From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash (south-eastern Kazakhstan)

Michael Frachetti1 & Norbert Benecke2

Does the riding of horses necessarily go with the emergence of Eurasian pastoralism? Drawing on their ne sequence of animal bones from Begash, the authors think not. While pastoral herding of sheep and goats is evident from the Early Bronze Age, the horse appears only in small numbers before the end of the rst millennium BC. Its adoption coincides with an increase in hunting and the advent of larger politically organised groups. Keywords: Kazakhstan, Eurasian steppe, Bronze Age, prehistoric period, historic period, pastoralism, horses, herd structure

Introduction
Pastoralist economies have been the base of Eurasian steppe societies for millennia. Ethnographers such as Vainshtein (1991), Khazanov (1994) and Masanov (1995) (among others) have long argued that Eurasian nomadism was rst facilitated by the domestication and riding of the horse, while they also describe mixed herds of sheep, goats, cattle and camels as key to the socio-economic success of steppe pastoralists (generally Bareld 1993). There is no doubt that the domesticated horse had a transformative affect on the economic, socio-political and military development of Eurasian societies (Anthony 2007). However, we have surprisingly limited archaeological evidence to trace the emergence of horses as the predominant domesticate in herds managed by Eurasian nomads especially in the earlier phases of the regions pastoralist prehistory (i.e. 3000-1000 BC). Recently, the site of Botai (in northern Kazakhstan) has become central to debates of horse domestication (Olsen et al. 2006). Botai and related Eneolithic/Early Bronze Age sites in the north-central steppe date from roughly 3500-3000 BC and reveal faunal assemblages extremely rich in horse remains often more than 99 per cent of all fauna. These lopsided gures illustrate that horses were undoubtedly the most important animals for northern steppe populations from the mid-fourth to the third millennium BC. Currently, debates continue as to whether these horses were all or mostly wild, all or mostly domesticated, or
1 2

Department of Anthropology, Washington University in St Louis, 1 Brookings Drive, CB 1114, St Louis, MO 63130, USA (Email: frachetti@wustl.edu) Leiter des Referats Naturwissenschaften Arch aozoologie, German Archaeological Institute, Scientic Department of the Head Ofce, Im Dol 2-6, Haus II, 14195 Berlin, Germany (Email: nb@dainst.de)

Received: 7 October 2008; Accepted: 26 January 2009; Revised: 9 February 2009

ANTIQUITY

83 (2009): 10231037

1023

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash

some combination of the two (Levine 2004, 2005; Olsen 2006; Anthony 2007). Recently published data from Botai, demonstrate that considerable numbers of the Botai horses were domesticated and used for procuring secondary products, specically mares milk (Outram et al. 2009). Morphological divergence in metacarpal size between Botai horses and Palaeolithic wild horses, physical evidence of bitting, and horse milk residues on ceramics further illustrate that Botais horses were already being controlled and employed in numerous domestic contexts by 3500 BC (Olsen 2003). These data suggest that horses were likely to have been ridden at this time as well (Anthony 2007). The polemic surrounding the timing and nature of northern-steppe horse domestication and riding has left a key point overlooked in broader discussions of pastoralism and animal domestication on the steppe. Namely, the steppe economy generated by the so-called Botai horse-culture takes a radical shift at the start of the third millennium BC and by 2500 BC a decidedly different mobile pastoralist strategy dominates on the steppe. Specically, the base of mobile pastoralist societies at this time became more heavily dependent upon domesticated sheep/goat and cattle (Benecke & von den Driesch 2003). Early Bronze Age sites in both the western and eastern steppe zones provide solid evidence for the rapid adoption of a sheep/cattle-based herd structure in the mid-third millennium BC and later (Antipina 1997; Kalieva & Logvin 1997; below). Yet in spite of the fact that many Bronze Age settlements have been excavated across the steppe, few provide a long-term continuous record of the development of such regional pastoralist economies. As a result, we are left to imagine Bronze Age steppe societies dened by horse riding, thereby evincing extensive and efcient mobility as the mechanism underpinning their land use, regional interaction and cultural diffusion. Recent excavations at the pastoralist settlement of Begash, located in Semirechye (SE Kazakhstan) (Figure 1), provide well-stratied assemblages of domesticated animal remains spanning from the Early Bronze Age to later historical periods (c . 2500 BC-AD 1900). The start date for Begash is conservatively dated here to c . 2500 BC (calibrated 14 C). One 14 C sample from Begashs lowest levels provides limited evidence for an earlier date for the foundation of the site, likely at about 2900-2700 cal BC (rst sigma range 3100-2450 cal BC; Frachetti & Maryashev 2007). The faunal record from Begash provides insight into the pastoralist subsistence economy from that key period in the third millennium BC when sheep and cattle herding became dominant across the steppe and traces the comparative herd structure of pastoral communities over an extended timeframe. Furthermore, Begashs faunal evidence illustrates the inter-relationship between strategic land use, pastoralist mobility and the integration of the horse alongside evolving economic and social strategies in the Semirechye region over the past 4500 years. These data clearly show the instrumental role of sheep, goat and cattle in the subsistence and occupational behaviour of Bronze Age steppe pastoralists as early as the mid-third millennium BC which is signicantly earlier than previously attributed to mountainous regions of Inner Asia such as Semirechye (Kuzmina 2008). The faunal analysis at Begash demonstrates that sheep/goat have consistently dominated the archaeozoological assemblage (followed by cattle) and that horses never comprise more
1024

Michael Frachetti & Norbert Benecke

Figure 1. South-eastern Eurasia, Semirechye, and the location of Begash (contours at 1000 & 2000m asl) (map data source: ESRI, 1994).

than 14 per cent of the sample while this is limited to the later historical periods (i.e. postMongolian phases). Although horse remains were not highly represented in the domestic context, this may belie the degree to which they were used for purposes other than food. A variety of taphonomic processes, such as treatment of dead horses, kill patterns, or actual small herd sizes, all may have contributed to the lack of horse remains in this domestic context. Nevertheless, Begash does provide a well-dated and comparative view of horse remains in the pastoralist domestic setting and in relation to the other animals that formed the productive base of Eurasian steppe pastoralism. Also important, the recovery of wild faunal remains throughout the cultural contexts of Begash sheds light on the extent and nature of non-herding strategies among the regions largely pastoralist groups for which horses are known to have been used in prehistory (e.g. hunting at Botai). At Begash, there is a correlation between a slow increase in horses and evidence for increased ranges of hunting and political expansions of the social landscape. Signicantly, faunal data from Begash contradict the notion that the emergence of Eurasian pastoralism was sparked by the rapid domestication and riding of the horse. In the eastern steppe at least, horse use seems to commence gradually and is not highly associated with early and middle Bronze Age pastoralists. Instead, the evidence from Begash suggests that the key species within Eurasian pastoralist economies started with
1025

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash
Table 1. General chronology of habitation phases at Begash. Calibrated date range (after Frachetti & Maryashev 2007: 229) Phase 1a 1b 2 3a 3b 4 5 6 1 sigma 2460-1950 BC 1950-1690 BC 1625-1000 BC 970-400 BC 390 BC-AD 30 AD 70-550 AD 1260-1410 AD 1680-1900 2 sigma 3500-1890 BC 2300-1500 BC 1690-920 BC 1010-400 BC 400 BC-AD 130 AD 60-550 AD 1220-1420 AD 1660-1950

and remained sheep/goat and cattle, as early as 2500 BC. Thus horse riding whenever it actually began was not essential to successful mobile herd management during the early to middle Bronze Age and did not have dramatic and immediate political, social and military consequences in the formative stages of Eurasian pastoralism. At Begash and to the extent it reects contemporary sites in the eastern Eurasian steppe horses only started to impact pastoralists wider exploitation of regionally diverse wild resources at the end of the second millennium BC, and subsequently, in the Iron Age, can be associated with documented large-scale politically oriented mobility. These data are distinctive from those presented by Anthony concerning the impact of the horse on contemporary Bronze Age groups from the western steppe (Anthony 2007: 222). Thus, the data from Begash draw into question the still common view that Eurasian pastoralism diffused eastward as a result of mounted horsemen in the Bronze Age (Kuzmina 2008).

Site background
The settlement at Begash is located in eastern Kazakhstan (Semirechye) in the piedmont zone of the Dzhungar Mountains (950m asl). The site nestles on a at ravine terrace enclosed by steep canyon walls on the north, west and south, and situated along a spring-fed stream that today has water year-round. The geography of the site indicates that Begash was a seasonally occupied winter-time settlement for regionally mobile pastoralists. Begash was excavated in 2002, 2005 and 2006 as part of the joint Kazakh-American Dzhungar Mountains Archaeology Project (DMAP) (Frachetti & Maryashev 2007; Maryashev & Frachetti 2007). The chronology of pastoralist settlement at Begash is divided into six phases (Table 1) documented by 34 AMS dates. The earliest occupation (Phase 1a) is placed in the rst half of the third millennium BC, with subsequent construction phases in the middle Iron Age (c . 400 cal BC) and later historic period. Between these construction phases (dened by remnant foundations and walls) the site appears to have been occupied intermittently, still by pastoralist communities, with cultural contexts that expand horizontally across the small settlement terrace. Thus, the overall chronology and occupation
1026

Michael Frachetti & Norbert Benecke

of Begash is considered largely continuous, albeit reecting variation in the regularity of use and mobility of the pastoralists throughout the phases of the sites occupation (Frachetti & Maryashev 2007).

Methods and results

This paper presents the results of archaeozoological studies from the excavation seasons of 2005 and 2006. Faunal materials were also recorded in 2002, and the analysis of that seasons data is available elsewhere (Frachetti 2004; the material from 2002 represents less than 5 per cent of the total faunal sample collected from all seasons). During all years of excavation, faunal remains were carefully collected by hand in all trenches of the site and screening of 50 per cent of the excavated soil was performed through a 10mm screen. We also wet sieved and oated representative hearth and midden samples from nearly all stratigraphic contexts. The collection available from Begash consists of nearly 22 000 remains. For the largest part, they represent discarded refuse from slaughtering animals and preparing food. On the whole, animal bone preservation is good at this site. The remains are characterised by a high degree of fragmentation with only very few bones being complete. The mean bone weight is 3.2g. For this reason species identication was only possible on a small part of the collection. Nearly 84 per cent of the specimens remained taxonomically unidentied. About 16 per cent of the specimens of the total collection show traces of burning. The animal remains are unevenly distributed over the phases. The smallest collection comes from Phase 1a with only 578 specimens. The Phases 3b, 4 and 5 have provided the largest collections with more than 3000 specimens each. The faunal remains of Begash mainly represent bones and teeth from various mammal species and a few bird bones have been found. Fish, reptiles and molluscs were not recovered at all.

Domestic mammals
The bulk of the sample from Begash represents domesticated fauna (almost 95 per cent of the identied bones). This clearly points to a signicant role of animal keeping in the economy of the settlement throughout its occupation, especially for providing food and raw materials. Six domesticated taxa were identied: sheep, goat, cattle, horse, camel and dog.

Sheep/goat
Sheep and goats are most prevalent among the domestic animals from Begash (Table 2). Altogether 2043 teeth and bone nds were assigned to both species. Interestingly, there is a decrease in the relative frequency of sheep and goats over time, from about 75 per cent in Phase 1a to just under 50 per cent in Phase 6 (Figure 2). Throughout prehistoric periods, however, sheep/goat consistently dominated the herd structure. The difculty in distinguishing bones of sheep and goat in heavily fragmented assemblages from archaeological sites is well known and we also met this problem in the material studied.
1027

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash
Table 2. Faunal remains from Begash, quantied in terms of the number of identied specimens (NISP), according to phases. Archaeological Phase Species Domestic mammals Sheep/Goat (Sheep) (Goat) Cattle Horse Camel Dog Wild mammals Red deer (Cervus elaphus) Red deer, antler Goitered gazelle (Gazella subgutturosa) Argali (Ovis ammon) Siberian Ibex (Capra sibirica) Wild pig (Sus scrofa) Fox (Vulpes vulpes) Mustelid (Mustela spec.) Birds Eagl owl (Bubo bubo) Chukar partridge (Alectoris chukar ) Hoopoe (Upupa epops) Starling (Sturnus vulgaris) Unidentied specimens Total 1a 76 (3) 20 4 1 1 1 475 578 1b 293 (24) (5) 108 8 4 14 1 1 1 3 1 1929 2363 2 401 (41) (3) 158 24 1 6 8 1 2 10 1 2111 2723 1 654 771 1 2 2 1 3a 61 (3) (1) 37 3 1 6 2 3b 527 (77) (3) 132 45 1 1 31 2 3 2 1 2 8 1 1 4223 4980 4 326 (32) (1) 160 45 5 11 2 4 2 2 2897 3454 5 223 (20) (2) 109 55 6 2 13 17 1 1 3622 4049 6 136 (6) (5) 94 38 1 1 16 3 4 3 2235 2531

Therefore, distinction between sheep and goat could only be made for a limited number of bone nds. As can be seen from the data compiled in Table 2, remains of sheep clearly outnumber those of goats. On average, sheep are ten times as frequent as goats. Osteometric data point to sheep with estimated withers heights between about 0.60 and 0.80m. A metrical comparison shows that the sheep of the older phases were of smaller size than those of Phases 5 and 6. The majority of female sheep were hornless as respective skull fragments show. The sheep/goat remains are highly fragmented with many bones showing traces of butchering. Thus, we suggest they represent discarded refuse from slaughtering and food preparation. Element distribution analysis reveals that all major elements of the skeleton are present in each of the collection from the various phases. Due to the small sample size there is only limited data available concerning age structure and sex ratio in sheep and goats. The kill-off pattern of sheep/goat was evaluated on the basis of dental eruption and wear as well as on epiphysial fusion (according to Habermehl 1975: 11423). The epiphysial fusion data for sheep/goat are summarised in Figure 3. They show
1028

Michael Frachetti & Norbert Benecke

Figure 2. Percentages of sheep/goat, cattle and horse from Begash according to phases of occupation, based on fragment counts (NISP).

a shift to older individuals among the slaughtered animals from the oldest to the most recent phases. There is evidence of foetal as well as neonatal bones only in Phases 1b and 2. Sex determination on pelvic bones indicates the presence of both females and males in sheep.

Cattle
Cattle are the second most frequent species in the fauna from Begash (Table 2). Altogether 818 teeth and bone nds could be assigned to this species. The percentage of cattle increases from the oldest to the more recent phases (Figure 2). The few osteometric data available from Begash point to animals in the lower to medium size range of prehistoric cattle. The cattle bones are generally characterised by a high degree of fragmentation. Several bones show traces of butchering, likely discarded refuse from slaughtering and food preparation. Element distribution analysis shows that all parts of the skeleton are represented, though data concerning age structure and sex ratio in the cattle assemblage are limited. The epiphysial fusion data point to a dominance of sub-adult and adult animals among the cattle slaughtered
1029

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash

Figure 3. Epiphysial fusion data for sheep/goat from Begash according to phases of occupation 1-6 (symbols), based on fragment counts (NISP).

in all phases (ageing according to Habermehl 1975: 104). The percentage of juveniles is about 10 per cent. Sex identication on two pelvic bones proves the presence of females.

Horse
With the exception of Phase 1a, equid remains were recovered from all phases of occupation at Begash. As is widely known, species identication on tooth and bone remains of the genus Equus can be difcult in regions where more than one species occur. In the area and period under consideration the presence of four equids may be expected in the faunal assemblages i.e. wild horse (Equus ferus), onager (Equus hemionus), domestic horse and domestic ass. Teeth and bones that can safely be assigned to onager have not been found at all. The same is true for the domestic ass, which may be expected from the Iron Age onwards (Benecke 2003: 74). All equid remains clearly represent horses. As hunting was only of marginal importance in the site studied, the majority of the horse bones most likely belong to domestic horses. Albeit statistically smaller, horses represent the third largest group among the domestic animals from Begash (Table 2). The relative frequencies show a steady increase of this species through time, from 2 per cent in Phase 1b to 14 per cent in the Phases 5 and 6. Whether the lack of horses in the earliest phase of occupation (Phase 1a) is an artefact of the small size of the total faunal collection or was a reality remains an open question. The second half of the third millennium BC, which roughly corresponds with Phase 1a, is considered as the period when horse domestication ourished in Western Asia (Benecke & von den Driesch
1030

Michael Frachetti & Norbert Benecke

2003). Nevertheless, percentages of horse remains at Begash remain below 6 per cent until approximately AD 50 (Phase 3b). According to the few metrical data available from the collection, the horses seem to have been rather uniform in terms of body size and shape with estimated withers heights ranging between 1.30 and 1.45m. The slaughter-pattern of horses has mainly been evaluated on the basis of epiphysial fusion (Habermehl 1975: 48). The epiphysial fusion data indicate that the majority of the horses were full grown at the time of slaughter. Bones with un-fused epiphyses exist in low numbers (N = 7) making up no more than 18 per cent of the bones evaluated (N = 39). The ageing derived from the analysis of epiphysial fusion is conrmed by the few data available for dental eruption and wear.

Bactrian camel
Nine bones of the collection from Begash have been identied as camel (Table 2). Out of these ve well-preserved postcranial bones allowed a closer determination of the species in question (Bactrian camel or dromedary). According to the criteria described by Steiger (1990), all determinable bones could be assigned to the Bactrian camel. There is no evidence for the presence of dromedary. Bactrian camel, therefore, seems to represent the only camel species exploited by the Bronze Age and Iron Age people in the area of Begash. Whether all bones really come from domestic animals is an open question. The environmental conditions in the surroundings of the site as judged from the list of wild mammals and birds are not consistent with biotopes typical for wild camel and therefore seem to favour the assumption that the nds from Begash more likely represent domestic animals. Compared to sheep/goat, cattle and horse the number of camel bones is low. They count for less than one per cent of all domestic mammals. All bones belong to adult animals. In contrast to sheep/goat, cattle and horse, the camel bones do not exhibit typical butchering marks. This seems to indicate that camels have generally not been exploited for their meat. The primary use of this species must have been as an animal for transport.

Dog
Domestic dog is represented by 19 bones and a partial skeleton. The percentage of dog counts for less than one per cent of all domestic mammals. No butchery marks have been discovered on any dog bone and some elements survived as whole bones. This seems to indicate that dogs were not exploited for their meat. There is also no evidence for skinning these animals. With the exception of a mandible with molar teeth coming from a young individual (three to ve months of age) all bones belong to animals that had reached maturity before they died. A partial skeleton of a dog was found in a feature dating to Phase 4 (III-B, h105-8). It consists of a mandible with permanent dentition, two cervical vertebrae (axis, epistropheus) and a few ribs. Measurements could be taken on four bones. Two of them a talus (length 33.5mm) and a metatarsus II (length 72.0mm) point to large-sized dogs and two bones a calcaneus (length 34.5mm) and a metatarsus IV (length 68.5mm) to animals of medium size. They would have been suited for tasks such as guarding, herding and possibly hunting.
1031

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash

Figure 4. Percentages of wild animals in the bone collections from Begash according to phases of occupation, based on fragment counts (NISP).

Wild mammals and birds

A part of the animal remains recovered from Begash belong to wild mammals and birds indicating that species of these animal groups contributed to the diet of the sites inhabitants and, obviously, were also exploited as a resource for raw materials. Seven wild mammal species and four wild bird species were identied (Table 2). Numerically, they count for 4-13 per cent of the identied specimens in the collections from the various phases (Figure 4). Red deer (Cervus elaphus) is the most frequent wild mammal in the collections from Begash. Altogether 103 teeth and bones as well as ve antler fragments could be assigned to this species. Most of the bones seem to come from full grown individuals. According to the epiphysial fusion data only three out of 34 specimens have unfused epiphyses. Red deer is generally regarded as a species living in deciduous or mixed forests in lowlands as well as in the mountains. At present, red deer still inhabits woodland areas in the Dzhungar Mountains (Melkopitayushchie Kazakhstana 3(4): 13-16). The goitered gazelle (Gazella subgutturosa) is the second most frequent ungulate species at Begash. With the exception of one bone nd from Phase 1b the species is missing in the
1032

Michael Frachetti & Norbert Benecke

collections of the oldest phases. Most of the remains represent full grown animals. There is only one bone coming from a juvenile individual. The goitered gazelle inhabits sands and gravel plains and limestone plateaus. Nowadays, the species is widely distributed in the southern and south-eastern part of Kazakhstan (Melkopitayushchie Kazakhstana 3(3): 19-21). Altogether 12 bones were assigned to the Argali sheep (Ovis ammon). They represent a horn core of a male individual and various elements of the postcranial skeleton. All bones come from full grown animals. The argali or mountain sheep inhabits the highlands of central Asia. Today this sheep is an endangered species which still occurs in low numbers in the highlands of the Tienshan Mountains (Melkopitayushchie Kazakhstana 3(3): 148-9). The Siberian ibex (Capra sibirica) is represented by nine bones of the postcranial skeleton. All specimens come from full grown animals. Siberian ibex is a widespread species in the mountains of central and northern Asia. Its typical habitat is alpine meadows. At present, Siberian ibex can still be found in the upper ranges of the Dzhungar Mountains (Melkopitayushchie Kazakhstana 3(3): 95-8). Eight bones of pig were recovered at Begash and due to their size were safely assigned to adult wild pig (Sus scrofa). At present, wild pig is a common game animal in the river deltas west of the Dzhungar Mountains and they also occupy the regions deciduous forests (Herre 1986: 39; Melkopitayushchie Kazakhstana 3(4): 156). Domesticated pigs are unknown from any Bronze Age sites in the region and the small sample size for each period suggests that pork was insignicant in the diet anytime at Begash. The only wild carnivore safely identied in the faunal assemblage from Begash is red fox (Vulpes vulpes). Altogether 25 bones were assigned to this species representing various elements of the skeleton. All remains come from full-grown animals. Currently, red fox is a widely and frequently distributed carnivore found throughout in the Dzhungar Mountains (Melkopitayushchie Kazakhstana 4(1): 76-8). Two postcranial bones of a small mustelid complete the list of wild mammals from Begash. Beside mammal remains a few bird bones were recovered at Begash representing four species (Table 2). The eagle owl (Bubo bubo) inhabits mountains and forests with cliffs and rocky areas, usually nesting on cliff ledges. The chukar partridge (Alectoris chukar ) is a Eurasian upland game bird breeding in dry, open and often hilly country. The hoopoe (Upupa epops) inhabits open ground with short grass or bare patches nesting in a hole in a tree or wall. The starling (Sturnus vulgaris) is a ubiquitous bird species that can be found in various habitats except dense forests, deserts or semi-deserts. Surprisingly, there is no evidence for the exploitation of sh. In the rivers of the Koksu Valley shing would have resulted in large quantities of sh, especially in those periods of the year when the rivers run high i.e. during spring and autumn. As screening and wet sieving was conducted systematically, the lack of sh remains is less likely due to recovery bias. According to recent isotopic studies on human bones, sh appear to have been an important food among human populations of the Eurasian steppe throughout the whole period extending from the Mesolithic to the Iron Age (OConnell et al. 2003). Therefore, it would be difcult to accept the view that the inhabitants of Begash did not exploit this easily accessible and abundant food resource. It seems more plausible that the lack of sh
1033

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash

remains in the faunal assemblages at Begash is due to poor preservation of sh bones at the site.

Discussion: herd structure and landscape use through time

According to the archaeozoological evidence from all phases of occupation at Begash, the local population relied most heavily on herding domestic sheep/goat and cattle. Wild animals are generally represented by low numbers of various taxa of prey (and furry) mammals and birds. From this we may conclude that hunting was only of limited signicance for providing food and raw materials at Begash. Elsewhere, Frachetti has proposed that vertical transhumance is the most likely strategy to accommodate the dietary needs of herd animals in the Dzhungar Mountains from at least the Early Bronze Age (Frachetti 2008a). Furthermore, the seasonal pattern of vertical mobility was variable throughout the long life of sites like Begash in response to geographic uctuations in ecology and periodic shifts in the sociopolitical landscape (Frachetti 2008b). A synthetic review of the faunal evidence presented above sheds further light on the factors that played into this land use variation and helps to esh-out the strategies of Eurasian pastoralists from the earliest to the latest phases of occupation of Begash. From the earliest phase of occupation at Begash (2500-1950 BC, Phase 1a) sheep/goat and cattle dominate the domesticated fauna, while ibex and deer are nominally represented among wild animals (Table 2). This assemblage is consistent with a pattern of wintering in midland elevations and summering in higher elevation pastures without extensive mobility (<30km per yr). The initial presence of goitered gazelle in Phase 1b (1950-1690 BC) suggests that Begashs pastoralists may have made sporadic hunting forays into more arid, low elevation territories at the start of the second millennium BC. Interestingly, we also nd the rst traces of horse remains in Phase 1b (albeit just under 2 per cent). Since relative percentages of domestic sheep/goat and cattle are little changed, there seems to be a correlation between the rst evidence of horse and a move toward varied exploitation of non-domesticated resources from across the landscape. Compared to remains of domesticated animals, wild animal percentages remain small throughout the life of the site. Yet starting in Phase 2 (1625-1000 BC), increasing diversity of wild taxa (Table 2) hint that limited hunting still augmented the established herding strategy and Begashs pastoralists concertedly widened their range of exploited territory. For example, the habitat preferences of wild pig, argali and fox, which appear in the sites fauna for the rst time, suggest that Begashs pastoralists were ranging from the deltaic river fan of the Koksu River to the highland pastures in the Dzhungar Mountains by the mid-second millennium BC a considerably wider range than in Phases 1a and 1b. However, recent archaeological survey by Frachetti and Maryashev in the western lowland regions beyond the Koksu River delta revealed Bronze Age habitation evidence only as far as the semiarid foothills, roughly 40km from Begash. Beyond this fading piedmont, archaeological settlements in the at semi-deserts and riparian lowlands to the west are devoid of Bronze Age remains. Thus, even sporadic hunting forays into these lowland habitats likely demanded an efcient means of returning to primary settlements such as Begash. Horses would have facilitated these regional excursions.
1034

Michael Frachetti & Norbert Benecke

The earliest evidence for consistent exploitation of the distant arid lowlands can be traced to the Iron Age (Phases 3 and 4) and medieval period (Phase 5). In these phases, the number of goitered gazelle which prefer arid, low-altitude habitats increases in the faunal record at Begash. In addition, Iron Age remains of argali (a mountain species) also illustrate continued exploitation of the highlands by the sites inhabitants as in the Bronze Age. Given that the wild faunal evidence is still relatively scanty, Iron Age pastoralists may have been expanding their geographic range for improved pasturing or to widen political networks. Yet in doing so, they were able to take advantage of a broader range of wild animals (especially in terms of species diversity) across the varied ecology of the Dzhungar piedmont and western lowlands. Interestingly, this widening geographic range correlates chronologically with an increase in the percentage of horses recovered at the site (Figure 2). However, as the relationship between sheep/goat and cattle herds shifts only slightly from the Early Bronze Age (when there is no evidence for horses) it would seem that the dominant herd animals were manageable with or without the added mobility provided by horses. Saka settlements in the Talgar region (c . 250km south) exhibit a comparable herd structure to that of Iron Age Begash, suggesting that low percentages of horses (when compared to other domesticates) was not unique to mountain sites at this time (Benecke 2003). The steady increase in horses in the faunal record does correlate with documented political and social expansions of eastern Eurasian mobile pastoralists in the mid-rst millennium BC. For example, the increase in horses in Phase 3 (750 BC-AD 50) corresponds with the growth of nomadic steppe confederacies such as the Saka and Wusun (Chang et al. 2003; Rogers 2007). Symbolic representation of political and military signicance of the horse is well represented in Iron Age burial kurgans in the region e.g. Arzhan (Cugunov et al. 2006) and Berel (Gorbunov et al. 2005). In addition, there is a marked expansion of the regions burial geography from localised cemeteries in the Bronze Age to highly visible and widely distributed kurgans in the Saka period (c . 700-50 BC) (Frachetti 2008a: 159). Chang has cogently argued that population expansion and increased political complexity are benchmarks of the late rst millennium BC in Semirechye (Chang 2002). Thus, the increase in horses at Begash coincides with a period of greater regional control, military strength and political status on the part of Iron Age nomadic groups. In sum, increased mobility afforded by horse riding may have facilitated all realms of pastoralist life, but marked increases in horse remains at Begash are more clearly associated with the expanding range of land use and documented shifts in the political landscapes of nomadic communities in the rst millennium BC.

Conclusion
In conclusion, the small number of wild animals in Begashs faunal record and the limited scope of a single sites data provide only a glimpse into the myriad variations in the subsistence strategies and social landscapes of Eurasian pastoralists. However, the ne resolution and chronological span of Begashs zooarchaeology elicits a revision of the commonly held association between horses, extensive mobility and the emergence of Eurasian pastoralist economies during the early and middle Bronze Age. The broadly durable statistical relationship between domestic species throughout the occupation of Begash illustrates that
1035

Research

From sheep to (some) horses: 4500 years of herd structure at the pastoralist settlement of Begash

the economic base of steppe pastoralism (in the south-eastern steppe) rested not in horses but in sheep/goat and cattle. Furthermore, the limited yet telling wild animal evidence is suggestive of a wider diversity of uses and impacts of domestic horses in Eurasian pastoralist strategies over the longue dure e. The domestic horse is documented at Begash by the start of the second millennium BC, but its impact on pastoralism is not clear. It is true that by increasing their use of the horse throughout the Iron Age and later periods, the inhabitants of Begash likely improved their mobility and access to pastures across various ecological niches for their primary herd animals. Nevertheless, the zooarchaeological record from Begash illustrates that the increase in horses through time correlates rst with opportunistic hunting forays at the end of the Bronze Age and then with expanding political engagements that undoubtedly reshaped the organisation of Eurasian pastoralist communities from the rst millennium BC onward. When compared to the relative stability of other domesticates at Begash, the small Bronze Age presence and limited expansion of horses in the faunal record before historic periods demands that we reconsider the degree to which domestic horses played a dominant role in emerging pastoralist lifeways or in aiding the diffusion of regional material culture throughout prehistory. Specically, there is not ample evidence for extensive horse riding during the second millennium BC at Begash. To the degree that Begash is indicative of other pastoralist settlements in the region, the faunal evidence directly challenges the image of middle to late Bronze Age pastoralists (2000-1000 BC) as derived from migrating horse-riders (Kuzmina 2008) and suggests that horse riding was not the most signicant catalyst for regional diffusion at this point in prehistory. This does not demote the importance of domestic horse riding as a key innovation for Eurasian populations in general or defray its historical impact on the region write large; rather these data suggest there were other mechanisms by which pastoralism, material culture, and ideology developed among regional populations in the third and second millennia BC (Frachetti 2008a). Acknowledgements
The authors would like to thank Dr Alexei Maryashev and Professor Karl Baipakov from the Institute of Archaeology, Almaty, Kazakhstan for continued collaboration and scientic exchange. Professors Fiona Marshall, Rowan Flad and Phillip Kohl graciously provided helpful comments on drafts of this article. Funding for this research was provided by National Science Foundation grant #0535341. Special thanks to Professor Charles Stanish and the Cotsen Institute of Archaeology at UCLA, where much of this article was written.

References
ANTHONY, D.W. 2007. The horse, the wheel, and language: how Bronze-Age riders from the Eurasian steppes shaped the modern world. Princeton (NJ): Princeton University Press. ANTIPINA, E.E. 1997. Metody rekonstruktsii osobennostei skotovodstva na yuge vostochnoi Evropy v epokhy bronzy. Rossiskaya Arkheologiya 3: 20-32. BARFIELD, T.J. 1993. The nomadic alternative. Englewood Cliffs (NJ): Prentice Hall.

BENECKE, N. 2003. Iron Age economy of the Inner Asian steppe. A bioarchaeological perspective from the Talgar region in the Ily River Valley (southeastern Kazakhstan). Eurasia Antiqua 9: 63-84. BENECKE, N. & A. VON EEN DRIESCH. 2003. Horse exploitation in the Kazakh Steppes during the Eneolithic and Bronze Age, in M.A. Levine, A.C. Renfrew & K. Boyle (ed.) Prehistoric steppe adaptation and the horse (McDonald Institute Monographs): 69-82. Cambridge: McDonald Institute for Archaeological Research.

1036

Michael Frachetti & Norbert Benecke

CHANG, C. 2002. The evolution of steppe communities from the Bronze Age through medieval periods in southeastern Kazakhstan (Zhetysu): the Kazakh-American Talgar Project 1994-2001. Sweet Briar (VA): Sweet Briar College. CHANG, C., N. BENECKE, F.P. GRIGORIEV, A.M. ROSEN & P.A. TOURTELLOTTE. 2003. Iron Age society and chronology in South-east Kazakhstan. Antiquity 77: 298-312. UGUNOV, K.V., H. PARZINGER, A. NAGLER & V. C TEREBENIN. 2006. Der Goldschatz von Ar zan: ein F urstengrab der Skythenzeit in der s udsibirischen Steppe. M unchen: Schirmer Mosel. Environmental Systems Research Institute. 1994. Digital chart of the world for use with ESRI desktop software, Arcdata. Redlands (CA), USA. FRACHETTI, M.D. 2004. Bronze Age pastoral landscapes of Eurasia and the nature of social interaction in the mountain steppe zone of eastern Kazakhstan. PhD dissertation, University of Pennsylvania. Ann Arbor (MI): University Microlms. 2008a. Pastoralist landscapes and social interaction in Bronze Age Eurasia. Berkeley (CA): University of California Press. 2008b. Variability and dynamic landscapes of mobile pastoralism in ethnography and prehistory, in H. Barnard & W. Wendrich (ed.) The archaeology of mobility: nomads in the Old and in the New World : 366-96. Los Angeles (CA): Cotsen Institute of Archaeology (UCLA). FRACHETTI, M.D. & A. MARYASHEV. 2007. Long-term occupation and seasonal settlement of eastern Eurasian pastoralists at Begash, Kazakhstan. Journal of Field Archaeology 32(3): 221-42. GORBUNOV, A.P., Z.S. SAMASHEV & .V. SEVERSKII. 2005. The treasures of frozen burial mounds of the Kazakh Altai: materials of the Berel burial site. Almaty: Il -Tekh-Kitap. HABERMEHL, K.-H. 1975. Die Altersbestimmung bei Haus- und Labortieren. 2. Auage. Berlin & Hamburg: Parey. HERRE, W. 1986. Sus scrofa Linnaeus, 1758 Wildschwein, in Handbuch der S augetiere Europas 2/II : 36-66. Wiesbaden: Paarhufer Artiodactyla. KALIEVA, S.S. & V.N. LOGVIN. 1997. Skotovody Turgaya v tretem tysyacheletii do nashei ery. Kystanai: Akademiia Nauk. KUZMINA, E.E. 2008. The prehistory of the Silk Road. Philadelphia (PA): University of Pennsylvania Press. KHAZANOV, A. 1994. Nomads and the outside world. Madison (WI): University of Wisconsin Press. LEVINE, M.A., 2004. Exploring the criteria for early horse domestication, in M. Jones (ed.) Traces of ancestry: studies in honour of Colin Renfrew (McDonald Institute Monograph): 115-26. Cambridge: McDonald Institute for Archaeological Research. 2005. Domestication and early history of the horse, in D.M. Mills & S.M. McDonnell (ed.) The domestic horse: the origins, development and management of its behaviour : 5-22. Cambridge: Cambridge University Press. MASANOV, N.E. 1995. Kochevaia tsivilizatsiia kazakhov osnovy zhiznedeiatel nosti nomadnogo obshchestva. Almaty: Sotsinvest Gorizont. OCONNELL, T.C., M.A. LEVINE & R.E.M. HEDGES. 2003. The importance of sh in the diet of Central Eurasian peoples from the Mesolithic to the Early Iron Age, in M.A. Levine, A.C. Renfrew & K. Boyle (ed.) Prehistoric steppe adaptation and the horse (McDonald Institute Monographs): 253-68. Cambridge: McDonald Institute for Archaeological Research. OLSEN, S.L. 2003. The exploitation of horses at Botai, Kazakhstan, in M.A. Levine, A.C. Renfrew & K. Boyle (ed.) Prehistoric steppe adaptation and the horse (McDonald Institute Monographs): 83-104. Cambridge: McDonald Institute for Archaeological Research. 2006. Early horse domestication in the Eurasian Steppes, in M.A. Zeder (ed.) Documenting domestication the intersection of genetics and archaeology: 245-69. Berkeley (CA): University of California Press. OLSEN, S.L., B. BRADLEY, D. MAKI & A. OUTRAM. 2006. Copper Age community organization in Northern Kazakhstan, in D.L. Peterson, L.M. Popova & A.T. Smith (ed.) Beyond the steppe and the sown: proceedings of the 2002 University of Chicago Conference on Eurasian Archaeology (Colloquia Pontica 13): 89-111. Leiden: Brill. OUTRAM, A.K, N.A. STEAR, R. BENDREY, S. OLSEN, A. KASPAROV, V. ZAIBERT, N. THORPE & R.P. EVERSHED. 2009. The earliest horse harnessing and milking. Science 323(5919): 1332-5. MARYASHEV, A.N & M.D. FRACHETTI. 2007. Issledovanie mogilnikov i poseleniya u sela Bigash v vostochnom Semirechye. Istoria i Arkeologiya Semirechya 3: 100-105. Melkopitayushchie Kazakhstana. 1969-1985. V chetyrekh tomakh. Alma-ata: Nauka. ROGERS, J.D. 2007. The contingencies of state formation in eastern Inner Asia. Asian Perspectives 46(2): 249-74. STEIGER, C. 1990. Vergleichend morphologische Untersuchungen an Einzelknochen des postkranialen skeletts der altweltkamele. unpublished xx dissertation, University of Munich. VAINSHTEIN, S.I. 1991. Mir Kochevnikov Tsentralnyi Azii. Moscow: Nauka.

1037

Research