Quaternary Science Reviews 65 (2013) 39e52

Contents lists available at SciVerse ScienceDirect

Quaternary Science Reviews

journal homepage: www.elsevier.com/locate/quascirev

Paleoanthropologically significant South African sea dated to 1.1e1.0 million years using a combination of UePb, TT-OSL and palaeomagnetism

Robyn Pickering a,*, Zenobia Jacobs b, Andy I.R. Herries c, Panagiotis Karkanas d, Miryam Bar-Matthews e, Jon D. Woodhead a, Peter Kappen f, Erich Fisher g, Curtis W. Marean g a School of Earth Sciences, University of Melbourne, McCoy Building, Cnr Swanston and Elgin Streets, Parkville, Melbourne, Victoria 3010, b Centre for Archaeological Science, School of Earth and Environmental Sciences, University of Wollongong, New South Wales 2522, Australia c Australian Archaeomagnetism Laboratory, Department of Archaeology, Environment and Community Planning, Faculty of Humanities and Social Sciences, La Trobe University, Victoria 3086, Australia d Ephoreia of PalaeoanthropologyeSpeleology of Southern , Ministry of Culture, Greece e Geological Survey of , 30 Malchei Israel St., Jerusalem 95501, Israel f Department of Physics, Faculty of Science, Technology and Engineering, La Trobe University, Victoria 3086, Australia g Institute of Human Origins, School of Human Evolution and Social Change, PO Box 872402, Arizona State University, Tempe, AZ 85287-2402, USA article info abstract

Article history: Deposits in sea caves found along the southern coastline of have produced a rich and Received 27 September 2012 detailed archaeological record of early modern humans. There is, however, little evidence for coastal Received in revised form deposits and human occupation older than MIS5e (w120 ka). Based on the correlation of four different 20 December 2012 chronological methods we present evidence for remnant cave deposits of 1.1e1.0 Ma from the quartzite Accepted 22 December 2012 sea cliff of , near Mossel Bay. Initial uraniumethorium ages at isotopic equilibrium indi- Available online cated an age of >500 ka for two flowstone layers, confirmed by uraniumelead dating of these flowstones from 1.099 0.012 to 1.047 0.011 Ma. TT-OSL (thermally transferred optically stimulated lumines- Keywords: Geochronology cence) provides an age of 1.02 0.088 Ma for the sand grains imbedded in the tufa underlying the fl UePb dating owstone and 0.720 0.066 to 0.665 0.056 for the overlying beach sediments, producing an internally TT-OSL dating consistent age sequence centring on 1.0e1.1 Ma. The normal palaeomagnetic signal of the younger Palaeomagnetism section of the flowstone is interpreted to represent the Jaramillo between 1.07 and 0.99 Ma. There is Pinnacle Point a clear hiatus in the middle of this flowstone, leading us to interpret the lower normal signal as the South African cave deposits Punaruu event at w1.115e1.1051 Ma. Together these four techniques point to an age of 1.1e1.0 Ma for these cave deposits at Pinnacle Point, far older than anticipated. The persistent presence of these 1.1 e1.0 Ma deposits means that the enigmatic lack of Earlier Stone Age (Acheulean) artefacts in the sea caves along this coastal region can no longer be explained entirely by the age of the caves or through removal of sediments by previous sea level highstands. We believe that these and other coastal caves from this region, if located high enough above sea level, may contain deposits of great antiquity, which could provide outstanding records of climate, environment, sea level change, and human occupation back into the early to middle . Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction of South Africa (Fig. 1), contain a series of caves and rockshelters, formed as a result of previous high sea-levels stands (Marean et al., Sea caves and rockshelters along the west and south coasts of 2007). These caves and rockshelters contain rich anthropogenic South Africa provide some of the richest archaeological records of sediments, as well as ancient dune and beachrock deposits and cave early modern humans worldwide. The quartzitic coastal cliffs of carbonate layers, both pure speleothem and impure tufa deposits. Pinnacle Point (PP), near the town of Mossel Bay on the south coast The younger deposits at PP have been the subject of much recent investigation (Marean et al., 2007; Brown et al., 2009; Bar- Matthews et al., 2010; Matthews et al., 2011), during which * Corresponding author. Tel.: þ61 (0)3 8344 6531; fax: þ61 (0)3 8344 7761. E-mail addresses: [email protected], [email protected], a number of cave sites were shown to contain deposits older than [email protected] (R. Pickering). the 500 ka limit of UeTh dating. This was initially surprising as

0277-3791/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.quascirev.2012.12.016 40 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52

Fig. 1. A map of South Africa (A) showing the position of Mossel Bay and the cave sites of PP13G and PPOH at Pinnacle Point (B). The position of PP13G at between 16 and 22 masl is shown in (C) with the section dated here enlarged in (D) (scale bar is 10 cm).

Pinnacle Points, along with the other archaeological sites along this age of the oldest deposits in these caves, and to determine when coast, usually only contain sediments younger than w120 ka. The the cave and rockshelter sites may have been available for human apparent lack of deposits, and indeed cave sites, older than 120 ka occupation in the past. A parallel aim was to test the concordance of has been the subject of much debate. One explanation could be that the UePb and TT-OSL age estimates, and their consistency with the the cave systems themselves are relatively young and postdate palaeomagnetic time scale. 120 ka, meaning that the preservation potential pre-120 was too low for a record to exist. Alternatively, Hendey and Volman (1986) 2. Pinnacle Points cave sites and their geology deposits argued that the pre-120 ka cave sediments from this coastal region were all removed by previous sea-level high stands. A third and The coastal cliffs at PP (Fig. 1) consist of heavily dissected different explanation is that this coastal region was not recognized exposures of the Skurweberg formation of the Palaeozoic as an attractive occupation site and the lack of pre-120 ka sedi- Table Mountain Sandstone Group (TMS) (Viljoen and Malan, 1993). ments is due to anthropogenic reasons. The most likely explanation Fault breccias formed by movement along shear zones within the is some combination of all three, in which there where few sites TMS are more susceptible to erosion by high sea levels, forming available, possibly un-used by early humans, all of which have been caves in the near vertical cliffs. There are numerous caves in varying subject to significant erosion from previous sea level high stands. In states of preservation along this coast, here we focus on two, small, any event, the prevailing understanding of these sites along the highly eroded cave remnants, sites PP13G and PPOH (Opera House). south and western coast of South Africa is that the probability of finding old occupation sites is unlikely. 2.1. PP13G Recent advances in uraniumelead (UePb) dating and palae- omagnetic analysis of speleothems (Woodhead et al., 2006; PP13G is a small cave that preserves an exposed sequence of Pickering et al., 2010; Herries and Shaw, 2011), and thermally- fossil beach sediments, a porous sand-rich tufa, a flowstone layer transferred optically stimulated luminescence (TT-OSL) dating of and an aeolianite deposit (Figs. 1 and 2). The cave is developed quartz grains (Wang et al., 2006b; Jacobs et al., 2011; Duller and between 22.16 and 16.37 m above present sea-level and in its Wintle, 2012) are opening up new opportunities to study these present configuration consists of two discrete small chambers earlier time periods by providing a precise and consistent chro- separated by a ledge of cemented sediments (Fig. 2C). The lower nological framework. Moreover, the ability of UePb to date spe- beach sediment is a bioclast-rich, moderate well sorted granule- leothems older than 500 ka and an increase in the sensitivity of sized deposit with up to boulder-sized quartzite blocks floating in magnetometers is allowing the ability to identify and directly date it. Microscopically the sediment consists of well rounded to sub- geomagnetic field events in speleothems throughout the Pliocene angular quartzite clasts and rounded bioclasts (30e40%). Bioclasts and Pleistocene (Dirks et al., 2010; Herries et al., 2010; Osete et al., tend to be elongated and there are thin long fragile fragments with 2012). To this end we employed a multiple chronometer study of only rounded edges (Fig. 2F). The sediment is very porous, only intercalated sandstone and speleothem deposits at Cave 13G slightly cemented by two phases of freshwater phreatic equant (PP13G) and Opera House Cave (PPOH). The goal was to resolve the calcite rims (Longman, 1980). R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 41

Fig. 2. Photograph of PP13G Cave showing the two chambers separating by a ledge of cemented sediments (C), including the location of the lower TT-OSL sample (X). The lower beach sediments (F) consists of elongated bioclasts (B) and rounded to subangular lithoclasts, quartzite (Qz) and calcitic cement (Cc). This grades into a tufa layer (E) showing a clastic-rich lower part overlain by alternating clastic-rich and calcite-rich layers, enlarged in (G) to show the different types of clasts and aggregates cemented with calcite (L ¼ beach lithoclast, Qz ¼ quartzite, B ¼ bioclast, V ¼ vug). The erosional contacts below and above the flowstone layer are shown in (D) and (A), where the flowstone layers of fibrous calcite (with arrows) are truncated by the upper beach sediments (sample 162570, XPL). The upper part of the upper beach sediments (B) consist of rounded bioclasts (B) and subangular quartzite clasts (Qz). 42 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52

The cemented beach sediment gives way gradually to an impure slightly different directions. Very often layers are marked by dis- (clastic-rich) porous calcareous tufa formation (Fig. 2E, G). The tufa tinct dark brown laminae consisting of needle-like crystals replaced is a complex formation consisting of micritic layers and aggregates by columnar calcite (Fig. 3). These fabrics are attributed to aragonite enclosing angular and rounded quartzite clasts, rounded quartz replaced by calcite (Railsback, 2000; Frisia et al., 2002). According sand, elongated subrounded to rounded bioclasts, and some to Frisia et al. (2002) the replacement calcite inherits the d13C and U micromammal bones. It also contains subrounded clasts of beach content of the precursor aragonite. Other layer contacts are defined sandstone presumably derived from the erosion of the underlying by clusters of scalenohedral calcite or are flat-topped with euhedral beach sediment. At its upper part the tufa becomes more pure with crystal terminations preserved as ghosts into replacement calcite. micritic aggregates cemented with equigranular sparite and fre- The latter are often associated with diagenetic calcite features. quently intervened by very thin layers of microcrystalline and These are euhedral calcites containing multiple late stages of pre- columnar calcite. In areas, the tufa is characterized by an oncoid- cipitation forming in voids and following different direction of like fabric (samples 46635 and 46638) and indications of bio- growth from the dominant of flowstone (e.g. sample 46603C logical induced calcite formation (Luetscher et al., 2011). The tufa middle). A layer consisting of zones of microcrystalline calcite was complex is capped by a thick translucent and laminated flowstone observed in sample 46603A. The microcrystalline calcite is also (at ca 17.60 m height). Microscopically the contact with the un- replaced by columnar calcite. Finally, the translucent flowstone is derlying travertine is rough and clearly erosional (Fig. 2A, D). The capped by a porous laminated calcareous tufa (samples top of flowstone consists of layers of columnar to fibrous calcite crystals 46605A and top of 46604) consisting of alternating layers of fan with uniform extinction (Fig. 2A). Each layer contact is marked by shaped small columnar crystals and brown complex corrugated clusters of scalenohedral calcite crystals and sometimes very thin microcrystalline layers. They often enclose or wrap around faint remnants of dark brown laminae with original needle-like quartzite clasts of various sizes. aragonite crystals replaced by columnar calcite. The flowstone is clearly truncated and smoothed on its upper surface (Fig. 2A). The overlying upper beach sediment is also coarse-grained and has 3. Dating methods a similar content to the lower beach sediment although bioclasts tend to be more rounded (Fig. 2B). It also contains subangular The cave sites at PP all contain four basic sediment types: spe- quartzite cobbles and small boulders. Its very base shows a complex leothem, ancient beach deposits, aeolian dune sand and anthro- cementation history, starting with vadose grey micrite (pendants pogenic sediments. The speleothem layers and the quartz grains and meniscus) and followed by freshwater phreatic isopachous within the aeolian and beach deposits are both intrinsically ame- equant calcite coarsening locally towards pore centre. Some pores nable to dating through U-series and Optically Stimulated Lumi- are later filled by dark grey micrite silt (geopetal structures). Finally nescence (OSL) respectively. Speleothem material, in this case fl a later phase of equant sparite is rimming the pore of the sediment owstone or horizontal layers of calcium carbonate which grow out (Longman, 1980). laterally from an overhead drip source, is inherently suitable to U- series dating techniques given the huge solubility differences be- tween 238U and its daughter isotopes, 230Th and 206Pb. Extensive 2.2. PPOH UeTh dating at the PP sites has produced a detailed formation history for the caves back to the limits of UeTh dating (w500 ka) PPOH may have been a substantially larger cave, but today it is (Marean et al., 2007; Bar-Matthews et al., 2010; Marean et al., a shallow 1e2 m deep rock shelter preserving only the very back of 2010). Several speleothem deposits, including PP13G and PPOH, the cave and a flowstone drape over the TMS. The flowstone in yielded UeTh ages at isotopic equilibrium, thus older than w500 ka PPOH is a translucent, layered, generally dense flowstone with few (Table 1), hinting at the antiquity of these deposits and making elongated microscopic pores along crystal contacts. It consists of them ideal targets for UePb methods. columnar to fibrous calcite with uniform extinction (Fig. 3). Some OSL dating provides a means of determining burial ages for layers are more porous consisting of columnar crystals growing in sediments that were exposed to sunlight immediately prior to deposition (Huntley et al., 1985; Aitken, 1988; Lian and Roberts, 2006; Jacobs and Roberts, 2007). This method has been applied widely at PP and other similar cave settings along the southern Cape coast to great effect (Jacobs et al., 2003a, 2003b, 2008, 2013) but has an upper limit of w200 ka. Given the presumed antiquity of the PP13G and PPOH deposits indicated by the UeTh ages, we used an alternative OSL-based method for quartz grains, namely thermally-transferred OSL (TT-OSL), which involves stimulating an electron trap that is less light sensitive (Wang et al., 2006b; Adamiec et al., 2008; Tsukamoto et al., 2008; Duller and Wintle,

Table 1 UeTh results for flowstones from PP13G and PP Opera House.

Site Sample U concentration 234U/238U 230Th/234U Age PP13G 46636b 0.38 1.0271 1.1799 Equilibrium 44640 0.53 1.8456 1.0859 Equilibrium PPOH 44603 0.57 1.0322 1.0300 Equilibrium 46604a 0.58 1.0845 1.0442 Equilibrium 46604b 1.12 1.0584 1.0957 Equilibrium 46604c 1.38 1.303 1.1324 Equilibrium 44605a 1.28 1.0282 1.0654 Equilibrium Fig. 3. Photomicrograph of Opera House Cave flowstone showing columnar calcite 44605b 1.37 1.0316 1.0692 Equilibrium (CC) replacing needle-like aragonite crystals (AR) in sample 46604, XPL. R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 43

2012). As with conventional OSL, this method is based on the has recently been published elsewhere (Woodhead et al., 2006; increase in number of trapped electrons in mineral grains (such as Bar-Matthews et al., 2010; Herries and Shaw, 2011; Jacobs et al., quartz) with increasing time after burial, in response to the energy 2011), so here we provide just a brief summary and sample spe- supplied by background levels of ionising radiation from environ- cific details. mental sources. To this end we sampled the PP13G flowstone for UePb dating 3.1. UePb dating of flowstone and the sand-sized quartz grains found within the underlying porous tufa and overlying aeolianite for TT-OSL dating. These are As 230Th itself is part of the U-series decay chain, material older both relatively new, recently established techniques, so we sampled than w500 ka is beyond the limit of the chronometer. UePb dating the entire PP13G sequence for palaeomagnetic analysis in order to can date material as young as a few hundred thousand years have another method to corroborate the UePb and TT-OSL results. (Richards et al., 1998), but the challenge with relatively young At PPOH the flowstone layer was sampled for UePb and palae- (w1 Ma) samples is detecting datable layers, meaning some kind of omagnetic analysis; no TT-OSL was attempted given the lack of pre-screening is necessary. We used laser-ablation ICP-MS to pro- sediments below or above the flowstone. Exactly where and how duce high-resolution trace element scans and layers with elevated the magnetic remanence is held in speleothems is an on-going 238U and low Pb concentrations are visually identified. question. It can form as a detrital, depositional remanence from Small (w50 mg) blocks of flowstone were are physically cut out the deposition of sediment in sourcewater or flooding of the surface of the chosen layers (Fig. 4) and etched in a weak HCl solution to of the speleothem, or as a chemical remanence on precipitation remove the outer microns and any Pb contamination. From here on (Herries and Shaw, 2011; Lascu and Feinberg, 2011). Synchrotron all sample handling took place in a class 350 clean laboratory. U and XFM data was thus used to investigate the magnetic remanence Pb isotopes are extracted and concentrated using standard ion acquisition and integrity of the palaeomagnetic signal in the PP exchange methods, following the procedure outlined in Woodhead speleothems. The detailed methodology of each of these techniques et al. (2006). U and Pb isotope ratios were then measured

Fig. 4. Laser ablation tracks (white line) across the samples from PP13G and PPOH, U concentrations shown in dark grey, Pb in light grey stipples and areas selected for dating by grey arrows eboth where U concentrations are over 1 ppm. Concentrations are plotted on a logarithmic scale and against the time of the scan, scaled to fit the sample. 44 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 separately on a Nu Instruments MC-ICP-MS (Nu080) again using taking the measured overdispersion into account. For purposes of the protocols outlined in Woodhead et al. (2006). ReOSL and BT-OSL age determination, the residual doses measured in the modern analogue sample collected from the exposed modern 3.2. TT-OSL dating of tufa and beach rock beach at the base of the cave and elsewhere along the coast (weighted means of 15 and 41 Gy, respectively) were subtracted The conventional OSL dating technique was not suitable for the from the corresponding CAM De values (Jacobs et al., 2011); the PP13G samples as the electron traps responsible for the ‘fast’ residual-corrected De values are listed in Table 2. An important component used in conventional OSL dating had been filled com- outcome of the comparison of the ReOSL and BT-OSL De values is pletely (‘saturated’). In such circumstances, the quartz grains can- the extent of agreement. These signals differ considerably in their not store any further radiation energy as trapped electrons, so only rate of depletion by sunlight, so the concordant De values indicate a minimum estimate of absorbed dose can be obtained. TT-OSL can that their electron traps were emptied to the same, low level when be used to determine the depositional age of sediments much older the sediments were last exposed to daylight. This is a necessary than the Last Interglacial in the study region, because the TT-OSL condition for application of the TT-OSL procedure, owing to the saturation limit for quartz greatly exceeds that for the ‘fast’ OSL duration of bleaching required to empty the relevant traps: 26 signal (see Jacobs et al., 2011). weeks for ReOSL and 50 weeks for BT-OSL (Jacobs et al., 2011). Two samples from PP13G were analysed for luminescence dat- ing in this study. Sample 46636 was collected from the tufa layer 3.3. Palaeomagnetic analysis, XFM Synchrotron data and magnetic between the lower beach deposit and flowstone (Fig. 2C). Study of remanence acquisition the microphotographs for this sample (Fig. 2E) suggest that we have dated both aeolian sand grains blown into the deposit during Samples for palaeomagnetic analysis were taken from the the formation of the tufa and re-worked quartz grains derived from flowstones in PP13G and PPOH, and from the underlying travertine the underlying beach deposit. The other sample (46639) was col- deposit at PP13G. All samples were oriented in-situ using a Suunto lected from the overlying aeolian dune deposit (Fig. 2C). The outer, KB14 compass and clinometer and the final declination was cor- light-exposed surfaces of the samples were removed under sub- rected for local field variation based on data from the nearby Her- dued red illumination in the laboratory, and the remaining material manus Magnetic Observatory and the International Geomagnetic was prepared for dating using standard laboratory procedures Reference Field (IGRF-11). The bulk mineralogy of speleothem (Aitken, 1988). samples was initially determined from the demagnetisation spectra The TT-OSL dating method, as proposed by Wang et al. (2006b), during alternating field and thermal demagnetisation. A single involves separating two signals: a recuperated OSL (ReOSL) signal specimen of the PPOH speleothem 162566 was also preliminary and a basic-transferred OSL (BT-OSL) signal, which together com- analysed on the X-ray Fluorescence Microprobe (XFM) at the Aus- prise the TT-OSL signal. Wang et al. (2006a, 2006b) proposed tralian Synchrotron with the aim of identifying the spatial distri- a regenerative-dose TT-OSL procedure using multiple aliquots, bution of iron within the speleothems. The data were collected in which was subsequently modified so that De estimates could be fast mapping mode using the Maia detector available at the XFM obtained from single aliquots (e.g. Wang et al., 2007; Tsukamoto beamline. The scan area was 23 10 mm2 with pixel size of et al., 2008). We have adapted the single-aliquot TT-OSL approach 5 5 mm2. The total sampling time for the resulting 9.2 mega-pixel specifically for the samples collected from our study area (i.e., the image dataset was approximately 4 h. southern Cape coast near Mossel Bay in South Africa; Fig. 1), using aliquots containing w1000 grains. Full experimental details and the 4. Results most appropriate measurement procedure for the samples descri- bed here are provided and discussed in Jacobs et al. (2011); further 4.1. UePb flowstone ages sample specific details of the dose rate determination are supplied in the supplementary on-line material (Text S1). Laser ablation trace element profiles of both the PP13G and We used the central age model (CAM) of Galbraith et al. (1999) PPOH flowstones revealed several layers with U concentrations to determine the relative spread (‘overdispersion’)inDe values over 1 ppm, meaning that UePb dating could be undertaken on remaining after making allowance for measurement uncertainties more than one suitable layer (Fig. 4). All U and Pb isotope data are (Galbraith et al., 2005). The ReOSL and BT-OSL analyses of both shown in the supplementary information Table S1. Ages are cal- samples (46639 and 46636) resulted in overdispersion estimates of culated using TeraeWasserberg Concordia plots (Fig. 6), taking into 16 5% and 8 3%, and 15 5% and 10 6%, respectively (Table 2). account initial (234U/238U) disequilibrium measured on large These overdispersion values fall within the range of values (w150 mg) matched samples. Ages are summarized in Table 3. reported for samples collected from the nearby Dana Bay, Klein Brak The PP13G flowstone has a UePb age of 1.099 0.012 Ma at the River and Hartenbos River that were measured in the same way base and 1.047 0.011 Ma in the middle (Table 3; Fig. 9). The ages (Jacobs et al., 2011). The radial plots of the De distributions for the for PP13G have remarkably small errors of as little as 1.1%. This is two different signals (ReOSL and BT-OSL) are shown in Fig. 5AeD. due to careful sample selection, U concentrations of over 1 ppm The final De values for each sample (i.e., those used to determine (typical for PP and likely the result of the calcite parent material the ReOSL and BT-OSL ages) were obtained using the CAM, which being relatively recent U-rich shellfish), a wide enough range of provides an estimate of the weighted mean and its standard error, 238U/206Pb ratios and precise measurement of the initial (234U/238U)

Table 2 Dose rate data, equivalent doses, and ReOSL and BT-OSL ages for sediment samples from PP13G.

Site Sample Dose rate (Gy/ka) Total dose Aliquots ReOSL OD (%) ReOSL BT-OSL OD (%) BT-OSL rate (Gy/ka) (N) D (Gy) age (ka) D (Gy) age (ka) Beta Gamma Cosmic e e PP13G 46639 0.46 0.03 0.33 0.01 0.01 0.01 0.84 0.05 10 601 36 16 5 720 ± 63 556 33 15 5 665 ± 68 46636 0.31 002 0.24 0.02 0.01 0.01 0.60 0.05 16 605 17 8 3 1015 ± 88 609 30 10 6 1021 ± 98 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 45

Fig. 5. De distributions for (A) 10 single aliquots of sample 46639 measured using TT-OSL dating procedures to obtain ReOSL De values, and (B) the same sample measured using TT- OSL dating procedures to obtain BT-OSL De values from the same 10 aliquots. (C) and (D) shows the same data, but for sample 46636. disequilibrium. The base of the PPOH flowstone is dated to speleothem to high temperatures often causes it to crack, expand 1.039 0.035 Ma, placing this flowstone’s age within error of the and eventually fracture (Herries and Shaw, 2011). However, a num- PP13G deposit. The middle of PPOH was dated to 1.076 0.115 Ma, ber of the samples failed to demagnetise even at 140 mT, indicating which, with the 10% error range, is still consistent with all other that a high coercivity mineral was carrying the remanence (Fig. 7; ages from PP13G and PPOH. This much larger error can be 162566). Thermal demagnetisation (TH) was thus undertaken on explained by the uncertainty on the initial 234U/238U measurement, sample 162566, which fractured into pieces around 400 C(Fig. 7). which for this sample is much larger than the others. Despite this, the sample has two distinct components, one with a low unblocking temperature (<120 C) and identifying goethite as 4.2. Luminescence ages the main remanence carrier of this sample. A higher temperature component is likely held by a very small proportion of magnetite. The luminescence ages for the tufa sand inclusions immediately Both components record a reversed polarity that is consistent with underlying the PP13G flowstone are 1.02 0.088 and 1.02 0.098 directions recorded during AF demagnetisation. for the ReOSL and BT-OSL signal respectively (Table 2; Fig. 9), The majority of other samples lost their remanence between consistent at the 1s level with the UePb ages (UePb age errors are 500 and 600 C and are carried by magnetite. The same mineralogy given as 2s). As these two OSL signals differ markedly in their was not always recorded in every layer of the same flowstone with sensitivities to light exposure and their thermal stability over some layers having a remanence held by goethite and others by geological timescales, the correspondence in ages argues against magnetite. However, the polarity was always consistent. The significant overestimation of age due to partial bleaching or un- remanence carried by magnetite is considered to be detrital in derestimation of age due to thermal instability of the ReOSL signal origin locking in the remanence soon after (i.e., within a few years) over the burial period (Jacobs et al., 2011). The uppermost aeo- calcite precipitation. The samples carried by goethite are of lianite at PP13G is dated to 0.720 0.063 (ReOSL) and chemical origin, but also forms and locks in the remanence signal at 0.665 0.056 Ma (BT-OSL) (Fig. 9). the time of calcite precipitation (Latham and Ford, 1993; see XRM data below). 4.3. Palaeomagnetic analysis and magnetic remanence acquisition The XFM Synchrotron data and the resulting elemental maps (Fig. 8) show some ‘blurring’ due to the macroscopic thickness of The samples were primarily demagnetised using alternating the speleothem which was used as received for this preliminary fields (AF) to isolate the primary remanence as heating of analysis. Consequently, the information depth for Fe was in the 46 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52

Fig. 6. TeraeWasserberg Concordia age plots for PP13G and PPOH, showing final ages corrected for initial (234U/238U) disequilibrium. order of 50 mm for Fe and 200 mm for As (estimated as twice the a very low unblocking temperature are held by goethite crystals attenuation lengths of, respectively, Fe-Ka and As-Ka fluorescence that are dispersed throughout the speleothem and were formed X-rays in calcite). The speleothem used in this analysis had one of during its precipitation; causing a chemical remanence (CRM) to the weakest remanences (lowest iron content) for the speleothem occur. While this is consistent with the magnetic mineralogy of at Pinnacle Point. This sample has evidence of a remanence held other speleothems that have been studied (Lascu and Feinberg, by both goethite and perhaps also magnetite or maghaemite, both 2011), it is rare to find speleothems whose remanence is carried of which appear to hold a reversed remanence. Iron is noted as by such a CRM. This is even more notable in South Africa due to occurring primarily along the banding structure of the speleo- the strong ferrimagnetic nature of most South African sediments, them (see also bottom panel in Fig. 8) which suggests it is pri- whose mineralogy is dominated by fine to ultra-fine, single marily of detrital origin and explains the ferrimagnetic remanence domain to superparamagnetic, ferromagnetic grain sizes due to with higher unblocking temperatures identified in many samples. the long term burning and weathering of the African landscape This iron appears to correlate with arsenic (and other metals), (Herries et al., 2006; Herries and Fisher, 2010). The occurrence of which has been suggested to relate to periods of aridity and goethite carried as a CRM indicates a complete lack of detrital erosion, explaining the increased influx of detrital iron into the material in the speleothem, making them potentially ideal for Ue caves (Zhou et al., 2008). The remanence of samples which have Pb analysis.

Table 3 UePb ages for PP 13G and PPOH.

Site Sample UePb (TeW) % Error Initial Present name Age 2SE 234U/238U 234U/238U PP13G e base 162569.1 1.099 0.012 1.1 1.628 0.030 1.028 0.002 PP13G e top 162569.2 1.047 0.011 1.1 1.747 0.020 1.038 0.001 PPOH e base 162565.1 1.039 0.035 3.4 1.573 0.056 1.030 0.001 PPOH e top 162565.2 1.076 0.115 10.7 1.446 0.125 1.021 0.001 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 47

Fig. 7. Demagnetisation spectra for palaeomagnetic samples of speleothem from PPOH and PP13G using alternating field (AF) and thermal demagnetisation (TH) and showing reversed (162566), intermediate (162568) and normal polarities (162567). 48 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52

Fig. 8. Experimental synchrotron X-ray fluorescence elemental maps showing (top panel) the spatial distribution of iron (green), arsenic (red), and strontium (blue) within a section of the reversed polarity PPOH speleothem, and (bottom panel) the banding structure of the speleothem as reflected in the elemental map of strontium alone with brighter colours representing higher Sr concentrations. The green area on the bottom of the top image shows the accumulation of more recent detrital iron in cracks within the flowstone and indicate the need to select samples without such microstructures or with a more porous nature. Iron within the speleothem correlates with arsenic and other metals shown up as pink layers and in some cases is also dispersed within the crystal matrix of the speleothem. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Both normal and reversed polarity directions were recorded on the UePb ages (1.058e1.004 Ma), the normal polarity recorded (Table 4). Intermediate samples were defined by paleopole di- in the upper section of the PP13G flowstone and at the base of the rections with paleolatitudes of <45 (P. Lat. Table 4) and as such PPOH flowstone represents the Jaramillo Normal SubChron dated tufa sample 162568 has a mean intermediate polarity for the block, to between 1.07 and 0.99 Ma (Ogg and Smith, 2004). The reversed although the sample is still trending in a normal polarity direction signal at the top of the PPOH flowstone corresponds to the end of (þ56 declination/53 inclination). The PP13G tufa records an in- the Matuyama Chron between 0.99 and 0.78 Ma, after the end of termediate normal (PP162568) to normal polarity with the over- the Jaramillo and consistent with the UePb age of 1.076 0.115 Ma. lying flowstone recording normal magnetic polarity throughout its The PP13G flowstone is 1.099 0.012 Ma at the base, so that depth (PP162567). A single subsample from the top of PP162567 even within error (1.111e1.087 Ma), this material is too old to be recorded a reversed magnetic polarity that is considered to repre- assigned to the Jaramillo SubChron (1.07e0.99 Ma). In PP13G the sent the polarity of the clastic crust preserved on top of this part of UePb dated layers are just over 1 cm apart and it is unlikely that the flowstone. The PPOH speleothem recorded normal polarity at 40 ka is recorded in such a condensed section. A clear hiatus layer is its base (PP162564), followed by reversed polarity at its top present, suggesting a prolonged phase of non-deposition. With this (PP162566) (Table 4). in mind, we interpret the normal signal for the base of the PP13G flowstone as belonging to the Punaruu event at w1.115e1.105 Ma 4.4. Combined dating results (Singer et al., 1999; Lai and Channell, 2007), but here it is directly dated to 1.099 0.012 Ma (1.111e1.087 Ma) (Fig. 9). The aeolinite A summary sketch of the sedimentary sequences at PPOH and dune capping the PP13G sequence has then been dated to between PP13G with the corresponding UePb and TT-OSL ages and paleo- 0.720 0.066 and 0.665 0.056 Ma and indicates deposition of magnetic results, compared to the known geomagnetic polarity material within the PP sea caves between w1.1 and w0.6 Ma. time scale is presented in Fig. 9 and in tabular form in Table 5. Based 5. Discussion

Table 4 Today PP13G and PPOH consist of the eroded remnants of what Palaeomagnetic data for PP13G and PPOH speleothem. must both have been much larger cave systems. The cycles of these Site Sample no. Dec. Inc. K No. P. Lat. Polarity caves forming, filling and eroding are numerous and complex and, PP13G 162567U 19.8 55.8 191.3 3 71.7 N as our new data suggest, have been a recurring feature of this re- PP13G 162567L 9.1 48.8 52.1 3 81 N gion for at least the last 1.1e1.0 million years. The preservation of PP13G 162568 55.5 67.1 65.8 4 44.5 I the detailed section at PP13G (Fig. 2) allows an insight into the type PPOH 162566 202.7 38.5 118.2 9 66.6 R and scale of the cycles of cave development, filling and erosion. The PPOH 162564 359 53.2 188.9 4 89.1 N PP13G cave itself was probably formed by more than one sea-level R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 49

Fig. 9. A summary sketch of the sedimentary sequences at PPOH and PP13G with the corresponding UePb and TT-OSL ages and paleomagnetic results, compared to the known geomagnetic polarity time scale. The UePb and TT-OSL ages pin the normal palaeomagnetic results to the Punaruu (w1.11e1.12 Ma) and Jaramillo events (1.07e0.99 Ma) and the reversed results to the end of the Matuyama (0.99e0.78 Ma). highstand at heights between 20 and 16 m over 1.1 million years Our initial UeTh equilibrium data hinted at the antiquity of the ago. This cave was then filled with beach sediment (lower beach) cave deposits at PP13G and PPOH, but the UePb, TT-OSL and during one of these highstands. Subsequently the beach sediment palaeomagnetism all now provide convincing evidence for a range was exposed, reworked and eroded forming a deposit that was of deposits dating to between w1.1 and 1.0 Ma and providing cemented with calcite to form a clastic-rich calcareous tufa. The a minimum age for the formation of the caves themselves (Fig. 9). tufa itself was also eroded before the overlying flowstone formed The TT-OSL and UePb ages are both in agreement and consistent between 1.11 and 0.99 Ma. The flowstone itself was also eroded and with each aspect of the sedimentary sequence, giving us confidence smoothed by the action of water and or abrasion processes. The in both emerging methods. In order to achieve high quality results younger capping beach sediments are similar to the lower beach certain a priori conditions must be met for both methods. UePb and were cemented mainly in a meteoric environment and are dating of speleothems will be most successful where U levels are capped by aeolinites dated to between 0.72 0.07 and over 1 ppm, making some kind of pre-screening for U concentra- 0.665 0.056 Ma. This whole sedimentary sequence was later tions invaluable. The petrography of the flowstone from PP13G and eroded by a new highstand close to 15e16 m, and the lower PPOH revealed relict aragonite grains, which are known to hold chamber was formed. This highstand may be related to MIS11, more U than the surrounding calcite grains. A similar co-existence when a 13 2 m asl level has been recorded in deposits 390 ka of relict aragonite and calcite has also been observed in flowstones close to PP (Roberts et al., 2012). The coarse and immature nature of of over 2 Ma (Pickering et al., 2010). TT-OSL will be most effective both beach sediments and their cementation mainly in the mete- where quartz grains were strongly bleached and the probability of oric environment rather suggests storm deposits in the back beach mixing with more ancient and/or unbleached material is low. Both environment and therefore it is impossible to know the precise conditions are likely to be met in coastal contexts, such as at PP, height of these highstands. where the sands are dune and/or beach and the speleothem parent

Table 5 UePb, TT-OSL and paleomagnetic ages for the sequence of beach sediments, tufa, flowstone and aeolianite from sites PP13G and PPOH.

Site Material Sample number Method Polarity Age (Ma) PP13G Aeolianite 46639 ReOSL 0.720 0.063 BT-OSL 0.665 0.056 Flowstone (top) 162569.2 UePb 1.047 0.011 Flowstone (top) 162567U Palaeomagnetism N Jaramillo: 1.07e0.99 Flowstone (middle) 162567L Palaeomagnetism N Punaruu: 1.115e1.1105 Flowstone (base) 162569.1 UePb 1.099 0.012 Tufa 162568 Palaeomagnetism I ? Tufa 46636 ReOSL 1.020 0.088 Tufa BT-OSL 1.020 0.098 PPOH Flowstone top 162566 Palaeomagnetism R End Jaramillo: <0.99 Flowstone (middle) 162565.2 UePb 1.076 0.115 Flowstone (middle) 162564 Paleomagnetism N Jaramillo: 1.07e0.99 Flowstone (base) 162565.1 UePb 1.039 0.035 50 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 materials are the U-rich carbonates of relatively young biogenic well as the erosion and retreat of the cliff face, have eroded much of calcite-derived calcretes. These results establish UePb and TT-OSL the PP13G and PPOH caves and deposits. as powerful new tools for age estimation in the Quaternary at Since many of the PP caves (including the two reported here) are time ranges beyond the reaches of conventional UeTh and OSL as high as 22 masl, it is extremely unlikely that the sediments were dating. Correlation with the more established method of palae- washed out by MIS5e at þ5e6 masl (Hearty et al., 2007). MIS11 sea omagnetism can also help to refine the age estimates produced by level height has been estimated to be as high as w21 m (Olson and these methods and ideally the cross correlation of multiple dating Hearty, 2009), which would be consistent with a washing out of the methods is recommended for dating such sedimentary sequences. South African sea caves at w400 ka. However, this MIS11 sea level The UePb dating of geomagnetic field events in speleothem is has recently been corrected downwards to þw6e13 masl (Raymo in part complicated by the need for such speleothem to be clean of and Mitrovica, 2012), which is consistent with height of beach detrital grains for UePb dating, but ideally containing such grains deposits near Mossel Bay dated to MIS11 (Jacobs et al., 2011). This for palaeomagnetic analysis. However, XRM Synchrotron data lower height estimate suggests that earlier/older deposits should (Fig. 8) combined with the palaeomagnetic demagnetisation survive, as is indicated by the dates presented here, together with spectra of the Pinnacle Point speleothems (Fig. 7) indicate that non-occupation deposits as old as 400 ka from PP13B at þ20e both detrital and chemical remanences occur that are locked-in to 19 masl (Marean et al., 2010). While the deposits at PP13G and the speleothem at, or extremely close to the time of deposition or PPOH are heavily eroded, our suit of ages show conclusively that precipitation. Detrital ferrimagnetic grains from flood events and caves acting as sediment traps along the South African coastline are sourcewater are cemented by calcite shortly after deposition. Prior old enough to preserve human occupations back to w400 ka and studies suggest this lock-in of remanence in speleothems is even w1.0 Ma. The lack of such occupation deposits may be the extremely short and on the order of a few years (Latham and Ford, product of current research sampling, geological bias related to the 1993; Lascu and Feinberg, 2011). In contrast magnetic crystals erosion of most deposits of this age by collapse related to the for- formed by chemical precipitation will be lock-in as the calcite mation of caves at lower sea levels and the retreat of the coastline forms and will also occur in calcite free of detrital grains more ideal itself, or a product of Acheulean hominins simply not utilising the for UePb analysis. As such, speleothems have the potential to caves. The answer to this puzzle requires considerable exploration provide an accurate and datable record of geomagnetic field and dating of cave deposits at þ20e19 masl along this coastal re- variation. gion; this study presents the geochronological tools required by The w1.1e1.0 Ma age for the cave deposits at PP indicates that such a venture. these caves formed prior to 1.1 million years ago. Out of the long list of well-known excavated caves and rock-shelters along the South 6. Conclusion African coast [Klasies River 1A and 2 (Singer and Wymer, 1982), (Butzer, 1973), PP13B (Marean et al., 2007), The w1.1e1.0 Ma deposits at PP13G and PPOH are now the (Henshilwood et al., 2001), Elands Bay Cave oldest dated material from a Cape coastal cave and possibly the (Parkington, 1976)], none of these sites have yielded occupations world’s oldest numerically dated sea caves. The combined use of older than 170 ka and none have Acheulean (Earlier Stone Age) UeTh, UePb, TT-OSL and palaeomagnetism allowed for a test of the archaeological material. This is in stark contrast to open-air sites, relatively new UePb and TT-OSL techniques and the precise, con- found in close proximity to these caves, where Acheulean material sistent ages of w1.1e1.0 Ma provide confidence in both methods. is abundant (Thompson, 2009). The only cave sites in South Africa UePb and TT-OSL may help fill the dating gap in the middle reaches with Acheulean material are the Cave of Hearths (Mason, 1988), of the Quaternary beyond limits of conventional UeTh and OSL Montague Cave (Keller, 1973), (Chazan et al., dating. The w1.1e1.0 Ma age for the cave deposits represent the 2008), , and Gladysvale (Herries, 2011) minimum ages for the origin of the caves and push back the life and these are all situated over 100 km from the coast. The lack of history of the cave sites at PP back by over 700 ka and w900 ka sediments with human occupation in South African sea caves earlier than the oldest human occupation. This suggests that other dating earlier than 170 ka displays at minimum 840,000 year gap caves and rockshelters along this coast at similar elevations may be between cave formation and the first human occupation. How do equally ancient, meaning that sites were available for occupation we explain this occupation gap? during the Acheulean. The puzzling lack of records of Acheulean It seems clear that a large number of cave deposits that may occupation from these coastal caves cannot be completely resolved have contained earlier archaeological material have been washed but cannot be explained solely by geological processes. out by high sea level stands during MIS 5e (w120 ka) and MIS 11 (w400 ka). However, our results eliminate the possibility that the Acknowledgements caves were first cut by the MIS11 high sea stand and that caves in the 22e16 m amsl (such as PP13G, PPOH and a number of other Acknowledgements to the MAP staff for their assistance and caves at Pinnacle Point including PP9; Matthews et al., 2011) lack competence in the field, the Dias Museum for use of their facilities, earlier sediments entirely because of that. It also indicates that SAHRA and HWC for permits. Funding received from the Swiss remnants of caves as old as 1.1e1.0 Ma age can survive in sea cliffs National Research Foundation (PBBEP2-126195 to RP), Australian despite the active state of marine erosion seen currently. Research Council (FT120100399 to AIRH), the US National Science These caves were all likely formed by several sea level high Foundation (grants # BCS-9912465, BCS-0130713, and BCS- stands at about this height, as attested by the complex config- 0524087 to CWM), the Hyde Family Foundation, the Institute of uration of the caves themselves and the deposits filling them. Human Origins at Arizona State University, and Arizona State Uni- Specifically, at PP13G the cave itself consists of two chambers at versity. The palaeomagnetic analysis was undertaken by AIRH at different heights, both filled with more than one beach formation, the University of Liverpool Geomagnetism Laboratory with the which themselves show evidence of several sea wave erosion support of Mimi Hill. The synchrotron analysis was undertaken on phases. Based on our dating of the entire sequence of tufa, flow- the XFM beamline at the Australian Synchrotron (beamtime gran- stone and beach sediment at PP13G and the flowstone at PPOH, we ted to AIRH, PK and RP). Thanks to Mark Foster and Ben Elliot for suggest that the caves containing these were formed by high sea assistance with rock climbing and rope access work needed for the stands older than 1.1e1.0 Ma. Subsequent sea level high stands, as exploration and sampling of PPOH and PP13G. We would like to R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52 51 thank two anonymous reviewers for their thorough and con- Jacobs, Z., Hayes, E.H., Roberts, R.G., Galbraith, R.F., Henshilwood, C.S., 2013. An structive comments. improved OSL chronology for the Still Bay layers at Blombos Cave, South Africa: further tests of single-grain dating procedures and a re-evaluation of the timing of the Still Bay industry across southern Africa. Journal of Archaeological Appendix A. Supplementary information Science. Jacobs, Z., Roberts, R.G., 2007. Advances in optically stimulated luminescence dating of individual grains of quartz from archeological deposits. Evolutionary An- Supplementary data related to this article can be found at http:// thropology 16, 210e223. dx.doi.org/10.1016/j.quascirev.2012.12.016. Jacobs, Z., Roberts, R.G., Galbraith, R.F., Deacon, H.J., Grün, R., Mackay, A., Mitchell, P., Vogelsang, R., Wadley, L., 2008. Ages for the of southern Africa: implications for human behavior and dispersal. Science 322, References 733e735. Jacobs, Z., Roberts, R.G., Lachlan, T.J., Karkanas, P., Marean, C.W., Roberts, D.L., 2011. Adamiec, G., Bailey, R.M., Wang, X.L., Wintle, A.G., 2008. The mechanism of ther- Development of the SAR TT-OSL procedure for dating Middle Pleistocene dune mally transferred optically stimulated luminescence in quartz. Journal of and shallow marine deposits along the southern Cape coast of South Africa. Physics D: Applied Physics 41, 135503. Quaternary Geochronology 6, 491e513. Aitken, M.J., 1988. An Introduction to Optical Dating. Oxford University Press. Keller, C.M., 1973. Montague Cave in Prehistory. University of California Anthro- Bar-Matthews, M., Marean, C.W., Jacobs, Z., Karkanas, P., Fisher, E.C., Herries, A.I.R., pological Records 28. Brown, K., Williams, H.M., Bernatchez, J., Ayalon, A., Nilssen, P.J., 2010. A high Lai, C., Channell, J.E.T., 2007. Geomagnetic excursions. In: Kono, M. (Ed.), Treatise on resolution and continuous isotopic speleothem record of paleoclimate and Geophysics: Geomagnetism. Elsevier, Amsterdam, pp. 373e416. paleoenvironment from 90 to 53 ka from Pinnacle Point on the south coast of Lascu, I., Feinberg, J.M., 2011. Speleothem magnetism. Quaternary Science Reviews South Africa. Quaternary Science Reviews 29, 2131e2145. 30, 3306e3320. Brown, K.S., Marean, C.W., Herries, A.I.R., Jacobs, Z., Tribolo, C., Braun, D., Latham, A.G., Ford, D., 1993. The Palaeomagnetism and Rock Magnetism of Cave and Roberts, D.L., Meyer, M.C., Bernatchez, J., 2009. Fire as an engineering tool of Karst Deposits Applications of Paleomagnetism to Sedimentary Geology. SEPM early modern humans. Science 325, 859e862. Special Publication No. 49, pp. 149e155. Butzer, K.W., 1973. Geology of Nelson Bay Cave, Robberg, South Africa. South African Lian, O.B., Roberts, R.G., 2006. Dating the Quaternary: progress in luminescence Archaeological Bulletin 28, 97e110. dating of sediments. Quaternary Science Reviews 25, 2449e2468. Chazan, M., Ron, H., Matmon, A., Porat, N., Goldberg, P., Yates, R., Avery, M., Longman, M.W., 1980. Carbonate diagenetic textures from nearshore diagenetic Sumner, A., Horwitz, L.K., 2008. Radiometric dating of the Earlier Stone Age environments. The American Association of Petroleum Geologists Bulletin 64, sequence in Excavation I at Wonderwerk Cave, South Africa: preliminary re- 461e487. sults. Journal of Human Evolution 55, 1e11. Luetscher, M., Hoffmann, D.L., Frisia, S., Spötl, C., 2011. glacier history from Dirks, P., Kibii, J.M., Kuhn, B.F., Steininger, C., Churchill, S.E., Kramers, J.D., alpine speleothems, Milchbach cave, . Earth and Planetary Science Pickering, R., Farber, D.L., Meriaux, A.S., Herries, A.I.R., King, G.C.P., Berger, L.R., Letters 302, 95e106. 2010. Geological setting and age of Australopithecus sediba from Southern Marean, C.W., Bar-Matthews, M., Bernatchez, J., Fisher, E., Goldberg, P., Africa. Science 328, 205e208. Herries, A.I.R., Jacobs, Z., Jerardino, A., Karkanas, P., Minichillo, T., Nilssen, P.J., Duller, G.A.T., Wintle, A.G., 2012. A review of the thermally transferred optically Thompson, E., Watts, I., Williams, H.M., 2007. Early human use of marine re- stimulated luminescence signal from quartz for dating sediments. Quaternary sources and pigment in South Africa during the Middle Pleistocene. Nature 449, Geochronology 7, 6e20. 905e911. Frisia, S., Borsato, A., Fairchild, I.J., McDermott, F., Selmo, E.M., 2002. Aragonitee Marean, C.W., Bar-Matthews, M., Fisher, E., Goldberg, P., Herries, A., Karkanas, P., calcite relationships in speleothems (Grotte de Clamouse, ): environ- Nilssen, P.J., Thompson, E., 2010. The stratigraphy of the Middle Stone Age ment, fabrics, and carbonate geochemistry. Journal of Sedimentary Research 72, sediments at Pinnacle Point Cave 13B (Mossel Bay, western Cape Province, 687e699. South Africa). Journal of Human Evolution 59, 234e255. Galbraith, R.F., Roberts, R.G., Laslett, G.M., Yoshida, H., Olley, J.M., 1999. Optical Mason, R.J., 1988. Cave of Hearths . Archaeological Research Unit, dating of single and multiple grains of quartz from jinmium rock shelter, University of the Witwatersrand, Johannesburg. northern Australia: part I, experimental design and statistical models. Matthews, T., Rector, A., Jacobs, Z., Herries, A.I.R., Marean, C., 2011. Environmental Archaeometry 41, 339e364. implications of micromammals accumulated close to the MIS 6 to MIS 5 tran- Galbraith, R.F., Roberts, R.G., Yoshida, H., 2005. Error variation in OSL palaeodose sition at Pinnacle Point Cave 9, southern Cape coast, South Africa. Palae- estimates from single aliquots of quartz: a factorial experiment. Radiation ogeography, Palaeoclimatology, Palaeoecology 302, 213e229. Measurements 39, 289e307. Ogg, J.G., Smith, A.G., 2004. In: Cradstein, F., Ogg, J.G., Smith, A.G. (Eds.), A Geologic Hearty, P.J., Hollin, J.T., Neumann, A.C., O’Leary, M.J., McCulloch, M., 2007. Global sea- Timescale. Cambridge University Press, Cambridge, pp. 63e86. level fluctuations during the Last Interglaciation (MIS 5e). Quaternary Science Olson, S.L., Hearty, P.J., 2009. A sustained þ21 m sea-level highstand during MIS 11 Reviews 26, 2090e2112. (400 ka): direct fossil and sedimentary evidence from Bermuda. Quaternary Hendey, Q.B., Volman, T.P., 1986. Last interglacial sea levels and coastal caves in the Science Reviews 28, 271e285. Cape Province, South Africa. Quaternary Research 25, 189e198. Osete, M.-L., Martín-Chivelet, J., Rossi, C., Edwards, R.L., Egli, R., Muñoz-García, M.B., Henshilwood, C.S., Sealy, J.C., Yates, R.J., Cruz-Uribe, K., Goldberg, P., Grine, F.E., Wang, X., Heller, F., 2012. The Blake geomagnetic excursion recorded in Klein, R.G., Poggenpoel, C., van Niekerk, K., Watts, I., 2001. Blombos Cave, a radiometrically dated speleothem. Earth and Planetary Science Letters 353e Southern Cape, South Africa: preliminary report on the 1992e1999 excavations 354, 173e181. of the Middle Stone Age levels. Journal of Archaeological Science 28, 421e448. Parkington, J., 1976. Coastal settlement between the mouths of the Berg and Olifants Herries, A.I.R., 2011. A chronological perspective on the Acheulian and its transition rivers, Cape Province. South African Archaeological Bulletin 31, 127e140. to the Middle Stone Age in southern Africa: the question of the Fauresmith. Pickering, R., Kramers, J.D., Partridge, T., Kodolanyi, J., Pettke, T., 2010. UePb dating International Journal of Evolutionary Biology 961401, 1e25. http://dx.doi.org/ of calciteearagonite layers in speleothems from hominin sites in South Africa 10.4061/2011/961401. by MC-ICP-MS. Quaternary Geochronology 5, 544e558. Herries, A.I.R., Adams, J.W., Kuykendall, K.L., Shaw, J., 2006. and mag- Railsback, L.B., 2000. An Atlas of Speleothem Microfabric. www.gly.uga.edu/ netobiostratigraphic chronology of the GD 2 locality of the Gondolin hominin- speleothems/ASindex1.html. bearing paleocave deposits, North West Province, South Africa. Journal of Hu- Raymo, M.E., Mitrovica, J.X., 2012. Collapse of polar ice sheets during the stage 11 man Evolution 51, 617e631. interglacial. Nature 483, 453e456. Herries, A.I.R., Fisher, E., 2010. Multi-dimensional modelling of magnetic mineral- Richards, D.A., Bottrell, S.H., Cliff, R.A., Strohle, K., Rowe, P.J., 1998. UePb dating of ogy as a proxy for fire use and spatial patterning: evidence from the Middle a speleothem of Quaternary age. Geochimica Et Cosmochimica Acta 62, 3683e Stone Age sea cave site of Pinnacle Point 13B (Western Cape, South Africa). 3688. Journal of Human Evolution 59, 306e320. Roberts, D.L., Karkanas, P., Jacobs, Z., Marean, C.W., Roberts, R.G., 2012. Melting ice Herries, A.I.R., Hopley, P., Adams, J., Curnoe, D., Maslin, M., 2010. Geochronology and sheets 400,000 yr ago raised sea level by 13 m: past analogue for future trends. palaeoenvironments of the South African early hominin bearing sites: a reply to Earth and Planetary Science Letters 357e358, 226e237. ‘Wrangham et al., 2009: shallow-water habitats as sources of fallback foods for Singer, B.S., Hoffman, K.A., Pringle, M.S., 1999. Dating transitionally magnetized hominins’. American Journal Physical Anthropology 143, 640e646. lavas of the late Matuyama Chron: toward a new 40Ar/39Ar timescale of re- Herries, A.I.R., Shaw, J., 2011. Palaeomagnetic analysis of the Sterkfontein palaeocave versals and events. Journal of Geophysical Research 104, 679e693. deposits: implications for the age of the hominin fossils and stone tool in- Singer, R., Wymer, J., 1982. The Middle Stone Age at Klasies River Mouth in South dustries. Journal of Human Evolution 60, 523e539. Africa. University of Chicago Press, Chicago and London. Huntley, D.J., Godfrey-Smith, D.I., Thewalt, M.L.W., 1985. Optical dating of sedi- Thompson, E., 2009. Acheulean artifact accumulation and early hominin land use, ments. Nature 313, 105e107. Garden Route Casino Road, Pinnacle Point, South Africa. Geoarchaeology 24, Jacobs, Z., Duller, G.A.T., Wintle, A.G., 2003a. Optical dating of dune sand from 402e428. Blombos Cave, South Africa: IIdsingle grain data. Journal of Human Evolution Tsukamoto, S., Duller, G.A.T., Wintle, A.G., 2008. Characteristics of thermally 44, 613e625. transferred optically stimulated luminescence (TT-OSL) in quartz and its po- Jacobs, Z., Wintle, A.G., Duller, G.A.T., 2003b. Optical dating of dune sand from tential for dating sediments. Radiation Measurements 43, 1204e1218. Blombos Cave, South Africa: Idmultiple grain data. Journal of Human Evolution Viljoen, J.H.A., Malan, J.A., 1993. Die Geologie van die Gebiede 3421 BB Mosselbaai 44, 599e612. en 3422 AA Herbertsdale. Department of Mineral and Energy Affairs, Pretoria. 52 R. Pickering et al. / Quaternary Science Reviews 65 (2013) 39e52

Wang, X.L., Lu, Y.C., Wintle, A.G., 2006a. Recuperated OSL dating of fine-grained Woodhead, J., Hellstrom, J., Maas, R., Drysdale, R., Zanchetta, G., Devine, P., Taylor, E., quartz in Chinese loess. Quaternary Geochronology 1, 89e100. 2006. UePb geochronology of speleothems by MC-ICPMS. Quaternary Geo- Wang, X.L., Wintle, A.G., Lu, Y.C., 2006b. Thermally transferred luminescence in chronology 1, 208e221. fine-grained quartz from Chinese loess: basic observations. Radiation Mea- Zhou, H., Greig, A., You, C.F., Lai, Z., Tang, J., Guan, Y., Yuan, D., 2008. High-resolution surements 41, 649e658. and precisely dated record of weathering and hydrological dynamics recorded Wang, X.L., Wintle, A.G., Lu, Y.C., 2007. Testing a single-aliquot protocol for recu- by manganese and rare-earth elements in a stalagmite from Central . perated OSL dating. Radiation Measurements 42, 380e391. Quaternary Research 69, 438e446.