The greening of Arabia: multiple opportunities for human occupation of the Arabian peninsula during the Late Pleistocene inferred from an ensemble of climate model simulations Article Accepted Version Jennings, R. P., Singarayer, J., Stone, E. J., Krebs-Kanzow, U., Khon, V., Nisancioglu, K. H., Pfeiffer, M., Zhang, X., Parker, A., Parton, A., Groucutt, H. S., White, T. S., Drake, N. A. and Petraglia, M. D. (2015) The greening of Arabia: multiple opportunities for human occupation of the Arabian peninsula during the Late Pleistocene inferred from an ensemble of climate model simulations. Quaternary International, 382. pp. 181-199. ISSN 1040-6182 doi: https://doi.org/10.1016/j.quaint.2015.01.006 Available at https://centaur.reading.ac.uk/39673/ It is advisable to refer to the publisher’s version if you intend to cite from the work. See Guidance on citing . To link to this article DOI: http://dx.doi.org/10.1016/j.quaint.2015.01.006 Publisher: Elsevier All outputs in CentAUR are protected by Intellectual Property Rights law, including copyright law. Copyright and IPR is retained by the creators or other copyright holders. Terms and conditions for use of this material are defined in the End User Agreement . www.reading.ac.uk/centaur CentAUR Central Archive at the University of Reading Reading’s research outputs online The greening of Arabia: multiple opportunities for human occupation of the Arabian Peninsula during the Late Pleistocene inferred from an ensemble of climate model simulations Richard P. Jenningsa, Joy Singarayerb, Emma Stonec, Uta Krebs-‐Kanzowd, Vyacheslav Khone,f, Kerim H Nisancioglug,h, Adrian Parkeri, Ash Partona, Huw S. Groucutta Tom Whitea, Nick. A. Drakej, Michael D. Petragliaa aSchool of Archaeology, Research Laboratory for Archaeology and the History of Art, University of Oxford, Oxford, OX1 3QY, United Kingdom bCentre for Past Climate Change, Department of Meteorology, University of Reading, Earley Gate, Reading, RG6 6BB cBRIDGE, School of Geographical Sciences, University of Bristol, BS8 1SS, United Kingdom dThe Alfred Wegener Institute, Bussestraße 24, D-‐27570 Bremerhaven, Germany eInstitute of Geosciences, University of Kiel, Kiel, Germany fA.M. Obukhov Institute of Atmospheric Physics RAS, Moscow, Russia gDepartment of Earth Science and the Bjerknes Centre, University of Bergen, Norway hUNI Research Climate, Allégaten 55, 5007 Bergen, Norway iDepartment of Social Sciences, Oxford Brookes University, Headington Campus, Oxford, OX3 0BP, United Kingdom jDepartment of Geography, King’s College London, K4U.06 Strand Campus, London WC2R 2LS, United Kingdom. Abstract Climate models are potentially useful tools for addressing human dispersals and demographic change. The Arabian Peninsula is becoming increasingly significant in the story of human dispersals out of Africa during the Late Pleistocene. Although characterised largely by arid environments today, emerging climate records indicate that the peninsula was wetter many times in past, suggesting that the region may have been inhabited considerably more than hitherto thought. Explaining the origins and spatial distribution of increased rainfall is challenging because palaeoenvironmental research in the region is in an early developmental stage. We address environmental oscillations by assembling and analysing an ensemble of five global climate models (CCSM3, COSMOS, HadCM3, KCM, and NorESM). We focus on precipitation, as the variable is key for the development of lakes, rivers and savannas. The climate models generated here were compared with published palaeoenvironmental data such as palaeolakes, speleothems and alluvial fan records as a means of validation. All five models showed, to varying degrees, that the Arabia Peninsula 1 was significantly wetter than today during the Last Interglacial (130 ka and 126/125 ka timeslices), and that the main source of increased rainfall was from the North African summer monsoon rather than the Indian Ocean monsoon or from Mediterranean climate patterns. Where available, 104 ka (MIS 5c), 56 ka (early MIS 3) and 21 ka (LGM) timeslices showed rainfall was present but not as extensive as during the Last Interglacial. The results favour the hypothesis that humans potentially moved out of Africa and into Arabia on multiple occasions during pluvial phases of the Late Pleistocene. Keywords: Climate models, Late Pleistocene, human evolution, Arabian Peninsula, precipitation 1. Introduction The timing and spatial distribution of the dispersal of Homo sapiens out of Africa is the subject of intense and continued scientific debate (e.g. Oppenheimer, 2009; Frumpkin et al., 2011; Boivin et al., 2013; Mellars et al., 2013). The consensus view is that genetic and archaeological evidence supports a major dispersal of human populations after 70,000 years ago (ka) during Marine Isotope Stage (MIS) 4 or early MIS 3 (Mellars et al., 2013). However, this has been challenged by recent archaeological discoveries and new genetic and environmental interpretations, which advocate multiple dispersals during MIS 5 (130-‐ 71 ka) and also later times (Boivin et al., 2013; Groucutt and Petraglia 2014; Parton et al., this issue). Central to this debate is the need to understand the environmental context of human dispersals. In regions where secure palaeoenvironmental records are incomplete or under development this is a difficult task. Here we present a different approach and assess the degree to which global climate models can inform us about past climate change and human dispersals. Climate models are sophisticated mathematical tools that have been used in similar contexts to address key questions such as the extinction of the Neanderthals and the arrival of modern humans in Europe (Davies et al., 2003), the identification of Neanderthal refugia in southern Iberia (Jennings et al., 2011) and the modelling of human populations structures in North Africa (Scerri et al., 2014). Eriksson and colleagues (2012) have recently used one model to explore Homo sapiens dispersals out of Africa but on a larger geographic scale than we now present. Instead, we focus solely on one key region, the Arabian Peninsula. 2 Those working in the Arabian Peninsula consider the region to be a critical area for understanding Homo sapiens dispersals and demography (Petraglia and Alsharekh, 2003; Marks et al. 2009; Parker et al., 2009; Armitage et al., 2011; Rose et al., 2011; Rosenberg et al., 2011, 2013; Petraglia, 2011; Groucutt and Petraglia, 2012, 2014; Crassard and Hilbert 2013; Crassard et al., 2013). Yet, the Saharo-‐Arabian desert belt, a hyperarid-‐arid expanse at 14-‐35°N, has often been seen as a major biogeographical barrier for human range expansions out of Africa. However, over the last decade it has been shown that the Sahara was not always desert and that rivers, lakes and savanna grasslands developed during pluvial episodes, linked to changes in insolation (e.g. Drake et al., 2011, 2013; Larrasoaña et al., 2013). As research develops in the Arabian Peninsula, researchers are returning to ancient lakebeds previously investigated in the 1970s (e.g. Petraglia et al., 2011; Rosenberg et al., 2011, 2013; Crassard et al., 2013). Although these palaeolakes were once thought to date to MIS 3, evidence now indicates that the lakes formed during MIS 5 (Rosenberg et al., 2011; Petraglia et al., 2011). Speleothem growth, a sign of increased humidity, has also been dated to this period in caves in south and southeast Arabia (Fleitmann et al., 2004, 2007), while Late Quaternary fluvial deposits have been identified in the United Arab Emirates (UAE) (Atkinson et al., 2013). Much of what we know about Arabia’s past climate is currently undergoing extensive revision. Just as in the Sahara, the idea that the Arabian Peninsula was wetter during certain periods of the Pleistocene is rapidly developing into the null hypothesis. Recent discoveries of Middle Palaeolithic sites in stratified contexts on the shores of some of these aforementioned palaeolakes strongly suggests that modern human populations, or possibly even Neanderthals, were in the interior of the Arabia Peninsula during MIS 5 (Groucutt and Blinkhorn, 2011; Petraglia et al., 2012; Crassard et al., 2013). If savanna grassland and lakes had developed 130 ka, the idea of a coastal dispersal (cf. Mellars et al., 2013), in which the interior of Arabia was bypassed, seems highly unlikely. Instead, the presence of savanna habitats suggests that human populations would have dispersed into the region during humid periods between 130-‐78 ka, given their presence in Skhul and Qafzeh in the Levant (Grün et al., 2005), at Jebel Faya in SE Arabia (Armitage et al., 2011; Bretzke et al. 2013) and in East Africa (Basell, 2008) at this time. 3 One of the key difficulties in testing the hypothesis that human populations inhabited the interior of Arabia during the Late Pleistocene is the lack of dated and stratified archaeological sites. For a landmass that is one quarter of the size of Europe, a small handful of dated lakebeds, speleothems and stratified sites do not provide enough evidence to link human presence to the palaeoclimatic framework. Here, we take a different approach and assess this data in the context of five climate models covering the Arabian Peninsula. We employ climate models because they are an informative way of developing insights into past climate change, especially where comprehensive palaeoclimate records are not available. Climate models take into account the major processes that shaped past climate change, such as astronomical forcing, ice sheet extent, sea-‐level, vegetation cover, and atmospheric greenhouse gas concentrations. The climate models provide spatial and temporal frameworks that can be tested using securely dated, independently derived sedimentological and palaeontological data (Braconnot et al., 2012; Heiri et al., 2014). In the present article we focus on modelling precipitation, as this variable is a useful measure for determining the amount of rainfall that potentially fell across the Peninsula during the Late Pleistocene, as water, of course, was vital for the range expansion of our species (Finlayson 2014). Multiple timeslices are examined, but with a particular focus on specific periods of MIS 5e, when rainfall levels are thought to have been at their highest (Parton et al., this issue). We first review the main weather systems bringing rainfall into the Peninsula today and describe the boundary conditions used in model experiments, as understanding both is necessary before using palaeoenvironmental and archaeological evidence to assess the models and form hypotheses on human demography in the Arabian Peninsula during these periods. 2. Present day rainfall patterns in the Arabia Peninsula The Arabian Peninsula supports some of the driest environments anywhere in the world. Areas of the Negev desert, north and northwest Saudi Arabia, and in the Rub Al-‐Khali, a vast expanse of sand desert in southern Saudi Arabia, receive mean annual rainfall levels of < 60 mm per annum, placing them in hyper-‐arid bioclimatic zones (Almazroui et al., 2012). Rainfall is more varied elsewhere in the peninsula (Figure 1), attaining an average of 75.4 mm in the Eastern Province of Saudi Arabia (with the highest yearly total recorded being 384 mm) (Barth and Steinkhol, 2004), up to 140 mm per annum in the UAE (Parker, 2006), 234 mm per annum in Bahrain (Elagib and Abdu, 2010) and up to 400 mm in the Al Jabal 4 and Al Akhdar mountains of Oman (Kwarteng et al., 2009). High levels of rainfall are also known in the Yemen highlands and in southwest Saudi Arabia, where up to 400 mm per annum is recorded and it can rain during every month of the year (Al-‐Subyani, 2005; Almazroui, 2011; Furl et al., 2014). However, decadal averages across the peninsula do not typically exceed 200 mm per annum outside the upland areas of SW Saudi Arabia, Yemen and Oman (Almazroui et al., 2012), meaning an arid to hyper-‐arid climate prevails across the vast majority of the peninsula. ***Figure 1 hereabouts An important reason for the observed rainfall variation in Arabia is orography, where rainfall increases with elevation. For instance, the Asir Mountains (3000 m above sea level) which run parallel to the Red Sea along the western side of the Peninsula, and the Hadramaut (1500 m above sea level) in the south of Arabia receive higher rainfall than interior areas. It has been noted that there are too few weather stations to record accurately precipitation levels in such regions owing to the complex orographic variability (Abo-‐Monasar and Al Zahrani, 2014). Elsewhere, at Wadi Yalalam in the west of Arabia, rainfall levels are higher in the wadi's upper reaches (220 mm), on the western slopes of the Hijaz escarpment, than at sea level (110 mm) where it discharges into the Red Sea at Tihamah (Al-‐Subyani, 2005). Similarly, although arid and semi arid zones (receiving < 300 mm) exist on the leeward side of the Al Jabal and Al Akhdar mountains in the interior of Oman, coastal regions remain humid due to monsoonal weather patterns (Kwarteng et al., 2009). In Jordan, annual rainfall levels are negligible on the southern and eastern sides of the Jordanian highlands, with only 32 mm recorded at Aqaba in the very south, but values exceed 500 mm per year on their upper western flanks, where Mediterranean conditions prevail (Freiwana and Kadioglu, 2008). As such, a strong N-‐S precipitation gradient exists in the Eastern Mediterranean, the Levant, and the southern Negev. Inter-‐annual rainfall variability is also a strong feature of the Arabian climate (Almazroui et al., 2012). Many climate systems bring mainly low levels of precipitation to the Arabian Peninsula, each varying in timing, location, and intensity. Sometimes a single rainfall episode can provide an area with its annual rainfall total, meaning decadal averages are required to gauge patterns (Elagib and Addin Abdu, 1997; Rheman et al., 2010). The Inter-‐ Tropical Convergence Zone (ITCZ), a major airflow that drives monsoonal activity across 5 sub-‐tropical latitudes of the world, is a key driver of rainfall across the Arabian Peninsula. The ITCZ moves on a continuous yearly cycle, reaching as far north as 25-‐30°N in Asia in July and as far south as 15°S over Africa in January (Henderson-‐Sellers and Robinson, 1991). Sub-‐tropical weather affects the south of the Arabian Peninsula in the summer months (May–September) as the ITCZ begins its move northwards. In particular, rains of the North African summer monsoon, which is a northward extension of the West African monsoon system, cross the Sahel and reach SW Arabia in July (Bosmans et al., 2014), while the Indian Ocean Monsoon reaches the southern coasts of Yemen and Oman in June (Kwarteng, 2009). The ITCZ movement is caused by seasonal land-‐sea thermal contrast and subsequent development of a low-‐pressure cell situated above the foothills of the Tibetan plateau (Fleitmann et al., 2004). Central and northern Arabia remains largely dry during these months. During the winter (October–April), the Azores high pressure of the North Atlantic Oscillation (NAO) and the East Atlantic/West Russian (EAWR) atmospheric circulation systems drive the ITCZ southwards, which in turn generates moisture-‐bearing westerlies. These take the form of storms, produced as cold air masses meet warm ocean waters, which generally move along the Mediterranean basin into the Middle East and down the Arabian Gulf as far as southern Oman (Krichak et al., 2000; Barth and Steinkhol, 2004; Kwarteng, 2009; Brayshaw et al., 2010; Trigo et al., 2010; Kalimeris, 2011). The northeast winter monsoon (Van Rampelbergh et al., 2013) also brings low levels of rain to the southern coast of Arabia, particularly in January as the ITCZ moves to its most southerly position. Other climatic systems also contribute rainfall to the Peninsula: winter air masses from the Mediterranean meet the Zagros Mountains and may develop into independent low-‐ pressure cells that bring precipitation to north and east Arabia (Barth and Steinkhol, 2004); tropical cyclones form in the Arabian Sea in May and again in October to November and bring rainfall to the Gulf States (Kwarteng 2009); and local convection over Arabia, where a strong contrast between weather cells can occur at different times of the year and lead to rain (Barth and Steinkhol, 2004; Kwarteng et al., 2009). For example, cool northeasterly air currents from the Siberian trough can meet warm air currents from the Red Sea Trough (also known as the Sudan Trough), resulting in low-‐pressure, warm and 6 humid air masses in November, March and April over Arabia. This may lead to low levels of rainfall (Almazroui et al., 2012; Barth and Steinkhol, 2004; Lionello 2012; Furl et al., 2014). 3. Model configuration Simulations from five climate models are used here to examine the impact of changes in solar insolation, ice sheet extent, sea-‐level, sea surface temperature (SST), vegetation cover, greenhouse gases and other variables on the climate systems that bring rainfall into the Arabian Peninsula today. An ensemble of models is used to improve model reliability and evaluate variability between different models. All of the models are part of the Palaeoclimate Modeling Intercomparison Project (PMIP) and common boundary conditions are used to ensure the model outputs are comparable (Table 1). 3.1 CCSM3 climate model This model differs from the other four models used in this study in that it is a downscaled version of the Community Climate Systems Model (CCSM3). Downscaling global climate models is undertaken to link atmospheric values generated in such models with higher-‐ resolution topographic and climate data (Jones et al., 2009). The CCSM3 model was downscaled by Hijmans et al., (2005) and made publically available on the WorldClim website. The CCSM 3 model is a fully-‐coupled, global atmosphere-‐land surface-‐ocean sea ice general circulation model. It comprises the atmosphere model CAM3 and the land model CLM 3 (Otto-‐Bleisner et al., 2006). These originally had resolutions of 1.40 of latitude and longitude but these were downscaled to a cell resolution of 30 arc seconds (c. 1km). Downscaling involved a global climate surface that was interpolated from topographic (SRTM) and weather station (World Meteorological Organisation) data (Hijmans et al., 2005). This served as baseline data for the downscaling of the CCSM3 model. In this article we present the results in the form of annual precipitation for 130 ka and for 21 ka. 3.2 COSMOS climate model COSMOS is a comprehensive fully coupled Earth System Model. The atmospheric model ECHAM5 (Roeckner et al., 2003), complemented by a land surface component JSBACH (Brovkin et al., 2009) used at T31 resolution (~3.75°), with 19 vertical layers. The ocean model MPI-‐OM (Marsland et al., 2003) including sea ice dynamics that is formulated using viscous-‐plastic rheology (Hibler, 1979), has a resolution of GR30 (3°× 1.8°) in the horizontal, with 40 uneven vertical layers. For this study we present equilibrated 7 simulations of selected time slices within the Last Interglacial/Eemian (130ka, 125 ka, and 115ka) (Pfeiffer and Lohmann, 2013) and of the Last Glacial Maximum (LGM) (21ka) (Zhang et al., 2013). As a control experiment we use a simulation equilibrated under pre-‐industrial conditions (Zhang et al., 2013). The boundary conditions (such as orbital parameters, green house gas concentrations, geometry of continental ice sheets and sea level) of the LGM and pre-‐ industrial simulation follow the guidelines of the PMIP. The Eemian simulations (130 ka, 125 ka, and 115ka) use the orbital configuration values of the respective time periods and a Greenland ice sheet which is diminished by reducing ice thickness by 1300 m at each grid point or, wherever today's ice elevation is less than 1300 m, removing ice completely and adjusting albedo accordingly. All other boundary conditions are chosen as in the pre-‐ industrial control experiment. 3.3 HadCM3 climate model The Hadley Centre climate model, HadCM3 (version 4.5), consists of a coupled atmospheric model, ocean model, and sea ice model components (Pope et al., 2000; Gordon et al., 2000). The atmosphere is a global grid-‐point hydrostatic primitive equation model with a resolution of 2.5° in latitude by 3.75° in longitude and 19 unequally spaced vertical levels. The land surface model (MOSES2.1; Essery et al. 2001) has a ‘tiled’ gridbox scheme with nine fractional surface types that exchange water, carbon, and energy with the atmosphere. There is a representation of freezing and melting of soil moisture and four soil depth layers. The spatial resolution of the ocean in HadCM3 is 1.25° by 1.25° by 20 unequally spaced layers in the ocean extending to a depth of 5200 m. The ocean model uses the mixing scheme of Gent and McWilliams (1990) with no explicit horizontal tracer diffusion. The sea ice model uses a simple thermodynamic scheme parameterization of ice drift and leads (Cattle and Crossley, 1995). For this study, two sets of experiments with HadCM3 were utilized. The first set of simulations cover time slices in the Eemian (130 ka, 125 ka, and 116 ka) and were completed as part of the PMIP3 initiative. The simulations use orbital configuration values for the relevant time slice (derived from Berger, 1978), and trace atmospheric greenhouse gases from ice core data on the EDC3 time-‐scale (see Table 1). Vegetation was fixed and based on pre-‐industrial estimates. Ice-‐sheets and sea level were similarly fixed at pre-‐ 8
Description: