Data sharing and publication

One of the key scientific contributions of this study centers on the online publication of well documented, standards-aligned, and “cleaned” (edited) datasets which can be used to facilitate replication of the analytical claims made in this paper and provide a foundation for future research further refining our understanding of the spread of Neolithic technologies. The eighteen primary participants in the project agreed to publish their full faunal databases online in the open access, peer reviewed data publishing system Open Context (http://opencontext.org). In keeping with current norms in scientific data curation practices, Open Context archives datasets with the California Digital Library, a university digital repository. All datasets received DOIs (persistent identifiers) and specific metadata documentation to help make them citable resources [53]–[67] (for links to data see Table S1; File S1). These edited datasets can be used to replicate analytical claims made in this paper and in future studies [68]. Details of the approaches used in data cleaning, documentation, semantic annotation and collaboration with data contributors are described elsewhere [69] (see Text S1).

Zooarchaeological Methods

From these shared databases we report on analyses of three combined datasets including 1) species frequencies, 2) biometrics, and 3) survivorship or age at death data for Ovis aries/orientalis, Capra hircus/aegagrus, Bos taurus/primigenius and Sus scrofa. The relative abundance of these taxa was calculated based on NISP (number of identified specimens) identified to the genus level. Because the assemblages used in the study differ in size, were recovered using different methods and were analyzed by researchers utilizing different practices regarding the level of taxonomic identification, we focused on the frequencies of the most abundant large mammal taxa.

Biometric data were recorded following standard protocols [70]. Given the small size of some assemblages, in order to compare breadth and depth measurements from multiple skeletal elements we use the Logarithmic Size Index (LSI) as a generalized measure of body size [71]. The LSI is calculated by transforming measurements from archaeological specimens into their common logarithms and then taking the difference between the resulting value and the logarithm of the same measurement from a standard animal. For our standard animals we utilize four published and widely used individuals including 1) a recent wild female mouflon (Ovis orientalis) from Iran [72]; 2) average measurements of a recent male and female bezoar goat (Capra aegagrus) from the Taurus mountains [72]; 3) the Ullerslev cow aurochs (Bos primigenius) from Mesolithic Denmark [73]; and a recent female wild boar (Sus scrofa) from eastern Turkey [74] (for LSI data see File S2).

Mean values and standard deviations were calculated for each taxon (Ovis, Capra, Bos, Sus) from each phased assemblage based on the LSI and were statistically compared using an ANOVA and Tukey Honestly Significant Difference tests (see Tables S3–S10). Although differences in the allometry of ancient populations and the standard animal are a source of variation using this method, LSI is a widely used and effective way to identify broad changes in body size at the population level [14], [20], [71], [75]–[77].

Survivorship data were calculated based on the state of fusion of the epiphyses of long bones. Incorporating age data based on tooth eruption and wear proved difficult as a result of multiple and idiosyncratic recording methods used by various database authors and were therefore not incorporated into this analysis. In particular, we focus on the statistic of % juvenile which represents the frequency of specimens with unfused long bone epiphyses out of the total sample of fused and unfused specimens. Although long bone epiphyses fuse at a range of ages in bovids and suids (from c. 0–3 months to 48+ months [78]–[80]), the majority of specimens represent skeletal parts that fuse between 12–24 months. The statistic % juvenile is, therefore, a useful general proxy for monitoring changes in the frequencies of skeletally mature versus immature individuals in an assemblage and therefore age of slaughter. Differences in the % juvenile among assemblages representing wild and domestic populations (as evidenced by biometric data and the work of previous authors; see Fig. S1) were statistically compared using Yates Chi Square tests corrected for continuity for each taxon. In addition, Spearman rank order correlation coefficients (r_s) were calculated in order to address the relationships between % juvenile and %NISP, % juvenile and mean LSI, and between % juvenile and time (years cal BC) for each taxon (Table S11).

Identifying domesticated animals

The domestication of livestock in Neolithic SW Asia took place over several millennia and involved a combination of cultural and biological processes including the application of intentional management strategies (e.g., control over movement, foddering, young male culling) and the reproductive isolation of managed herds from their wild ancestors [81]–[83]. These processes, which likely began in the tenth millennium cal BC, resulted in the appearance of ‘domesticated’, phenotypically distinctive, husbanded animal populations by the mid to late ninth millennium cal BC in the Fertile Crescent region [14], [17]. In central and western Turkey, where domesticated animals first appear in the eighth and seventh millennia cal BC, respectively [19], [76], [84], three lines of evidence are widely employed to distinguish the management of domestic livestock populations from the hunting of wild populations including 1) changes in body size; 2) demographic evidence for intentional herd management; and 3) appearance of taxa outside of their natural range. Although identifying the nature of the earliest Neolithic animal management strategies in central Turkey still poses considerable challenges [19], these lines of evidence provide a clear picture of the westward spread of domestic animals in the later Neolithic (i.e., seventh millennium cal BC).

Measurements of the breadth and depth of long bone epiphyses and some podial bones (e.g., talus, calcaneus) can be used to identify the decrease in body size and reduction in sexual dimorphism commonly associated with the later stages of the domestication process in sheep, goat, cattle and pig [14], [77], [85]–[87]. Although the relationship between body size, animal status (wild versus domestic), and human management is complex, especially in the early stages of the domestication process [88]–[90], change in body size is a useful proxy for identifying the spread of domesticated animal populations especially in the later Neolithic when they exhibit significant phenotypic changes compared to their wild ancestors (Fig. 3) [91]. In this study the term ‘domesticated’ is used to refer to animals exhibiting phenotypic changes associated with long-term, intensive human management. It is therefore distinguished from the long-term ‘domestication process’, during the early stages of which phenotypically wild animals came under human management. Biometric data from earlier Neolithic sites in SE Turkey representing morphologically wild populations of sheep, goat, aurochs and boar were used to provide a baseline for identifying changes in body size in later central and western Anatolian Neolithic populations. Although it is acknowledged that it would be preferential to use local wild populations in each subregion of Anatolia as a benchmark for defining size change [19], this is not possible with currently available datasets.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Changes in mean size through time for Ovis, Capra, Bos, and Sus (based on LSI of breadth and depth measurements)(for data see Tables S3–6).

Vertical lines represent standard deviations. Colors reflect geographic location of site (after Fig. 1). Values to the left of the vertical axis represent means for each region. Key sites are labeled (after Fig. 1).

https://doi.org/10.1371/journal.pone.0099845.g003

Animal husbandry regimes frequently involve the systematic culling of specific demographic groups within a herd [87]. These practices, which often target young males not needed for herd reproduction, result in higher frequencies of juveniles in archaeofaunal assemblages compared to most hunting practices and can therefore be used to identify systems of animal husbandry [92]. This is particularly true for sheep and goats where it has been used to identify early husbandry strategies [15], [88] but is also relevant for cattle and pigs—although swine management strategies exhibit more variability due to the multiparous nature of this species [93].

Finally, the appearance of taxa outside of their natural habitat may suggest that they were transported under conditions of human management [87], [94]. The natural range of wild sheep and goats includes the mountains and plains of eastern, south-central, and southwestern Turkey [95]. Western and NW Turkey are outside of this range and the appearance of these taxa in these regions indicates that they were introduced as domesticates. This line of evidence is less useful for Sus and Bos, which have less restricted early Holocene ranges. Wild boar is currently found in every province of Turkey and aurochs are expected to have had an equally wide distribution in the early Holocene [51], [76]. In combination, biometric evidence for size change, increased juvenile culling, and the appearance of taxa outside of their natural habitat provide a reliable means to identify domestic animals and animal husbandry in Neolithic archaeofaunal assemblages in central and western Anatolia dating from the eighth to the sixth millennia cal BC.

Regionalization and species preferences

As animal economies based on the management of domestic livestock moved outside of the Euphrates and Tigris basins where they were characterized by highly variable combinations of caprines (i.e., sheep [Ovis aries] and goats [Capra hircus]), cattle (Bos taurus), and pigs (Sus scrofa) (Fig. 2; Table S2), they took on regional characteristics in response to local of ecogeographic conditions and cultural preferences [8], [9], [19], [76], [96]. The earliest assemblages examined in this study indicate that wild sheep and goats were intensively targeted by Epipaleolithic hunters in the Taurus mountains at Karain B, Öküzini and Direkli Caves. In central Turkey the earliest Neolithic economies include large numbers of wild boar, as seen at Pınarbaşı A and Boncuklu [25]. However, with the exception of the site Musular [97], where aurochs were heavily exploited [75], all later Neolithic animal economies in this region focused intensively on caprines (central Turkey mean_caprineNISP = 87%) with modest amounts of Bos and very little Sus (Fig. 2; Table S2). In the Lakes region and western Turkey, although caprines are abundant, pigs and cattle represent larger portions of the animal economies (Lakes region mean_Bos+SusNISP = 32%; western Turkey mean_Bos+SusNISP 34%)(Fig. 2;Table S2). This is also the case in southern Turkey where cattle and pigs together nearly match caprines in abundance (South Turkey mean_Bos+SusNISP = 48%) (Fig. 2;Table S2). Finally, in the NW region cattle increase to their highest levels (NW Turkey mean_BosNISP = 29%)—a trend that has been linked with an increase in the consumption of dairy products [51], [98].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Relative frequencies of OC (Ovis aries/orientalis+Capra hircus/aegagrus), Bos taurus/primigenius, and Sus scrofa in Neolithic assemblages (see Table S2 for data).

Sites labeled and colored after Fig. 1. X's represent mean values for each region.

https://doi.org/10.1371/journal.pone.0099845.g002

Biometric evidence for domesticates

Biometric data has long been used to identify changes in phenotype associated with the process of animal domestication [85], [87], [93]. In Neolithic Turkey, biometric data for sheep and goats each show a sharp decrease in body size (compared to late Pleistocene/early Holocene wild caprines), which is first evident in the second half of the eighth millennium cal BC at the site of Çatalhöyük, central Turkey (Fig. 3; Tables S3–S4; for statistical comparison see Tables S7–S8). This represents the first appearance of domesticated (i.e., exhibiting a domestic phenotype) caprines in Turkey outside of the Fertile Crescent region. Although not evident at the early Neolithic site of Suberde, small-sized, domesticated sheep and goats appear in the Lakes region and in western and southern Turkey in the early seventh millennium cal BC and in NW Turkey in the mid to late seventh millennium cal BC (Fig. 3; Tables S3–S4).

In the Lakes region the mean size of goats at Erbaba, Bademağacı, and Höyücek is notably larger than that for contemporary sites in central, western and NW Turkey, while for sheep the Erbaba and Çatalhöyük West assemblages are significantly larger than all other later Neolithic assemblages and even overlap with the wild caprines from late Pleistocene Karain B and Öküzini caves (Fig. 3; Table S3–S4, Tables A–E in File S1). All of these sites are located within the natural habitats of wild sheep and/or goats, and it is possible that this pattern reflects hunting and/or interbreeding with local wild caprine populations by Neolithic agro-pastoralists in these regions.

For Bos, a dramatic decrease in body size compared to Anatolian aurochs is evident in the early seventh millennium cal BC in the Lakes region at Bademağacı ENI and in western Anatolia at Ulucak VI (Fig. 3; Tables S5, S9). Small sized, domestic cattle appear later in central Turkey where they are only evident in the later levels (South P-T and TP Area) at Çatalhöyük dating to the mid to late seventh millennium cal BC. At the same time, domestic cattle also arrive with the earliest Neolithic settlements in NW Turkey where they exhibit the smallest mean size for any region [76], [91], [99].

Patterns of size change for suids differ from those observed in other taxa. Surprisingly, the earliest small-sized domestic pigs appear in western Turkey at Ulucak VI in the early seventh millennium cal BC and they are evident slightly later in the Lakes region at Bademağacı and at Yumuktepe in southern Turkey (Fig. 3; Tables S6, S10). Our data indicate that domestic pigs were never incorporated into Neolithic economies in central Turkey, where Sus remains are rare and comparable in size to early Holocene wild boar into the fifth millennium cal BC. The only exception to this comes from two specimens from the latest deposits (West Mound) at Çatalhöyük that are within the size range of Neolithic Anatolian domestic pigs (Fig. 3; Table S6). Given the extremely low frequency of Sus in these levels (<1% total NISP) and the lack of juveniles, these specimens may represent gifts or imported oddities rather than evidence for onsite pig husbandry. Alternately, it is also possible that they represent particularly small wild boar or intrusive specimens representing domesticates from later periods. Domestic pigs are also notably absent in the earliest Neolithic settlements in NW Turkey including Ilıpınar X, Barcın VI, Fikirtepe, Menteşe ancien, and Hoca Çesme. However, they appear suddenly in the NW after 6000 cal BC as seen at Ilıpınar IX, Menteşe récent, and Yenikapı.

Age at death

Age at death data combined with biometrics (Fig. 4) provide strong supporting evidence that Neolithic bovids and suids exhibiting a domestic phenotype were under intensive human management. Neolithic assemblages in which small sized domesticates are present exhibit increased juvenile culling compared to those dominated by phenotypically wild bovids and suids—a pattern that fits the expectations of models for early herd management [14], [87], [100], [101] (Table S11). For Ovis, Bos and Sus the frequency of juveniles is significantly higher in assemblages with small-sized domesticates compared to those representing phenotypically wild, hunted populations, clearly reflecting the effects of intentional human management strategies (Chi-square test: Ovis, X² = 21.64, df = 1, p<0.0001; Bos, X² = 6.38, df = 1, p = 0.0115; Sus, X² = 25.06, df = 1, p<0.001). For example, in central Turkey the first appearance of domestic cattle in the mid seventh millennium cal BC corresponds with an increase in the frequency of juvenile individuals from 13% to 35%. As a result of the intensive culling of juvenile goats in some Epipaleolithic and early Neolithic (PPNA, EPPNB) assemblages, Capra is the exception to this pattern showing no change in the frequency of juveniles among assemblages representing domestic versus wild animals (X² = 0.3, df = 1, p = 0.5839).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Plots showing relationship between body size (mean LSI) and %juvenile (based on long bone fusion) for Ovis, Capra, Bos, and Sus (for data see Tables S3–6).

Colors reflect geographic location of site (after Fig. 1). “W” indicates assemblages representing wild populations. Key sites are labeled (after Fig. 1). For Capra, dark blue marks represent Zagros sites Asiab, ZC Shanidar, and Shanidar Mousterian.

https://doi.org/10.1371/journal.pone.0099845.g004

More generally, Ovis and, to a lesser extent, Bos both exhibit steady increases in the frequencies of juveniles over time across all regions (Fig. S1). Assemblages with the highest frequencies of domestic sheep (those in central Turkey) and domestic cattle (those in NW and western regions) also exhibit the most intensive juvenile culling (Tables S2, S3, S5, S11). For Capra and Sus, which show more demographic variability in assemblages produced by hunting, there are no overall changes in the representation of juveniles over time, although assemblages with domestic pigs exhibit more intensive juvenile culling (mean_juv = 51%) than those representing wild boar (mean_juv = 32%) (Fig. 4; Fig. S1; Tables S2, S4, S6, S11).

Chronological trajectories for the spread of domestic animals

The patterns emerging from the shared datasets used in this project suggest that the movement of domesticated animals and husbandry techniques westward across Neolithic Turkey followed several different trajectories, clearly indicating that livestock did not move as part of a standard agro-pastoral ‘package’ [50]. Although domestic sheep and goats appear first in central Turkey by the mid eighth millennium cal BC, domestic cattle appear a millennium later and domestic pigs are never incorporated into Neolithic economies in this region [76]. The combination of domestic caprines, cattle and pigs occurs for the first time in Turkey, outside of the SE, in the early seventh millennium cal BC, where all four livestock species are evident in the Izmir region at Ulucak VI and then slightly later in SW Turkey at Bademağacı ENI, and in southern Turkey at Yumuktepe [102] (Fig. 1). Since a completely autochthonous domestication event in western Turkey is unlikely, this pattern suggests a rapid westward movement of domestic animals across southern Turkey [4]. The geography and timing of this early expansion suggest it may have followed a coastal route, either by land or by sea [20], although the initial phases of this process in southern Turkey are currently poorly documented [40], [103]. However, the absence of domestic cattle and pigs in central Turkey at this time suggests that this region was not directly involved in the earliest spread of domestic livestock either to the Lakes region or western Turkey and reaffirms the distinctive nature of the Central Anatolian Neolithic tradition [104]–[106].

Finally, the earliest Neolithic communities in NW Turkey utilized domestic caprines and cattle—with a marked increase in the latter including a new emphasis on culling juveniles—but did not keep domestic pigs [8], [76]. This suggests that the origin of the Neolithic in the NW was not a simple extension of the same colonization processes that brought farming communities to SW and western Turkey but instead represents a distinctive process, perhaps involving different populations of farmers and/or interaction with local foragers [47], [49], [107]. The initial absence of pig husbandry at sites pertaining to the archaic Fikirtepe tradition in NW Turkey may reflect central Anatolian influences, which have been noted in the early Neolithic material culture in this region, although they may also reflect responses to local environmental and economic conditions [6]. In either case, the adoption of pig husbandry in the early sixth millennium cal BC in the NW suggests increasing interaction with neighboring groups along the Aegean coast where pig husbandry had been practiced for almost a millennium [4]–[6], [8], [76], [84].

The spread of domestic animals and Neolithic traditions

The broad patterns evident in the faunal data can be linked with recent interpretations of material culture evidence for the spread of Neolithic technologies in western and NW Turkey. It has been argued that the westward spread of farming across Turkey was characterized by multiple distinctive impulses [4]–[6], [42]. The first of these relates to the westward movement of an Aceramic tradition (contemporary with the Levantine Late PPNB) with Near Eastern features including red plaster floors, known primarily from Ulucak VI, Bademağacı and Aceramic Hacılar and dating to the late eighth and early seventh millennia cal BC [4], [6], [34], [41]. Subsequent impulses involve the movement of two distinctive ceramic traditions including first, the overland expansion of the so-called Dark Faced Burnished Ware (DFBW) tradition westward and into the Marmara region in the mid seventh millennium, followed by the slightly later movement of the Red Slipped Burnished Ware (RSBW) horizon along the coast and into interior SW Turkey [5], [6], [42], [44], [47], [107], [108].

The faunal patterns identified in this project fit broadly within these distinctive Neolithic material culture traditions. The movement of domestic caprines, cattle and pigs along the southern and western regions of Turkey evident at Ulucak VI, and slightly later at Bademağacı ENI and Yumuktepe corresponds with the initial Aceramic expansion of Neolithic lifeways and with the later RSBW horizon. The presence of convincing material culture parallels between the early settlements in these regions as well as similarities in their animal economies supports the presence of a distinctive late eighth-seventh millennium cal BC coastal, and perhaps inland, social network linking southern, SW and western Turkey [4], [5], [11], [37], [48], [107], [109].

In contrast, the movement into NW Turkey of a distinctive animal economy characterized by domestic caprines and cattle but the absence of domestic pigs may be associated with the spread of the distinctive DFBW tradition from the interior Anatolian plateau. Here similarities between the Dark Burnished wares from central Turkey and the archaic Fikirtepe tradition are supported by a shared absence of domestic pigs, a distinctive feature of both regions, as well as the shared exploitation of dairy products and other well documented material culture correlates [5], [6], [11], [76], [98], [110], [111]. In addition, domestic cattle first appear in central Turkey in the mid seventh millennium cal BC, approximately contemporaneous with their introduction into the Marmara region where they become the focus of the animal economy.

Incorporation of local wild fauna into domestic herds

Our data also shed light on processes that took place during the western spread of Neolithic domestic animals. In the Lakes region of SW Turkey, goats exhibit significantly larger mean sizes compared to Neolithic goats from other regions; the same pattern is evident for the sheep from Erbaba. Moreover, in central Anatolia, following the appearance of domesticates in the mid seventh millennium cal BC, cattle exhibit mean sizes larger than those for any other region of Anatolia examined in this study (including SW, West, South, NW). Neolithic settlements in central and SW Turkey are located within the natural habitat of wild sheep, goats, and aurochs, and it is suggested that these patterns in the biometric data reflect intensive interaction between Neolithic human populations and local wild fauna [32]. Although it is possible that the biometric data simply reflect the opportunistic hunting of locally available bezoar, mouflon, and aurochs, it is also possible that these patterns reflect the intentional (or accidental) incorporation of local wild animals into domestic livestock populations.

For cattle, it has been suggested that intentional breeding of domesticate cows with bull aurochs may have been a regular part of a ritual-laden management strategy at Çatalhöyük [112]. Moreover, recent paleogenetic work on suids in Turkey has shown that the distinctive maternal lineage of domestic pigs involved in the initial Neolithic expansion into Europe, likely derives from western/SW Anatolian boar populations [103]. This suggests that Neolithic communities in western Turkey incorporated local wild sows into their domestic herds which were then introduced into Neolithic Europe in the late seventh millennium cal BC [103], [113]. In light of this genetic evidence for the incorporation of local boar into domestic herds, combined with biometric evidence for interaction with wild sheep, goats, and aurochs, it can be hypothesized that SW (and perhaps also central) Turkey represents a particularly important area of interaction between humans, domestic animals and wild populations where the genotype and phenotype of Neolithic livestock populations were altered prior to expansions westward into Europe. Paleogenetic studies of caprines and also cattle from sites in this region dating to the eighth and seventh millennia cal BC are necessary to explore this issue further.