Loess deposits in the low latitudes of East Asia reveal the ~20-kyr precipitation cycle

As an important part of the global monsoon system, the Asian summer monsoon (ISM, WNPSM, EASM) forms due to seasonal land–sea thermal contrasts driven by Northern Hemisphere insolation, transporting moisture inland and sustaining East Asian ecosystems. Loess records from the Chinese Loess Plateau (CLP) have classically been used to infer EASM variability, with magnetic susceptibility (MS) and carbonate or snail-shell δ18O often showing dominant ~100-kyr cycles since the Middle Pleistocene and suggesting ice-volume forcing. However, other CLP proxies and stalagmite δ18O from East Asia indicate dominant ~20–23-kyr precessional cycles and minimal ~100-kyr variance, implying insolation forcing and creating the "Chinese 100-kyr problem"—a discrepancy between loess and speleothem records. Model simulations support precession-driven strengthening of summer monsoon winds and precipitation during high insolation, yet loess and stalagmite records disagree. To address this, records from beyond the mid-latitude CLP are needed. This study targets low-latitude Xiashu loess at Madang (Jiangxi Province) in the monsoon core zone to reassess EASM cyclicity and its drivers over the last ~350 kyr.

Mid-latitude CLP loess studies show MS and carbonate/snail δ18O dominated by ~100-kyr cycles since the Middle Pleistocene, interpreted as EASM changes paced by global ice volume. In contrast, a spliced EASM record from western CLP and microcodium δ18O from central CLP exhibit dominant precessional (~20–23-kyr) variability. East Asian stalagmite δ18O records are spatially and temporally consistent at orbital scales and primarily reflect monsoon intensity, showing strong ~20-kyr cycles with little ~100-kyr variance, supporting precession-driven forcing. Climate model simulations similarly show strengthened southwesterlies/southeasterlies, inland rainfall increase, and more negative precipitation δ18O during high Northern Hemisphere insolation. Proposed explanations for the "Chinese 100-kyr problem" include: marginal location of the CLP where precessional precipitation effects are muted; MS flux revealing 20-kyr importance despite raw MS dominance at 100-kyr; MS integrating multiple processes beyond precipitation; suppression of stalagmite δ18O glacial–interglacial variability by moisture pathway effects; post-depositional smoothing amplifying ~100-kyr and damping ~20–40-kyr signals; and thresholds for magnetic enhancement under reduced glacial precipitation. Records from the monsoon marginal zone show wet-dry variations similar to speleothems over the past ~300 kyr, but core-zone loess records were lacking prior to this study.

Field site and sampling: The Madang loess profile (29°59′29″N, 116°39′44″E) in Pengze County, Jiangxi Province, is 15.9 m thick and divided into seven loess–paleosol units corresponding to L1/S1–L4 of the CLP. A total of 291 samples were collected for grain size, magnetic susceptibility (MS), frequency-dependent susceptibility (FDS), dithionite-citrate-bicarbonate extractable iron (FeD), total iron (FeT), hematite content, and redness. Six OSL samples were collected from units L1 and S1. Chronology and correlation: OSL ages (30.1 ± 2.2 ka at 40 cm to 85.5 ± 6.0 ka at 350 cm) were used to define the MS time series and correlate it with the Luochuan CLP profile. Nine depth–age tie points enabled linear interpolation/extrapolation to build the Madang age model, yielding a basal age of 346.0 ka and a top age of 42.9 ka with ~1 kyr sampling resolution. Cross-correlation between Madang and Luochuan MS (using a program for unevenly spaced time series) gave r_xy = 0.88. Spectral analyses: Wavelet and power spectrum analyses were applied to MS and FeD/FeT time series to identify dominant periodicities and their confidence levels (95% robust intervals). MS and FDS measurements: Using a Bartington MS2 system, low-frequency (470 Hz) and high-frequency (4700 Hz) MS were measured on dried, weighed samples; FDS was defined as the low–high frequency difference. FeD and FeT determinations: FeD was obtained via the DCB method: ~0.1 g powdered sample reacted with sodium dithionite in buffered sodium citrate–bicarbonate at ~pH 7.3, shaken at 80 °C for 90 min, centrifuged, and analyzed colorimetrically (Shimadzu UV-1800, 535 nm) against Fe2O3 standards. FeT was measured after HF/HNO3 digestion in PTFE bombs with ICP-OES (Agilent 5110). OSL dating: Under red light, samples were pretreated (HCl, H2O2), fine grains (4–11 µm) isolated and treated with H2SiF6 to extract quartz; purity verified by OSL-IR depletion ratio. Equivalent dose (De) determined using SAR protocol on Risø TL/OSL-DA-20 readers with specified preheat/cut-heat and stimulation conditions. Dose rates derived from ICP-MS/AES U, Th, K concentrations, water content (±10%), alpha efficiency 0.04 ± 0.02, and cosmic ray corrections as functions of elevation and depth.

- The Madang MS time series exhibits a dominant ~101-kyr cycle (95% confidence), consistent with eccentricity-band variability and similar to integrated CLP MS over 0–350 ka. - The FeD/FeT time series shows a distinct ~18-kyr cycle (95% confidence), indicating dominance by the precessional (~20-kyr) band and reflecting precipitation-driven chemical weathering. - The MS and FeD/FeT records at Madang have different dominant cycles, unlike many CLP records where both often show ~100-kyr dominance. - Chronology and correlation: Nine depth–age tie points yield a profile spanning 346.0–42.9 ka with ~1 kyr resolution. MS cross-correlation between Madang and Luochuan is strong (r_xy = 0.88). - FeD/FeT behaves as a precipitation-dominated proxy at low latitudes: modern calibrations in South China show strong FeD/FeT–MAP correlation (r = 0.83 for MAP 900–1720 mm) and negligible correlation with MAT (r = -0.08). - At Madang, MS likely reflects temperature or redox processes: MS correlates positively with Antarctic temperature proxy (r_xy = 0.61) and negatively with the LR04 benthic δ18O stack (r_xy = -0.70), with high MS during interglacials and low MS during glacials. - FeD/FeT correlates with East Asian summer monsoon intensity indicators and forcing: generally higher FeD/FeT aligns with more negative stalagmite δ18O at Sanbao Cave (weak negative correlation r_xy = -0.21) and with the July insolation difference between 30°N and 30°S (r_xy = 0.73). - Comparisons with marginal-zone records (western CLP and Tengger Desert) show coherent precessional (~20–23-kyr) wet–dry cycles across the broader EASM region. - Model simulations for the past 425 kyr show MAT dominated by ~100-kyr cycles and MAP increasingly influenced by ~20-kyr from north to south, consistent with Madang’s separation of temperature (~100-kyr MS) and precipitation (~20-kyr FeD/FeT).

The core-zone Madang loess reveals that EASM precipitation varied predominantly at precessional (~20-kyr) timescales, while temperature varied mainly at ~100-kyr timescales. This separation of cycles helps reconcile the "Chinese 100-kyr problem" by showing that different proxies in different climatic settings can record distinct aspects of the monsoon system: FeD/FeT captures precipitation-driven chemical weathering under warm, wet low-latitude conditions, whereas MS is more sensitive to temperature-driven redox conditions and/or post-depositional smoothing. Two hypotheses are advanced: (1) both MS and FeD/FeT could reflect precipitation, but differential post-depositional smoothing at Madang preserves the 20-kyr FeD/FeT signal while damping the MS precessional band; (2) MS at wet low latitudes is predominantly controlled by temperature-induced redox status, with excessive moisture potentially suppressing magnetic enhancement. The strong resemblance between Madang and Luochuan MS suggests a regionally coherent factor (likely temperature or shared diagenetic processes), while FeD/FeT at Madang tracks precession-paced insolation, consistent with increased land–sea thermal contrast driving stronger summer monsoons and higher precipitation. Comparisons with stalagmite δ18O and marginal-zone loess further support coherent precessional precipitation variability across the EASM domain. Although the Madang record lacks a strong 40-kyr component seen in some simulations for the Yangtze River Valley, the dominant precessional precipitation signal and eccentricity-band temperature signal together depict alternating climate modes (warm–dry, warm–wet, cold–dry, cold–wet) over the last ~350 kyr and imply future drying until ~AD 5700 followed by gradual wetting to ~AD 31000 under insolation forcing trajectories.

The Madang low-latitude loess profile provides a high-resolution ~350-kyr record from the EASM core zone, demonstrating that precipitation (FeD/FeT) varied mainly with precession (~20-kyr), while temperature (MS) varied with ~100-kyr cycles. These findings indicate that EASM precipitation in this region is primarily forced by precession-dominated insolation changes, helping resolve the "Chinese 100-kyr problem" by attributing contrasting proxy cyclicities to differing climatic controls and site conditions. The record underscores a broad coherence of precessional wet–dry cycles across the EASM region and highlights the importance of low-latitude loess archives for disentangling temperature and precipitation signals at orbital timescales.

- Age model constructed by correlating Madang MS to Luochuan and adopting an orbitally tuned chronology introduces potential circularity; however, validation with abundant OSL dates at Luochuan mitigates severe distortion. - Linear interpolation between limited depth–age tie points may introduce temporal uncertainties, contributing to weak correlations with independent records (e.g., stalagmite δ18O). - Spatial and temporal heterogeneity of EASM precipitation means local precipitation at Madang may not strictly covary with regional monsoon intensity, weakening proxy–proxy correlations. - Local microgeomorphology, ecosystem changes during loess accumulation, and potential downward migration of iron under redox conditions may affect FeD/FeT, though no visible iron pans were observed. - MS in low-latitude, wet settings may be influenced by reducing conditions and local landforms (drainage/ventilation), complicating its interpretation purely as a precipitation proxy. - The Madang record lacks the 40-kyr precipitation cycle simulated for parts of the Yangtze River Valley, indicating possible proxy sensitivity or regional differences not fully captured.