Rethinking the evidence for early horse domestication at Botai

Despite its transformative impact on human history, the early domestication of the horse (Equus caballus) remains exceedingly difficult to trace in the archaeological record. In recent years, a scientific consensus emerged linking the Botai culture of northern Kazakhstan with the first domestication of horses, based on compelling but largely indirect archaeological evidence. A cornerstone of the archaeological case for domestication at Botai is damage to the dentition commonly linked with the use of bridle mouthpieces, or “bit wear.” Recent archaeogenetic analyses reveal, however, that horse remains from Botai are not modern domesticates but instead the Przewalski’s horse, E. przewalskii—warranting reevaluation of evidence for domestication. Here, we compare osteological traits hypothesized to have been caused by horse transport at Botai with wild Pleistocene equids in North America. Our results suggest that damage observed in Botai horse teeth is likely generated by natural disturbances in dental development and wear, rather than through contact with bridle equipment. In light of a careful reconsideration of the mid-Holocene archaeological record of northern Eurasia, we suggest that archaeological materials from Botai are most effectively explained through the regularized mass harvesting of wild Przewalski’s’ horses—meaning that the origins of horse domestication may lie elsewhere.

The innovation of horse transport-chariots and mounted horseback riding-transformed nearly every aspect of the ancient world. Horse carts and riding altered the trajectories of processes such as pastoral subsistence, mobility, warfare, communication, trade, agriculture, disease and biological exchange 1 . Despite the centrality of horse domestication to human history, however, a clear understanding of the initial domestication of Equus caballus has proven particularly elusive-in large part because of the vagaries of the archaeological record, which have left few material traces with which to understand early human-horse relations. As a result, archaeological models for horse domestication often relied largely on indirect measures such as the frequency of Equus bones at archaeological sites, or changes in their morphological variability, to distinguish human activity 2,3 .
Emerging techniques in archaeological science have helped to fill these analytical gaps, providing direct lines of evidence for human-horse interactions that can be identified in ancient material assemblages. Most significantly, the discovery of "bit wear"-damage caused by a bridle mouthpiece to the equine dentition 4,5 -helped crystallize the emerging discipline of equine paleopathology [6][7][8][9] . In the late 2000s, an archaeological consensus appeared to converge on sites of the Botai culture in northern Kazakhstan dating to the 4th millennium BCE, as the birthplace of horse domestication-based in no small part on the identification of apparent "bit wear" on a Botai tooth 10 .
In the last several years, though, continued innovation of new scientific methods -and their application to Botai materials has upended core assumptions of the Botai domestication model. Genomic sequencing demonstrated that Botai equids are in fact not the progenitor of E. caballus but a sister taxon, E. przewalskii 11 . This animal, the Przewalski's horse, is today a wild species with threatened conservation status, and has never been managed or used for transport in the historic era. In light of continuing progress in our understanding of physiological and anatomical links between equine skeletal changes and human activity since the emergence of Botai consensus [12][13][14] , this discovery warrants a reconsideration of the argument for horse transport at Botai.
Here, through careful comparisons with the fossil record of Pleistocene wild horses in North America, we demonstrate that apparent instances of bit damage on Botai horses are likely generated by natural disturbances in dental development paired with natural wear. In light of new evidence and reassessment of other key tenets of the argument for horse management at Botai, we suggest that this important assemblage instead represents the final chapter of a millennia-long tradition of mass harvesting of wild horses. We offer an alternative model for the origins of the domestic horse that places E. caballus within a broader Eurasian and African trajectory of equid domestication and transport.  19 . Around this time, many areas of northern Eurasia also experienced an apparent spike in the frequency of horse bones found at archaeological sites, which some interpret as evidence for human-mediated dispersal of domestic E. caballus 2,3 . At some sites, this pattern in skeletal frequencies is bolstered by the occasional recovery of horse-themed works of portable art, and identification of horse bones in ritual or funerary contexts 20 . None of these pre-Sintashta lines of evidence, however, uniquely distinguish a domestication relationship between humans and horses. Early arguments for Eneolithic domestication in the Black Sea steppe or Eastern Europe were bolstered by the discovery of anthropogenic bit wear damage to the lower second premolar of a horse specimen from the site of Deriyevka in Ukraine, initially thought to date as early as ca. 4000 BCE 5 . However, direct radiocarbon dating of this horse revealed it to be an intrusion from the early Iron Age, ca. 700-200 BCE 21 , dashing any direct links between the site and human management of domestic horses, and dimming enthusiasm for Deriyevka as the epicenter for domestication.
In the early 2000s, increased scholarly attention coalesced around another archaeological site (and related localities of the same general period) as a potential candidate for Eneolithic horse domestication: Botai. This site is located along a tributary of the Ishim River along Kazakhstan's northwestern border with Russia ( Fig. 1). Here, researchers uncovered a pithouse village along with a massive assemblage composed nearly entirely of horse remains. Subsequent analysis of the site, and related assemblages from the sites of Krasni-Yar and Vasilovka, revealed intricately decorated horse bones, tools made from horse bones 22 , and pithouses filled with decayed vegetative matter hypothesized to be horse dung 23 . At some Botai locations, posthole structures raised the possibility of corrals 22 . Analysis of some loose lower premolars from Botai even suggested damage caused by a bit, although not in the typical form expected from a metal mouthpiece 24 . Conveniently positioned both chronologically and geographically, Botai seemed to slot nicely into extant explanatory models for Indo-European movements and help explain the broader cultural transitions observed in Eurasia during the 4th millennium BCE.
Nonetheless, initial response to the Botai domestication hypothesis was far from one of consensus. As archaeologists developed an ever-expanding toolkit for identifying evidence of human activity in the archaeozoological record, a number of concerns emerged which cast serious doubt on the site's viability as an early domestic horse assemblage. One of the most serious among these concerns was the apparent inconsistency of age and sex patterns in the Botai horse assemblage with those of a pastorally-managed herd. As articulated by Levine, through her detailed ethnographic study of contemporary pastoral groups 8 , pastoral management of a breeding herd typically entails slaughter of two categories of animal: (1) adults that have reached the age of reproductive senescence, these being mostly female, and (2) juvenile animals that have yet to reach breeding age, these being mostly male. The resulting bimodal mortality profile has proven to be a consistent feature of pastoral horse assemblages from the Bronze Age onwards 25,26 . In contrast, demographic analysis of Botai horses by Levine 8 and a reanalysis by Olsen 22 revealed that the Botai horses are primarily breeding-age adults, split into a roughly equal balance of male and female horses. Most recently, this finding was confirmed a third time through DNA study by Fages et al. 27 , who www.nature.com/scientificreports/ found a marked contrast between the sex ratios found in Botai and later ritual assemblages of domestic horses.
Although it is possible that the inclusion of a large number of ritually-sacrificed, adult male transport horses into an otherwise predominantly female assemblage could have artificially skewed the age and sex patterns 22 , at face value, the overall mortality patterns found at Botai appear fundamentally inconsistent with pastoral management.
Other lines of evidence also seemed to cast a shadow of doubt on the domestic status of Botai horses. As a more nuanced understanding of the process of equine tooth wear emerged, Olsen 22 demonstrated that irregularities in occlusal morphology identified on teeth from Botai were not necessarily indicative of bit use, as similar features can be identified on wild equids. And while vertebrae from early riding horses often preserve regular traces of damage to the lower back 9 , pathological study of the small number of available vertebral elements from Botai 28 showed no recognizable damage associated with horse transport. Finally, bone arrowheads were actually found in direct association with at least some specimens-demonstrating that the animals were killed through hunting 22 . Benecke and von den Driesch 29 found that in terms of size and morphological variance, Botai specimens were not separable from earlier hunted assemblages of wild horses-which have formed a key aspect of northern Eurasian subsistence since the Paleolithic. As a result of these considerations, even researchers that favored a Botai domestication model were compelled to admit that managed/ridden horses must have comprised a small percentage of the assemblage 20,22 .
The emergence of a Botai consensus. Amidst these conflicting signals, two new discoveries in the 2000s tilted the needle in favor of popular consensus for domestication at Botai. The first of these was the publication of a new dataset of isotopic measurements from lipid residues in ceramic sherds by Outram et al. 10 . An earlier study using the same approach 30 identified that the ceramic vessels at Botai were used primarily to process horse fats-a conclusion that was perhaps unsurprising, given the overwhelming emphasis on horses in the faunal assemblage. In the updated study, Outram and colleagues sought to differentiate fats linked with milk products from those linked to meat and adipose tissue. New reference datasets showed that while horse milk and meat fats had an overlapping distribution in carbon isotope values, the seasonally-sensitive hydrogen isotope deuterium could help identify residues linked to the summer months, when milking typically occurs. Together, these observations formed the basis of an argument that Botai ceramics once held milk, rather than other horse fats.
A second crucial component to this study was a reanalysis of jaw bones and dentition, using revised criteria developed by Bendrey 6 that sought osteological features more reliably associated with horse transport than those identified by Brown and Anthony 24 but criticized by Olsen 22 . These new criteria, derived through careful comparison of natural history collections, included new bone formation to the diastema of the lower jaw, a feature thought to occur in very low levels in wild animals, and the erosion of cementum and enamel along the anterior margin of the lower second premolar. While such cementum erosion occurs in both wild and domestic animals, by comparing wild and domestic animals with known life and work histories, Bendrey 6 found that metal bits often produce a version of this damage that is parallel-sided in shape, and is often more invasive than natural wear (causing exposure of dentine).
Application of these new criteria to Botai revealed some instances of new bone formation on the lower jaw that were higher than those observed in wild comparative collections (a score of "2" in the categorical scoring system). Additionally, one lower second premolar seemed to demonstrate long, parallel-sided cementum erosion and enamel wear resulting in dentine exposure-a new "smoking gun" for horse bridling and riding during the 4th millennium BCE (Fig. 2). These new lipid values and paleopathological finds were further supported by metric comparisons of metapodial bones, which appeared to show that the Botai horses were the domestic E. caballus 10 . Together, these new finds underscored an emerging consensus that Botai represented the first step in the domestication story, which persists to the present day.

Recent innovations in archaeological science.
Although the consensus forged by these discoveries was robust, recent discoveries necessitate a reevaluation of some or all of the Botai domestication argument. www.nature.com/scientificreports/ While morphological distinction among "caballine" horses (Equus caballus and its close relatives) is notoriously challenging, improving techniques in ancient DNA now enable full genomic sequencing of ancient specimenspermitting for the first time a reliable species identification of archaeological equids. In a key finding, Gaunitz and colleagues 11 revealed that despite the previously identified morphological patterns in metapodial shape, all analyzed Botai specimens were in fact E. przewalskii, and not the progenitor of the domestic horse. Attempting to reconcile this discovery with the previous consensus for domestication, Gaunitz and coauthors concluded that Botai must represent an earlier domestication event that was later abandoned, leading to the subsequent feralization and rewilding of the Przewalski's horse. The contextual nature of most of the existing evidence for domestication at Botai, however, may not support this interpretation. Many assumptions underpinning this earlier work-such as the idea that vegetal matter found in pithouses was horse dung, that postholes were used to form corrals, or that summer seasonality of isotope values indicated horse milk-were contingent upon the presence of domestic horses, rather than empirically demonstrated. Other recent biomolecular work, such as a recent study analyzing dental calculus of human remains from Botai 31 , found no evidence for dietary milk proteins of any kind. Consequently, it seems that the argument for Botai domestication of E. przewalskii now hinges heavily, if not solely, on the mandibular bone formation and apparent bitting damage found on a single lower second premolar.
Re-examining evidence for horse transport at Botai. How robust is the evidence for horse transport at Botai? A careful examination of published imagery of this tooth shows evidence of natural damage and disturbances in dental development that may undercut the argument linking the apparently parallel-sided anterior enamel and dentine exposure observed in a single second premolar from Botai with a bridle bit. Specifically, although the enamel exposure visible along the anterior margin of this tooth does indeed appear parallel-sided at first glance, it actually consists of two adjoining, oval-shaped areas of exposure. These exposed areas of enamel correspond precisely with bands or "valleys" of reduced cementum deposition visible along the buccal and lingual margins of the tooth (Fig. 2). Reduced cementum deposition at the occlusal/coronal area of the tooth presumably would make the anterior side of the premolar more susceptible to cementum erosion from natural wear processes, resulting in the exposure of two oval-shaped areas of enamel. The exposed area of enamel is also interrupted by a series of small circular pits in the enamel, producing small areas of exposed dentine, which Outram et al. 10 argue to be indicators of bit use-a marker of severe and invasive wear that is not observed in unworked horses 6 . However, this photo shows that each of the pits in the Botai tooth margin are exceedingly small in diameter-far smaller than any we have observed in archaeological horse remains from Mongolia, even those controlled with organic bits 32 . Small, circular pitting is also a characteristic sign of pit-form enamel hypoplasia-a dental defect caused by disruption of normal enamel matrix secretion, which results in pockets of thin or poorly developed enamel. These observations raise the possibility that, in a large assemblage of wild horses, natural processes (rather than human activity) could produce similar features to those identified in the Botai assemblage.

Methods
To characterize the frequency of damage found on the Botai tooth (cementum banding, as well as the frequency of enamel hypoplasia, and pitting of the enamel along the anterior margin of the second premolar) in wild populations, we analyzed a sample of 72 lower second premolars and 81 upper second premolars of wild equids (Equus spp.) derived from Pleistocene fossil localities in North America-where no domestic horses existed prior to the arrival of European colonists in the last millennium. Our analyzed sample included material from Bluefish Caves in northern Yukon, various localities in the Edmonton area and the Wally's Beach site in Alberta, Natural Trap Cave from the Bighorn Mountains of northern Wyoming, various localities in New Mexico and Texas, and Cedral and San Josecito Cave in northeastern Mexico (Fig. 1). All of these sites preserve abundant equid remains, which are late Pleistocene in age 33  Enamel hypoplasia and cementum banding. For those specimens showing hypoplastic defects, we noted the location of the defect on the tooth crown. We identified the type of defect(s) present on the anterior margin of the tooth (i.e., pit-form, plane-form, or furrow-form; following Witzel et al. 34 and Hillson 35 ) and whether they presented dentine exposure. We treated isolated teeth as separate individuals, unless there was evidence that isolated teeth were associated (i.e., left and right teeth of the same mandible or palate), in which case the associated teeth were treated as a single individual. We calculated the percentage of specimens (i.e., individuals) presenting enamel hypoplasia with and without dentine exposure on the anterior margin of the lower and upper second premolars (Table 1; Supplementary Table 1).
We note that one obstacle to the study of enamel hypoplasia in the cheek teeth of equids and other hypsodont ungulates is the presence of cementum covering the tooth crown 36,37 . In living horses, the cementum is thicker at the occlusal/coronal area (i.e., the extra-alveolar area) of the tooth, particularly on the buccal side of the lower cheek teeth and the lingual side of the upper cheek teeth 38,39 . Cementum in the occlusal/coronal area of the tooth is primarily comprised of tertiary cementum and is attached to the gingival mucosa 38 . Cementum significantly decreases in thickness below the occlusal/coronal area (i.e., at the intra-alveolar area of the tooth) in a cervical www.nature.com/scientificreports/ direction where, depending on the ontogenetic development of the tooth, it might be absent 38,39 . Aged horses with teeth in extreme stages of wear often show abnormally large increases in tertiary cementum 38 .
To control for the impact of variable cementum coverage, we qualitatively scored the extent to which cementum covers the tooth crown using a 0-5 scoring system (following Barrón-Ortiz et al. 40 ) (Supplementary Table 1). In this system, 0 indicates that the tooth crown is not covered by cementum, 1 denotes that 1-25% of the tooth crown is covered by cementum, 2 indicates that 26-50% of the tooth crown is covered by cementum, 3 denotes that 51-75% of the tooth crown is covered by cementum, 4 indicates that 76-95% of the tooth crown is covered by cementum, but that the cementum present below the occlusal region of the tooth consists of a thin layer, and 5 denotes that the entire tooth crown is covered by a thick layer of cementum. We excluded any specimens with a score of 5 from our analysis, because the cementum covering the tooth made it difficult to consistently evaluate whether enamel hypoplasia was present in the tooth. For teeth with only one side of the tooth crown completely covered by cementum we scored the exposed sides of the tooth. Cementum did not pose a serious problem to the examination of enamel hypoplasia for many of the teeth we studied, because post-mortem weathering and degradation of cementum exposed the enamel underneath.
For cementum banding, we evaluated the presence/absence and number of cementum bands (when present) at the occlusal/coronal area of the tooth (Supplementary Table 1). We treated isolated teeth as separate individuals and associated teeth were treated as a single individual. We excluded teeth in which the occlusal/coronal cementum was damaged or weathered from cementum analysis, resulting in a total sample of 52 lower and 78 upper second premolars. From this sample, we calculated the proportion of specimens displaying two or more well-defined cementum bands (Table 2).

Mandible bone formation.
Although preservation from Pleistocene horses is generally too poor to assess diastema damage within this same assemblage, we studied six horse mandibles from the well-preserved Wally's Beach site in southern Alberta (Table 3; specimens housed at the RAM) to determine whether they present diastema bone formation comparable to levels observed in the Botai assemblage. We used the scoring system proposed by Bendrey 6 to determine the extent of new bone formation in the diastema.

Results
Our results provide clear indication that pitting similar to that observed on the Botai horse is particularly common on the anterior margin of the lower second premolar of wild North American horses. Although only a small handful (n = 4; 2.5%) of specimens exhibit any form of visible enamel hypoplasia on the anterior margin of the upper premolar, on the lower premolar, a total of 14 specimens out of 72 (19.4%) teeth analyzed exhibit enamel hypoplasia (Table 1). Of these, 11 (15.3%; Table 1) lower second premolars display pit-form hypoplastic defects accompanied by circular or oval-shaped dentine exposures; a condition that is consistent with the damage observed in the Botai horse tooth (Fig. 3).   www.nature.com/scientificreports/ Furthermore, our evaluation of cementum banding, indicates that among wild horses, there is often a wide gradient in this trait-ranging from teeth that do not show any distinct cementum bands to those that show well-defined cementum bands. Teeth with at least two or more well-defined cementum bands, as is the case in the Botai tooth, appear to be slightly more common in the upper second premolar (25.6%) than in the lower second premolar (15.4%) ( Table 2).
We were able to assess four out of six horse mandibles from the Wally's Beach site for evidence of mandibular bone formation. Three of the four mandibles showed bone changes in the diastema indicative of a score of "1" in the scoring system developed by Bendrey 6 . These mandibles show faint changes, including the development of a well-defined ridge along the diastema and roughening of the bone. The remaining specimen that was wellpreserved enough for us to study, showed well-developed pathological bone formation on the left side of the mandible, earning a score of "2" (Table 3). At Botai, 4/15 mandibles showed a similar score 10 .

Discussion
Comparing our results to the Botai tooth provides strong indication that the osteological features taken for evidence of horse transport, may have been produced solely through natural processes. Although our analysis suggests that it is uncommon for enamel hypoplasia pits to co-occur in multiples on the anterior margin of the tooth, one wild horse in our sample, from Bluefish Cave I (MgVo-1) in northern Yukon, exhibited two such pits, nearly identical to those attributed to bit wear at Botai, along with visible cementum banding (Fig. 3). Based on available data, we are unable to speculate on whether enamel hypoplasia itself might occur in higher frequency or severity in domestic and wild samples. However, the high frequency of enamel hypoplastic defects with dentine exposed in North American horses suggests that dentine exposure-in the form of circular or oval pits-cannot be considered reliable evidence of transport damage. In fact, the apparent frequency of this type of damage in the Botai assemblage (1/9 or 11.1% 10 ) is commensurate or even below the frequency of this type of feature observed in wild North American populations (15.3%; Table 1).
Because of the different processes involved in bit wear and hypoplastic defects, the Botai tooth can potentially be evaluated in a non-destructive manner using recently developed visualization and virtual histological techniques 40 . In contrast to other types of enamel hypoplasia, pit-form hypoplastic defects are characterized by a scatter of isolated pits or single pits surrounded by unaffected enamel 34,35 . The rim of each pit, as seen on the surface of the tooth, is usually smooth 35 . In longitudinal histological sections, the rim of each pit appears smoothly curved and the striae of Retzius in the enamel forming the walls of the pit are also curved 34,35,41 . Frequently accentuated lines, i.e., Wilson bands, are also seen associated with pit-form hypoplastic defects 34,35 . Pits that are caused by other factors such as flaking of poorly mineralized enamel [42][43][44] , cavitary lesions resulting from pre-eruptive enamel resorption 41 and enamel erosion from mechanical wear or abrasion (e.g. bitting damage) lack a smooth and curved rim and they cross-cut normal striae of Retzius (Fig. 4). Consequently, damage caused by bits should be easily distinguishable from pit-form hypoplastic defects using histological approaches. Table 3. Diastema bone formation score in six Pleistocene horse mandibles from the Wally's Beach site, Alberta. www.nature.com/scientificreports/ Interestingly, our results suggest a discrepancy in the frequency of enamel hypoplasia in the anterior margin of the upper and lower second premolars of the wild equid samples (Table 1). These teeth form at approximately similar times during ontogeny 45 . Therefore, disruptions in enamel secretion would be (in theory) expected to be correlated to some degree in upper and lower second premolars for a given individual. Although we were not able to analyze associated upper and lower second premolars in our sample, there is no indication of preservation bias, collecting bias, or sampling biases between upper and lower teeth in the samples we studied. One possible explanation for this discrepancy in the frequency of enamel hypoplasia in the upper and lower teeth is different dental developmental geometries. Dental developmental geometry is a factor that is known to affect the expression and visibility of enamel hypoplasia 35 .
Our new data suggest that, without careful consideration of other enamel and cementum patterns, criteria used to identify bit wear through enamel exposure can sometimes produce spurious results. Standard practice suggests that cementum erosion on the anterior margin of the occlusal/coronal area of the lower second premolar, leading to the exposure of a band of enamel that is parallel-sided in shape and that has a height greater than 5 mm (measured from the occlusal surface towards the mandible), is indicative of bit damage 6 . The Botai tooth presents a band of enamel exposure that is 18.0 mm in height 10 , well above the cutoff value of 5 mm. However, as indicated previously, despite appearances of parallel sides, the enamel exposure consists of two adjoining, ovalshaped areas of exposure. These exposed areas of enamel correspond with two well-defined bands of reduced cementum deposition, observed along the buccal and lingual sides of the tooth. We propose that reduced cementum deposition in this region of the tooth made the anterior side of the premolar more susceptible to cementum erosion from natural wear, resulting in the exposure of two tall, oval-shaped areas of enamel.
Multiple, well-defined cementum bands are relatively common in the lower (15.4%) and upper (25.6%) second premolars of wild equid teeth we studied ( Table 2). In some cases, cementum bands appear to be the result of reduced cementum deposition, similar to the condition observed in the Botai tooth. However, in other cases they appear to result from increased cementum deposition on top and below the cementum band. Based on these observations, two alternative, although not mutually exclusive, mechanisms can potentially explain the formation of cementum bands. The first of these is disruption of the secretory activity of cementoblasts, the cells that form cementum-if tooth eruption is more or less constant, disruption in the secretory activity of cementoblasts would lead to a reduction in the deposition of tertiary cementum for the duration of the disruption. Alternatively, cementum bands may be caused by changes in the rate of tooth eruption-if the deposition of tertiary cementum is more or less constant, then changes in the rate of tooth eruption would affect how much tertiary cementum gets deposited around the occlusal/coronal area of the tooth crown at a given time (less cementum if the rate of tooth eruption is increased, and more cementum if the rate of tooth eruption is decreased). In either case, it would seem that the mechanism of cementum band formation is common enough in wild horses to warrant careful consideration by archaeologists interested in anthropogenic tooth damage.
Although we were unable to confidently assess the relative frequency of mandibular bone formation on the diastema in our sample due to the poor preservation of this element in fossil contexts, one of four analyzable specimens from our best-preserved assemblage (Wally's Beach, Alberta) exhibited well-developed pathological www.nature.com/scientificreports/ bone formation-earning a score of "2" using the scoring system outlined by Bendrey 6 ( Table 3; Fig. 5). Although the precise cause for this bone formation in the Wally's Beach horse is not immediately clear, diet and age likely played a factor-based on dental wear, this horse is elderly (likely 15 + years based on the triangular shape of the incisor grinding surface and the absence of cups/interior enamel spots in the central incisors). The other three Wally's Beach specimens well-preserved enough to be analyzed exhibited a score of "1", while two more were too damaged to be confidently assessed ( Table 3). The diastema score from Wally's Beach is equal to those previously taken to represent evidence of bit damage at Botai 10 , and occurs in a roughly comparable frequency (1/4) to the Botai assemblage (4/15).
Alternative explanations for Botai. In light of our new data, arguments for horse domestication at Botai no longer appear to be supported by the available archaeological evidence. Without the presumption of horse transport, many aspects of the Botai assemblage are more efficiently explained by interpretation of the site as the result of regularized mass-harvesting of wild horses. For example, Botai's location at a river crossing is consistent with wild equid hunting tactics that date back deep into the Pleistocene. At Paleolithic sites across Europe, entire bands of horses-either mostly-female harem groups, all-male bachelor bands, or both-were commonly ambushed alongside natural water features where they were more effectively trapped and slaughtered 46 . This strategy appears to have been employed by the earliest hominin horse hunters, dating back nearly a half million years or more 47,48 . Group harvesting at Botai could easily explain unresolved questions in the assemblage, including apparent presence of entire carcasses, the predominance of prime-aged adult animals, and the recovery of bone arrowheads in situ with deceased horse remains, as well as the utter absence of other domestic fauna at Botai 22 . The relatively equal ratios of male and female animals found at Botai could imply that the site was used to harvest both bachelor bands and harem groups over its use history. Summer seasonality identified by Outram et al. 10 using isotope data could reflect horse milk production, but many mass harvesting sites of horses also display consistent summer seasonality, even over centuries or millennia of reuse 49 . If chosen for a favorable topographic position or location on a key ecological corridor or migration route, horse mass harvesting sites may have been regularly utilized in a particular time of year over long stretches of time.
A reappraisal of the pre-Botai archaeological record of humans and horses also supports this view. Many of the cultural modifications found in the Botai artifact assemblage-the decoration of horse bones, the use of horse bones as tools, and even the occasional ritual inhumation of horse remains-are fully consistent with hunter-gatherer cultures in which horse hunting plays an important role. Horses are the most commonly depicted animal in Eurasian Paleolithic cave paintings 50 , and were a favorite muse for hunter-gatherer artists across the Pleistocene and into the Holocene-appearing on bones, ivory, or stone objects-and probably many organic artifacts that have not survived-as decorations or as dedicated votives 51 . The discoveries of ritual features and artwork at Botai or Eneolithic sites from the Black Sea region, while important, fail to effectively delineate a domestication relationship from the rich hunting tradition that preceded it.
How, then, to explain the apparent resurgence in the frequency of horses-a Pleistocene animal whose populations began to dwindle after the Last Glacial Maximum-during the 4th millennium BCE? A convergence of climate and human factors unrelated to horse domestication may have contributed. The late Neolithic witnessed a cooling trend across much of northern Eurasia 52-54 that could have boosted the population size and expanded the viable habitat of cold-adapted equids. These favorable climatic changes may also have been buoyed by the behavior of pastoralists migrating into new areas of Europe and Asia, who likely slashed forests and increased the cover of grassland habitats across a wide region 55 .
Removing Botai from archaeological narratives streamlines and improves other emerging models for Eurasian prehistory. For example, if Botai people were horse hunters and horses were not yet domesticated ca. 3500 BCE, the absence of human genomic links between Botai and pastoralist Yamnaya people 56 , and the absence of domestic horses south of the Caucasus prior to 2000 BCE 57 are consistent with predictions, rather than lingering puzzles. www.nature.com/scientificreports/ Finally, it is important to note that the interpretation of Botai is not simply academic, but has profound practical implications for conservation. Equus przewalskii is now a badly endangered taxon that only a few decades ago was considered extinct in the wild. Thanks in part to careful conservation efforts in Mongolia and elsewhere, this animal has been painstakingly reintroduced from a host population of less than a dozen individuals 58 . Since the conclusion of Gaunitz et al. 11 that Przewalski's horses are not truly "wild" is not supported by the available archaeological data, their protection must remain an essential priority for conservation of grassland biodiversity in the Eurasian steppe.
Alternative models for horse domestication. If Botai does not represent either E. caballus or the origins of horse transport, where, then, are we left to turn for answers to the initial domestication of the horse? Future work will be necessary to answer this question with any confidence. However, we suggest that more careful attention should be paid to the domestic horses and chariots of Sintashta, and the preceding cultures of the Black Sea steppe, where horse milk proteins have recently been directly identified in human teeth dated to the 3rd millennium BCE 31 . We also point to the likely significance of cultural connections between Transcaucasia and adjoining areas of the Near East, where donkeys and onagers were used for transport by the 3rd millennium BCE or before 59,60 . In settling on a new model for horse domestication, it will be essential to revisit the findings of a rapidly changing discipline of archaeological science with fresh eyes and a willingness to reevaluate old conclusions based on new discoveries.

Conclusion
The continual emergence of new lines of evidence for understanding ancient human-horse interactions necessitates vigilant reevaluation of models for the domestication of the horse. Based on comparison of materials from the site of Botai with wild equid remains from North America, we suggest that skeletal changes to the diastema and lower premolars observed in this assemblage are a result of natural processes rather than early transport. In light of these finds and other recent biomolecular discoveries, this site and its related assemblages appear to be best explained through mass harvesting of wild horses rather than early domestication. Future work will continue to require careful reanalysis of existing assumptions and revisitation of inherently ambiguous and indirect archaeological data.

Data availability
All data used in the analysis are provided directly in the manuscript and its supporting information. www.nature.com/scientificreports/