Paleoclimate during Neandertal and anatomically modern human occupation at Amud and Qafzeh, Israel: the stable isotope data

Abstract

The δ¹³C(en) and δ¹⁸O(en) values of goat and gazelle enamel carbonate indicate that Neandertals at Amud Cave, Israel (53–70 ka) lived under different ecological conditions than did anatomically modern humans at Qafzeh Cave, Israel (approximately 92 ka). During the Last Glacial Period, Neandertals at Amud Cave lived under wetter conditions than those in the region today. Neither faunal species ate arid-adapted C₄ plants or drought-stressed C₃ plants. The variation in gazelle δ¹⁸O(en) values suggests multiple birth seasons, which today occur under wetter than normal conditions. The magnitude and pattern of intra-tooth variation in goat δ¹⁸O(en) values indicate that rain fell throughout the year unlike today.

Anatomically modern humans encountered a Qafzeh Cave region that was more open and arid than Glacial Period Amud Cave, and more open than today's Upper Galilee region. Goat δ¹³C(en) values indicate feeding on varying amounts of C₄ plants throughout the year. The climate apparently ameliorated higher in the sequence; but habitats remained more open than at Amud Cave. Both gazelles and goats fed on C₃ plants in brushy habitats without any inclusion of C₄ plants. The magnitude of intra-tooth variation in goat δ¹⁸O(en) values, however, suggest that some rain fell throughout the year, and the relative representation of woodland dwelling species indicates the occurrence of woodlands in the region.

Climate differences affecting the distribution of plants and animals appear to be the significant factor contributing to behavioral differences previously documented between Neandertals and anatomically modern humans in the region. Climate forcing probably affected the early appearances of anatomically modern humans, although not the disappearance of Neandertals from the Levant.

Introduction

Throughout the Pleistocene, Western Asia served as a geographical crossroads for the movement of hominins and animals between Africa and Eurasia, as well as a refugium during time periods with severe climates in the north and in the south (Bar-Yosef, 1992a, Shea, 2008, Belmaker, 2009a, Frumkin et al., 2011). Human remains recovered from the region show a range of morphologies including archaic Homo sapiens (here called Neandertals) and anatomically modern humans (e.g., Wolpoff, 1996, Trinkaus, 2005). With the possible exception of level B at Tabun (see Grün, 2006), Neandertal sites in Israel (Amud and Kebara) fall chronologically within MIS 4 (Marine Isotope Stage 4: 70 ka–60 ka) and MIS 3 (60 ka–40 ka), whereas sites attributed to anatomically modern humans (Qafzeh and Skhul) date to MIS 5 (128 ka–80 ka) (Valladas et al., 1987, Valladas et al., 1988, Valladas et al., 1999, Schwarcz et al., 1988, Stringer et al., 1989, Grün et al., 1991, McDermott et al., 1993, Mercier et al., 1993, Mercier et al., 1995, Rink et al., 2001). Anatomically modern humans apparently entered Israel from Africa during the Last Interglacial, a warm period in East Africa that included times of severe aridity (Scholz et al., 2007), whereas Neandertals migrated from Europe, presumably in response to increasingly cold and dry conditions at the onset of the last glacial cycle (Tchernov, 1989, Bar-Yosef, 1989, Bar-Yosef, 1992b, Bar-Yosef and Vandermeersch, 1993, Scholz et al., 2007; but see Belmaker and Hovers, 2011, Frumkin et al., 2011).

Lieberman and Shea (1994) demonstrated behavioral differences in hunting strategies between Neandertals and anatomically modern humans based on associations with different relative combinations of similar fauna and stone tools. They proposed that behavioral differences were directly related to morphological differences between the populations although they allowed for the potential of climate change as an important variable. Explanations for the apparent replacement of anatomically modern humans by Neandertals during MIS 4, and the subsequent replacement of Neandertals by modern humans, include direct inter-population (or inter-species) competition (Shea, 2003), intra-population pressure (Stiner et al., 1999, Speth and Tchernov, 2002), overhunting (Speth and Clark, 2006) using behaviors similar to those in modern human foragers (Speth, 2006), or, in other regions, cognitive differences between Neandertals and anatomically modern humans (Klein, 1999). More recently, Shea (2008) proposed that abrupt and significant climate changes at 75 ka and 45 ka forced the disappearances of anatomically modern humans and Neandertals, respectively. Even so, there is a general assumption that the ecological situation changed little throughout the Levantine sequence (Lieberman and Shea, 1994, Shea, 2008) based on overall similarities in the fauna across sites (Tchernov, 1988).

Local climate in Israel, in contrast to Europe or Africa, is difficult to ascertain because it falls “in the transition zone between a humid climate in the north and an extremely arid climate in the south and southeast” (Vaks et al., 2003:182). The general assumptions are that Israel was relatively cooler during the Last Glacial Period, relatively warmer during the Last Interglacial, and that today’s highly seasonal rainfall pattern occurred throughout the complete sequence (Bar-Yosef, 1992a). Here, we test the assumptions about rainfall patterns and environments using stable carbon and oxygen isotope data from the tooth enamel of gazelles and goats recovered from Amud Cave and Qafzeh Cave (Fig. 1). This study takes advantage of properties of tooth development to make inferences about seasonality and palaeoclimate at these sites. We conclude that significant ecological differences associated with climate changes are the most likely explanation for variation in foraging strategies between the two groups. Such variation could have occurred in the absence of any competition or cognitive disparity between Neandertals and anatomically modern humans, and irrespective of morphological differences.

Israel (with the exception of the Negev) is situated between 29.5° and 33.5°N along the southeastern Mediterranean coast. Today, it can be divided into four longitudinal physiographic zones (from west to east), which account for differences in local climatic conditions across the region. These include: 1) the Coastal Plain; 2) the Galilee Mountains (from north to south); 3) the Jordan Rift Valley; and 4) the Golan Heights. Israel experiences a semi-arid Mediterranean climate with a rainy winter and dry summer, although the diversity of geographical factors causes precipitation to be spatially and temporally variable (Goldreich, 2001). For example, maximum values around 1000 mm/yr occur in the central Upper Galilee and in the Golan Heights, while at lower elevations the amount of rainfall decreases to less than 500 mm/yr in the Coastal Plain and in the Jordan Valley. Mean annual temperature increases from north to south ranging from approximately 16 °C in the north to 23 °C in the south (Goldreich, 2001, Ayalon et al., 2004).

The interrelationship between the amount of available moisture and temperature is a critical factor for the distribution of different plant groups (Zohary, 1982, Murphy and Bowman, 2009). In Israel, three phytogeographic regions meet: 1) the Mediterranean, 2) the Irano-Turanian, and 3) the Saharo-Sindian (Fig. 1; Cleave, 1999). There is a distinct correlation between these phytogeographic districts and the photosynthetic pathways (C₃ vs. C₄) utilized by plant species in each area (Shomer-Ilan, 1983, Vogel et al., 1986). C₃ grasses make up the largest vegetation group in the Mediterranean region where most rain falls regularly between November and April. In contrast, C₄ grass species are found predominantly in the Saharo-Sindian and Irano-Turanian regions, with higher annual temperatures and variable annual precipitation at some sites. Today, Amud Cave lies on the eastern edge of the Mediterranean or just within the Irano-Turanian phytogeographic zone, and Qafzeh Cave lies within the core Mediterranean phytogeographic zone (Madella et al., 2002, Rabinovich and Hovers, 2004).

The average annual temperature during the Last Glacial Interval (MIS 4-3) in Israel was about 5 °C colder than today based on alkenone temperatures from deep-sea cores in the Mediterranean (Emeis et al., 2000) and fluid inclusions in speleothems from Soreq Cave (McGarry et al., 2004). Faunal studies at several sites demonstrated that Palearctic-European elements, which invaded the region several times over the last 400 k.yr. (Stiner et al., 2009), returned at the beginning of MIS 4 (Tchernov, 1981, Tchernov, 1988, Tchernov, 1989, Tchernov, 1994). Until recently, the general opinion was that the interval was drier than today (e.g., Shea, 2008), based largely on the oxygen isotope compositions of speleothems (Bar-Matthews et al., 1997; but see Frumkin et al., 1999, Frumkin et al., 2000). A recent compilation and assessment of well-dated paleolake levels and periods of speleothem growth periods, however, strongly suggested that this region of the Levant, in general, was wetter than at present (Frumkin et al., 2011; see also Belmaker and Hovers, 2011). Horowitz and Gat (1984) suggested the plant and animal communities within oak woodlands, which today are largely restricted to the wetter, northern-central area of the Mediterranean phytogeographic region, could have been more extensively distributed.

The conditions in local areas are also uncertain. For example, the large mammal fauna recovered from Amud Cave are dominated by gazelle (Gazella gazella), an open woodland species, and by goats (Capra sp.), which today are restricted to bushy, rocky habitats (Rabinovich and Hovers, 2004). In contrast, the micromammal fauna suggest wetter conditions (Belmaker and Hovers, 2011). These reports reveal an important problem area. It is possible that Amud Cave represented an arid region within an overall wetter Levant during the Last Glacial Interval or, alternatively, the large mammal fauna represent individual habitats within the region surrounding the cave.

The Last Interglacial (MIS 5) has generally been considered relatively humid (e.g., Shea, 2008) based on speleothem isotope data (Bar-Matthews et al., 2000, Bar-Matthews et al., 2003). In contrast, the δ¹³C values in the same speleothems suggested to Frumkin and colleagues (Frumkin et al., 2000) that MIS 5e (128,000–120,000 ka) was arid with somewhat subsequently wetter conditions during MIS 5d-a. Recent consideration of other proxy climate measures also suggested aridity, not humidity, during MIS 5 (Frumkin et al., 2011), although sapropel (S5) data suggest that the period was less stable climatically than today (Cane et al., 2002, Weldeab et al., 2002). African and Indo-Arabian fauna dominate many assemblages in the region overall, suggesting a northern shift of the Saharo-Sindian desert belt (Tchernov, 1981, Tchernov, 1988, Tchernov, 1994). At Qafzeh Cave, however, woodland-adapted species dominate the large mammal fauna throughout the sequence (red deer [Cervus elephas], fallow deer [Dama mesopotamica], and auroch [Bos primigenius]; Rabinovich and Tchernov, 1995, Rabinovich et al., 2004) even though the microfauna from the lower levels suggest dry conditions (Belmaker, 2009b). As yet, there is no report on microfauna from the upper levels. Thus, the local environmental setting at Qafzeh Cave is uncertain just as it is at Amud Cave.

Analyses of the ¹³C/¹²C and ¹⁸O/¹⁶O ratios in mammalian tooth enamel, δ¹³C(en) and δ¹⁸O(en) values respectively, are independent of and complementary to other methods of paleoclimate reconstructions. The δ¹³C(en) value indicates diet because the value is a function of the average δ¹³C of total diet (Ambrose and Norr, 1993; see Kellner and Schoeninger [2007] for a recent review) as determined by the proportion of C₃ and C₄ plants in the diet (DeNiro and Epstein, 1978). The worldwide range of δ¹³C values of C₃ plants is roughly 15‰ although even the highest values do not overlap those in C₄ plants (Kohn, 2010). The C₃ plants with the lowest δ¹³C values grow in closed canopy forests and those toward the higher end grow in open habitats (van der Merwe and Medina, 1991). In addition, the δ¹³C values of C₃ plants are negatively correlated with mean annual precipitation on a worldwide basis (Kohn, 2010) and the same is true in the Eastern Mediterranean (Hartman and Danin, 2010). So, the C₃ plants with the highest δ¹³C values grow under arid or semi-arid conditions (Kohn, 2010). Today in the Eastern Mediterranean, C₄ plants occur only in areas with rainfall below 350 mm/yr and are, therefore, indicators of aridity (Vogel et al., 1986, Hartman and Danin, 2010). In sum, high δ¹³C(en) values in mammalian herbivores can indicate arid conditions whether their diet is composed of C₃ or C₄ plants or a mixture of the two. Serially sampled intra-tooth variation tracks an annual or seasonal record of dietary change in feeding in open or closed habitats, in arid, semi-arid, or mesic environments, or in the ratio of C₃:C₄ plants in the diet throughout the time of mineralization that varies (Balasse, 2002, Balasse et al., 2002, Balasse et al., 2003) or remains stable (Balasse, 2003).

Given the range of variation in C₃ and C₄ plants, there is some disagreement over what δ¹³C(en) value indicates the overall range of δ¹³C(en) values expected for herbivores eating 100% C₃ plants and those that include some C₄ biomass in their diet. We used several methods to estimate those values.

Modern East African herbivores eating a 100% C₃ diet show a range in δ¹³C(en) values from −15.9‰ to −10.6‰ where the lower values come from animals feeding in closed canopy forests, somewhat higher values are from animals feeding in open habitats, and the highest values come from animals feeding under highly arid conditions; the species eating 100% C₄ plants show a range from −3.6‰ to 2.2‰ (Cerling et al., 2003). Using the highest δ¹³C(en) value observed among C₃-feeders (expected under arid conditions) and the lowest value for pure C₄-feeders, a dietary mixture of 80% C₃ and 20% C₄ plants would result in an estimated δ¹³C(en) value of −9‰, and a 50:50 mixture of C₃ and C₄ plants should produce a value around, −7‰.

An alternative approach uses an assumed offset between the δ¹³C(diet) and δ¹³C(en) values of ∼14‰ based on data from experimental and free-ranging herbivores (Lee-Thorp et al., 1989, Balasse, 2002, Passey et al., 2005). Worldwide, the δ¹³C values for C₃ plants in open (not arid) environments have values as high as −26‰ (Kohn, 2010); therefore, we assume that δ¹³C(en) values around −12‰ indicate feeding 100% on C₃ plants in open habitats. The highest δ¹³C value for C₃ plants in Israel today, measured during the dry season, is around −25‰ and the lowest average for C₄ plants is around −14.5‰ (Hartman and Danin, 2010). Therefore, when the C₃ plants grow under dry (arid) conditions, the herbivore δ¹³C(en) value should be around −11‰ for 100% C₃-feeders, −9‰ for 80% C₃-feeders, and −6‰ for 50% C₃-feeders. Using the highest δ¹³C values for C₃ plants worldwide (around −24‰; Kohn, 2010), and the lowest δ¹³C values for C₄ plants in Israel today, herbivore δ¹³C(en) values would be around −10‰, −8‰, and −5‰ for 100%, 80%, and 5% C₃-feeding, respectively, on plants growing under arid conditions.

At the other end of the range, the lowest reported δ¹³C value for C₃ plants in Israel today (Hartman and Danin, 2010) is around −28‰. The estimated δ¹³C(en) value is around −14‰. Such a value should represent an animal feeding in woodlands since closed canopy forests do not occur in Israel today. The overall similarity in estimates for δ¹³C(en) values using all three sets of data (African herbivore δ¹³C(en) values, worldwide plant δ¹³C values, and Israeli plant δ¹³C values) allowed us to estimate the relative diet proportions of C₃ and C₄ plants from animal δ¹³C(en) values. In summary, we assume that δ¹³C(en) values around −16‰ indicate feeding in closed canopy forests, values around −14‰ indicate feeding in open woodlands, values around −12‰ indicate feeding in open habitats (bushy or C₃ grass covered), values around −10‰ indicate feeding in arid C₃ habitats, and values above −10‰ indicate dietary intake of some C₄ plants in an arid habitat.

Depending on physiology, the δ¹⁸O(en) value in mammals indicates either aridity or ambient temperature. This occurs because both the phosphate and carbonate fractions in mammalian tooth enamel reflect the δ¹⁸O value of their body water (Bryant et al., 1996, Iacumin et al., 1996), which depends on the oxygen stable isotope composition of precipitation δ¹⁸O(ppt) and species physiology (Luz et al., 1984, Luz et al., 1990, Huertas et al., 1995, Kohn and Cerling, 2002). In herbivorous species that are able to satisfy their water needs largely from food, the relationship between δ¹⁸O(en) values and δ¹⁸O(ppt) values varies depending on aridity (Levin et al., 2006). Such species usually rely on leaves from a variety of browse types (e.g., tree and bush leaves, and forbs; Maloiy et al., 1979, National Research Council Committee and goats, 2006). Leaf water δ¹⁸O values can be significantly higher than δ¹⁸O(ppt) values due to preferential loss of ¹⁶O during evaporation through stomata, which increases with decreasing relative humidity (Gonfiantini et al., 1965, Dongmann et al., 1974). This dependence of δ¹⁸O(en) on relative humidity has been demonstrated in Australian marsupials (Ayliffe and Chivas, 1990), East African herbivorous mammals (Levin et al., 2006), and North American deer in arid regions (Luz et al., 1990; Figure 2). In such species, intra-tooth variation in δ¹⁸O(en) values mainly reflect seasonal variation in relative humidity. In contrast, under conditions of normal to high relative humidity, leaf water δ¹⁸O values of leaves usually track δ¹⁸O(ppt) values.

In mammalian species that obtain the majority of their body water by drinking surface water, δ¹⁸O(en) values correlate with δ¹⁸O(ppt) values (Levin et al., 2006), which in turn correlates with ambient temperature (Dansgaard, 1964, Rozanski et al., 1992). At a given location, δ¹⁸O(ppt) values increase by about 0.6‰/°C (Rozanski et al., 1992, Kohn and Welker, 2005) with higher values in summer and lower values in winter. For example, bulk δ¹⁸O(en) values of goats (Capra ibex and Capra sp.), which obtain the majority of their body water by drinking surface water (Smith and Sherman, 1994, National Research Council Committee and goats, 2006), correlate linearly with δ¹⁸O(ppt) values across a wide geographic range (Huertas et al., 1995) (Figure 2). Intra-tooth variations in δ¹⁸O(en) values, measured on serial subsamples in goats and similar species, track seasonal and annual changes in δ¹⁸O(ppt) values (Stuart-Williams and Schwarcz, 1997, Fricke et al., 1998, Balasse et al., 2002). The serial subsamples show generally lower δ¹⁸O values during the winter and higher values in summer due to the correlation between δ¹⁸O(ppt) values and ambient temperature (Rozanski et al., 1992, Balasse et al., 2003, Kohn and Welker, 2005).

In Israel today, 80–90% of rain falls between November and April with virtually no rain falling during the period of high summer temperature (Ayalon et al., 2004). There is little annual variation in δ¹⁸O(ppt) values at sites with rainfall exceeding 200 mm/year (e.g., ≤1‰ across nine years at Soreq Cave) even though daily values can differ markedly (Ayalon et al., 2004: Figure 4). Under such conditions, water-drinking mammals should record only cool season rainfall with little variation in δ¹⁸O(en) values within a single tooth due to an averaging of δ¹⁸O values as enamel is mineralized (see below). In contrast, places like Ma’ale Ephrayim (alternatively, Maale Efraim), which experiences 100–200 mm of rainfall each year, displays far greater intra- and inter-annual variation (Ayalon et al., 2004). Goats from such a region could display dissimilar average δ¹⁸O(en) values across animals when individual goats derive from different years, although intra-tooth variation is still expected to be small.

Put most simply, the variation or lack of it in δ¹⁸O (en) values of a tooth from a non-domestic mammal that drinks surface water can indicate whether its drinking water varies isotopically throughout the year (intra-tooth variation in δ¹⁸O[en]) values or not (constant δ¹⁸O[en] values). If it varies, it can indicate rain falling in both colder and warmer times of the year (Balasse et al., 2002). If constant, it suggests rain falling during a single time of the year or that there is little annual change in temperature (Kohn et al., 1998). The δ¹⁸O(en) values within a tooth from mammal species that obtain their body water from their food can indicate if there had been seasonal variation in relative humidity under conditions of low relative humidity.