Implications of summer breeding frogs from Langebaanweg , South Africa : Regional climate evolution at 5 . 1 mya

FUNDING: DST–NRF Centre of Excellence in Palaeosciences; DST–NRF Centre of Excellence for Invasion Biology; National Research Foundation (South Africa); University of the Free State No direct palaeoclimatic proxies have been available to indicate the seasonality or amount of rainfall on the west coast of southern Africa during the Early Pliocene. The Benguela Upwelling System (BUS) is today one of the factors responsible for the present-day aridity on the west coast of southern Africa. The initiation of the BUS is frequently linked to the entrenchment of aridity and the establishment of the current winter rainfall pattern on the west coast; however, marine proxies are inconclusive regarding the effects of past fluctuations in the BUS and sea surface temperatures on the rainfall regime. Neither the fossil evidence nor the fact that plants using the C3 photosynthetic pathway predominate at this time, provide direct evidence of winter rainfall at Langebaanweg. We challenge certain assumptions which are commonly made in the literature regarding the timing of inception of a winter rainfall regime on the west coast and the onset of aridity in the Langebaan region, and provide new evidence as to seasonality of rainfall at Langebaanweg in the Early Pliocene. Herein, the identification of frog species from the genus Ptychadena from Langebaanweg provides new and compelling evidence for a summer rainfall regime, or of at least significant summer rainfall, at 5.1 mya in the southwestern Cape of South Africa.


Introduction
Southern Africa is a predominantly summer rainfall region but the Winter Rainfall Zone (WRZ) situated along the southwestern and southern tip of the African continent, lying at the boundary between southern hemisphere temperate and tropical climate systems, is an exception (Figure 1). Evidence for the inception, geographical extent, intensity and fluctuations of the WRZ prior to the last glacial period is currently fragmented through both time and space, and a coherent understanding of the evolution of this ecosystem, unique in sub-Saharan Africa, has not been attained. 1 Although progress has been made in understanding climate change and rainfall seasonality in the Late Quaternary, 2-5 the fluctuations of these variables over the Neogene are complex and incompletely understood. The Neogene terrestrial fossil record is sporadic and incomplete as a result of the lack of preservation of terrestrial organic materials and thus direct climatic proxies, deposits which are problematic in terms of dating, and a lack of understanding of the geomorphological evolution of the subcontinent. 5,6 Much therefore remains to be elucidated about the inception and evolution of the WRZ. In this paper, we challenge certain assumptions which are commonly made in the literature regarding the inception of a winter rainfall regime on the west coast and the onset of aridity in the Langebaan region, and provide new evidence for rainfall seasonality at Langebaanweg in the Early Pliocene.
The Early Pliocene (~5.1 million years ago (mya)) site of Langebaanweg (LBW) situated on the southwestern coast of South Africa (Figure 1), is a remarkably rich and world-renowned vertebrate fossil site, representing a time period when fossil assemblages are particularly rare in sub-Saharan Africa. 7 LBW is the site of first appearance in the fossil record for numerous taxa, including large and small mammals, birds, reptiles and amphibians. 7-10 LBW sheds light on the critical period during which tectonism led to the latitudinal migration of climate belts, which resulted in widespread reorganisation in atmospheric and ocean circulation. 11 These events were characterised by a more humid episode, with modern aridity becoming established only in the later Pliocene, between ~4 mya and 2.8 mya. 11 Congruently, the fauna at LBW indicates more humid conditions than at present, but the precise nature of the vegetation in the area remains uncertain and until recent research on the frog fauna there were no effective proxies indicating the amount of rainfall, or seasonality, at the fossil site. 10,[12][13][14] A recent study of the rich and diverse anuran (frog) community from LBW served as an effective and direct climatic proxy, and showed that this group, sensitive as they are to rainfall and moisture, supported the higher humidity inferred from studies of other faunal groups from the site, and indicated relatively high rainfall. 12,15 The huge contrast between the frog community in the Langebaanweg region today, as compared with 5.1 mya, is clearly indicated by the fact that species richness on the relatively dry west coast of South Africa is low today (1-10 species) 16,17 , whereas at LBW, six families and some 19 taxa have been differentiated 12 . The LBW frog fauna is put in context when compared to modern species richness elsewhere in the southwestern Cape, which varies from 11 to 30 species. 16,18 Currently, southern African Anura are distributed throughout three different rainfall zones: the Summer Rainfall Zone (SRZ), the Winter Rainfall Zone (WRZ) and the Year-round Rainfall Zone (YRZ) which is found on the South African south coast. A number of amphibian taxa from the WRZ and the SRZ overlap in the YRZ. 18 All frog families and genera previously identified at LBW were ambiguous in terms of the seasonality of rainfall as all but one of the species (a SRZ taxon) identified currently occur within all three rainfall regimes (Table 1). In the present study, we take a step further, utilising the recent identification at LBW of two species of the genus Ptychadena (family Ptychadenidae) to shed further light on the seasonality of rainfall at this site. Additionally, this discovery further elucidates the biogeography of the family Ptychadenidae.

Climatic history of the Winter Rainfall Zone
The WRZ receives predominantly winter rainfall from eastward migrating cold fronts embedded in polar cyclonic systems originating over the South Atlantic. 19,20 During summer, the South Atlantic Anticyclone is well developed and migrates to the south, blocking both the westward propagation of easterly waves that bring summer rainfall to much of southern Africa, and the polar frontal systems. The influence of the polar frontal systems diminishes northwards and is linked to decreasing rainfall and increasing aridity. 14 The cold, nutrient-rich Benguela Upwelling System (BUS) supports a diverse and rich range of marine life along the southwestern African coast. 21 The South Atlantic Anticyclone promotes the BUS and is thus the main cause of the present aridity along the southwestern African coast, the Namib, and the dryness of the adjacent interior of the sub-continent. 5,20,[22][23][24][25] In general, the climate on the west coast is determined by the complex seasonal shifts and interplay of the prevailing winds over the Benguela region. 5 Fluctuations in summer rainfall are linked to upwelling intensity in the BUS 26 and along the south coast 5 . The situation is complex, as indicated by the fact that Chase et al. 5 found that at a hyrax-midden site falling within the present WRZ, but close to the border with the SRZ, summer rainfall variability over the Holocene during the last 7000 years had the greatest impact on water availability, and was the primary determinant of humidity. Analysis of herbivore tooth enamel indicates that there were incursions of C 4 vegetation in the mid-Pleistocene, 27 which could be related to decreased atmospheric pCO 2 conditions during interglacial periods, or possibly to changes in rainfall seasonality.
Recent research on marine proxies for sea surface temperature (SST) in the southern Cape Basin -such as calcareous dinoflagellate cysts and alkenones -suggests the inception of the BUS at about 10.5-10 mya. 28,29 How the BUS became established is comprehensively summarised in Neumann and Bamford 6 . Some authors 29,30 have suggested that tectonic uplift in southwestern Africa would have led to a cooling in the BUS at 12 mya, and after 5 mya, but the timing and cause of such tectonic events, and even their occurrence, is controversial 6,31 . The initiation of the BUS has frequently been linked in the literature to summer aridity and the entrenchment of the current winter rainfall pattern on the west coast 6,24,32,33 , and this in turn has been taken to have influenced other processes, such as the diversification of plants (see Altwegg et al. 33 -although a recent analysis 1 failed to find proof that seasonal aridity at ~8 mya triggered floristic radiation and diversification in the Cape Floral Region).
The connection between the BUS and seasonality of rainfall is unclear as estimates of the evolution of SSTs are not available 34 and the marine record indicates that the BUS and SSTs have fluctuated considerably over time in the late Neogene. High glacio-eustatic sea levels and the presence of warm-water molluscan taxa in the middle to late Pliocene 35,36 add further complexity to deciphering the evolution of oceanic and climatic conditions in the current WRZ.   There is also evidence of an oceanic subsurface warming between 6.5 mya and 5.0 mya which may be related to variability in the strength of North Atlantic deepwater formation on the temperatures of the intermediate waters of the South Atlantic. 34 The climatic data during LBW times are ambiguous in terms of oceanographic conditions, with SSTs apparently in sharp decline, but with low productivity (suggesting reduced upwelling) indicated by a drop in total organic carbon. 34 The fossil-bearing members of the Varswater Formation present at LBW conversely contain an abundance of authigenic phosphate, which is widely interpreted as being indicative of strong upwelling (Roberts et al. 14 and references therein). The presence of some mollusc taxa from LBW which are currently found northwards from East London and eastwards from False Bay suggest warmer ocean temperatures than present. 37

Identification and provenance of Ptychadena at Langebaanweg
The majority of fossils from LBW were recovered from two highly fossiliferous members which form part of the Varswater Formation, namely the Muishondfontein Pelletal Phosphate Member (MPPM) and the Langeberg Quartzose Sand Member (LQSM). See Roberts et al. 14 for further details on the geology of the site.
An examination of LQSM sediments (~5.1 mya) collected from a mine dump during phosphate-mining operations several decades ago was recently undertaken. The interpretation of the LQSM in the area from which the dump derived (the so-called 'east stream') was that it represented a river floodplain. 38 In the course of this study, an additional anuran genus (accession number SAM-PQL-70839) was identified -Ptychadena (Family: Ptychadenidae) -on the basis of a single sacrum ( Figure 2). Subsequently a further ptychadenid, which appears to belong to a relatively larger species, was recovered from MPPM deposits which were excavated from the main dig site at the West Coast Fossil Park from square G5 (accession number SAM-PQL-71524) ( Figure 2). The larger Ptychadenid showed several morphological differences to the smaller taxon, including a greater distance between the anterior and posterior articular ends of the sacrum (best seen in ventral view, Figure  2b and 2f), a larger neural spine, and a wider spacing of the sacral articular condyles.
The sacrum of a modern Ptychadenid, Ptychadena mascareniensis, from the Iziko South African Museum (ZR-045512), was scanned using microcomputed tomography and then measured for comparative purposes (see Figure 2 for positioning of measurements  39 Four unique morphological synapomorphies are noted, including the symmetrical fusion of the eighth presacral vertebra with the sacral vertebra (sacrum), forming a composite sacrum. 40 The LBW ptychadenid sacrum displays the fusion of the presacral vertebra and sacrum, together with the distinctive anterolateral orientation and morphology of the transverse processes of members of the genus Ptychadena, including a characteristic foramen above the centrum that serves as the outlet for the spinal nerve (as noted by Blackburn et al. 41 ). In addition, Ptychadena species differ from many ranids in that the neural arches of the composite sacrum lack deep grooves and fossae on the dorsal surface and exhibit a single minute low-lying neural spine. 40 Blackburn et al. 41 provide conclusive evidence, through a thorough investigation of extant and fossil taxa, that the Ptychadenidae are the only major clade of frogs with the above sacral morphologies, making the identification of the LBW taxa indisputable.
LBW extends our knowledge of the fossil distribution of ptychadenids to include the southwestern Cape at 5.1 mya, and represents the most southerly occurrence of this genus to date in the African fossil record.
The fossil frog bones reported on in this paper are curated by the Iziko South African Museum, Cape Town, and each bone reported on has been allocated an identifying accession number.

Other fossil sites containing Ptychadena species
The antiquity of the Ptychadena lineage is indicated by the earliest appearance of Ptychadena in the fossil record in the late Oligocene sediments (~25 mya) from the Nsungwe Formation in southwestern Tanzania. 40 This record represents the earliest of fossil anurans from Africa below the equator, and is the earliest definitive record of any family within the diverse ranoid clade Natatanura (sensu Blackburn et al. 41 39 , with the exception of two species found in the Nile Delta 40 . The distribution of the genus Ptychadena in southern Africa is delimited in areas where rainfall is above 60 mm in January and the minimum temperature is above 8 °C in September. These two variables represent 36.4% and 29.0% of the distribution of the genus in southern Africa, respectively (G.J.M. unpublished data). These two parameters describe the distribution of extant species from East London to the northeast, under the influence of the SRZ and increasing temperatures (Figure 1).
In the current WRZ, all modern species breed in winter, with the exception of two invasive species from the SRZ. 45 Some genera and families are found in both the WRZ and SRZ, and the remaining southern African families and genera are found in the SRZ and breed in summer (Table 1). Six amphibian families have been identified from LBW (Hyperoliidae, Brevicipitidae, Pxyicephalidae, Pipidae, Heleophrynidae and Bufonidae), and 19 different taxa, 10 of which remain unidentified, have been differentiated. 12 Genera identified include Hyperolius, Breviceps, Tomopterna, Xenopus, Heleophryne and Amietophrynus. 12 All the frog families and genera identified from LBW 12 currently occur in all three rainfall regimes (SRZ, WRZ and YRZ), with endemism to the WRZ only existing at species level in these genera 16 . Matthews et al. 12 tentatively identified Kassina (a genus restricted to the SRZ) -an identification which has subsequently been confirmed by computed tomography scans of comparative material, but the family to which it belongs (Hyperoliidae) contains other genera which are found in the current WRZ. Interestingly, histological studies done on the aquatic Xenopus (Family: Pipidae; common name: the African clawed frog) femora from LBW (paper in preparation) suggest marked seasonality as clear LAGS (lines of arrested growth) are evident.
Phylogeographic structuring in the distribution of anuran taxa with respect to the winter and summer rainfall regions has been recorded, 46 but little is known about the evolution of this pattern which clearly has deep roots. In the literature, the southern African amphibian fauna is commonly divided into two main groups -the so-called 'tropical' (a term used to describe the speciose tropical frog fauna distributed widely in the northeastern parts of southern Africa) and the unique 'Cape' frog faunas of the southwestern Cape which are characterised by extremely high rates of endemism and contain a large number of species with narrow niche envelopes. 16,17,47,48 These two groups correspond to the SRZ (the tropical frog fauna) and WRZ (the Cape frog fauna), and their breeding periods are delineated by mean annual precipitation in the summer or winter rainfall months, respectively. 17,48 The two rainfall regimes have such a high turnover in species that it is evident at a global scale. 49 Nonetheless, there are some genera that straddle both rainfall zones, but the population genetics of these taxa suggest deep genetic divisions, such that populations are confined to one zone or the other. 46

The palaeoenvironment at Langebaanweg
Fossil pollen is frequently used in palaeoenvironmental reconstruction to ascertain whether C 3 plants (those which fix and reduce inorganic CO 2 into organic compounds using the C 3 photosynthetic pathway) or C 4 plants (those which employ the C 4 photosynthetic pathway) dominated a palaeo-landscape. Grasses categorised as C 4 vegetation are typically associated with a summer rainfall regime, and C 3 grasses with the WRZ. Franz-Odendaal et al.'s 32 research on the large mammals from LBW indicated that the expected grazers (e.g. alcelaphine, hippopotamus and rhinoceros), as well as browsing species, showed d13C values which indicated that LBW was a C 3 dominated environment. This finding was corroborated by a study of phytoliths from the site. 50 The small mammals from LBW, such as the rats and mice, suggest a fynbos component to the vegetation 9 , as does fossil pollen 51,52 . The presence of cool-growing C 3 grasses at LBW during the deposition of MPPM sediments was taken to indicate that the present-day climatic regime of winter wet/summer dry was established early in the Pliocene epoch. 32 Using the vegetation to establish the rainfall regime is, however, questionable, given that fynbos and C 3 grasses both have a C 3 signature, and the contribution of C 3 grasses versus that of C 3 fynbos at LBW is undetermined. In addition, there are indications that the contribution from C 3 grasses to the general C 3 signature at LBW was small, as Stynder 10 suggests that fossil bovid species indicate that grass was scarce, and the environment may have been heavily wooded. The presence of woodland is supported by the presence of certain bird taxa, but arid and semi-arid landscapes are also represented. 7,9 The precise nature of the Early Pliocene vegetation at LBW remains elusive, has no modern analogue, and cannot contribute in any verifiable way to ascertaining the rainfall regime, or provide information on seasonality, at LBW. The fact that fynbos was present on the west coast during the Early Pliocene has been used as evidence of a WRZ; however, the premises on which this association is based are arguable and are discussed further in the next section.

Inception of the WRZ and aridification of the west coast
Of the eight frog genera found at LBW, six (Hyperolius, Breviceps, Tomopterna, Heleophryne, Amietophrynus and Xenopus) are found in summer, winter and year-round rainfall areas. The evidence of seasonality provided by Kassina is not clear-cut as, although this taxon is currently found in the SRZ, other members of the Hyperoliidae, such as Semnodactylus, are found in the current WRZ and YRZ. Ptychadena is, however, currently not found in the more southerly and western areas of South Africa, or in any parts of the WRZ 16 , suggesting that this family has evolved and adapted to live in the SRZ.
The ptychadenids are a marker SRZ species, as they are found only in this rainfall zone, and the presence of Ptychadena at LBW thus provides strong evidence of a summer rainfall regime, or at least of a rainfall regime which included summer rainfall. The faunal evidence from LBW clearly indicates that at 5.1 mya the region was still receiving a relatively high rainfall, which fell partly, if not entirely, in summer.
This finding has important implications for studies involving the evolution of west coast flora and fauna, and the interpretation of phylogenetic studies which use molecular clock estimates to interpret the timing of lineage divergence. As mentioned previously, such studies have typically used the formation of the BUS as a proxy for the inception of the winter rainfall regime along the West Coast and the beginning of aridification. 6,24,32,33 Certain fynbos taxa such as Restionaceae, Ericaceae and Proteaceae formed part of the vegetation in southwestern Africa during the Palaeogene 53 and fossil pollen research indicates that fynbos taxa have formed part of west coast ecosystems since well before the inception of BUS and a winter rainfall regime 14,15,23,51,52,54 . Hoffman et al.'s 1 habitat reconstruction for the dated phylogenies of 12 plant clades from the Cape Floral Region in southern Africa indicated that they evolved in aseasonal rainfall environments. In addition, this research indicated that the initial development of aridity in the Cape Floral Region was not linked to the onset of strong rainfall seasonality -contradicting the common assumption that aridity and the development of a strongly seasonal winter rainfall regime on the west coast occurred at the same time and are effectively linked. This de-linking of these two variables indicates that commonly made assumptions in the literature are incorrect, and illustrates how much still remains to be elucidated about the evolution of the WRZ.
On a global scale, the middle Miocene is considered to have been a time of reduced seasonality and relatively stable climate with elevated temperatures and high humidity and rainfall. 1 54 note that between 10 mya and 6 mya, pollen types from plants of tropical affinity disappeared, and those from the Cape flora gradually increased. The onset of aridity on the west coast has thus been postulated to pre-date the LBW fossils by 3-5 mya. This postulate has, however, been based on palaeoclimatic proxies situated much further to the north of LBW in Namibia 31,55 and the Orange River 54 . The LBW frog community indicates a relatively high rainfall 12 , as do some other fossil taxa 13 , and the fact that woodland appears to have been widespread 10 . The evidence from LBW strongly contradicts assertions that aridification of the west coast in the region was well established by 5.1 mya, although such a scenario was certainly applicable further northwards, with aridity in the Namib dating back to 16 Ma. 55 However, relatively moister conditions appear to have prevailed in the southern, as opposed to the central, Namib desert during the Miocene and the flora appears to have remained C 3 in both regions until the Pliocene. 56 The marine evidence mentioned previously is inconclusive regarding the effects of fluctuations in the BUS and SSTs on the rainfall regime during the Early Pliocene. Much uncertainty therefore exists as to the exact status of the strength, and effect, of the BUS and SSTs, and the inferences drawn from other palaeoclimatic proxies are contestable. Neither the fossil evidence, the fossil pollen data, nor the fact that C 3 plants predominate, provide direct evidence of winter rainfall at LBW, or indeed of the period of inception of the WRZ.

Conclusions
The majority of anuran taxa at LBW are currently found in the SRZ and WRZ, and consequently are not diagnostic in terms of the rainfall regime. However, the identification of two Ptychadena species from LBW, as well as the presence of Kassina, provides new evidence of a summer rainfall regime, or of at least some significant summer rainfall, at ~5.1 mya on the southwestern coast of southern Africa. This evidence is a significant contribution to our understanding of the evolution of seasonality on the west coast in the LBW region and challenges the commonly held assumption that the WRZ has been established on the west coast since the later Miocene/Early Pliocene. The presence of a Ptychadena species also contributes to our knowledge of the palaeo-distribution of the family, which is currently widespread in sub-Saharan Africa, and has endured since the Oligocene in Tanzania.