossils through the extinction interval at the Bethulie section revealed an anomalous negative excursion within the boundary laminites [34] and [56]. This anomaly is interpreted as having been caused by a relatively rapid release of methane or CO₂ into the atmosphere over a period of some 150 kyr. The greenhouse effect from these emissions, whether they originated from the Siberian volcanism [47] or oceanic overturn [32], could have caused the observed global aridification of continental interiors [48].

The collecting data shows that the archosauriform, Proterosuchus, is the first vertebrate taxon to appear in the Karoo Basin following the End-Permian mass extinction (Fig. 5) and, as such, can be regarded as the terrestrial index fossil for the Early Triassic in South Africa. Unlike Moschorhinus, which survived the main extinction event only to disappear soon after, Proterosuchus first appears during the Earliest Triassic (at approximately 7 m above PTB) and extends well into the overlying Katberg Formation [49]. The reasons why Proterosuchus flourished and the similarly sized Moschorhinus declined may be due to lifestyle differences. An amphibious lifestyle is regarded as ancestral for all archosauromorphs, thus, it is likely that the archosauriform Proterosuchus was at least semi-aquatic, and therefore, may have lived in and around the more permanent water bodies. If this is the case, Proterosuchus too would have become part of the preferential preservation of skeletons in and around ephemeral ponds and lakes.

The amphibian genus Micropholis first appears in the fossil record at approximately 18 m above the PTB. These specimens have often been found in association with the procolophonoid ‘Owenetta’ kitchingorum. The species ‘Owenetta’ kitchingorum is widely thought to extend the range of the genus from the Permian into the Triassic (e.g., [51]), but recent studies indicate that it should be placed in another or its own genus [14] and [40]. When this is done, it will result in the absence of Owenetta from the Triassic.

Currently, there are thought to be three Late Permian cynodont genera in the South African Karoo Basin, viz. Procynosuchus, Cynosaurus, and Nanictosaurus [49]. However, our collecting data shows that none of these genera cross the PTB. Three new genera appear in the Earliest Triassic, in the upper Palingkloof Member of the Balfour Formation. They include Progalesaurus [51], Galesaurus and Thrinaxodon liorhinus (Fig. 5). Progalesaurus, the sister taxon of Galesaurus, was recently recovered from the Earliest Triassic Palingkloof Member, approximately 36 m above the PTB [51]. It is currently represented by a single type specimen and, thus, has not been plotted on the range chart in Fig. 5.

The biostratigraphical field data shown in Fig. 5 indicates that of the seven Late Permian dicynodont taxa recovered from the PTB sections in South Africa, only one taxon, Lystrosaurus curvatus, actually survived the extinction. In contrast, our fossil recovery shows that 75% of the Late Permian therocephalian genera survived the extinction. However, to date, we have not found a single Late Permian amphibian, eosuchian, procolophonoid or cynodont genus that survived into the Earliest Triassic portion of the Palingkloof Member in the South African Karoo Basin, although new genera representing all these groups appear later in the Early Triassic strata of the Katberg Formation.

The dicynodonts were clearly more affected by the end-Permian extinction compared to any other therapsid clade, apart from the Gorgonopsia, which are absent from the Early Triassic. Of the nine gorgonopsian genera found in the lower Dicynodon Assemblage Zone, our collecting shows that only two are represented in the uppermost strata, suggesting that the gorgonopsians were in decline well before the end of the Permian [29]. The dicynodonts on the other hand were still relatively diverse and abundant during the latest Late Permian, but were almost completely eradicated during the end-Permian extinction event (although it should be noted that of the eight dicynodont genera present at the time of the Dicynodon Assemblage Zone, only five persisted up to the PTB beds). To date, the field data show that the only dicynodont taxon represented on both sides of the PTB in South Africa is Lystrosaurus curvatus. At present, it can be regarded is a boundary interval marker as its narrow range is restricted to approximately 20 m, spanning the boundary interval itself (Fig. 5, [7]).

Along with Lystrosaurus curvatus, the therocephalian genera Ictidosuchoides, Moschorhinus and Tetracynodon also survived the end-Permian extinction event. Moschorhinus is the only large terrestrial carnivore (apart from the archosauriform Proterosuchus) present during the Earliest Triassic. However, none of these four surviving taxa have been found above the Earliest Triassic portion of the Palingkloof Member of the Balfour Formation, and although they survived the extinction, they do not appear to have flourished during the Early Triassic as all four taxa disappear within 30 m of the PTB (Fig. 5, [16]).

The recovery taxa are all new genera that have not been found in Late Permian strata. Archosauriforms are the first group to appear in Earliest Triassic strata, closely followed by amphibians, procolophonoids, dicynodonts and then cynodonts. These taxa clearly had adaptations/lifestyles that allowed them to flourish on the arid braidplains of the Early Triassic Karoo Basin. It is noteworthy that the recovery in South Africa was faster compared to that in the South Urals basin in Russia [6] and [20]. Benton et al. [5] noted that small fish-eaters and insectivores were absent from the Early Triassic in the South Urals basin after the end-Permian extinction event, whereas aquatic nymph eaters such as Micropholis and insectivores such as Galesaurus and Thrinaxodon appear much sooner after the extinction event in South Africa.

Currently there are four taxonomically valid Lystrosaurus species in South Africa: L. maccaigi, L. curvatus, L. murrayi and L. declivis [18]. Lystrosaurus curvatus, L. murrayi and L. declivis are the only dicynodont species present in the Earliest Early Triassic (Fig. 5). The rest of the dicynodonts including the other Lystrosaurus species, L. maccaigi, became extinct at or immediately below the PTB. Although Lystrosaurus curvatus survived the extinction event, it disappeared soon after the extinction. In contrast, L. murrayi and L. declivis flourished after the extinction to become two of the most abundant and successful herbivores during the Early Triassic. It is clear that L. murrayi and L. declivis as well as the small cynodonts, small amphibians and small procolophonoids were well-adapted to the Early Triassic Karoo environment and had morphological adaptations and/or lifestyles that allowed them to flourish in relatively arid conditions [52]. Several survival strategies have been proposed over the years and the various theories are summarised below.

Although the amphibians Micropholis and Lydekkerina, the archosauriform Proterosuchus, the procolophonoids ‘Owenetta’ kitchingorum and Procolophon, and the cynodonts Thrinaxodon liorhinus and Galesaurus appear to originate in Earliest Triassic strata, it is argued that there must have been at least seven separate ghost lineages that crossed the boundary to give rise to these taxa [14], [41] and [51]. Lystrosaurus murrayi and L. declivis appear to be very closely related taxa [18] and may have evolved from a Triassic L. curvatus type ancestor (although the relationships between the various Lystrosaurus species using cladistic analysis require examination). With the addition of ghost lineages to our field data, 48% of all taxa cross the PTB in the Karoo Basin, but out of these taxa only four genera (17%) actually cross the boundary. Furthermore, these four genera, Lystrosaurus (curvatus), Ictidosuchoides, Tetracynodon and Moschorhinus, disappear close to the boundary in the Earliest Triassic portion of the Palingkloof Member in what could be described as a second wave of extinctions.

Recent work based on the existence of ghost lineages, led Modesto et al. [41] to suggest that the effect of the end-Permian extinction on terrestrial vertebrates was not as great as previously thought (e.g., [5], [17], [25] and [43]). Their studies on Late Permian and Early Triassic parareptiles from the Karoo Basin [38], [39], [40], [41] and [46] have indicated that their survivorship rate may be as much as 71% [41]. This estimation was based on a study of six parareptile species including Barasaurus besairiei, Owenetta rubidgei, Saurodectes rogersorum, 'Owenetta' kitchingorum, Coletta seca, Sauropareion anoplus and the family Procolophonidae. Although Saurodectes, "Owenetta" kitchingorum, Coletta, and Sauropareion have only been recorded from Lower Triassic strata, a phylogenetic analysis shows that all four of these taxa have ghost lineages that must have crossed the PTB [41]. The lineages of Coletta and Sauropareion, which are transitional forms between Owenettidae and Procolophonidae [14] and whose ranges are based on single specimens, share a common ancestor with the Procolophonidae [14] and [41]. The ranges of all three taxa extend into the Late Permian on the basis of two Russian procolophonoids, namely, Suchonosaurus and Kinelia, based on a maxilla and a dentary, respectively [11]. More recently, the Permian Barasaurus besairiei from Madagascar has been shown to cross the PTB as well [27], which would suggest a parareptile survival rate of 86% (six out of seven taxa).

However, according to our field data of the first appearance datum of ‘Owenetta’ kitchingorum (Fig. 5), it is possible that the common Late Permian ancestor (ghost taxon) of Saurodectes and ‘Owenetta’ kitchingorum [14] crossed the PTB before separating into the two distinct genera (also mentioned as a possibility in [40]). This would result in Barasaurus, the ghost taxon of Saurodectes and ‘Owenetta’ kitchingorum, Coletta, Sauropareion and the Procolophonidae crossing the PTB, which amounts to a 71% survival rate and agrees with Modesto et al. [41]. It should be noted, however, that no procolophonid has been found in the Late Permian Karoo Basin of South Africa. It is possible that, in the Karoo Basin, only three lineages (50%), namely, Coletta seca, Sauroparieon anoplus and the ghost taxon of Saurodectes rogersorum and “Owenetta” kitchingorum crossed the PTB.

It is clear from our field data that there was a massive faunal turnover at the PTB in the Karoo Basin with the number of actual genera crossing the boundary being extremely low (only 17%) and no fewer than nine new taxa (41%) appearing during the Earliest Triassic. A mass extinction is commonly defined as having a 40% extinction rate (e.g., [4] and [13]) and from our data, it can be seen that nine out of 23 taxa do not cross the PTB, which is equivalent to 39%. Thus, even when ghost lineages of the Earliest Triassic taxa are considered, the end-Permian extinction in the South African Karoo Basin still qualifies as a mass extinction.

It can be concluded that the taxa that are first recorded from Earliest Triassic strata (and whose lineages may have arisen during the Latest Permian) were in some way pre-adapted to the increasingly arid environment. These taxa include the archosauriform Proterosuchus, the amphibians Micropholis and Lydekkerina, the parareptiles ‘Owenetta’ kitchingorum and Procolophon, the dicynodonts Lystrosaurus murrayi and L. declivis, and the cynodonts Thrinaxodon liorhinus and Galesaurus, all of which may have adopted either burrowing or amphibious lifestyles [30]. Thus, it is possible that both fossorial and amphibious lifestyles enabled terrestrial vertebrates to flourish in the Earliest Triassic continental environments of Gondwana.