Modern Homo sapiens have participated in a large number of ecosystem transformations, but it is difficult to detect the origin or early consequences of these behaviors. Archeology, geochronology, geomorphology, and paleoenvironmental data from northern Malawi document the changing relationship between the presence of foragers, ecosystem organization, and alluvial fan formation in the Late Pleistocene. After about the 20th century, a dense system of Mesolithic artifacts and alluvial fans was formed. 92,000 years ago, in the paleo-ecological environment, there was no analogue in the previous 500,000-year record. Archaeological data and principal coordinate analysis show that early man-made fires relaxed the seasonal restrictions on ignition, affecting vegetation composition and erosion. This, combined with climate-driven precipitation changes, eventually led to an ecological transition to the early pre-agricultural artificial landscape.
       Modern humans are powerful promoters of ecosystem transformation. For thousands of years, they have changed the environment extensively and intentionally, sparking debate about when and how the first human-dominated ecosystem emerged (1). More and more archaeological and ethnographic evidence shows that there is a large number of recursive interactions between foragers and their environment, which indicates that these behaviors are the basis of our species evolution (2-4). Fossil and genetic data indicate that Homo sapiens existed in Africa approximately 315,000 years ago (ka). Archaeological data show that the complexity of behaviors occurring across the continent has increased significantly in the past approximately 300 to 200 ka spans. The end of the Pleistocene (Chibanian) (5). Since our emergence as a species, humans have begun to rely on technological innovation, seasonal arrangements, and complex social cooperation to thrive. These attributes enable us to take advantage of previously uninhabited or extreme environments and resources, so today humans are the only pan-global animal species (6). Fire played a key role in this transformation (7).
       Biological models indicate that the adaptability to cooked food can be traced back to at least 2 million years ago, but it was not until the end of the Middle Pleistocene that conventional archeological evidence of fire control appeared (8). The ocean core with dust records from a large area of ​​the African continent shows that in the past millions of years, the peak of elemental carbon appeared after about 400 ka, mainly during the transition from interglacial to glacial period, but also occurred During the Holocene (9). This shows that before about 400 ka, fires in sub-Saharan Africa were not common, and human contributions were significant in the Holocene (9). Fire is a tool used by herdsmen throughout the Holocene to cultivate and maintain grasslands (10). However, detecting the background and ecological impact of the use of fire by hunter-gatherers in the early Pleistocene is more complicated (11).
       Fire is called an engineering tool for resource manipulation in both ethnography and archaeology, including improving livelihood returns or modifying raw materials. These activities are usually related to public planning and require a lot of ecological knowledge (2, 12, 13). Landscape-scale fires enable hunter-gatherers to drive away prey, control pests, and increase habitat productivity (2). On-site fire promotes cooking, heating, predator defense, and social cohesion (14). However, the extent to which hunter-gatherer fires can reconfigure the components of the landscape, such as the structure of the ecological community and the topography, is very ambiguous (15, 16).
       Without outdated archaeological and geomorphological data and continuous environmental records from multiple locations, understanding the development of human-induced ecological changes is problematic. Long-term lake deposit records from the Great Rift Valley in Southern Africa, combined with ancient archaeological records in the area, make it a place to investigate the ecological impacts caused by the Pleistocene. Here, we report on the archeology and geomorphology of an extensive Stone Age landscape in south-central Africa. Then, we linked it with paleoenvironmental data spanning >600 ka to determine the earliest coupling evidence of human behavior and ecosystem transformation in the context of man-made fires.
       We provided a previously unreported age limit for the Chitimwe bed in the Karonga District, located at the northern end of the northern part of Malawi in the southern African Rift Valley (Figure 1) (17). These beds are composed of red soil alluvial fans and river sediments, covering about 83 square kilometers, containing millions of stone products, but no preserved organic remains, such as bones (Supplementary text) (18). Our optically excited light (OSL) data from the Earth record (Figure 2 and Tables S1 to S3) amended the age of the Chitimwe bed to the Late Pleistocene, and the oldest age of alluvial fan activation and stone age burial is about 92 ka ( 18, 19). The alluvial and river Chitimwe layer covers the lakes and rivers of the Pliocene-Pleistocene Chiwondo layer from a low-angle unconformity (17). These deposits are located in the fault wedge along the edge of the lake. Their configuration indicates the interaction between lake level fluctuations and active faults extending into the Pliocene (17). Although tectonic action may have affected the regional topography and piedmont slope for a long time, the fault activity in this area may have slowed down since the Middle Pleistocene (20). After ~800 ka and until shortly after 100 ka, the hydrology of Lake Malawi is mainly driven by climate (21). Therefore, neither of these is the only explanation for the formation of alluvial fans in the Late Pleistocene (22).
       (A) The location of the African station relative to modern precipitation (asterisk); blue is wetter and red is drier (73); the box on the left shows Lake Malawi and surrounding areas MAL05-2A and MAL05-1B The location of the /1C core (purple dot), where the Karonga area is highlighted as a green outline, and the location of the Luchamange bed is highlighted as a white box. (B) The northern part of Malawi basin, showing the hillshade topography relative to the MAL05-2A core, the remaining Chitimwe bed (brown patch) and the excavation location of the Malawi Early Mesolithic Project (MEMSAP) (yellow dot) ); CHA, Chaminade; MGD, the village of Mwanganda; NGA, Ngara; SS, Sadara South; VIN, literary library picture; WW, Beluga.
       OSL center age (red line) and error range of 1-σ (25% gray), all OSL ages related to the occurrence of in situ artifacts in Karonga. Age relative to the past 125 ka data shows (A) kernel density estimates of all OSL ages from alluvial fan sediments, indicating sedimentary/alluvial fan accumulation (cyan), and lake water level reconstruction based on principal component analysis (PCA) characteristic values Aquatic fossils and authigenic minerals (21) (blue) from the MAL05-1B/1C core. (B) From the MAL05-1B/1C core (black, a value close to 7000 with an asterisk) and the MAL05-2A core (grey), the counts of macromolecular carbon per gram normalized by the sedimentation rate. (C) Margalef species richness index (Dmg) from MAL05-1B/1C core fossil pollen. (D) Percentage of fossil pollen from Compositae, miombo woodland and Olea europaea, and (E) Percentage of fossil pollen from Poaceae and Podocarpus. All pollen data are from the MAL05-1B/1C core. The numbers at the top refer to the individual OSL samples detailed in Tables S1 to S3. The difference in data availability and resolution is due to different sampling intervals and material availability in the core. Figure S9 shows two macro carbon records converted to z-scores.
       (Chitimwe) The landscape stability after fan formation is indicated by the formation of red soil and soil-forming carbonates, which cover the fan-shaped sediments of the entire study area (Supplementary text and Table S4). The formation of Late Pleistocene alluvial fans in the Lake Malawi Basin is not limited to the Karonga area. About 320 kilometers southeast of Mozambique, the terrestrial cosmogenic nuclide depth profile of 26Al and 10Be limits the formation of the Luchamange bed of alluvial red soil to 119 to 27 ka (23). This extensive age restriction is consistent with our OSL chronology for the western part of the Lake Malawi Basin and indicates the expansion of regional alluvial fans in the Late Pleistocene. This is supported by data from the lake core record, which indicates that the higher sedimentation rate is accompanied by about 240 ka, which has a particularly high value at ca. 130 and 85 ka (supplementary text) (21).
       The earliest evidence of human settlement in this area is related to the Chitimwe sediments identified at ~92 ± 7 ka. This result is based on 605 m3 of excavated sediments from 14 sub-centimeter space control archaeological excavations and 147 m3 of sediments from 46 archaeological test pits, controlled vertically to 20 cm and horizontally controlled to 2 meters (Supplementary text and Figures S1 to S3) In addition, we also surveyed 147.5 kilometers, arranged 40 geological test pits, and analyzed more than 38,000 cultural relics from 60 of them (Tables S5 and S6) (18). These extensive investigations and excavations indicate that although ancient humans including early modern humans may have lived in the area about 92 ka ago, the accumulation of sediments associated with the rise and then stabilization of Lake Malawi did not preserve archaeological evidence until Form the Chitimwe bed.
       Archaeological data support the inference that in the late Quaternary, the fan-shaped expansion and human activities in northern Malawi existed in large numbers, and the cultural relics belonged to the types of other parts of Africa related to early modern humans. Most artifacts are made of quartzite or quartz river pebbles, with radial, Levallois, platform and random core reduction (Figure S4). Morphological diagnostic artifacts are mainly attributed to the Mesolithic Age (MSA)-specific Levallois-type technique, which has been at least about 315 ka in Africa so far (24). The uppermost Chitimwe bed lasted until the early Holocene, containing sparsely distributed Late Stone Age events, and was found to be related to the late Pleistocene and Holocene hunter-gatherers throughout Africa. In contrast, stone tool traditions (such as large cutting tools) usually associated with the Early Middle Pleistocene are rare. Where these did occur, they were found in MSA-containing sediments in the late Pleistocene, not in the early stages of deposition (Table S4) (18). Although the site existed at ~92 ka, the most representative period of human activity and alluvial fan deposition occurred after ~70 ka, well defined by a set of OSL ages (Figure 2). We confirmed this pattern with 25 published and 50 previously unpublished OSL ages (Figure 2 and Tables S1 to S3). These indicate that out of a total of 75 age determinations, 70 were recovered from sediments after approximately 70 ka. Figure 2 shows the 40 ages associated with in-situ MSA artifacts, relative to the main paleoenvironmental indicators published from the center of the MAL05-1B/1C central basin (25) and the previously unpublished MAL05-2A northern basin center of the lake. Charcoal (adjacent to the fan that produces OSL age).
       Using fresh data from archaeological excavations of phytoliths and soil micromorphology, as well as public data on fossil pollen, large charcoal, aquatic fossils and authigenic minerals from the core of the Malawi Lake Drilling Project, we reconstructed the MSA human relationship with Lake Malawi. Occupy the climate and environmental conditions of the same period (21). The latter two agents are the main basis for reconstructing relative lake depths dating back to more than 1200 ka (21), and are matched with pollen and macrocarbon samples collected from the same location in the core of ~636 ka (25) in the past. The longest cores (MAL05-1B and MAL05-1C; 381 and 90 m respectively) were collected about 100 kilometers southeast of the archaeological project area. A short core (MAL05-2A; 41 m) was collected about 25 kilometers east of the North Rukulu River (Figure 1). The MAL05-2A core reflects the terrestrial paleoenvironmental conditions in the Kalunga area, while the MAL05-1B/1C core does not receive direct river input from the Kalunga, so it can better reflect the regional conditions.
       The deposition rate recorded in the MAL05-1B/1C composite drill core started from 240 ka and increased from the long-term average value of 0.24 to 0.88 m/ka (Figure S5). The initial increase is related to changes in the orbital modulated sunlight, which will cause high-amplitude changes in the lake level during this interval (25). However, when the orbital eccentricity drops after 85 ka and the climate is stable, the subsidence rate is still high (0.68 m/ka). This coincided with the terrestrial OSL record, which showed extensive evidence of alluvial fan expansion after about 92 ka, and was consistent with the susceptibility data showing a positive correlation between erosion and fire after 85 ka (Supplementary text and Table S7) . In view of the error range of the available geochronological control, it is impossible to judge whether this set of relationships evolves slowly from the progress of the recursive process or erupts rapidly when reaching a critical point. According to the geophysical model of basin evolution, since the Middle Pleistocene (20), rift extension and related subsidence have slowed down, so it is not the main reason for the extensive fan formation process that we mainly determined after 92 ka.
       Since the Middle Pleistocene, climate has been the main controlling factor of lake water level (26). Specifically, the uplift of the northern basin closed an existing exit. 800 ka to deepen the lake until it reaches the threshold height of the modern exit (21). Located at the southern end of the lake, this outlet provided an upper limit for the lake’s water level during wet intervals (including today), but allowed the basin to close as the lake’s water level fell during dry periods (27). The reconstruction of the lake level shows the alternating dry and wet cycles in the past 636 ka. According to evidence from fossil pollen, extreme drought periods (>95% reduction in total water) associated with low summer sunshine have led to the expansion of semi-desert vegetation, with trees restricted to permanent waterways (27). These (lake) lows are correlated with pollen spectra, showing a high proportion of grasses (80% or more) and xerophytes (Amaranthaceae) at the expense of tree taxa and low overall species richness (25). In contrast, when the lake approaches modern levels, vegetation closely related to African mountain forests usually extends to the lakeshore [about 500 m above sea level (masl)]. Today, African mountain forests only appear in small discrete patches above about 1500 masl (25, 28).
       The most recent extreme drought period occurred from 104 to 86 ka. After that, although the lake level returned to high conditions, open miombo woodlands with a large amount of herbs and herb ingredients became common (27, 28). The most significant African mountain forest taxa is Podocarpus pine, which has never recovered to a value similar to the previous high lake level after 85 ka (10.7 ± 7.6% after 85 ka, while the similar lake level before 85 ka is 29.8 ± 11.8%). The Margalef index (Dmg) also shows that the species richness of the past 85 ka is 43% lower than the previous sustained high lake level (2.3 ± 0.20 and 4.6 ± 1.21, respectively), for example, between 420 and 345 ka ( Supplementary text and figures S5 and S6) (25). Pollen samples from approximately time. 88 to 78 ka also contains a high percentage of Compositae pollen, which can indicate that the vegetation has been disturbed and is within the error range of the oldest date when humans occupied the area.
       We use the climate anomaly method (29) to analyze the paleoecological and paleoclimate data of cores drilled before and after 85 ka, and examine the ecological relationship between vegetation, species abundance, and precipitation and the hypothesis of decoupling the inferred pure climate prediction. Drive baseline mode of ~550 ka. This transformed ecosystem is affected by lake-filling precipitation conditions and fires, which is reflected in the lack of species and new vegetation combinations. After the last dry period, only some forest elements recovered, including the fire-resistant components of African mountain forests, such as olive oil, and the fire-resistant components of tropical seasonal forests, such as Celtis (Supplementary text and Figure S5) (25). To test this hypothesis, we modeled lake water levels derived from ostracode and authigenic mineral substitutes as independent variables (21) and dependent variables such as charcoal and pollen that may be affected by increased fire frequency (25).
       In order to check the similarity or difference between these combinations at different times, we used pollen from Podocarpus (evergreen tree), grass (grass), and olive (fire-resistant component of African mountain forests) for principal coordinate analysis ( PCoA), and miombo (the main woodland component today). By plotting PCoA on the interpolated surface representing the lake level when each combination was formed, we examined how the pollen combination changes with respect to precipitation and how this relationship changes after 85 ka (Figure 3 and Figure S7). Before 85 ka, the gramineous-based samples aggregated toward dry conditions, while the podocarpus-based samples aggregated toward wet conditions. In contrast, the samples after 85 ka are clustered with most samples before 85 ka and have different average values, indicating that their composition is unusual for similar precipitation conditions. Their position in PCoA reflects the influence of Olea and miombo, both of which are favored under conditions that are more prone to fire. In the samples after 85 ka, Podocarpus pine was only abundant in three consecutive samples, which occurred after the interval between 78 and 79 ka began. This suggests that after the initial increase in rainfall, the forest seems to have recovered briefly before it finally collapsed.
       Each point represents a single pollen sample at a given point in time, using the supplementary text and the age model in Figure 1. S8. The vector represents the direction and gradient of change, and a longer vector represents a stronger trend. The underlying surface represents the water level of the lake as a representative of precipitation; the dark blue is higher. The average value of PCoA feature values ​​is provided for the data after 85 ka (red diamond) and all data from similar lake levels before 85 ka (yellow diamond). Using the data of the entire 636 ka, the “simulated lake level” is between -0.130-σ and -0.198-σ near the average eigenvalue of the lake level PCA.
       In order to study the relationship between pollen, lake water level and charcoal, we used nonparametric multivariate analysis of variance (NP-MANOVA) to compare the overall “environment” (represented by the data matrix of pollen, lake water level and charcoal) before and after the 85 ka transition. We found that the variation and covariance found in this data matrix are statistically significant differences before and after 85 ka (Table 1).
       Our terrestrial paleoenvironmental data from the phytoliths and soils at the edge of the West Lake are consistent with the interpretation based on the lake proxy. These indicate that despite the high water level of the lake, the landscape has been transformed into a landscape dominated by open canopy forest land and wooded grassland, just like today (25). All locations analyzed for phytoliths on the western edge of the basin are after ~45 ka and show a large amount of arboreal cover reflecting wet conditions. However, they believe that most of the mulch is in the form of open woodland overgrown with bamboo and panic grass. According to phytolith data, non-fire-resistant palm trees (Arecaceae) exist only on the shoreline of the lake, and are rare or absent in inland archaeological sites (Table S8) (30).
       Generally speaking, wet but open conditions in the late Pleistocene can also be inferred from terrestrial paleosols (19). Lagoon clay and marsh soil carbonate from the archaeological site of Mwanganda Village can be traced back to 40 to 28 cal ka BP (previously calibrated Qian’anni) (Table S4). The carbonate soil layers in the Chitimwe bed are usually nodular calcareous (Bkm) and argillaceous and carbonate (Btk) layers, which indicates the location of relative geomorphological stability and the slow settlement from the far-reaching alluvial fan Approximately 29 cal ka BP (Supplementary text). The eroded, hardened laterite soil (lithic rock) formed on the remnants of ancient fans indicates open landscape conditions (31) and strong seasonal precipitation (32), indicating the continuous impact of these conditions on the landscape.
       Support for the role of fire in this transition comes from the paired macro charcoal records of drill cores, and the inflow of charcoal from the Central Basin (MAL05-1B/1C) has generally increased from about. 175 cards. A large number of peaks follow in between approximately. After 135 and 175 ka and 85 and 100 ka, the lake level recovered, but the forest and species richness did not recover (Supplementary text, Figure 2 and Figure S5). The relationship between charcoal influx and the magnetic susceptibility of lake sediments can also show patterns of long-term fire history (33). Use data from Lyons et al. (34) Lake Malawi continued to erode the burned landscape after 85 ka, which implies a positive correlation (Spearman’s Rs = 0.2542 and P = 0.0002; Table S7), while the older sediments show the opposite relationship (Rs = -0.2509 and P <0.0001). In the northern basin, the shorter MAL05-2A core has the deepest dating anchor point, and the youngest Toba tuff is ~74 to 75 ka (35). Although it lacks a longer-term perspective, it receives input directly from the basin where the archaeological data is sourced. The charcoal records of the northern basin show that since the Toba crypto-tephra mark, the input of terrigenous charcoal has steadily increased during the period when archaeological evidence is most common (Figure 2B).
       Evidence of man-made fires may reflect deliberate use on a landscape scale, widespread populations causing more or larger on-site ignitions, alteration of fuel availability by harvesting understory forests, or a combination of these activities. Modern hunter-gatherers use fire to actively change foraging rewards (2). Their activities increase the abundance of prey, maintain the mosaic landscape, and increase the thermal diversity and heterogeneity of succession stages (13). Fire is also important for on-site activities such as heating, cooking, defense, and socializing (14). Even small differences in fire deployment outside of natural lightning strikes can change forest succession patterns, fuel availability, and firing seasonality. The reduction of tree cover and understory trees is most likely to increase erosion, and the loss of species diversity in this area is closely related to the loss of African mountain forest communities (25).
       In the archaeological record before the MSA began, human control of fire has been well established (15), but so far, its use as a landscape management tool has only been recorded in a few Paleolithic contexts. These include about in Australia. 40 ka (36), Highland New Guinea. 45 ka (37) peace treaty. 50 ka Niah Cave (38) in lowland Borneo. In the Americas, when humans first entered these ecosystems, especially in the past 20 ka (16), artificial ignition was considered to be the main factor in the reconfiguration of plant and animal communities. These conclusions must be based on relevant evidence, but in the case of direct overlap of archaeological, geological, geomorphological, and paleoenvironmental data, the causality argument has been strengthened. Although the marine core data of the coastal waters of Africa have previously provided evidence of fire changes in the past about 400 ka (9), here we provide evidence of human influence from relevant archaeological, paleoenvironmental, and geomorphological data sets.
       The identification of man-made fires in paleoenvironmental records requires evidence of fire activities and temporal or spatial changes of vegetation, proving that these changes are not predicted by climate parameters alone, and the temporal/spatial overlap between changes in fire conditions and changes in human records (29) Here, the first evidence of widespread MSA occupation and alluvial fan formation in the Lake Malawi basin occurred at approximately the beginning of a major reorganization of regional vegetation. 85 cards. The charcoal abundance in the MAL05-1B/1C core reflects the regional trend of charcoal production and deposition, at approximately 150 ka compared with the rest of the 636 ka record (Figures S5, S9, and S10). This transition shows the important contribution of fire to shaping the composition of the ecosystem, which cannot be explained by climate alone. In natural fire situations, lightning ignition usually occurs at the end of the dry season (39). However, if the fuel is dry enough, man-made fires may be ignited at any time. On the scale of the scene, humans can continuously change the fire by collecting firewood from under the forest. The end result of any type of man-made fire is that it has the potential to cause more woody vegetation consumption, lasting throughout the year, and on all scales.
       In South Africa, as early as 164 ka (12), fire was used for the heat treatment of tool-making stones. As early as 170 ka (40), fire was used as a tool for cooking starchy tubers, making full use of fire in ancient times. Prosperous Resources-Prone Scenery (41). Landscape fires reduce the arboreal cover and are an important tool for maintaining grassland and forest patch environments, which are the defining elements of human-mediated ecosystems (13). If the purpose of changing vegetation or prey behavior is to increase man-made burning, then this behavior represents an increase in the complexity of controlling and deploying fire by early modern humans compared with early humans, and shows that our relationship with fire has undergone a shift in interdependence (7). Our analysis provides an additional way to understand the changes in the use of fire by humans in the Late Pleistocene, and the impact of these changes on their landscape and environment.
       The expansion of the Late Quaternary alluvial fans in the Karonga area may be due to changes in the seasonal combustion cycle under conditions of higher than average rainfall, leading to increased erosion of the hillside. The mechanism of this occurrence may be the watershed-scale response driven by the disturbance caused by the fire, the enhanced and sustained erosion of the upper part of the watershed, and the expansion of alluvial fans in the piedmont environment near Lake Malawi. These reactions may include changing soil properties to reduce permeability, reduce surface roughness, and increase runoff because of the combination of high precipitation conditions and reduced arboreal cover (42). The availability of sediments is initially improved by peeling off the covering material, and over time, soil strength may decrease due to heating and reduced root strength. The exfoliation of the topsoil increases the sediment flux, which is accommodated by the fan-shaped accumulation downstream and accelerates the formation of red soil on the fan-shaped.
       Many factors can control the landscape’s response to changing fire conditions, most of which operate within a short period of time (42-44). The signal we associate here is obvious on the millennium time scale. Analysis and landscape evolution models show that with the vegetation disturbance caused by repeated wildfires, the denudation rate has changed significantly on a millennium time scale (45, 46). The lack of regional fossil records that coincide with the observed changes in charcoal and vegetation records hinders the reconstruction of the effects of human behavior and environmental changes on the composition of herbivore communities. However, large herbivores that inhabit more open landscapes play a role in maintaining them and preventing the invasion of woody vegetation (47). Evidence of changes in different components of the environment should not be expected to occur simultaneously, but should be seen as a series of cumulative effects that may occur over a long period of time (11). Using the climate anomaly method (29), we regard human activity as a key driving factor in shaping the landscape of northern Malawi during the Late Pleistocene. However, these effects may be based on the earlier, less obvious legacy of human-environment interactions. The charcoal peak that appeared in the paleoenvironmental record before the earliest archaeological date may include an anthropogenic component that does not cause the same ecological system changes as recorded later, and does not involve deposits that are sufficient to confidently indicate human occupation.
       Short sediment cores, such as those from the adjacent Masoko Lake Basin in Tanzania, or the shorter sediment cores in Lake Malawi, show that the relative pollen abundance of grass and woodland taxa has changed, which is attributed to the past 45 years. The natural climate change of ka (48-50). However, only a longer-term observation of the pollen record of Lake Malawi >600 ka, along with the age-old archaeological landscape next to it, is it possible to understand the climate, vegetation, charcoal, and human activities. Although humans are likely to appear in the northern part of the Lake Malawi basin before 85 ka, about 85 ka, especially after 70 ka, indicate that the area is attractive for human habitation after the last major drought period ended. At this time, the new or more intensive/frequent use of fire by humans is obviously combined with natural climate change to reconstruct the ecological relationship> 550-ka, and finally formed the early pre-agricultural artificial landscape (Figure 4). Unlike earlier periods, the sedimentary nature of the landscape preserves the MSA site, which is a function of the recursive relationship between the environment (resource distribution), human behavior (activity patterns), and fan activation (deposition/site burial).
       (A) About. 400 ka: No human beings can be detected. The humid conditions are similar to today, and the lake level is high. Diverse, non-fire resistant arboreal cover. (B) About 100 ka: There is no archaeological record, but the presence of humans may be detected through the influx of charcoal. Extremely dry conditions occur in dry watersheds. The bedrock is generally exposed and the surface sediments are limited. (C) About 85 to 60 ka: The water level of the lake increases with the increase in precipitation. The existence of human beings can be discovered through archaeology after 92 ka, and after 70 ka, the burning of highlands and the expansion of alluvial fans will follow. A less diverse, fire-resistant vegetation system has emerged. (D) About 40 to 20 ka: Environmental charcoal input in the northern basin has increased. The formation of alluvial fans continued, but began to weaken at the end of this period. Compared with the previous record of 636 ka, the lake level remains high and stable.
       The Anthropocene represents the accumulation of niche-building behaviors developed over thousands of years, and its scale is unique to modern Homo sapiens (1, 51). In the modern context, with the introduction of agriculture, man-made landscapes continue to exist and intensify, but they are extensions of patterns established during the Pleistocene, rather than disconnections (52). Data from northern Malawi shows that the ecological transition period can be prolonged, complicated and repetitive. This scale of transformation reflects the complex ecological knowledge of early modern humans and illustrates their transformation to our global dominant species today.
       According to the protocol described by Thompson et al., on-site investigation and recording of artifacts and cobblestone characteristics on the survey area. (53). The placement of the test pit and the excavation of the main site, including micromorphology and phytolith sampling, followed the protocol described by Thompson et al. (18) and Wright et al. (19). Our geographic information system (GIS) map based on the Malawi geological survey map of the region shows a clear correlation between Chitimwe Beds and archaeological sites (Figure S1). The interval between the geological and archaeological test pits in Karonga area is to capture the widest representative sample (Figure S2). Karonga’s geomorphology, geological age and archaeological surveys involve four main field survey methods: pedestrian surveys, archaeological test pits, geological test pits and detailed site excavations. Together, these techniques allow sampling of the main exposure of the Chitimwe bed in the north, central, and south of Karonga (Figure S3).
       The on-site investigation and recording of artifacts and cobblestone features on the pedestrian survey area followed the protocol described by Thompson et al. (53). This approach has two main goals. The first is to identify the places where the cultural relics have been eroded, and then place archaeological test pits uphill in these places to restore the cultural relics in situ from the buried environment. The second goal is to formally record the distribution of artifacts, their characteristics, and their relationship with the source of nearby stone materials (53). In this work, a three-person team walked at a distance of 2 to 3 meters for a total of 147.5 linear kilometers, traversing most of the drawn Chitimwe beds (Table S6).
       The work first focused on Chitimwe Beds to maximize the observed artifact samples, and secondly focused on long linear sections from the lake shore to the highlands that cut across different sedimentary units. This confirms a key observation that the artifacts located between the western highlands and the lakeshore are only related to the Chitimwe bed or more recent Late Pleistocene and Holocene sediments. The artefacts found in other deposits are off-site, relocated from other places in the landscape, as can be seen from their abundance, size, and degree of weathering.
       The archaeological test pit in place and the excavation of the main site, including micromorphology and phytolith sampling, followed the protocol described by Thompson et al. (18, 54) and Wright et al. (19, 55). The main purpose is to understand the underground distribution of artifacts and fan-shaped sediments in the larger landscape. Artifacts are usually buried deep in all places in Chitimwe Beds, except for the edges, where erosion has begun to remove the top of the sediment. During the informal investigation, two people walked past Chitimwe Beds, which were displayed as map features on the Malawi government geological map. When these people encountered the shoulders of the Chitimwe Bed sediment, they began to walk along the edge, where they could observe the artifacts eroded from the sediment. By tilting the excavations slightly upwards (3 to 8 m) from the actively eroding artifacts, the excavation can reveal their in-situ position relative to the sediment containing them, without the need for extensive excavation laterally. The test pits are placed so that they are 200 to 300 meters away from the next closest pit, thereby capturing changes in the Chitimwe bed sediment and the artifacts it contains. In some cases, the test pit revealed a site that later became a full-scale excavation site.
       All test pits start with a square of 1 × 2 m, face north-south, and are excavated in arbitrary units of 20 cm, unless the color, texture, or content of the sediment changes significantly. Record the sedimentology and soil properties of all excavated sediments, which pass evenly through a 5 mm dry sieve. If the deposition depth continues to exceed 0.8 to 1 m, stop digging in one of the two square meters and continue digging in the other, thereby forming a “step” so that you can enter deeper layers safely. Then continue to excavate until the bedrock is reached, at least 40 cm of archeologically sterile sediments are below the concentration of artifacts, or the excavation becomes too unsafe (deep) to proceed. In some cases, the deposition depth needs to extend the test pit to a third square meter and enter the trench in two steps.
       Geological test pits have previously shown that Chitimwe Beds often appear on geological maps because of their distinctive red color. When they include extensive streams and river sediments, and alluvial fan sediments, they do not always appear red (19). Geology The test pit was excavated as a simple pit designed to remove the mixed upper sediments to reveal the underground strata of the sediments. This is necessary because the Chitimwe bed is eroded into a parabolic hillside, and there are collapsed sediments on the slope, which usually do not form clear natural parts or cuts. Therefore, these excavations either took place on the top of the Chitimwe bed, presumably there was underground contact between the Chitimwe bed and the Pliocene Chiwondo bed below, or they took place where the river terrace sediments needed to be dated (55).
       Full-scale archaeological excavations are carried out in places that promise a large number of in-situ stone tool assemblies, usually based on test pits or places where a large number of cultural relics can be seen eroding from the slope. The main excavated cultural relics were recovered from sedimentary units excavated separately in a square of 1 × 1 m. If the density of artifacts is high, the digging unit is a 10 or 5 cm spout. All stone products, fossil bones and ochre were drawn during each major excavation, and there is no size limit. The screen size is 5mm. If cultural relics are discovered during the excavation process, they will be assigned a unique bar code drawing discovery number, and the discovery numbers in the same series will be assigned to the filtered discoveries. The cultural relics are marked with permanent ink, placed in bags with specimen labels, and bagged together with other cultural relics from the same background. After analysis, all cultural relics are stored in the Cultural and Museum Center of Karonga.
       All excavations are carried out according to natural strata. These are subdivided into spits, and the spit thickness depends on the artifact density (for example, if the artifact density is low, the spit thickness will be high). Background data (for example, sediment properties, background relationships, and observations of interference and artifact density) are recorded in the Access database. All coordinate data (for example, findings drawn in segments, context elevation, square corners, and samples) are based on Universal Transverse Mercator (UTM) coordinates (WGS 1984, Zone 36S). At the main site, all points are recorded using a Nikon Nivo C series 5″ total station, which is built on a local grid as close as possible to the north of UTM. The location of the northwest corner of each excavation site and the location of each excavation site The amount of sediment is given in Table S5.
       The section of sedimentology and soil science characteristics of all excavated units was recorded using the United States Agricultural Part Class Program (56). Sedimentary units are specified based on grain size, angularity, and bedding characteristics. Note the abnormal inclusions and disturbances associated with the sediment unit. Soil development is determined by the accumulation of sesquioxide or carbonate in underground soil. Underground weathering (for example, redox, formation of residual manganese nodules) is also frequently recorded.
       The collection point of OSL samples is determined on the basis of estimating which facies may produce the most reliable estimation of sediment burial age. At the sampling location, trenches were dug to expose the authigenic sedimentary layer. Collect all the samples used for OSL dating by inserting an opaque steel tube (about 4 cm in diameter and about 25 cm in length) into the sediment profile.
       OSL dating measures the size of the group of trapped electrons in crystals (such as quartz or feldspar) due to exposure to ionizing radiation. Most of this radiation comes from the decay of radioactive isotopes in the environment, and a small amount of additional components in tropical latitudes appear in the form of cosmic radiation. The captured electrons are released when the crystal is exposed to light, which occurs during transportation (zeroing event) or in the laboratory, where the lighting occurs on a sensor that can detect photons (for example, a photomultiplier tube or a camera with a charged coupling device) The lower part emits when the electron returns to the ground state. Quartz particles with a size between 150 and 250 μm are separated by sieving, acid treatment and density separation, and used as small aliquots (<100 particles) mounted on the surface of an aluminum plate or drilled into a 300 x 300 mm well The individual particles are analyzed on an aluminum pan. The buried dose is usually estimated using a single aliquot regeneration method (57). In addition to assessing the radiation dose received by grains, OSL dating also requires estimating the dose rate by measuring the radionuclide concentration in the sediment of the collected sample using gamma spectroscopy or neutron activation analysis, and determining the cosmic dose reference sample Location and depth of burial. The final age determination is achieved by dividing the burial dose by the dose rate. However, when there is a change in the dose measured by a single grain or group of grains, a statistical model is needed to determine the appropriate buried dose to be used. The buried dose is calculated here using the central era model, in the case of single aliquot dating, or in the case of single-particle dating, using a finite mixture model (58).
       Three independent laboratories performed OSL analysis for this study. The detailed individual methods for each laboratory are shown below. In general, we use the regenerative dose method to apply OSL dating to small aliquots (tens of grains) instead of using single grain analysis. This is because during the regenerative growth experiment, the recovery rate of a small sample is low (<2%), and the OSL signal is not saturated at the natural signal level. The inter-laboratory consistency of age determination, the consistency of the results within and between the tested stratigraphic profiles, and the consistency with the geomorphological interpretation of the 14C age of carbonate rocks are the main basis for this assessment. Each laboratory evaluated or implemented a single grain agreement, but independently determined that it was not suitable for use in this study. The detailed methods and analysis protocols followed by each laboratory are provided in the supplementary materials and methods.
       Stone artifacts recovered from controlled excavations (BRU-I; CHA-I, CHA-II, and CHA-III; MGD-I, MGD-II, and MGD-III; and SS-I) are based on the metric system and quality characteristics. Measure the weight and maximum size of each workpiece (using a digital scale to measure the weight is 0.1 g; using a Mitutoyo digital caliper to measure all dimensions is 0.01 mm). All cultural relics are also classified according to raw materials (quartz, quartzite, flint, etc.), grain size (fine, medium, coarse), uniformity of grain size, color, cortex type and coverage, weathering/edge rounding and technical grade (complete or fragmented) Cores or flakes, flakes/corner pieces, hammer stones, grenades and others).
       The core is measured along its maximum length; maximum width; width is 15%, 50%, and 85% of length; maximum thickness; thickness is 15%, 50%, and 85% of length. Measurements were also performed to evaluate the volume properties of the core of hemispherical tissues (radial and Levallois). Both intact and broken cores are classified according to the reset method (single platform or multi-platform, radial, Levallois, etc.), and flaky scars are counted at ≥15 mm and ≥20% of the core length. Cores with 5 or fewer 15 mm scars are classified as “random”. The cortical coverage of the entire core surface is recorded, and the relative cortical coverage of each side is recorded on the core of the hemispherical tissue.
       The sheet is measured along its maximum length; maximum width; width is 15%, 50%, and 85% of length; maximum thickness; thickness is 15%, 50%, and 85% of length. Describe the fragments according to the remaining parts (proximal, middle, distal, split on the right and split on the left). The elongation is calculated by dividing the maximum length by the maximum width. Measure the platform width, thickness, and outer platform angle of the intact slice and proximal slice fragments, and classify the platforms according to the degree of preparation. Record cortical coverage and location on all slices and fragments. The distal edges are classified according to the type of termination (feather, hinge, and upper fork). On the complete slice, record the number and direction of the scar on the previous slice. When encountered, record the modification location and invasiveness in accordance with the protocol established by Clarkson (59). Renovation plans were initiated for most of the excavation combinations to evaluate restoration methods and site deposition integrity.
       The stone artifacts recovered from the test pits (CS-TP1-21, SS-TP1-16 and NGA-TP1-8) are described according to a simpler scheme than controlled excavation. For each artifact, the following characteristics were recorded: raw material, particle size, cortex coverage, size grade, weathering/edge damage, technical components, and preservation of fragments. Descriptive notes for the diagnostic features of the flakes and cores are recorded.
       Complete blocks of sediment were cut from exposed sections in excavations and geological trenches. These stones were fixed on site with plaster bandages or toilet paper and packaging tape, and then transported to the Geological Archaeology Laboratory of the University of Tubingen in Germany. There, the sample is dried at 40°C for at least 24 hours. Then they are cured under vacuum, using a mixture of unpromoted polyester resin and styrene in a ratio of 7:3. Methyl ethyl ketone peroxide is used as a catalyst, resin-styrene mixture (3 to 5 ml/l). Once the resin mixture has gelled, heat the sample at 40°C for at least 24 hours to completely harden the mixture. Use a tile saw to cut the hardened sample into 6 × 9 cm pieces, stick them on a glass slide and grind them to a thickness of 30 μm. The resulting slices were scanned using a flatbed scanner, and analyzed using plane polarized light, cross-polarized light, oblique incident light, and blue fluorescence with the naked eye and magnification (×50 to ×200). The terminology and description of thin sections follow the guidelines published by Stoops (60) and Courty et al. (61). The soil-forming carbonate nodules collected from a depth of> 80 cm are cut in half so that half can be impregnated and performed in thin slices (4.5 × 2.6 cm) using a standard stereo microscope and petrographic microscope and cathodoluminescence (CL) Research microscope. The control of carbonate types is very cautious, because the formation of soil-forming carbonate is related to the stable surface, while the formation of groundwater carbonate is independent of the surface or soil.
       Samples were drilled from the cut surface of the soil-forming carbonate nodules and halved for various analyses. FS used the standard stereo and petrographic microscopes of the Geoarchaeology Working Group and the CL microscope of the Experimental Mineralogy Working Group to study the thin slices, both of which are located in Tübingen, Germany. The radiocarbon dating sub-samples were drilled using precision drills from a designated area of ​​approximately 100 years old. The other half of the nodules is 3 mm in diameter to avoid areas with late recrystallization, rich mineral inclusions, or large changes in the size of calcite crystals. The same protocol cannot be followed for the MEM-5038, MEM-5035 and MEM-5055 A samples. These samples are selected from loose sediment samples and are too small to be cut in half for thin sectioning. However, thin-section studies were performed on the corresponding micromorphological samples of adjacent sediments (including carbonate nodules).
       We submitted 14C dating samples to the Center for Applied Isotope Research (CAIS) at the University of Georgia, Athens, USA. The carbonate sample reacts with 100% phosphoric acid in an evacuated reaction vessel to form CO2. Low-temperature purification of CO2 samples from other reaction products and catalytic conversion to graphite. The ratio of graphite 14C/13C was measured using a 0.5-MeV accelerator mass spectrometer. Compare the sample ratio with the ratio measured with the oxalic acid I standard (NBS SRM 4990). Carrara marble (IAEA C1) is used as the background, and travertine (IAEA C2) is used as the secondary standard. The result is expressed as a percentage of modern carbon, and the quoted uncalibrated date is given in radiocarbon years (BP years) before 1950, using a 14C half-life of 5568 years. The error is cited as 1-σ and reflects statistical and experimental error. Based on the δ13C value measured by isotope ratio mass spectrometry, C. Wissing of the Biogeology Laboratory in Tubingen, Germany, reported the date of isotope fractionation, except for UGAMS-35944r measured at CAIS. Sample 6887B was analyzed in duplicate. To do this, drill a second sub-sample from the nodule (UGAMS-35944r) from the sampling area indicated on the cutting surface. The INTCAL20 calibration curve (Table S4) (62) applied in the southern hemisphere was used to correct the atmospheric fractionation of all samples to 14C to 2-σ.