Tortoises as a dietary supplement: A view from the Middle Pleistocene site of Qesem Cave, Israel

Highlights

• Dietary reconstructions reflect the human capacity for adaptation to the environment.

• Faunal remains allow for the study of behavioural variation and its evolutionary significance.

• Tortoises represent an important combination of edible and non-edible resources.

• Tortoises served mainly as supplementary resources at Qesem Cave, Israel (420-300 ka).

• Data on small game at Qesem shed light on livelihood strategies and eco-social behaviour.

Abstract

Dietary reconstructions can offer an improved perspective on human capacities of adaptation to the environment. New methodological approaches and analytical techniques have led to a theoretical framework for understanding how human groups used and adapted to their local environment. Faunal remains provide an important potential source of dietary information and allow study of behavioural variation and its evolutionary significance. Interest in determining how hominids filled the gaps in large prey availability with small game or what role small game played in pre-Upper Palaeolithic societies is an area of active research. Some of this work has focused on tortoises because they represent an important combination of edible and non-edible resources that are easy to collect if available. The exploitation of these slow-moving animals features prominently in prey choice models because the low handling costs of these reptiles make up for their small body size. Here, we present new taphonomic data from two tortoise assemblages extracted from the lower sequence of the Middle Pleistocene site of Qesem Cave, Israel (420-300 ka), with the aim of assessing the socio-economic factors that may have led to the inclusion of this type of resource in the human diets. We show that hominid damage on large tortoise specimens from Qesem Cave is not unusual and that evidence such as cut marks, percussion marks and consistent patterns of burning suggests established sequences of processing, including cooking in the shell, defleshing, and direct percussion to access the visceral content. These matters make it possible not only to assess the potential role of tortoises as prey, but also to evaluate collecting behaviour in the resource acquisition systems and eco-social strategies at the Acheulo-Yabrudian Cultural Complex (AYCC) in the southern Levant.

Introduction

The evolution of the human diet has long been a core research topic in many archaeological and palaeoanthropological studies. Although it is generally assumed that Homo sapiens subsisted on a varied diet throughout its history, studies related to earlier hominids do in fact provide controversial data. Analyses of the stable isotopes of carbon and nitrogen carried out in the 1990s reinforced the idea of nutritional rigidity, especially in Neanderthal fossils. These techniques were primarily applied to hominids who lived in northern latitudes under cold climatic conditions, as these environments seemed to preserve collagen better than temperate ones. The considerable amount of dietary uniformity revealed by these studies cemented the view of Neanderthals as top predators and consumers of large game meat (e.g., Richards et al., 2000, Bocherens et al., 2005, Thorp and Sponheimer, 2006). However, this approach had a strong methodological limitation, as the lack of specimens from warmer periods and more southern latitudes may skew the overall dietary picture (Hardy, 2010). Adaptation to the local environment is usually observed in the diet of hunter-gatherers, which tends to fit the biogeographic region where they live (e.g., Ho et al., 1972, Gould, 1981, Tanaka, 1998). Thus, geography and climate seem to have played an important role in nutrition, and therefore analyses must consider the ecological conditions in which past communities lived. Refined techniques made it possible for stable isotope studies of carbon and nitrogen to be extended to southern European specimens and to address this gap. The results obtained by Salazar-García et al. (2013) in some Mediterranean sites showed similar isotope values to those living in colder environments, implying again, a diet that was predominantly based on terrestrial foods. Similarly, other data from Middle Pleistocene localities in south-eastern France, such as Payre, suggest the same conclusions with a clear preference for large herbivores in the Rhône Valley (Ecker et al., 2013). However, Hardy (2010) highlights the risk of relying on δ¹⁵N alone to define nutritional habits, since isotope-based analysis on bulk collagen represents only the protein contribution to the diet, which is rapidly dominated by meat protein —thus, the relationship between the dietary meat ratio and δ¹⁵N in bone collagen is nonlinear (Bocherens, 2009). Such analyses should always be complemented by data from other disciplines for a complete dietary picture.

Dental (macro- and micro-) wear studies have also shed light on the human diet, suggesting a more complex diet during the Middle and early Late Pleistocene. Lalueza-Fox et al. (1996) analysed the non-occlusal microwear pattern of a sample of 153 M teeth from various modern hunter-gatherer, pastoralist, and agriculturalist groups with different diets, and compared them with 20 Middle and Late Pleistocene fossils, mainly from Europe. The results indicated that Neanderthal specimens in the sample had a pattern close to that of modern carnivorous hunter-gatherers (such as Inuit and Fueguians), while archaic Homo sapiens seemed to have a more abrasive diet. In spite of this, they also noted certain variability in the Neanderthal specimens from Saint-Césaire, Marillac, and Malarnaud, which fell within the distribution of mixed diet hunter-gatherers. The notion of Neanderthal nutritional homogeneity was phased out in subsequent microwear studies by increasing the number of samples and their geographic diversity, leading Pérez-Pérez et al. (2003) to propose higher intra-group variability among Neanderthal deme specimens. These authors presented data that positioned the dietary habits of Neanderthals as highly variable; their nutritional strategies adapted and changed according to environmental circumstances, not unlike the way populations of anatomically modern humans behave. More recently, Fiorenza et al. (2011) applied Occlusal Fingerprint Analysis to 19 Neanderthal specimens (Krapina, Tabun, Monsempron, Shanidar, Amud, Vindija, and Le Moustier) and 12 anatomically modern humans (Qafzeh, Mladec, Barma Grande, Sungir, and Pataud) to establish analysis groups based on the eco-geographical context and the palaeoenvironmental information of the region. The results pointed to a dietary variation dependent on geographical location but independent of the taxonomic attribution of the fossils studied. Patterns of wear in the fossils were compared with existing groups of hunter-gatherers, demonstrating diets that were rich in meat for fossils from the coldest areas (Le Moustier, Monsempron, and Vindija) and greater nutritional diversity in fossils from Mediterranean habitats (Shanidar, Tabun, and Amud). The occlusal microwear texture analyses developed by El Zaatari et al. (2011) and Hlusko et al. (2013) reinforce this view and suggest that wear patterns are consistent with a diet rich in meat as well as plant foods in accordance to palaeoecological conditions. These data are not inconsistent with the isotopic signature indicating that meat provided the bulk of the dietary protein, but they demonstrate the inclusion of available plant foods.

Studies of vegetal remains are a growing body of work that sheds light on the inclusion of alternative food sources in the diet of pre-modern humans (Hardy and Moncel, 2011, Henry et al., 2011, Henry et al., 2014, Hardy et al., 2012, Hardy et al., 2013, Hardy et al., in press). Some of the earliest indirect evidence comes from the Acheulian site of Gesher Benot Ya'aqov in Israel, where seven species of nuts and pitted stones for nut cracking were recovered (Goren-Inbar et al., 2002). Macrobotanical remains in the form of charred seeds have been found in sites of subsequent periods, such as Kebara Cave in Israel (Lev et al., 2005) and Gorham's Cave in Gibraltar, which include cone scales or nutshells derived from stone pine (Pinus pinea) (Ward et al., 2013). But the most direct evidence for use of plants comes from studies of residues on stone tools (Hardy and Moncel, 2011, Henry et al., 2011, Henry et al., 2014) and dental calculus, which contain trapped phytoliths and starch grains (Henry et al., 2011, Hardy et al., 2012). The results presented by Henry et al. (2014) suggest that both Neanderthals and early modern humans from Europe and the Near East ate a similarly wide array of plant foods, including those considered low ranked, such as underground storage organs (USO) and grass seeds. In the case of Shanidar, Neanderthals seem to have cooked and consumed grass seeds from the Triticeae tribe (Henry et al., 2011).

Animal bones provide another important source of dietary information. In fact, most studies of paleodiet come from zooarchaeology, which by virtue of its long history has generated a significant body of data allowing researchers to draw behavioural inferences related to the evolution of human subsistence strategies. The recovery of numerous herbivore remains associated with lithic tools has led to focused research on the dietary contribution of ungulates (e.g., Stiner et al., 2000, Patou-Mathis, 2000, Valensi and Psathi, 2004, Gaudzinski-Windheuser and Roebroeks, 2014). Studies of large mammal assemblages have shown variation based on the eco-geographical location of the archaeological sites and the climate (Lorenzen et al., 2011) —a situation that reflects the hominid capacity to adapt to different landscapes and environments (e.g., Blasco et al., 2013a). An example is that in settlements of northern latitudes reindeer are normally abundant, while in southern latitudes deer and horses tend to dominate; other large ungulates such as rhinos, aurochs, bison, and even elephants also appear in the prey spectrum, highlighting the potential caloric input of these animals as a significant component in human subsistence (e.g., Stiner, 1994, Gaudzinski and Roebroeks, 2000, Gaudzinski-Windheuser and Roebroeks, 2014, Pike-Tay and Cosgrove, 2002, Rosell et al., 2012, Blasco et al., 2013a, Valensi et al., 2013). In contrast, settlements located in rocky and mountainous areas are dominated by animals belonging to these environments, such as tares, mouflon, ibex, and chamois (Yravedra and Cobo-Sánchez, 2015). Nevertheless, Yravedra and Cobo-Sánchez (2015) emphasise that some archaeological sites with favourable ecological conditions for ibex and chamois do not show the exploitation of such animals, indicating behavioural complexity and large versatility among human groups. According to Stiner (2013), only unique aspects of specific localities made other options of foraging feasible beyond ungulates. For this author, the economic differences between pre- and Upper Palaeolithic societies would be more apparent in how these cultures fill gaps in the large game availability with small animals.

There is growing evidence for the consumption of small game and even the use of marine resources in specific eco-geographical regions of the Middle and early Late Palaeolithic (e.g., Blasco, 2008, Stringer et al., 2008, Blasco and Fernández Peris, 2009, Colonese et al., 2011, Cortés-Sánchez et al., 2011, Hardy and Moncel, 2011, Blasco and Fernández Peris, 2012a, Blasco and Fernández Peris, 2012b, Cochard et al., 2012, Hardy et al., 2013, Blasco et al., 2013a, Blasco et al., 2014a). Although these data remain limited by comparison with previous results on large game, they raise questions about behavioural variation and its evolutionary significance in the larger scale. From this perspective, certain advances are modifying the traditional view of ancient diets being based exclusively on large game: studies of tooth microwear, analytical methods designed to ascertain vegetable components in the diet, and the discovery of localities where small game seems to have played a role as a supplementary meat source. Perceptions of paleodiet are becoming more nuanced and complex. In line with this evidence, tortoise specimens, the topic of this paper, may highlight variation in resource use in the Pleistocene, allowing questions about prey choice and hominin's adaptability to be explored.

Testudine bones are common in a number of Pleistocene sites. The use of turtles and tortoises for food seems to go as far back as to the Early Pleistocene. Several studies of aquatic remains from the site of FwJj20 in northern Kenya revealed anthropogenic damage on aquatic animals, including catfish and turtles (Braun et al., 2010, Archer and Braun, 2013, Archer et al., 2014). Archer et al. (2014) argued that the aquatic fauna may have been a highly nutritious resource for hominids in the Turkana Basin about 1.95 Ma, becoming a valuable nutritional alternative to an exclusively terrestrial resource base. Aquatic resources, among them turtles, could have offered specific advantages over terrestrial food, including a reduction in the required investment for the procurement, a decrease in the technological costs, and an associated reduction of predation risk (Archer et al., 2014). Knowledge of the ethology of these animals could lead to their classification as an easy target. Turtles are particularly vulnerable to predation when they are mating at the surface of the water and/or nesting on the beach, because they are immobile during the course of these episodes (Frazier, 1980). Female turtles pursue seasonally specific nesting grounds, and more than a quarter of a given population of females may reproduce in a given year, making a substantial proportion of individuals vulnerable to predation at these times (Congdon et al., 1983, Archer et al., 2014). Stewart (1994) maintained that aquatic resources equal or exceed terrestrial resources in importance in hunter-gatherer diets, especially in stressful seasons. For example, Tanaka (1976) documented that plants and other non-mammalian food compose between 81% and 96% of the San diet. These ethnographic data were used by Stewart (1994) to propose that, although the subsistence strategies cannot be generalised to all the environments and behavioural patterns of Pleistocene hominids, aquatic animals may have been alternative (and nutritious) food resources, especially during periods when other meat resources may have been of poor quality.

Several Eurasian Early Pleistocene sites also preserve testudine remains; however, we do not know the taphonomic history of most assemblages, partly because most such studies are focused on large- and medium-sized mammals. Nevertheless, anthropogenic damage has been described at Sima del Elefante (Atapuerca, Spain), where 8 tortoise (Testudo hermanni) carapace fragments from levels TE11 and TE14c showed incisions and scraping marks on their internal surfaces (Blasco et al., 2011). The use of this type of small game was interpreted as an indication of generalist behaviour among these early European hominids, demonstrating the capacity of these groups to adapt to the environment. In the case of the Acheulian site of Gesher Benot Ya'aqov (Israel), a small sample of freshwater turtles (Mauremys caspica)—3 carapace plates corresponding to 2 individuals—was recovered following a similar distribution pattern to that determined for the rest of anthropogenically processed ungulates (Hartman, 2004, Alperson-Afil et al., 2009). Although its distribution reveals no distinct patterns, no hominid damage was described on these bones (Rabinovich and Biton, 2011).

Tortoise specimens are common in South African archaeological sites ranging from Middle Pleistocene to the Holocene (e.g., Avery et al., 2004). They are found in both Middle Stone Age (MSA) and Later Stone Age (LSA) sites in very high densities (e.g., 49–304 specimens/m³ at Byneskranskop [Cruz-Uribe and Schrire, 1991]). However, they usually have been analysed from a palaeodemographic and palaeoenvironmental perspective and only rarely taphonomically. Tortoise assemblages from the Pinnacle Point Cave 13B and Blombos Cave are ones of the few MSA localities that have been studied using microscopic taphonomic methods (Thompson, 2010, Thompson and Henshilwood, 2014a, Thompson and Henshilwood, 2014b). Pinnacle Point Cave 13B involved 4418 tortoise specimens, including 8 individuals of the Angulate Tortoise (Chersina angulata) and 4 of the Areolated tortoise (Homopus areolatus), plus indeterminate tortoise remains. Thompson (2010) found that hominid damage such as cut marks (n = 27), percussion marks for opening the shell (n = 19), and preferential burning patterns from cooking were rare but present. Both shell and limb fragments seem to show similar proportions of burned and unburned fragments (24% and 17% burned, respectively), indicating that much of the burning on tortoise fragments was the result of post-depositional thermal alteration. In spite of this, some specimens showed burning concentrated on the external surface (n = 141 on the external; n = 35 on the internal), leading Thompson (2010) to suggest that cooking in the shell was occasionally used as a part of tortoise processing. On the other hand, Blombos Cave yielded 3191 tortoise remains (3190 attributed to Chersina angulata and only one to the Leopard Tortoise [Geochelone pardalis]). Anthropogenic alteration was detected in the form of burning signs (59.3% of all burned specimens were charred on the external surface of the shell); cut marks on the internal side of the shell and on limb bones (n = 44); and hammerstone marks frequently associated with cracks or fractures (n = 71) (Thompson and Henshilwood, 2014a). This evidence supports the dietary significance of this prey as a highly ranked resource in the subsistence systems during the MSA.

Evidence for the exploitation of tortoises during the European Middle Pleistocene is registered at several levels of Bolomor Cave (350-100 ka, Valencia, Spain), particularly at level IV, dated to >120 ka (Blasco, 2008, Blasco and Fernández Peris, 2012a, Blasco and Fernández Peris, 2012b, Blasco et al., 2013a). Level IV provided 526 tortoise (Testudo hermanni) bones with a minimum number of individuals (MNI) of 19. Most of the alterations reflect anthropogenic damage, including cut marks, percussion notches, impact flakes, and burning patterns, which indicate a clear association between hominids and this small prey. Taphonomic data on burned tortoise shells from Bolomor level IV are consistent with the common pattern described ethnographically by turning the animal upside down and roasting it in its shell (e.g., Werner, 1990). The consumption of testudines in the Iberian Peninsula appears more frequently in younger deposits, particularly in Mousterian contexts. An example is the archaeological site of Abrigo de la Quebrada (Valencia, Spain), where cut marks on the internal side of one carapace plate were documented at level IV, dated to MIS 3 (Sanchis et al., 2013), and the Abric del Pastor (Alicante, Spain), where the human processing of tortoises is evidenced by 7 cut-marked bones (costal and peripheral plates, humerus, radius and scapula –levels IVc-IVd), and a percentage ranging between 5.4 and 42.8 of charred specimens (levels III-IVd) (Sanchis et al., 2015). The late Mousterian levels of Gruta da Oliveira (Torres Novas, Portugal), dated between 70 and 35 ka, yielded 3394 tortoise (T. hermanni) specimens (MNI = 24). The highest proportion of fragments comes from layers 14 and 16, where they comprise almost half (44%) of the assemblage. At this site, 67% of tortoise bones showed evidence of burning, with a clear trend towards the external side; 4% displayed impact damage (e.g., impact flakes, percussion marks, and notches); and one shell fragment from layer 19 bore cut marks (Nabais, 2012). A different case was observed at Cueva de los Torrejones (Guadalajara, Spain), assigned to MIS 3. This site yielded a hyoplastron with several incisions on its external surface, leading Arribas et al. (1997) to propose the development of butchery processes involving the dismemberment of the tortoise forelimbs. Nevertheless, other possible scenarios cannot be ruled out, because cut marks are located on the exposed surface of the shell without a clear link to the removal muscles.

Additional suggestions for the exploitation of tortoises and turtles have been made for a series of Middle Palaeolithic sites in Italy, such as Grotta Breuil and Grotta dei Moscerini (Stiner, 1994, Stiner et al., 2000). Moscerini, from around 115–110 to 65 ka, is notable because two species of testudines, one terrestrial (Testudo graeca) and one aquatic (Emys orbicularis), were utilised by human groups (Stiner, 1994). Testudine remains bearing anthropogenic damage come mainly from level group M6 and comprise a total of 39 fragments (MNI = 5). Shells and limb bones tend to have been heavily broken up while fresh; and many carapace fragments show evidence of battering with a stone hammer, such as impact cones and percussion depressions on external surfaces of the carapaces or very near break edges. However, there is no distinctive pattern of differential burning (Stiner, 1994).

The Near East shows an abundant dataset of tortoises (Testudo graeca/Testudo cf. graeca) during the Middle and Early-Late Pleistocene; however, no cut marks have been described thus far (Stiner, 2005, Yeshurun et al., 2007, Speth and Tchernov, 2002, Stiner et al., 2009, Stiner et al., 2011). At Misliya Cave (approximately 200 ka [Valladas et al., 2013] in Mount Carmel, Israel), a small fraction of small mammals and tortoises was documented, specifically 50 T. graeca bones (MNI = 4; Yeshurun et al., 2007). This assemblage of Spur-thighed Tortoises is a small proportion of the faunal assemblage, which consists almost exclusively of high-ranked ungulate prey. This led Yeshurun et al. (2007) to support the idea proposed by Stiner et al., 1999, Stiner et al., 2000 that no overexploitation of the environment has taken placed in early Middle Palaeolithic populations in Levant and there was no need to frequently broaden the diet in order to add other alternative resources (e.g., small game). Nevertheless, Yeshurun et al. (2007) were cautious in their interpretation, noting that other aspects of the early Middle Palaeolithic habitation of Misliya Cave do not necessarily agree with the small-game-based interpretation of Stiner et al., 1999, Stiner et al., 2000. An example is the very well-preserved hearth over 20-cm thick, which indicates repeated burning in the same place and intensive occupation (Yeshurun et al., 2007).

In the case of Hayonim Cave (Galilee, Israel), dated to 230–140 ka (Mercier et al., 2007), an interesting view on the exploitation of the tortoise is offered from burning and breakage documented in a large assemblage (n = 5212) (Stiner, 2005). Tortoise assemblages seem to be the result of anthropogenic accumulations, with a relative abundance of damage linked to the prey having been roasted on coals and cracked open with stone hammers to access viscera contained inside the shell. Stiner (2005) describes high rates of impact modifications on peripheral bones of the carapace (14% of the shell edge elements), suggesting that shells were set on one edge and struck on the opposite. Beyond this pattern, many limb bones appear to have been ripped or torn away from the carcass according to fractures observed on the proximal ends of the humeri (13% at Layer E). On the other hand, burning on both sides of shell elements suggests that the tortoises of Mousterian layers of Hayonim Cave were frequently deposited in the hearth ash (22% at Layer E). Although the use of fire to cook is proved at the site, few remains were burned to the point of calcination, indicating that bones were either discarded in live fire beds or that post-depositional processes such as trampling destroyed most calcined bones (Stiner, 2005).

Speth and Tchernov (2002) also suggested the human use of Spur-thighed Tortoises for food at Kebara Cave (Mount Carmel, Israel) between about 60 and 48 ka. These authors studied a total of 2345 bones from Middle Palaeolithic levels, most of them from the midden (units IX–XI), where carnivore activity was almost absent. Differential burning on tortoise shell parts revealed valuable information on cooking methods. According to Speth and Tchernov (2002), burning was most frequent on fragments of the carapace (10.2%), compared to 4% for plastral elements and 0.9% for limb bones. In addition, 53.7% of the burned carapace remains were altered exclusively on the exterior surface and 39.5% were mainly on peripheral plates. This pattern was interpreted as the product of cooking rather than accidental exposure of discarded remains to fire, demonstrating that most tortoises were cooked upside down in their shell—the same way that was also described above in level IV of Bolomor Cave in Spain (Blasco, 2008) and Pinnacle Point Cave 13B and Blombos Cave in the Western Cape of South Africa (Thompson, 2010, Thompson and Henshilwood, 2014a, Thompson and Henshilwood, 2014b).

Qesem Cave (Israel), dated at 420–200 ka (Gopher et al., 2010, Mercier et al., 2013, Falguères, in press), is another archaeological site in the Levant preserving a small sample of tortoise specimens (Stiner et al., 2009, Stiner et al., 2011, Blasco et al., 2014b) and is the focus of our study here. Stiner et al., 2009, Stiner et al., 2011 documented a total of 88 fragments throughout the stratigraphic sequence, of which 17 bore burning damage (19%). Although the degree of burning was consistently low, it was more frequent on the plastron than on the carapace or limbs. This bias led Stiner et al. (2011) to suggest the same cooking of these animals that was observed in the early Middle Palaeolithic assemblage from Hayonim Cave (Stiner, 2005). Our intention here is to complete the taphonomic study of the tortoise bones from Qesem Cave by presenting new data from the lower stratigraphic sequence, specifically from two particular contexts: one related to the hearth area (dated at around 300 ka; Shahack-Gross et al., 2014, Falguères, in press) and another coming from one deeper sedimentary deposit located under the rock shelf (>300 ka). These data go beyond burning marks, including cut marks and percussion, and may help understanding the presence of tortoises and their potential role as small prey at Qesem Cave. Tortoises may provide insight into the subsistence behaviour of individuals within the Acheulo-Yabrudian Cultural Complex (AYCC) and shed light on other possible socio-economic aspects that led to the inclusion of this type of resource in the diet.