Permanent Human Occupation of the Western Tibetan Plateau in the Early Holocene

Abstract

Archaeological investigations worldwide have focused on when and how humans permanently settled in high-altitude environments. Recent evidence from Xiada Co, Qusongguo, and Dingzhonghuzhuzi in western Tibet, where lithic artifacts and radiocarbon dates with original deposits were first accessed, provides new insights into human activities in this extreme environment during the early Holocene. This paper examines the mobility and land-use patterns of foragers in western Tibet from the perspectives of lithic analysis. Assemblages from three sites suggest homogenous technologies and raw material use, as well as potential interaction network of hunter-gatherers within the plateau during the early Holocene. It further argues that the material exponents and travel cost models of site location supported permanent occupation of the western Tibetan Plateau in this period.

1. Introduction

Archaeological investigations worldwide have been interested in examining when humans settled at high altitudes (more than 2500 m above sea level). In the past decade, archaeological studies have demonstrated that Denisovans occupied the northeastern corner of the Tibetan Plateau at an elevation of 3280 masl as early as 160,000 years ago [1]. Some have even claimed that they simultaneously appeared on the Tibetan hinterland [2]. Others have argued that humans lived permanently in the Central Andean highlands by 12,000 BP [3], and new evidence from Fincha Habera in Ethiopia has indicated that people had lived there between 47,000 and 31,000 years ago [4]. Although whether these early settlements on the Ethiopian and Andean plateaus were permanent is being debated [5,6], a growing body of evidence suggests that humans have occupied high altitudes at least as late as the early Holocene. However, in the case of the Tibetan plateau, the dominant opinion remains that permanent occupation of the Tibetan Plateau only manifested after 3600 BP; that is, the introduction of farming, specifically barley or wheat-dominated agriculture [7]. The critics have argued that this overlooks the contribution of earlier (around 5000 years) millet farming on the southeastern margin and those of hunter-gatherers to permanent occupation [8,9]. For example, Meyer et al. [10] conducted travel cost analysis and compared it with ethnographic hunter-gatherer data and argued that the Chusang site indicates permanent occupation of the Tibetan plateau at least 8000 years ago. However, Zhang et al. [11] argued that this evidence is not sufficient.

We argue that the crux of this debate depends on how we define permanent occupation and what archaeological evidence can provide the support. At high elevations, very few individual sites were occupied year-round, and when discussing permanent occupation at high elevations, we should be referring to permanent use of the upland landscape rather than individual sites. In the mode of permanent high-altitude landscape occupation, foragers lived and moved within the high-altitude environment for most of the year to sustain their livelihood. We may see different parts of mobile settlement systems on the uplands in archaeological records; in the mode of non-permanent high-altitude landscape occupation, where hunter-gatherers moved between lower and higher elevation environments depending on resource availability, we may see sites of seasonally moving groups or logistical task groups sent from lower elevation dwellings.

1. Xiada Co (4350 masl); 2. Dingzhonghuzhuzi (4285 masl); 3. Qusongguo (4305 masl); 4. Nwya Devu (4600 masl); 5. Chusang (4230 masl); 6. Tshem gzhung kha thog (4100 masl); 7. Jiangxigou 1&2, 93-13 (3330 masl); 8. 151 (3397 masl); 9. Heimahe 1 (3202 masl), Heimahe 3 (3210 masl); 10. Layihai (2600 masl); 11. Baishiya (3280 masl); 12. Oshhona (4100 masl). 13. General location of Beshkent, Javan, Mullo-Nijaz, Makoni-Mor; 14. Dzamathang (3101 masl). Previous studies on high-elevation land-use patterns have mainly focused on the eastern Tibetan Plateau (Figure 1). Archaeological investigations on the northeastern Tibetan Plateau have revealed traces of dozens of hunter-gatherer sites along the lakeshores and river terraces 14,000 to 6000 years ago [12,13,14,15,16,17]. Evidence of microblade, short-use hearths, and faunal remains facilitate our understanding of different occupation patterns. Travel cost modeling and ethnoarchaeological analogies have also suggested the possible existence of seasonal movement between low- and high-elevation environments in this period [18]. It is believed that these sites are probable nodes within a single foraging system that developed on the margins of the Tibetan Plateau and served as the basis of its colonization [19].

The northeast Tibetan Plateau connects the plateau and lowlands and is dotted with river valleys with relatively rich biological production, wildlife diversity, and water resources. It provides access to the central and northern high-altitude areas of the Tibetan Plateau. However, investigating when humans permanently settled in the harsh environment of the western Tibetan Plateau, which is separated by extremely high mountains and limited movement paths to surrounding lower elevations, is quite challenging. The Nwya Devu site, dating from about 30,000 BP, is located in the northern Tibetan hinterland. It is the least populated part of the presently inhabited Tibetan Plateau, and sheds new light on the debate regarding the permanent occupation of the plateau [20].

Since the National Cultural Heritage survey was conducted in 1992, more than 20 microblade sites have been found in the very high altitudes (>4000 masl) of western Tibet [21]. However, despite the sporadic appearance of surface-collected artifacts, few of these sites have been systematically dated or excavated. In 2019, Sichuan University initiated a project on Early Settlement Archaeology in western Tibet to revisit those surface-collected sites found 15 years ago and have identified a series of early sites with deposits of in situ archaeological remains. Our fieldwork in Xiada Co, Qusongguo, and Dingzhonghuzhuzi in the Ngari region provides potential evidence regarding the occupation of the western Tibetan Plateau during the early Holocene. We conducted test excavations at several sites and obtained new AMS radiocarbon dates from stratigraphy for the very first time (Figure 2 and

Table 1. Original information of the C14 dating of sites.

No.	Material	BP	Uncertainty
Xiadaco_L2	charcoal	7860	±30
Xiadaco_L4	charcoal	7830	±30
Qusongguo_1	charcoal	9080	±30
Qusongguo_2	charcoal	9220	±30
Dingzhonghuzhuzi	charcoal	8150	±30

). In this paper, we present the preliminary results obtained from these sites and attempt to answer the following questions based on the technological analysis of the lithic assemblages:

(1)

What are the technological characteristics of the lithic artifacts from the three sites, and what are the similarities or differences between them?

(2)

What technological strategies and activities may have been carried out at these sites that indicate patterns of mobility and land use on the western Tibetan Plateau during this period?

2. Archaeological Sites

2.1. Xiada Co (79.36° N, 33.4° E, 4400 masl)

Xiada Co is a freshwater lake that is 5.9 km long and has an area of about 8 km². It is located in flat terrain between the Gangdise in the north and the Himalayas in the south, at an altitude of 4400 m—the primary channel that connects Rutog with Kashmir, dotted with alluvial fans shaped by mountain gullies. The lake is replenished by rainwater carried by rivers and partially from water that melts from the nearby glacier [24]. The site was first discovered in 1992 by Huo and Li, who collected 92 stone artifacts on the surface of the northeast shore during the survey. In 2019, we found a new location on the second terrace of the north shore with a dense distribution of slightly weathered lithic artifacts. We excavated a 1 × 1 m test pit at this new location and identified four stratigraphic layers (Figure 3b):

Layer 1: Surface; light-yellow coarse sand with gravel;

Layer 2: Dark fine sand with a small amount of small gravel, about 10–30 cm thick. The layer contained abundant lithic artifacts, charcoal, faunal remains, and fire-affected rock fragments;

Layer 3: Compact, yellow fine sand; about 5 cm in thickness;

Layer 4: Yellow coarse sand with a lot of gravel. A fewcharcoal fragments and lithic artifacts were found in this layer.

We collected lithic artifacts from the pit and from the surface approximately 120 m² around the test pit (Figure 3a). Charcoal samples in layer 2 and 4 were dated to 8772–8575 Cal BP and 8718–8653 Cal BP (95.4%).

2.2. Qusongguo (31.12° N, 80.63° E, 4325 masl)

Qusongguo is located near the village of Menci in the Menci Basin of the upper reaches of the Sutlej River (Figure 1). It is located at the foot of the famous Kardong site, where three tributaries (Quga, Langqin, and Quna) of the Sutlej River converge. The site was first discovered in 2004 by the Sichuan University archaeological team with only surface collections. In the resurvey conducted in 2020, we discovered a new location on the second terrace of the Quga River, and most of the surface artifacts showed signs of light weathering. At this location, we identified and excavated a near-circular concentration of cobbles, which was likely a hearth (Figure 4b). We excavated a 1 × 1 m test pit next to the hearth feature, and identified four stratigraphic layers (Figure 4c):

Layer 1: Surface layer, containing fine yellow-brown silt with gravel;

Layer 2: Dark silt containing lithic artifacts, charcoal fragments, and fire-affected rock fragments;

Layer 3: Compact, gray-white silt with fire-affected rock fragments;

Layer 4: Grayish-yellow coarse sand with gravel. No archaeological remains were found in this layer.

Charcoal samples in Layer 2 were dated to 10,281–10,260 Cal BP and 10,496–10,441 Cal BP (95.4%). We collected lithic artifacts from the pit and from the surface approximately 100 m² around the feature and the test pit (Figure 4a). No faunal remains were recovered at Qusongguo.

2.3. Dingzhonghuzhuzi (32.41° N, 80.02° E, 4305 masl)

Dingzhonghuzhuzi is located on an alluvial fan in the valley of Shiquan River. The site was first discovered in 1992 by the Sichuan University archaeological team. Previous surface findings include axes, ceramic sherds, scrapers, choppers, and a microblade assemblage dominated by conical cores. In 2020, a new location was discovered with dense distribution of lithic artifacts without any ceramic sherds or ground-stone tools. The exposed lithic artifacts were located in a layer of dark soil, and the charcoal for dating came from this layer (Figure 5). We excavated a 1 × 1 m test pit at this location. Samples found in this layer were dated to 9223–9177 Cal BP (95.4%). As there was only one shallow-buried layer, we believe that these collections, the layer of black silt and charcoal in this area can be considered contemporaneous.

3. Lithic Assemblages

In this section, we examine the technological characteristics of lithic assemblages from the three sites and compare them. We regard the surficial and in situ artifacts as a single assemblage for three reasons: (a) the surface collection area was about 100 square meters around the test pits, and most of the small artifacts had sharp edges, indicating they had not been strongly displaced; (b) there was only one period for the cultural layer at the three sites; (c) surface and in situ lithics showed a similar technological type, raw material, and size.

We identified lithic raw materials macroscopically. The lithic analysis approach combines two complementary methods: reduction sequence and attribute analysis [25,26,27]. The first evaluates the core reduction sequences and stages of knapping. For each assemblage, artifacts were identified and divided into simple core-flake assemblage and microblade assemblage based on distinct technologies. Each assemblage was counted and assigned to four groups: cores (including lumps with single scars), blanks (including flakes and microblades), chips/chunks, and retouched tools (

Table 2. Count and frequency of all artifacts from the three sites.

	Xiada Co				Qusongguo				Dingzhonghuzhuzi
Category	Surf.	L2	L4	Total	Surf.	T1	H1	Total	Surface
Core-Flake Assemblage	65	13	2	80 (21.2%)	64	1	1	66 (26.1%)	187 (75%)
Core	7	1		8 (2.1%)	5			5 (0.2%)	8 (3.2%)
Complete flake	33	3		36 (9.6%)	27			27 (10.7%)	112 (45.3%)
Flake fragment	25	9	2	36 (9.6%)	32	1	1	34 (14.8%)	67 (27.1%)
Microblade Assemblage	7	5	2	14 (3.7%)	53	1	5	59 (23.3%)	6 (0.2%)
Microblade core	3			3 (0.8%)	6			6 (2.3%)	4 (1.6%)
Microblade	3	2	1	6 (1.6%)	15		4	19 (3.6%)
Technical piece	1	3	1	5 (1.3%)	32	1	1	34 (13.4%)	2
Tool	83	2	1	86 (22.9%)	10			10 (4.0%)	23 (9.3%)
Chips and Chunks	62	97	37	196 (52.1%)	108	2	8	118 (46.6%)	31 (12.1%)
Total	217	117	42	376	235	4	14	253	247

). The second provides quantitative data on technological choices to characterize each assemblage and facilitate comparison between them. Attributes were recorded for individual artifacts ≥ 1 cm(after Scerri et al. [28], with some alternation) including metric characteristics (Figure 6) and patterns of the striking platform, dorsal surface, ventral surface, and termination. Retouched tools were described following the criteria proposed by Inizan et al. [29].

3.1. Lithic Assemblages from Xiada Co

In 2019, a total of 376 lithic artifacts were collected from Xiada Co, including 159 from the test pit and 217 from the surface. Various raw materials were carefully selected, wherein siliceous rocks ((Figure 7(1–3))and sandstone (Figure 7(4)) constituted the majority, Both materials could be derived from the rock formations in the mountain gullies and alluvial fans toward the north of the lake. Other artifacts were made of obsidian (Figure 8(13)), brown chert (Figure 8(4)), red chert, and crystal, which may not have been sourced locally. High-quality raw materials, such as obsidian, chert, crystal, and local siliceous rocks, were only exploited by knappers in microblade production. Core-flake production tended to focus on the local siliceous rock and sandstone.

Among the simple core-flake assemblage, only eight cores were identified, mainly characterized by single platforms with no preparation and unidirectional scars on the flaking surface (Figure 7). Three cores are still covered with cortex. Morphometrical features of the flakes are not standardized. Flakes frequently exhibited cortical (25%) and unidirectional (46.4%) scar patterns, and were unique or parallel-aligned in morphology on the dorsal surface. While we found cortical platforms, plain platforms (79%) dominated the flakes. Together, these features displayed a continuous, unidirectional use of the same flaking surface and platform without preparation during reduction. All the flakes had thick platforms (2.3–17.6 mm in thickness, mean value = 7.3 mm), and most of them demonstrated prominent bulbs (70%), which could have been caused by direct percussion using a hard hammerstone.

We also identified a few artifacts related to the microblade technology, including wedge-shaped and boat-shaped microblade cores, microblades, and overpassed flakes (flaking surface rejuvenation pieces). Microblade cores were carefully prepared, all of which belonged to the early stages of reduction, and could be further detached (Figure 8).

Tools were well represented in the assemblage (n = 86, more than 20% of all the collected lithic artifacts; Figure 9). Endscrapers accounted for the majority of the complete tools (n = 46, 67%), followed by convergent and single-edged sidescrapers, bifacial pieces, and notched pieces (

Table 3. Types of tool assemblages from the three sites.

Tools	Xiada Co	Qusongguo	Dingzhonghuzhuzi
Endscraper	46	5	15
Single-edged sidescraper	3	1	4
Double-edged sidescraper	2
Convergent sidescraper	10	2	2
Bifacial piece	5	1	1
Notched piece	3	1	1
Tool fragment	17
Total	86	10	23

). Flakes and a few chunks were selected as blanks. Morphological studies suggested that flake tools were different in size from the unmodified flakes, indicating a potential preference for larger blanks at Xiada Co (mean value in length, width, thickness = 57.1 mm, 49.6 mm, 15.7 mm of retouched pieces; and 41.5 mm, 33.4 mm, 10.9 mm of unretouched pieces). Most of the retouched edges were shaped unifacially in a direct position, and alternate and bifacial retouches were also present. We found several tools covered by continuous and invasive removals (n = 36), whereas others had random and light removals on their edges. Formal tools featured circular endscrapers (Figure 8(3,4)) and endscrapers made on thick flakes with a steep retouching (similar to the “Rabot” in Borde’s list; Figure 9(1,2)).

3.2. Lithic Assemblages from Qusongguo

We collected 253 lithic artifacts from Qusongguo in 2020, including four pieces from the test pit (T1), 14 from the feature (H1), and 235 from the surface. Diverse raw materials were used for the lithic production, of which siliceous rock, sandstone, and andesite were dominant and could be procured locally. Sources of obsidian, crystal, and chert have not been identified so far, and they may have been exotic. All materials were exploited to produce both simple flakes and microblades, except obsidian and crystal, which were only observed in the microblade assemblage.

Six artifacts were defined as simple cores without preparation of platforms and flaking surfaces. Most of them had opposed platforms (n = 3) and multiple platforms (n = 2). One of the cores and 13% of the flakes retained more than half of the cortex. Flakes were varied in morphology, which suggested a low degree of standardization. They were characterized by plain platforms (95%), unidirectional (50%) and bidirectional (10%) scars patterns, relatively thick platforms (3.5–14.7 mm in thickness, mean value = 7.9 mm), and prominent bulbs (73%). Attributes observed on the flakes and the cores indicate that unidirectional, direct percussion using a hard hammerstone was the main method in core reduction. Additionally, opposed and multi-striking platforms might have also been used (Figure 10).

Microblade cores were in the exhausted stage, one of which was wedge-shaped, while it was difficult to identify the shape of the remaining cores. A few technical pieces of microblade production were observed, including core tablets, overpassed flakes, and elements that could be considered as knapping “errors” or by-products of the early reduction phases (Figure 11).

Only ten pieces of tools were identified at this site, including endscrapers, convergent and single-edged sidescrapers, notched pieces, and bifacial pieces (Figure 12). All tools were made of flakes. Most of them displayed direct removals on their edges, and bifacial and alternate ones were also present. Removals on the retouched edges were short and discontinuous, and we did not identify any formal tools.

3.3. Lithic Assemblages from Dingzhonghuzhuzi

A total of 247 lithic artifacts were collected from Dingzhonghuzhuzi. Several different raw materials were used for lithic production. Consistent with the other two sites, only high-quality chert, crystal, and obsidian examples, which might have been sourced from distant places, were found in the microblade assemblages. Potentially local siliceous rocks, sandstones, hornfels, and andesites were used for the simple core-flake production.

Among the simple core-flake assemblages there were eight cores in total: two single-platform cores, two double-platform cores, and four multi-platform cores. Only one contained a small amount of cortex. None of the cores displayed signs of prepared platforms. The multi-platform cores displayed polyhedral morphologies and multidirectional negatives of removals. Most of the flakes were complete (91%), characterized by quadrangular shapes and feathered distal ends (89%). Dorsal scar patterns on flakes were mainly unidirectional (41%), perpendicular (29%), and centripetal (30%). This was consistent with the cores, indicating that unidirectional and multidirectional knapping by rotating the cores were adopted for the reduction process. Almost all the flakes had wide and thick platforms (mean value = 22.5 in width, 9.5 mm in thickness), and most of them had plain platforms (95%) as well as prominent bulbs (76%). This suggests that the flakes were detached from unprepared platforms and subjected to direct percussion with a hard hammerstone. A few flakes (15%) had more than 50% cortex on the dorsal surface, while others were almost completely deprived of cortex. To conclude, these features exhibited a stable production of flakes and suggest proven knapping skills in Dingzhonghuzhuzi (Figure 13).

Four microblade cores and two pieces of rejuvenation flakes related to microblade technology were also identified. Morphologically, conical and wedge-shaped cores were found in the assemblage. Microblade cores found at the site were either broken along the joints during detachment or were already in the exhausted stage (Figure 14).

Further, we identified 23 tools in the collection, most of which were endscrapers (65%), followed by convergent and single-edged sidescrapers, notched pieces, and bifacial pieces (Figure 15). Flakes and chunks were selected as blanks. The proportion of flakes (n = 21, 13.5% of all the flakes including tool blanks) that was transformed into various types of retouched tools was relatively low. Most of the retouched edges presented direct, short, and discontinuous removals. We also observed thick endscrapers with steep retouching, similar to those found at Xiada Co (Figure 15(2)).

3.4. Comparative Analysis of the Lithic Technology

We found that local resources such as siliceous rocks and sandstones were used as primary raw materials at all three sites. Hornfels and andesites were also observed at Dingzhonghuzhuzi. These raw materials were used for simple core-flake production and occasionally microblade production. Meanwhile, high-quality raw materials including obsidian, crystal, and chert, were only observed in the microblade assemblages from the three sites (Figure 16). These materials could have been exotic.

It can be postulated that all assemblages from the sites were clearly oriented toward the production of both simple flakes and microblades. The main method adopted for the production of simple core-flakes was unidirectional direct percussion using a hard hammerstone. Both striking platforms and flaking surfaces of these cores were not prepared. Some differences were observed in the assemblages of Dingzhonghuzhuzi, evidenced by the use of multidirectional knapping. We also found evidence of wedge-shaped microblade cores at the three sites, with a boat-shaped core found in Xiada Co and a conical core in Dingzhonghuzhuzi. The number of microblade cores and related products collected from the sites is far from enough. The collected evidence has been merely used to glean morphological information of microblade cores. Consequently, we need more material and experimental research to replicate the chaîne opératoire and discuss the relationship between various technologies. In general, lithic assemblages from the three sites displayed a general similarity to the ones corresponding with the northeast Asian late Upper Paleolithic period.

Endscrapers and sidescrapers constituted the majority of toolkits at the three sites. A small number of bifacial pieces were also observed in each assemblage. Tools were mainly manufactured from flakes, wherein most of the edges displayed unifacial and direct retouching. Similar formalized tools, such as endscrapers made on thick flakes with steep retouching, were found among the collections of both Xiada Co and Dingzhonghuzhuzi.

Lithic analysis suggests that the relatively contemporaneous sites of Xiada Co, Qusongguo, and Dingzhonghuzhuzi on the western Tibetan Plateau shared similar technologies during the early Holocene (ca. 10,000–8000 BP). Based on the overall consistency of the material culture, we further explored the technological strategies, mobility patterns, and land-use patterns of the inhabitants across the landscape during this period.

4. Discussion

4.1. Technological Strategies and Site Functions

We employed the theoretical framework of technological organization, wherein the sequence of lithic technology from procurement to discarding is related to behavioral patterns, which allows the reconstruction of site functions and mobility strategies [30,31,32,33]. In general, highly mobile groups typically prefer more organized production following a curatorial strategy to keep their tool supplies portable and maintainable. In contrast, less mobile groups may develop under conditions where raw material and time are not major concerns. Such groups may develop an expedient strategy for lithic production. We made the following interpretations of site functions by examining technological strategies of the assemblages and archaeological contexts from the test excavations.

At the Xiada Co site, local siliceous rocks and sandstones from the northern alluvial fan were used as raw materials. Simple core-flake products in various stages of reduction were found. Cores including lumps with single scar, flakes, chips, and chunks were also observed. Several flakes and cores contained more than 50% cortex. Microblade cores were in the stages of preparation or detachment. These features, together with the abundant chips found in the test pit, indicate that the initial reduction of these cores might have been carried out on site. Further, larger flakes were intentionally selected as blanks of tools, which constituted a considerable percentage of the assemblages. Both formal tools with well-shaped edges and flakes with random light retouching were observed in the assemblages, suggesting the coexistence of expedient and curated strategies for tool manufacturing. From the examination of the 50 cm thick ash layer containing abundant charcoals and faunal remains, we inferred that Xiada Co might have been a multifunctional site during the early Holocene, playing the role of a workshop or basecamp for the hunter-gatherer settlement system. Cores and blanks that could be used might have been taken away from the camp for other activities, leaving early products and chips as remnants. The curated tools can be considered as provisions, such as personal gear, prepared for task groups. Certain tools could have been prepared for specific and repeated uses in the base camp, such as butchering or processing the resources.

At Qusongguo, locally available raw materials were flaked expediently. Simple core-flake products included a small number of cores and flakes in both primary and secondary stages. We also observed some marginally retouched flakes. As noted above, obsidian, crystal, and chert—used for microblade production—were probably sourced from outside, brought to the region by people. Microblade cores were highly exhausted, and there were few microblades with irregular shapes—most of them were defected during reduction. Further, technical pieces reflected that the striking platforms and cores’ flaking surfaces had been adjusted and rejuvenated several times. These features, taken together, highlight an overwhelming utilization of high-quality raw materials and the very strong demand for microblade products. Furthermore, the limited size of the hearth, the smaller quantity of charcoal fragments, as well as the lack of faunal remains, suggest that Qusongguo might have been an ephemeral overnight camp of a task group. People who lived here focused on microblade production, possibly for hunting tasks. They carried obsidian and chert (possibly also prepared microblade cores) and discarded the useless by-products and exhausted cores after their short stay.

Lithic analysis suggests that the inhabitants of Dingzhonghuzhuzi attached importance to both simple core-flake and microblade production. Simple core-flake products from Dingzhonghuzhuzi include cores, flakes, and few chips and chunks. Abundant, potentially local-originated rocks were used as raw materials for the flakes. Most of the complete flakes were in the secondary stage (according to the cortex coverage). This, along with the scarcity of chips and chunks, suggests that the systematic initial reduction of cores occurred somewhere else. Flakes with similar sizes and sharp edges and the frequent change of platforms and flaking surfaces during reduction suggest a greater utilization of cores. Few flakes were worked into retouched pieces. They were most probably used directly, without further processing. In general, the popular strategy would have been to produce flakes in large quantities before careful selection. The size of microblade cores was quite small and they were found in their exhausted or disposal stage, which reflects an intensive exploitation of high-quality raw materials. Based on the above observation, it can be inferred that Dingzhonghuzhuzi might have served as a site for processing certain types of resources (such as animals or plants). The basecamp and workshop of this settlement system might have been near the site.

4.2. Land-Use Patterns and Mobility Strategies of Early Holocene Hunter-Gatherers in Western Tibet

The analysis above reflects different mobility patterns of the occupants of the three sites. It can be inferred that the hunter-gatherers might have established their basecamps on the plateau. To examine the land-use patterns of the western Tibetan Plateau during this period, we referred to the ethnographic studies of hunter-gatherers, such as Kelly [34]. The compilation of the round-trip distance of logistical move averaged 42 km, and all fell under 80 km. The total annual distance of residential moves averaged 148 km and rarely exceeded 1000 km. For the three sites to be considered as residential or logistical camps of people sent from the lowlands, the distance between the high- and low-elevation environments for hunter-gatherers would be the distance from the archaeological sites to points on the 2500 m topographic contour. According to the ethnographic records, the nearest distance is roughly correlated with the land-use patterns of hunter-gatherers as follows(

Table 4. Round-trip distances and their supported land-use patterns.

Round-Trip Distance	Supported Land-Use Patterns of Hunter-Gatherers in Highlands
≤80 km	Both permanent and non-permanent use (logistical or residential moves between high- and low-elevations) of highlands
80–1000 km	Residential move from lowlands or permanent occupation of highlands
>1000 km	Permanent occupation of highlands

).

The western Tibetan Plateau and the lowlands in the east are thousands of kilometers apart. Logistical or residential moves to the eastern lowlands within a year were far beyond hunter-gatherers’ mobility capacity. Lowlands along the west and northern fringes of the plateau were closer; however, the straight-line round-trip distance from any of the sites is more than 500 km. Therefore, the logistical challenges made moving between highlands and lowlands unlikely. We conducted a preliminary least-cost travel analysis to simulate the possible route from the western Tibetan Plateau to the surrounding lowlands following the procedure proposed by Meyer et al. Seven nearest potential locations were identified for posited temporary use of Xiada Co by the low-elevation hunter-gatherers. Considering terrain factors, we employed Tobler’s hiking function [35] and Tripcevich’s vertical factor table [36] to construct a cost model. Additionally, we utilized the formula derived from experiment, which relates the time required to walk 1 m to the slope [37], represented as Y = 0.0002X^2 + 0.002X + 0.6086 (as Y is the time in seconds required to walk 1 m, X is the slope), to estimate the walking times between various points from Xiada Co to the destination sites. As shown in Figure 17 and

Table 5. Simulated distance and time required for a one-way trip from the Xiada Co site to the seven random points.

Path	Distance/km	Time/h
D1	820.5764	205.1441
D2	973.235	243.3088
D3	1256.113	314.0282
D4	625.2065	156.3016
D5	741.7871	185.4468
D6	695.7309	173.9327
D7	869.3044	217.3261

, the minimum round-trip distance between Xiada Co and the nearest lowland is 1250.4 km. The distance would be similar for the other two sites as well. Additionally, challenges posed by the harsh environment of the Ngari area, such as average altitude of 4000 m and inaccessible mountains because of snow buildup, shortened the time available for travel during a year and increase the travel cost. Considering these constraints, it is also unlikely that the early humans who occupied the sites moved between low and high elevations in an annual around.

On the other hand, a hypothesis of non-permanent occupation of the highlands also needs supporting evidence of material culture and dating results from archaeological sites in the surrounding lowlands. That is, hunter-gatherers on the western Tibet Plateau during this period should also have left tracks on the lowlands adjacent to the edge of the plateau. In northern Afghanistan and eastern Kyrgyzstan, microblade cores were dominated by a bullet shape that is typical of Central Asia [38]. In India, a series of Mesolithic industries were found all over the country including discoveries in the Spiti Valley [39]. However, evidence from lowlands in the trans-Himalaya regions remains scant. Lithic artifacts of this period found in northern India were characterized mainly by geometric microliths, which is another significant tradition other than the microblade technology dominant in east Asia [40,41]. To the north of the Tibetan Plateau, microblades have been found almost all over the territory of the Xinjiang Autonomous Region, including findings from the foothills of the Kunlun Mountains and the eastearn Pamir Plateau. There is no lack of wedge-shaped microblade cores, but the accurate age of most sites remains unknown [42,43]. Evidence for similar lithic technology with available dates from western Tibet’s adjacent areas could be found in Tajikistan. Current discoveries with similar wedge-shaped microblade cores include sites such as Oshhona, Beshkent, Makoni-mor, and Mullo-nijaz located along the margin of the Pamir Plateau and the surrounding lowlands. Of these, Oshhona has been dated to 8509–7880 Cal BP and 8163–7690 Cal BP [38,44,45]. They are chronologically close to the three sites described in this paper. However, the geographical distance and topographical patterns are potential obstacles that hindered frequent travel between high- and lowlands across the Pamir Knot.

In the studies on the Andean highlands, scholars have suggested the concept of the south-central dry puna as an ecological “megapatch”, wherein dispersed but interconnected settlement systems were established by hunter-gatherers [46,47]. Homogenization of cultural traditions is expected in the highlands. While there could have been communication between high- and low-elevation areas, it is likely that there was a cultural distinction between highlands and lowlands. Forager sites in the Andean puna might have shared several similarities in campsite locations, hunted animals, raw material acquisition, and plant resource use. Once the occupants of the Andean puna gained initial knowledge of key resource acquisition and procession, they would have duplicated this experience elsewhere in the highlands. Similarities in lithic assemblages and formal tool types suggest common activities and shared culture over vast and rugged landscapes. The concept of the megapatch could help explain cultural similarities over large areas of land because hunter-gatherer groups adapted to the highland landscapes and ecosystems through long-term occupation.

This concept sheds light on the hunter-gatherer sites in western Tibet. Available evidence suggests an overall consistency in lithic technology and formal tool types. Hunter-gatherers of the western Tibetan Plateau during the early Holocene adopted a combination of simple core-flake and microblade technologies. Inhabitants of Xiada Co, Qusongguo, and Dingzhonghuzhuzi used raw materials in the same way. Raw materials that were local, exotic, and had different textures were designated for different technology groups. Siliceous rocks and sandstones were among the locally sourced raw materials used to produce simple flakes and occasionally microblades. High-quality raw material, such as obsidians and crystals, were only used for the production of microblades. Perhaps the inhabitants had formulated a structured use of the raw materials.

The lakeside terrace where Xiada Co is located, the Sutlej River valley area where Qusongguo is located, and the piedmont alluvial fan area in the Shiquan River Basin where Dingzhonghuzhuzi is located are near environments abundant in water, animals, plants, as well as lithic raw material resources. Obsidians could also have been widely sourced across the Tibetan Plateau [48]. These early Holocene hunter-gatherers appeared to rely on the resources in the highlands, have an empirical knowledge of the distribution of these important resources, and share similar strategies for acquiring and processing them.

Additionally, adoption of microblade technology and the variety of raw materials at the three sites suggest the existence of a high-mobility strategy during the early Holocene. The demand for diverse raw materials and scarce obsidian (also mahogany obsidian) may have caused hunter-gatherers to travel between different landscapes. Transportable microblades and formal tools may have offered increased flexibility and efficiency to the hunter-gatherers, acting as an adaptive solution to the highly patchy environment of the western Tibetan Plateau. The occupants may have been involved in larger exchange systems and may have developed social networks among the hunter-gatherer groups on the plateau to obtain high-quality raw materials.

Xiada Co, Qusongguo, and Dingzhonghuzhuzi were occupied for varied durations. Although the material evidence indicates that the sites were not used on a year-round basis, it does not deny the permanently occupation of the Tibetan Plateau. On the contrary, short-term occupation of multiple localities may have been used as a strategy for living in the highlands. This could have helped hunter-gatherers to maximize the acquisition and utilization of resources. As Osorio et al. [46] proposed, the three sites, occupied spanning a period of 2000 years, may reflect a highland way of life that developed over generations of permanent occupation within this megapatch.

Studies on lake cores, paleo-shorelines, and fluvial terraces permit a reconstruction of the Holocene environmental changes on the western Tibetan Plateau [49,50]. Remains of pollen, diatoms, ostracods, and carbonate stable isotopes indicate a warmer and more humid era compared to today. This is due to the abrupt climate change that occurred from 10.0 to 9.5 ka BP until about 6.0 ka BP. At this juncture, we propose that the early foragers already permanently occupied the interior plateau along the mountains and rivers in western Tibet during this period. Archaeological sites on the plateau, including Xiada Co, Qusongguo, and Dingzhonghuzhuzi, did not exist in isolation during the early Holocene. Early hunter-gatherers may have established settlement systems in diverse landscapes and moved flexibly between spots. What we see now are probably nodes within a settlement system. However, questions remain unanswered. Are the surface collections of microblade products in Ngari related to the system? What is the nature of the settlement systems on the western Tibet Plateau? How were the early Holocene hunter-gatherers interconnected in wider networks? Therefore, more supporting evidence from well-dated contemporaneous sites and further investigation of raw material sources on the Tibetan Plateau are required. Our research on the geological sources of the raw materials is ongoing.

5. Conclusions

A comparative lithic analysis revealed that the hunter-gatherers on the western Tibetan Plateau employed similar technologies during the early Holocene. This study discusses the mobility and land-use patterns of hunter-gatherers on the plateau by examining technological strategies and site functions. It was found that Xiada Co might have been used as a basecamp where occupants stayed for a longer time or paid repeated visits in a short period. Qusongguo might have been used as a temporary camp by a task group. Finally, Dingzhonghuzhuzi might have been a site with specific functions, occupied for a longer time.

Based on the evidence collected from the three sites, we conclude that the hunter-gatherers of this period were well versed with the western plateau’s landscapes. Considerable cultural homogeneity of the three sites reflects a long-term adaptation to the highland environment. Occupants were capable of using various types of landforms to organize activities and acquire key resources from the highland, which indicates localized problem-solving. Additionally, potential interaction networks may have existed, wherein highland occupants interacted through flexible mobility strategies. At the same time, the travel cost modeling and lack of similar technology from contemporaneous lowlands suggest a low possibility of non-permanent land-use patterns. Therefore, we propose that the hunter-gatherers living on the western Tibetan Plateau occupied the highlands on a permanent basis during the early Holocene period in a favorable environment. The three sites might have been different parts of their settlement systems that developed across the plateau. Occupants formulated ways of living on the plateau after generations of arduous efforts. As found in archaeology, flexible short-term activities could have been a subsistence strategy to facilitate permanent occupation on the plateau. Further investigation is needed for a more detailed and comprehensive picture of their life.