Sources of lead in a Tibetan glacier since the Stone Age

Lead (Pb) is a toxic heavy metal whose exposure via ingestion and inhalation causes adverse health impacts, even at low levels. Although Pb use dates to the Neolithic and was smelted in ancient China, major atmospheric emissions surged during the Industrial Revolution and after the introduction of leaded gasoline in the 1920s. While North America and Europe saw declines after phasing out leaded gasoline, Asian emissions (especially China) have remained high. The Tibetan Plateau (TP), among the most remote and least industrial regions, offers an ideal archive to resolve natural versus anthropogenic Pb contributions. Previous Guliya ice core trace element (TE) records indicated two periods with non-crustal metal enrichments (~1850–1950 and ~1970–2015) but source attribution was uncertain. This study aims to determine the onset and sources of anthropogenic Pb recorded in the Guliya ice cap by measuring high-precision Pb isotopes (including low-abundance 204Pb) from the Stone Age to 2015, and by applying a Bayesian mixing model to deconvolve natural and anthropogenic sources through time.

Prior work documents millennia-long human Pb use and variable regional onsets of contamination in natural archives (e.g., ~6500 BC in North American sediments; ~3000 BC in central China lake sediments). Guliya TE records previously showed two periods of elevated non-crustal metals, with tentative links to European coal (1850–1950) and, post-1970, to regional fossil fuels and biomass burning. Other TP and Himalayan archives (Mt. Everest ice core, northeastern Tibet lakes and peat) show relatively constant, dust-dominated Pb isotopic backgrounds before the 1950s, whereas non-TP archives (Greenland, Belukha, Colle Gnifetti) register large pre-1950 variations driven by nearby industrial emissions, indicating Tibet/Himalayas were largely shielded from European/Russian Pb. Published Pb isotopic fingerprints exist for Chinese coals, fuels, Pb/Zn ores, and selected regional aerosols, though many lack 204Pb, limiting source resolution in past studies. This work builds on those records by adding 204Pb and quantitative mixing model analyses.

Study site and materials: Ice cores were drilled on the Guliya ice cap (35°17′N; 81°29′E) in the western Kunlun Mountains at ~6200 m a.s.l. (plateau; GP1992, GP2015) and on the summit at ~6700 m a.s.l. (GS2015). A total of 89 discrete ice samples were analyzed: 37 from GP1992 (1500–1990), 46 from GP2015 (1970–2015), 4 from GP2015 (~36–12 ka BP), and 2 from GS2015 (~13.4 ka and ~3 ka BP). Seventeen dust potential source area (PSA) samples from around the TP and adjacent deserts were also analyzed. Timescales: GP1992 chronology used annual dust layer counting and the 1963 beta-activity horizon. GP2015 last-1000-year chronology used annual layer counting with reference horizons (2015, 1992, 1963). Holocene dating used radiocarbon of plant fragments and δ18Oatm; pre-15 ka BP chronology used 10Be and 36Cl peaks aligned to Laschamp (~41 ka BP) and Mono Lake (~33 ka BP) events with polynomial interpolation between. Sample preparation: Archived, previously TE-analyzed samples were decontaminated (Class 100 clean room), melted, acidified to 2% v/v Optima HNO3 for 30 days to leach trace elements, then stored frozen until analysis. Pb isotope analysis: 204Pb, 206Pb, 207Pb, 208Pb were measured at Texas A&M (Radiogenic Isotope Geosciences Facility) on a Thermo Neptune Plus MC-ICPMS. Samples were spiked with Tl (~1 ppb; 203Tl–205Tl) and aspirated via APEX Omega desolvating nebulizer. Mass bias was corrected using Tl; additional bracketing corrections used NBS 981 (assumed 206Pb/204Pb = 16.9406; 207Pb/204Pb = 15.4957; 208Pb/204Pb = 36.7184) with consistent ~0.02%/amu correction. In-run precisions (2σ) for 206Pb/204Pb, 207Pb/204Pb, 208Pb/204Pb ranged ~0.002–0.4% (median ~0.01%); higher-abundance ratios (206Pb/207Pb, 206Pb/208Pb) averaged ~0.002%. Blanks were <1 pg/g 206Pb; duplicate analyses were performed on 7 samples. Data analysis: MATLAB scripts extracted means/errors, performed fractionation corrections, smoothing (binomial filters), changepoint detection (converted to evenly spaced series of 200 points; mean-statistic), and non-parametric Wilcoxon rank-sum tests. K-means clustering (Calinski–Harabasz and silhouette criteria) was applied to normalized features; 206Pb/204Pb, 208Pb/204Pb, 206Pb/207Pb, 206Pb/208Pb were identified as most discriminative. Three-isotope plots compared Guliya samples with grouped PSAs and anthropogenic sources (Chinese coal, fuels, Pb/Zn ores; UK/Europe coal and gasoline; regional urban aerosols). Bayesian mixing model: MixSIAR (R) was used with 204Pb-based ratios (206Pb/204Pb, 207Pb/204Pb, 208Pb/204Pb), after univariate (Shapiro–Wilk, Kolmogorov–Smirnov) and multivariate normality (MVN package) checks. Sources included three natural PSAs (TP, Southwest TP, Taklamakan) and three Chinese anthropogenic sources (coal, liquid fuels, Pb/Zn ores). Very long MCMC runs were employed; mixing polygon simulations were checked prior to modeling. Temporal comparisons were made with Chinese sectoral Pb emission inventories (liquid fuels, coal, non-ferrous smelting) and Xinjiang coal energy production time series.

- Onset and trajectory of anthropogenic Pb: Pb isotope ratios shifted below Stone Age natural background beginning in 1949, with rapid decreases in 1960 and again after 1974, marking the emergence and intensification of anthropogenic Pb at Guliya. Comparable rapid changes were observed in Mt. Everest Pb isotopes and a 1970 Pb concentration spike in Mt. Logan, Canada, linked to Asian emissions. - Clustering and enrichment: K-means divided samples into Cluster 1 (pre-1974 plus some post-1974; more radiogenic; mean Pb EF ≈ 1.0) and Cluster 2 (exclusively post-1974; less radiogenic; mean Pb EF ≈ 1.5), indicating natural versus anthropogenic influence, respectively. - Natural sources pre-1974: Cluster 1 Pb isotope ratios overlap Tibetan Plateau PSAs in three-isotope space. Wilcoxon rank-sum p-values (206Pb/204Pb p=0.9532; 207Pb/206Pb p=0.5290; 208Pb/206Pb p=0.2572) support a TP natural dust origin. Regional archives (Everest ice, northeastern TP lakes/peat) show stable, dust-dominated backgrounds pre-1950, unlike extra-TP sites affected by industrial Pb. - Anthropogenic source attribution post-1974: Three-isotope comparisons indicate best matches with Chinese coal and Pb/Zn ores, but shifts toward Chinese fuels are evident. European/British coal and gasoline isotopic signatures are too unradiogenic to explain Guliya data and are excluded. Urban aerosols from Pakistan, Kazakhstan, and India (pre-ban) are also too unradiogenic, suggesting limited transport to Guliya. - Temporal alignment with Chinese emissions: Decreases in Guliya 206Pb/204Pb track rises in Chinese gasoline Pb emissions (start ~1960; rapid growth post-1978 economic reforms). Slight isotope ratio increases align with 1991 low-leaded gasoline introduction and 2000 phase-out; however, continued declines to ~2007 imply ongoing vehicle-related Pb (continued use and/or legacy remobilization). After ~2007, Xinjiang coal combustion rose sharply, coincident with a reversal to more radiogenic ratios, implicating coal. - Quantitative mixing (MixSIAR): Fuel contributions show changepoints at 1974 (to ~25%) and 2004 (increase after a brief 2002 dip), identifying gasoline as the primary anthropogenic source until ~2007. Natural sources (TP and Taklamakan) contribute ~70% before 1974, declining to ~50% in 1974 and ~35% by 2000; Southwest TP contributions were generally low but doubled (~10% to ~20%) between 1757 and 1932, consistent with slightly more radiogenic Pb and circulation shifts. Coal contributions are likely underestimated by MixSIAR due to overlap between Chinese coal and TP PSA isotopic signatures and high coal variability. - Industrial Revolution period (~1750–1935): Two smoothed "bumps" in isotope ratios coincide with TE enrichments but are attributed primarily to changes in natural dust provenance (incursions from SW Tibet) and atmospheric circulation, not European coal. - No ancient civilization Pb detected: Samples from pre-Bronze Age and 1500–1949 share similar isotope ratios, indicating ancient Pb use is not recorded at Guliya. - Representative cluster means (Table 1): Cluster 1 vs Cluster 2 averages include 206Pb/204Pb ≈ 18.684 vs 18.395; 207Pb/206Pb ≈ 0.839 vs 0.851; 206Pb/207Pb ≈ 1.193 vs 1.175; 208Pb/207Pb ≈ 2.480 vs 2.461, consistent with increased anthropogenic influence post-1974.

The study resolves when and from where anthropogenic Pb began impacting a remote Tibetan glacier. High-precision Pb isotopes, including 204Pb, reveal a distinct post-1949 decline in radiogenicity with rapid steps in 1960 and 1974, marking the initial and accelerated anthropogenic signal. Source fingerprinting via three-isotope comparisons, temporal alignment with Chinese sectoral emissions, and Bayesian mixing indicate that Chinese gasoline dominated the anthropogenic Pb reaching Guliya through the early 2000s, despite national phase-out in 2000, likely due to lagged transitions and remobilized legacy Pb. Around 2007, a shift in Guliya isotope trends aligns with rapidly increasing coal combustion in Xinjiang, implicating regional coal as the dominant source thereafter; contributions from Pb/Zn ores likely increased as well. Pre-1950 variations reflect changes in natural dust provenance and circulation, not extraregional industrial emissions, affirming the TP’s isolation from European/Russian Pb in that era. These findings refine prior TE-based interpretations by leveraging 204Pb and probabilistic mixing, demonstrating that even highly dust-influenced ice can sensitively record anthropogenic signals and changes in regional energy use and policy. The results underscore the value of remote TP archives for tracking transboundary pollution histories and validating emission controls.

This work presents a high-resolution Pb isotopic record from the Guliya ice cap spanning ~36 ka BP to 2015, establishing a robust natural baseline and pinpointing the rise of anthropogenic Pb. Anthropogenic influence begins to emerge in 1949, intensifies by 1960 and 1974, and is dominated by Chinese gasoline through ~2007, after which coal combustion becomes the principal source, consistent with regional energy trends in Xinjiang. Incorporating 204Pb measurements and a Bayesian mixing framework (MixSIAR) enabled more sensitive and quantitative source apportionment than prior studies relying only on higher-abundance isotopes. Future research should (i) expand 204Pb-inclusive isotopic characterization of Chinese coal, fuels, ores, and aerosols to reduce mixing model uncertainties, (ii) increase sample density for the 1750–1950 interval to better resolve circulation-driven natural source shifts, (iii) integrate additional tracers (e.g., Sr, Nd isotopes; REE patterns) to separate coal from natural dust where signatures overlap, and (iv) couple isotope records with atmospheric transport modeling to refine source regions and seasonality.

- Mixing model constraints: Chinese coal Pb isotopic compositions are highly variable and overlap with TP natural PSA signatures, likely causing underestimation of coal contributions in MixSIAR. Many published anthropogenic datasets lack 204Pb, limiting fingerprint distinctiveness. - Sample representation: Only 15 representative samples cover 1750–1930, limiting resolution of Industrial-era variability. Summit core coverage is limited (n=2), precluding robust summit–plateau comparisons. - Transport inference: Back-trajectory frequencies for pre-1950 transport paths are low, adding uncertainty to long-range source attributions. - Analytical/model assumptions: MixSIAR assumes (multi)variate normal source distributions and complete source sets; the model will allocate small contributions to all sources, reflecting uncertainty (e.g., non-zero anthropogenic contributions inferred for Stone Age samples). - Archive characteristics: High dust loads at Guliya damp EF magnitudes, and preferential dissolution effects could, in theory, bias individual samples (one highly radiogenic outlier was excluded due to unresolved cause).