Prehistoric population expansion in Central Asia promoted by the Altai Holocene Climatic Optimum

The study investigates how middle-to-late Holocene climate variability influenced early human migrations and cultural dynamics across the Central Asian steppe, particularly in the Altai–Sayan region. Although global and hemispheric datasets indicate long-term Holocene temperature trends, continental interiors show conflicting reconstructions, known as the Holocene temperature conundrum. Arid Central Asia records disagree on the timing of peak Holocene warmth (early, middle, or middle-late Holocene). Given the Altai–Sayan’s role as a corridor of trans-Eurasian exchanges (crops, livestock, people) and a climatic transition zone between the mid-latitude westerlies and the East Asian summer monsoon, resolving its Holocene thermal and hydroclimatic history is essential. The authors aim to provide improved temporal constraints on regional temperature and moisture changes using multiproxy lake records and to evaluate their relationship with prehistoric human population dynamics and cultural exchanges during the Bronze Age.

The paper reviews global and regional Holocene climate reconstructions highlighting discrepancies between long-term cooling (e.g., composite temperature reconstructions) and pollen-based warming trends, with seasonality biases and model deficiencies proposed as causes (the Holocene temperature conundrum). In arid Central Asia, proxy records (palynological and geochemical) disagree on when the Holocene Thermal Maximum (HTM) occurred. The HTM’s magnitude and timing vary regionally: paleoecological records from the Mediterranean suggest warmer summers circa 9–5 kyr BP, likely driven by Arctic amplification. The Altai–Sayan region’s cultural significance and sensitivity to both westerlies and monsoon dynamics underscore the need to reconcile regional signals. Prior archaeological and genetic studies document intensified nomadic pastoralism and cultural expansions around 5 ka but lack clear environmental context. The authors position their work within this debate by combining multiproxy lake archives with archaeological radiocarbon syntheses to assess climate–society linkages.

Study area and cores: Sediment cores were retrieved in 2015 from Kanas Lake (core KNS15D; 48°43′19.55″N, 87°1′3.98″E; water depth 20 m; core length 2.25 m) near the lake outflow, and Tiewaike Lake (core TWK15A; 48°49′36.58″N, 87°00′55.54″E; water depth 5.6 m; core length 5.6 m) in the Altai Mountains, NW China. Chronology: Accelerator mass spectrometry (AMS) 14C dating was performed (KNS15D: 7 terrestrial plant macrofossils and 3 bulk organic matter (BOM); TWK15A: 14 BOM and 7 aquatic plant macrofossils). To minimize reservoir effects, BOM dates from KNS15D were excluded from the age model; reservoir effects in TWK15A were evaluated. Dates were calibrated with IntCal20 using Bacon age–depth modeling. Multiproxy analyses: (1) Geochemistry: High-resolution XRF core scanning (2-mm step) measured Al, Si, S, Cl, K, Ca, Ti, Mn, Fe (1 mA, 10 kV, 15 s) and Zn, Rb, Sr, Zr (2 mA, 30 kV, 25 s), reported as cps (semi-quantitative). Conventional XRF on 154 discrete samples (1-cm spacing) quantified Rb, Sr, Zn, Ti, Zr, Mn, SiO2, Al2O3, Fe2O3, CaO, K2O (major metals 1–2% uncertainty; trace metals RSD <5%); scanning vs conventional XRF validated by correlation analyses. Principal component analysis (PCA) was applied to centered log-ratio transformed XRF-scanning data to summarize compositional variance; stratigraphically constrained cluster analysis (CONISS) defined geochemical zones. (2) Silicon isotopes and BSi: Forty-six δ30Si_diatom measurements were made on KNS15D across ~14.1 kyr. Diatom siliceous frustules were isolated via density separation, oxidative cleaning, and contamination checks (microscopy, XRF; Al2O3/SiO2 ≈0.16; <1.6% contamination). Alkaline fusion (NaOH), cation-exchange purification, and MC-ICP-MS (Thermo Neptune Plus) analyses were conducted with bracketing standards (NBS28) and secondary standard (Diatomite) for accuracy (long-term diatomite 1.27‰ ± 0.07, 2 SD, n=195). Biogenic silica (BSi) was measured following established protocols. (3) Stable isotopes and elemental composition of organic matter: TOC, TN, δ13Corg, δ15Norg were measured after acid decarbonation (1N HCl at 60 °C), rinsing, and freeze-drying, using an EA-IRMS system (precision better than 0.01‰). Because of high BSi, TOC% was corrected for authigenic silica content: TOC_corrected = TOC_measured × 100/(100 − BSi%). Grain-size distributions (0.02–2000 μm) were measured at 1-cm intervals. (4) Pollen analysis: 49 samples from TWK15A were processed (HCl–NaOH–HF), with Lycopodium spore tablets added for concentration estimates; ≥500 grains counted per sample under ×400 magnification. (5) Statistical modeling: Generalized additive models (GAMs; mgcv) with REML smoothing were fitted to time series to estimate trends and identify significant change intervals via first derivative deviating from zero (gratia). (6) Archaeological synthesis: Summed probability distributions (SPDs) of calibrated radiocarbon dates from archaeological sites across northern Asia and Europe (CARD and published datasets) were generated in OxCal v4.4.4 (IntCal20). To reduce oversampling bias, dates within each site were binned at 100-yr intervals and averaged following Timpson et al. Spatial attributes (lat, lon, altitude) were compiled from publications and WorldClim 2.1 DEM data.

• Defined the Altai Holocene Climatic Optimum (AHCO) as a significantly warm and wet interval between ~6.5 and 3.6 kyr BP, with a thermal peak at ~4.7–4.3 kyr BP.
• Multiproxy evidence from Kanas Lake (high TOC, BSi, and elevated δ18O_diatom; reduced detrital input per PCA PC1) and Tiewaike Lake (increased tree/shrub pollen including Betula; δ13Corg minima; higher C/N, Ti, δ15Norg, and Rb/Sr indicating enhanced terrestrial input) indicates increased diatom productivity, longer growing seasons, denser vegetation, and reduced catchment erosion during 6.5–3.6 kyr BP.
• δ30Si_diatom peaks and concurrent BSi declines during 4.7–4.3 kyr BP reflect enhanced lake thermal stratification and periodic Si/nutrient limitation under very warm conditions. Centennial-scale cooling events are inferred at ~4.2 and ~3.6 kyr BP.
• The AHCO aligns with broader regional HTM signals: warm conditions at Bosten Lake (Pediastrum simplex/duplex and Δ47), brGDGT-derived MAAT peaks at Lugu and TCQH Lakes (southeastern Tibetan Plateau), tropical East African composites, and summer/winter temperature maxima at Lake Baikal, Alaska, and the Ural Mountains (~6–4 kyr).
• Strengthened Asian summer monsoon during the HTM is supported by pollen-based precipitation reconstructions from Tianchi Crater Lake and Sihailongwan, and Sahiya Cave δ18O indicating monsoon intensity maxima (~5.5–3.6 kyr with a 4.8–3.8 kyr peak).
• Archaeological SPDs show increased human population/activity across the Altai–Sayan and adjacent steppe after ~6.5 kyr BP, especially post-5 kyr, with expansions associated with Afanasievo (~5.1–4.5 kyr) and Chemurchek (~4.5–3.7 kyr) cultures. Latitudinal and altitudinal distributions of dates indicate broader spatial reach during the AHCO.
• Contrasting hydroclimate in lower basins of western Asia during the HTM (e.g., Caspian Sea regression ~5.6–3.7 kyr; ~600-yr megadrought in Kyrgyzstan ~5.8–5.2 kyr) suggests spatially heterogeneous impacts, possibly influencing migration directions (limited NE Yamnaya movement into Altai, significant westward Yamnaya expansion ~4.8–4.3 kyr).
• Post-3.6 kyr cooling corresponds with shifts in cultural developments toward oasis-desert regions (e.g., Xiaohe Culture ~4.1–3.4 kyr), consistent with late Holocene wetting trends in arid Central Asia.

By providing high-resolution multiproxy lake records from the Altai Mountains, the study resolves conflicting views of Holocene temperature evolution in arid Central Asia and demonstrates a pronounced regional climatic optimum (AHCO) from ~6.5 to 3.6 kyr BP, peaking ~4.7–4.3 kyr BP. The synchronous increases in diatom productivity (BSi, δ18O_diatom), organic matter accumulation (TOC), and arboreal pollen, alongside reduced detrital inputs, indicate warmer, wetter conditions, longer growing seasons, and denser vegetation. These environmental improvements likely enhanced rangeland productivity and stability, creating favorable conditions for the expansion of nomadic pastoralism and cultural interactions across the northern Central Asian steppe, as reflected in rising archaeological SPDs and the spread/intensification of Afanasievo and Chemurchek cultural horizons. The AHCO coincides with broader regional HTM signals and strengthened Asian monsoon activity, linking land–sea thermal contrasts to continental interior warmth. The identification of centennial cool events (~4.2 and ~3.6 kyr BP) further contextualizes cultural dynamics and shifts in settlement patterns. Overall, the findings directly address the research question by tying a clearly defined regional climatic optimum to increased prehistoric population density, mobility, and exchange networks across Central Asia.

The study reconstructs a robust Holocene climate history from Kanas and Tiewaike Lakes and defines the Altai Holocene Climatic Optimum (AHCO) at ~6.5–3.6 kyr BP, with a thermal peak at ~4.7–4.3 kyr BP. This warm and humid interval, characterized by enhanced lake productivity, reduced catchment erosion, and expanded forest cover, aligns with regional HTM expressions and strengthened monsoon activity. Archaeological SPDs indicate that these favorable conditions promoted population expansion, intensified pastoralism, and cultural exchanges across the Altai–Sayan and broader Central Asian steppe during the Bronze Age. Future research should (1) expand high-resolution, seasonally sensitive multiproxy records across elevation and latitude transects in arid Central Asia; (2) integrate additional quantitative temperature and precipitation proxies to disentangle seasonal versus annual signals; (3) refine chronological controls to better resolve centennial-scale variability; and (4) couple paleoenvironmental data with regional archaeological datasets and agent-based models to quantify climate–society interactions.

• The XRF core-scanning elemental data are semi-quantitative and can be influenced by water content, surface roughness, and grain-size effects, although validated against conventional XRF measurements.
• Potential radiocarbon reservoir effects in lacustrine settings required exclusion or correction of BOM dates; residual uncertainties may remain, especially in TWK15A despite assessment.
• Summed probability distributions (SPDs) of radiocarbon dates provide a coarse proxy for human activity and are sensitive to sampling intensity, site preservation, and methodological choices, even after binning to reduce oversampling bias.
• Proxies may carry seasonal biases (e.g., diatom productivity reflecting growing-season conditions), complicating inference of annual mean temperatures.
• Findings are derived from two lake systems and regional syntheses; broader spatial coverage would improve generalizability and resolve local versus regional signals.