Humans permanently occupied the Andean highlands by at least 7 ka

High-elevation environments above 2500 metres above sea level (m.a.s.l.) were among the planet's last frontiers of human colonization. Research on the speed and tempo of this colonization process is active and holds implications for understanding rates of genetic, physiological and cultural adaptation in our species. Permanent occupation of high-elevation environments in the Andes Mountains of South America tentatively began with hunter–gatherers around 9 ka according to current archaeological estimates, though the timing is currently debated. Recent observations on the archaeological site of Soro Mik'aya Patjxa (8.0–6.5 ka), located at 3800 m.a.s.l. in the Andean Altiplano, offer an opportunity to independently test hypotheses for early permanent use of the region. This study observes low oxygen (δ18O) and high carbon (δ13C) isotope values in human bone, long travel distances to low-elevation zones, variable age and sex structure in the human population and an absence of non-local lithic materials. These independent lines of evidence converge to support a model of permanent occupation of high elevations and refute logistical and seasonal use models. The results constitute the strongest empirical support to date for permanent human occupation of the Andean highlands by hunter–gatherers before 7 ka.

. The yellow inset indicates the area of geographical analysis for hypothesis 3. Base imagery from NASA [12].    a All radiocarbon dates calibrated using SHCAL13 [17] using Bchron [18] as implemented in R statistical computing environment [19]. See the electronic supplementary material,  These interpretations come with some uncertainty. Past δ 18 O saturation levels were not necessarily the same as present levels. However, ice core studies in the southcentral Andean highlands show that Early Holocene δ 18 O values were virtually the same as modern values with differences under +2‰ [21]. Diagenetic alternation can also confound interpretation of δ 18 O values in archaeological bone carbonates. The examination of carbon and nitrogen mass fractions and carbon : nitrogen (C : N) ratios in the organic fraction of bone can offer some indication of compositional integrity of structural carbonate in the bioapatite because the organic matrix tends to protect bioapatite crystallites [22]. Unaltered bone exhibits bone mass fractions of greater than 35 wt% carbon and 11-16 wt% nitrogen [23][24][25][26] with atomic C : N ratios in the range of 2.9-3.6 [27].  3. H2: human bone stable carbon isotope chemistry (δ 13 C)

Results
Stable carbon isotope values (δ 13 C) in human bone can also offer clues to the structure of mobility patterns across elevational gradients. As elevation increases, δ 13 C saturation in the plant tissues that humans directly or indirectly consume increases [28,29]. rarely produce values greater than −23‰, and high-elevation plants commonly do. Consistent with the high-elevation values, 17 carbonized plant samples from secure pit-feature contexts at Soro Mik'aya Patjxa produced values between −24 and −20‰ including a +1‰ correction for fractionation owing to carbonization [33]. Modern camelids living above 3500 m.a.s.l. in northwest Argentina and northern Chile exhibit δ 13 C values between −28 and −18‰ [29,32] including a +5‰ correction for trophic fractionation [34] and +1.5‰ correction for post-industrial fossil fuel effects [30,31]. The δ 13

Results
Stable carbon isotope values, δ 13  All values reported here account for well-known fractionation effects related to carbon incorporation into bone carbonate, carbon incorporation into collagen and carbonization (see the electronic supplementary material). Diagenetic processes are unlikely to confound this interpretation, given the carbon and nitrogen mass fraction values and C : N ratios observed in the collagen of burials 9 and 16 (see H1 results).

H3: travel distance to 2500 m elevation contour
Kelly's [35] compilation of average logistical move distances among eight ethnographic foraging groups averages 42 km round-trip with all falling under 80 km round-trip (figure 4). Residential move distances for ethnographic foraging groups average 148 km per year with all but one observation falling below 1000 km. Given these ethnographic hunter-gatherer travel distances, if the round-trip distance between Soro Mik'aya Patjxa and the nearest location on the 2500 m topographic contour is greater than 80 km, then the logistical use model (model 1) would be refuted. If the distance is greater than 1000 km, then the logistical and seasonal use models (models 1 and 2) would be refuted. Conversely, if the distance is less than 80 km, then all three models are supported, and if the distance is less than 1000 km, then only the seasonal and permanent models (models 2 and 3) are supported.

Results
Soro Mik'aya Patjxa is located in the interior highlands of the Andean Altiplano at 16°14 S, 69°44 W, 3860 m.a.s.l. (figure 1). Least-cost travel analysis indicates that the minimum one-way travel distance between Soro Mik'aya Patjxa and the nearest location on the 2500 m elevation contour would have been 122 km with a travel time estimate of 41 h (figure 5; see the electronic supplementary material, Material and methods). This translates to a minimum round-trip distance of 244 km-over three times the 80 km maximum logistical move distance observed among ethnographic hunter-gatherers. However, it is within the maximum annual residential move distance. Thus, the geographical evidence suggests that Soro Mik'aya Patjxa was unlikely to have served as a logistical camp tethered to a low-elevation base (model 1). Rather, the geographical position is more consistent with a site that was part of a settlement system that either operated seasonally or year-round at high elevation (models 2 or 3).

H4: demographic composition
If a given high-elevation site served as a logistical camp operating from a low-elevation base (model 1), we would expect the site's demographic profile to be highly skewed such that children, old adults and either males or females [36]-depending on the logistical task-are under-represented. Specifically, young children between about 4 and 8 years old would probably have been absent from logistical forays, given that they would have been a burden to carry and would have been unlikely to maintain adult walking speeds over long distances even though they might be able to attain such speeds [37]. The ratios of women : men or vice versa would be statistically greater than 50%. By contrast, if a given site was part of a settlement system that was seasonally or permanently situated within that environment (model 2 or 3), we would expect the demographic profile to include the full spectrum of age classes and both sexes in statistical parity (i.e. statistically indistinguishable from 50 : 50).
Although the age profiles derived from burial assemblages tend to be biased reflections of living population structures [38][39][40], the mere presence of multiple young children or old adults would constitute sufficient evidence for their presence at the site and thus grounds to refute the logistical mobility model (model 1). Sex ratios as a test of the working hypotheses are more sensitive to interpretive error, given that task group organization and burial practices may or may not be gendered. This component of the demographic evidence is therefore interpreted with greater caution.

Results
Demographic observations on 16 individuals reveal a profile that is similar to that observed in numerous non-industrial societies [41], including age classes ranging from young children (4-6 years old) to old adults (50+ years old) and a female : male ratio in statistical parity (x 2 1 = 0.69, p = 0.41; figure 6 and table 1; electronic supplementary material). This diverse demographic profile is most consistent with a population that occupied the highlands on either a seasonal or permanent basis (model 2 or 3) and inconsistent with logistical use of the highlands (model 1).

H5: lithic raw-material provenance
If hunter-gatherers moved logistically into high-elevation environments (model 1), we would expect them to have habitually transported and deposited non-local materials at high-elevation sites such as Soro Mik'aya Patjxa. If they moved seasonally into high-elevation environments (model 2), we would expect low levels of non-local lithic materials. If foragers occupied high-elevation environments permanently (model 3), then we would expect a near absence of non-local raw materials from low-elevation environments [8,10].
The trade of non-local materials could complicate such interpretations, but an absence of non-local materials would nonetheless support the model of permanence. Another complicating factor is rawmaterial isomorphism across broad geographical extents. While it may be parsimonious to interpret archaeological materials with local analogues as locally acquired, it may be difficult or impossible to completely rule out the possibility of non-local acquisition.

Results
Of the 539 Middle and Late Archaic lithic artefacts examined from the Ilave Basin, 534 (99.1%) were made from raw materials that we located in a raw material survey of the region (electronic supplementary material, table S3). The few samples that did not have clear analogues in our comparative sample included unexceptional materials that may simply reflect extremes in the range of variation in the local

Summary and discussion
From three general models for hunter-gatherer use of high-elevation environments in the Andes, we deduced five sets of archaeological hypotheses that we tested against data from the site of Soro Mik'aya Patjxa. All observations are consistent with the expectations for permanent use of high-elevation environments (table 2). No support was found for the logistical use hypotheses, and only partial support was found for the seasonal use hypotheses. We, therefore, conclude that the hunter-gatherers of Soro Mik'aya Patjxa occupied the highlands on a permanent basis by at least 7 ka. Archaeological data are notoriously ambiguous, and any one of the tests reported here is subject to varying degrees of uncertainty. Although careful consideration of the data in the light of competing interpretations can go some way towards identifying the most likely explanations [42], alternative explanations may be difficult or impossible to reject entirely. Regarding hypotheses 1 and 2, palaeoclimatic variation and diagenetic processes could potentially explain the observed stable oxygen and carbon isotope values even though the values are more likely to reflect water and food intake by the Soro Mik'aya Patjxa individuals. For hypothesis 3, it is possible that Archaic Period hunter-gatherers made logistical forays to Soro Mik'aya Patjxa from low-elevation base camps even though the extreme distances entailed suggest that long-term use of the highlands was more likely. For hypothesis 4, it is certainly possible that young children trekked over 122 km with more than 1300 m in elevation gain from low-elevation base camps to Soro Mik'aya Patjxa even if that possibility is unlikely. As for hypothesis 5, it is possible that some of the lithic materials examined actually came from distant low-elevation regions even though such an interpretation is less parsimonious than one of local acquisition. Careful consideration of a given line of archaeological evidence can lead to the most likely explanation, but it rarely guarantees freedom from error.
Despite uncomfortably high levels of uncertainty for any one line of evidence, the use of multiple independent lines of evidence can exponentially reduce uncertainty in the greater effort to select the best behavioural models from a candidate set. We considered five lines of archaeological evidence from Soro Mik'aya Patjxa, and although it is difficult to ascribe numerical weight to each result, the convergence of all five lines of evidence on one solution exponentially diminishes the possibility of erroneous model selection. This study, therefore, to our knowledge, offers the most secure evidence for early permanent use of Andean highlands to date and adds to a growing body of empirical support for the theoretical expectation of permanent highland use after 9.0 ka [6,8,10]. A number of other high-elevation sites corroborate the interpretation of early permanent use of the highlands albeit with fewer or less conclusive lines of evidence. Low-elevation lithic materials were rare or absent from the highland sites of the Junín Puna and Rio Osmore regions after 9 ka [8,10]. The geographical positioning of many hunter-gatherer sites in interior, high-elevation locations further opposes the logistical use model (model 1) [43][44][45]. Such sites are well beyond typical logistical foraging ranges but within annual mobility ranges. Given that at least some of those sites date to the Terminal Pleistocene [45,46], the geographical evidence suggests that hunter-gatherer groups may have begun to use the highlands on at least a seasonal basis beginning in the Late Pleistocene. Archaeological evidence for the timing and structure of hunter-gatherer use of the highlands remains limited, but continued research considering multiple lines of evidence will enhance empirical resolution.
For now, the findings of this study offer a critical parameter value towards modelling rates of cultural and genetic change in the Andean highlands-a region known for the domestication of alpaca, potatoes and quinoa; the emergence of state-level political and economic complexity; and rapid genetic evolution in response to high-elevation stressors. The findings presented here show that estimates for the rates of these cultural and genetic changes can be constrained by a terminus ante quem date of 7 ka for permanent occupation of the Andean highlands.