Ecological risk assessment of rare earth elements in agricultural soils adjacent to an abandoned coal gangue pile in Chongqing, Southwest China Authors: Maochao Zhang, Lihua Peng, Hong Li, Liusan Cheng, Mingqiu Wang Published: 2026 ABSTRACT AbstractRare earth elements plus yttrium (REYs) are increasingly recognized as emerging contaminants, yet their ecological risks in agricultural soils remain are poorly understood. This study assessed REY contamination and ecological risk in agricultural soils adjacent to an abandoned coal gangue pile in a post-coal-mining area of Chongqing, Southwest China. Geochemical characteristics, together with stratigraphic and paleogeographic context and multivariate statistical analysis, suggest that coal gangue likely serve as a conduit for REYs sourced from volcanic materials associated with the Emeishan Large Igneous Province (ELIP) and their weathering products into nearby soils. Pollution indices indicate site-wide REY contamination. The pollution load index (PLI) values of 1.72–3.30 (mean 2.19) classified all soils as polluted. Element-specific geoaccumulation index (Igeo) values were predominantly within 0–1 in 96% of cases, indicating uncontaminated to moderately contaminated conditions for individual REYs, with a few subsoil cases reaching moderate contamination. Despite a generally consistent contamination level across the 0–30 cm profile, the 0–20 cm horizon exhibited slightly higher ΣREY concentrations and elevated PLI and RI values relative to the 20–30 cm layer. The potential ecological risk index (RI) is dominated by moderate risk conditions, with strength risk cases occurring mainly in the upper 0–20 cm layer. Elemental contributions to RI ranked, from highest to lowest, as Lu, Tb, Eu, Ho, and Tm, which together accounted for about 63% of the index. The HREYs (Lu, Tb, Ho, Tm) contributed about 51% of the RI, underscoring their dominant role in shaping ecological risk. Future studies should extend toward integrated, region-wide assessments of REY contamination, ecological risk, and potential human exposure in agricultural soils affected by coal gangue piles derived from Late Permian Longtan Formation coal mining, in order to support risk-based environmental management in post-coal-mining regions.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-51210-w.Keywords: Rare earth elements, Potential ecological risk, Agricultural soils, Coal gangueSubject terms: Ecology, Ecology, Environmental sciences, Natural hazards Introduction Rare earth elements comprise the lanthanides together with yttrium (REYs)1,2. Scandium (Sc) is also often included in this group due to its similar chemical behavior and common genetic association with REY-rich ore deposits3. However, most geochemical studies exclude Sc from REY classifications due to its small ionic radius2. Based on electron configuration and geochemical behavior, REYs are subdivided into light rare earth elements (LREYs: La to Eu) and heavy rare earth elements (HREYs: Gd to Lu and Y)4,5. Although yttrium has a relatively low atomic weight, it is assigned to the HREY group due to its occurrence, ionic radius, and geochemical properties being closer to HREYs than to LREYs3. Promethium (Pm), although part of the lanthanide series, is excluded from environmental and geochemical research because it lacks stable isotopes and occurs only negligibly in nature6,7. In recent decades, REYs have been widely applied in agriculture, metallurgy, medicine, petrochemicals, electronics, renewable energy, and aerospace industries8,9. Such extensive applications of REYs have resulted in their elevated release into the environment9,10. Due to their environmental persistence, bioaccumulation, and chronic toxicity11–13, REYs are increasingly regarded as emerging contaminants of concern for both ecological integrity and public health14–17. Soils act as a major sink for toxic elements in terrestrial ecosystems and represent a key medium for the transfer of contaminants into the food chain, thereby posing potential risks of human exposure18–21. For non-occupationally exposed populations, dietary intake constitutes the primary pathway of exposure to trace elements22. Experimental evidence has linked REY exposure to several adverse health outcomes, including neural tube defects, nephrogenic systemic fibrosis, anemia, and endomyocardial fibrosis14,23. Furthermore, REYs can enter the human body through multiple pathways and induce dysfunction in various organs and systems via changes in genetic, epigenetic, and signaling pathways24. Although no cases of human poisoning via the food chain have been reported, concerns persist regarding chronic low-level REY exposure, as these elements can accumulate in the blood, brain, and bone22,25. In recognition of these risks, continuous monitoring and health risk assessments are widely recommended22. Acknowledging these concerns, China established regulatory measures for REY content in agricultural products as early as 2006, incorporating total REY concentrations (ΣREYs) as an indicator in the Farmland Environmental Quality Standards for Edible Agricultural Products (HJ 332–2006)26. Mining and mineral processing represent the primary anthropogenic sources of REYs, and many studies have therefore examined REY contamination near rare earth mining and beneficiation sites27,28. However, Late Permian coal-bearing strata in Southwest China are naturally enriched in REYs, and some coal seams in Chongqing exhibit concentrations far exceeding global averages29–33. Coal gangue, the major solid waste from mining, is widely recognized as a persistent pollution source that releases multiple contaminants through weathering and leaching. Compared with heavy metal(loid)s (e.g., Cd, Pb, As, Cr), which have been extensively studied34,35, the mobilization and ecological risks of REYs from gangue remain insufficiently addressed. Chongqing, Southwest China, is both a historic coal mining hub and an ecologically sensitive region. As the core area of the Three Gorges Reservoir, it plays a vital role in freshwater protection, ecological security, and biodiversity. Although all coal mining activities in Chongqing ceased in 2020, large volumes of coal gangue remain distributed across the hilly agricultural landscape. Intense monsoonal precipitation accelerates the weathering of these deposits and promotes the mobilization of associated elements. Under such conditions, coal gangue may serve as a long-term source of REYs, with the potential to enter surrounding farmlands. This raises concerns regarding soil quality, crop safety, and possible human exposure. Despite these potential risks, the extent and ecological consequences of REY contamination in agricultural soils adjacent to an abandoned coal gangue pile remain poorly constrained. Addressing this knowledge gap is critical, given both the ecological sensitivity of the region and the elevated REY contents documented in the local coal-bearing strata. In this study, a comprehensive assessment was undertaken to examine REY contamination and ecological risks in agricultural soils adjacent to an abandoned coal gangue pile in Chongqing, Southwest China. Three widely recognized indices were employed to achieve this: the pollution load index (PLI), the geoaccumulation index (Igeo), and the potential ecological risk index (RI). PLI provides an integrated measure of multi-element pollution across the soil profile, Igeo identifies element-specific enrichment relative to geochemical background values, and RI incorporates both contamination levels and toxic-response coefficients to evaluate ecological risk. Based on these complementary frameworks, the study aimed to (1) quantify REY concentrations and their depth distributions across topsoil (0–10 cm), midsoil (10–20 cm), and subsoil (20–30 cm), (2) assess the degree of REY contamination, and (3) evaluate the associated ecological risks posed by REYs. Materials and methods Study area and sample collection The study area is located in Chongqing, Southwest China, as shown in Fig. 1, within a coal gangue accumulation zone derived from an abandoned coal mine. The coal-bearing strata in this region belong to the Upper Permian Longtan Formation, which is extensively distributed in Southwest China36. The Longtan Formation disconformably overlies the Middle Permian Maokou Formation and is conformably overlain by the Upper Permian Changxing Formation37,38. (a) Regional location of the study area in China; (b) Distribution of soil sampling sites relative to the coal gangue pile; (c) Overview of the coal gangue pile on the upper hillslope; (d) Coal gangue check dam and adjacent cultivated farmland; (e) Exposed limestone bedrock representing soil parent material. Sampling sites are coded as “LxxY”, where “Lxx” identifies the sampling location shown in (b) and “Y” denotes the depth interval (A: 0–10 cm; B: 10–20 cm; C: 20–30 cm), consistent with Table 6. (a) and (b) are modified from Zhang et al.34. The map was designed using ArcGIS Desktop (version 10.6; Esri, Redlands, CA, USA, https://www.esri.com). The study area is situated on a gentle hillslope that grades downslope into a small valley. The coal gangue pile is located on the upper slope, with a continuous block of agricultural land immediately downslope and a residential area further downslope. This slope-to-valley arrangement positions the farmland as the primary receptor zone for REYs mobilized from the gangue pile through weathering and leaching. Given this spatial configuration, the continuous agricultural land directly downslope of the gangue pile was selected as the sampling area (Fig. 1c–e). The sampling sites were evenly distributed across this entire farmland block to ensure representative spatial coverage (Fig. 1b). Field observations indicate that the soils in the sampling area are intensively cultivated, exhibiting a loose to moderately friable surface structure. The soil matrix is predominantly yellowish-brown, with localized occurrences of small rock fragments derived from the underlying limestone parent material (Fig. 1d–e). Soil sampling was conducted in mid-May 2021 (Fig. 1). Composite soil samples were systematically collected from three depth intervals: 0–10 cm (topsoil), 10–20 cm (midsoil), and 20–30 cm (subsoil), yielding a total of 21 composite soil samples. Each composite soil sample consisted of a homogenized mixture of five subsamples (approximately 1.0 kg each), collected from the center and four corners of a defined 1 m × 1 m quadrat. Two representative limestone samples, collected from the exposed bedrock at the base of the soil profiles, were included to characterize the parent material of the agricultural soils in the study area. All soil and limestone samples were sealed in polypropylene bags, labeled, and transported to the laboratory for analysis. Sample analysis Soil samples were air-dried at room temperature and sieved through a 2 mm nylon mesh to remove pebbles and plant debris. Subsequently, they were ground to a fine powder using an agate mortar and passed through a 0.149 mm nylon mesh sieve. Limestone samples were processed using the same protocol. All samples were stored in sealed plastic bags until analysis. Soil and limestone samples were digested using a mixed HF–HNO3 procedure following Qi et al.39. The detail procedure is as follows: Firstly, 50 mg of sample was accurately weighed into a polytetrafluoroethylene (PTFE) crucible, to which 1 mL of HF and 1 mL of HNO3 were added. The crucible was then sealed in a steel digestion vessel and heated in an oven at 190 °C for more than 30 h. After cooling, the crucible was removed, placed on a low-temperature hot plate, and evaporated to dryness. Subsequently, 1 mL of HNO3 was added and the sample was again evaporated to dryness. A Rh internal standard solution (200 ng), 2 mL of HNO3, and 3 mL of deionized water were then added to the crucible, which was resealed in the steel vessel and heated at 140 °C for 5 h. After cooling, the solution was transferred, homogenized, and 0.4 mL was pipetted into a centrifuge tube and diluted to 10 mL for ICP-MS analysis. The concentrations of REYs and associated trace elements (Nb, Ta, Zr, and Hf) were determined using a Quadrupole Inductively Coupled Plasma Mass Spectrometer (Q-ICP-MS), model ELAN DRC-e, manufactured by PerkinElmer, Canada, at the State Key Laboratory of Ore Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences. Analytical precision was monitored through repeated measurements, and the relative standard deviation (RSD) for most trace elements was better than 10%. Method detection limits (MDLs), calculated from repeated blank measurements, are provided in Table S1. Since direct geochemical data for coal gangue were unavailable, REY and HFSE data for coal seam roof-and-floor materials (Coal R&F), coal seams, and tuff interlayers were compiled from published literature30,31,40–42 (Supplementary Table S2) and used as proxies to approximate the geochemical characteristics of coal gangue. In addition, data for the REY and HFSE in high-Ti and low-Ti Emeishan basalts43–47 were compiled (Supplementary Table S3) to provide comparative reference values. Characteristic parameters of REYs Lanthanides (La–Lu) exhibit very similar chemical properties due to the filling of the 4f electron shell, and are often treated as migrating geochemically as a coherent entity48,49. However, processes such as redox reactions, adsorption–desorption, and remineralization can induce measurable fractionation among these elements50–53. Normalized REY patterns are widely applied as an effective tool to identify source contributions54. By minimizing Oddo–Harkins effects, these patterns facilitate the recognition of REY fractionation and improve understanding of their distribution in natural environments55–58. Post-Archean Australian Shale (PAAS) and the North American Shale Composite (NASC) are widely used reference compositions for normalizing REY concentrations in sediments and soils. In this study, PAAS was selected because its composition is comparatively uniform and well characterized, providing a robust approximation to the average post-Archean upper continental crust50. While NASC remains valuable for regional assessments, its use may introduce inconsistencies arising from sampling limitations and regional heterogeneity59,60, making PAAS the more appropriate reference for the present analysis. To quantify REY fractionation in soils, cerium and europium anomalies (δCe and δEu) were calculated based on PAAS-normalized concentrations. The anomaly indices were determined using the following equations: where the subscript N denotes PAAS-normalized concentrations, obtained by dividing the measured elemental concentration in the sample by its corresponding PAAS reference value. Values greater than 1 indicate a positive anomaly, whereas values less than 1 indicate a negative anomaly61. Pollution assessment of REYs in soil Geoaccumulation index The geoaccumulation index (Igeo), proposed by Müller62, was developed to evaluate the degree of contamination of an element in sediments relative to its geochemical background. It has since been widely applied for assessing anthropogenic influences on surface soils63,64. In this study, the Igeo was applied to quantify the extent of REY accumulation in soils. The index is calculated using Eq. (3). where Cn is the measured concentration of the selected REY, and Bn is the local natural background concentration of the REY in soil. Background values of REYs in Chinese soils were adopted from Chen et al.65. A factor of 1.5 was introduced to account for the possible variations of the background data due to lithological variations. The Igeo values were interpreted according to the classification scheme proposed by Müller62 (Table 1). Table 1Classification of the geoaccumulation index (Igeo).ClassValueSoil quality0Igeo ≤ 0Uncontaminated10 < Igeo ≤ 1Uncontaminated to moderately contaminated21 < Igeo ≤ 2Moderately contaminated32 < Igeo ≤ 3Moderately to strongly contaminated43 < Igeo ≤ 4Strongly contaminated54 < Igeo ≤ 5Strongly to extremely contaminated6Igeo > 5Extremely contaminated Classification of the geoaccumulation index (Igeo). Pollution load index The Pollution Load Index (PLI), proposed by Tomlinson et al.66, provides a quantitative measure of cumulative multi-element contamination. It is derived from the contamination factor (CF): where CREY is the measured concentration of a given REY in the sample, BREY is the corresponding geochemical background value, CF is the contamination factor, and n is the number of REYs considered. PLI values ≤ 1 indicate no pollution, whereas values > 1 denote progressive contamination and potential environmental degradation66. Potential ecological risk assessment The potential ecological risk index (RI), proposed by Hakanson67, was employed in this study to evaluate the potential ecological risks caused by REY contamination in soils. This method integrates both the concentration of individual elements and their relative toxicities. The relevant equations are as follows: where is the potential ecological risk factor of the ith element; is the measured concentration of the ith element; and is the background value of the ith element. denotes the toxic-response factor assigned to each REY, with values of La = 1, Ce = 1, Pr = 5, Nd = 2, Sm = 5, Eu = 10, Gd = 5, Tb = 10, Dy = 5, Y = 2, Ho = 10, Er = 5, Tm = 10, Yb = 5, and Lu = 2068. The classification criteria for both and RI are summarized in Table 2 (after Chen et al.68). Table 2Classification of potential ecological risk based on and RI. RangePollution levelRI rangeRisk level < 20MildRI < 110Mild20 ≤ < 40Moderate110 ≤ RI < 220Moderate40 ≤ < 80Strength220 ≤ RI < 440Strength80 ≤ < 160Strong440 ≤ RIStrong160 ≤ Very strong Classification of potential ecological risk based on and RI. Statistical analysis Descriptive statistics and correlation analysis were performed using Jamovi v2.7 for Windows69. Pearson correlation coefficients were applied to assess linear relationships among REYs, and principal component analysis (PCA) was applied to distinguish potential sources by grouping elements with similar concentration patterns. Results and discussion REY concentrations in the soils REYs were quantitatively determined in agricultural soils adjacent to an abandoned coal gangue pile. A depth-stratified sampling approach was employed, with samples collected from the 0–10 cm (topsoil), 10–20 cm (midsoil), and 20–30 cm (subsoil) intervals to assess potential vertical trends in REY distribution. Concentrations of individual REYs are listed in Supplementary Table S4. The ΣREY concentrations in the soils ranged from 289.8 to 530.5 mg kg−1, with a mean of 365.6 mg kg−1. At the location scale (Fig. 1b), ΣREYs show discernible site-to-site variability (Table 6), with consistently lower values at L03 (289.8–318.0 mg kg−1 across depths) and the maximum observed at L21C (530.5 mg kg−1). Slight enrichment was observed in the 10–20 cm horizon, with mean values of 366.9 mg kg−1 (0–10 cm), 375.7 mg kg−1 (10–20 cm), and 354.1 mg kg−1 (20–30 cm) (Table 3; Fig. 2a). These concentrations substantially exceeded the reported global average ΣREYs in surface soils (137.1 mg kg−1; commonly 100–200 mg kg−1; Arciszewska et al.9) and the background concentrations in Chinese soils (167 mg kg−1; Chen et al.65), demonstrating significant REY enrichment in the agricultural soils of the study area. The ΣREYs in the soils also exceeded those of the limestones and the Emeishan basalts (Supplementary Table S3), but lower than values reported for coals, Coal R&F, and tuff in the study area (Table 3; Fig. 2b). Collectively, these comparisons indicate that the soils are enriched in REYs relative to background lithologies. Table 6Summary of ΣREYs, LREY/HREY, PLI and RI for soils at different depths (A: 0–10 cm; B: 10–20 cm; C: 20–30 cm) across sampling locations (Fig. 1b).SamplesΣREYsLREY/HREYPLIRIL01A402.13.62.4229.8L02A359.03.32.2212.7L03A292.43.51.7169.8L11A392.63.72.3215.0L12A373.93.52.2216.7L21A373.23.12.3225.1L22A374.93.32.3220.7L01B442.53.52.6255.6L02B362.73.42.2211.4L03B289.83.51.7168.0L11B338.23.32.1202.3L12B397.33.62.3221.5L21B429.03.32.6257.1L22B370.63.22.3222.8L01C340.23.42.1205.8L02C347.03.32.1206.5L03C318.03.71.8179.1L11C306.33.91.7168.4L12C323.33.51.9187.1L21C530.53.43.3325.9L22C313.43.31.9185.7Sample identification follows the format “LxxY”. “Lxx” (the first three characters) refers to the sampling location labeled in Fig. 1b, and “Y” denotes the depth interval: A (0–10 cm), B (10–20 cm), and C (20–30 cm). Summary of ΣREYs, LREY/HREY, PLI and RI for soils at different depths (A: 0–10 cm; B: 10–20 cm; C: 20–30 cm) across sampling locations (Fig. 1b). Sample identification follows the format “LxxY”. “Lxx” (the first three characters) refers to the sampling location labeled in Fig. 1b, and “Y” denotes the depth interval: A (0–10 cm), B (10–20 cm), and C (20–30 cm). Table 3The main REY geochemical indicators in soils and potential source rocks of the study area.Sample typeΣREYs (mg/kg)LREY/HREYδEuδCeMinMaxMeanTopsoil (0–10 cm)292.4402.1366.93.40.991.00Midsoil (10–20 cm)289.8442.5375.73.40.971.01Subsoil (20–30 cm)306.3530.5354.13.51.031.02Soil (0–30 cm)289.8530.5365.63.41.001.02Bedrock (Limestone)4.325.414.93.41.111.02High-Ti basalt112.5395.0249.63.21.460.94Low-Ti basalt86.7357.4133.61.81.450.91Coals52.41766.2402.32.90.670.93Coal R&F155.3845.7432.83.50.941.00Tuff//787.83.51.020.94Soils of China//176.73.80.990.95The sources of the REY and high-field-strength element (HFSE; Nb, Ta, Zr, Hf) data compiled for this study are detailed in Supplementary Tables S2 and S3. As presented in Supplementary Table S2, data for coal seam roof-and-floor materials (Coal R&F), coal seams, and tuff interlayers, serving as proxies for coal gangue, were sourced from published literature (Dai et al.31,40,41; Zou et al.30,42). Furthermore, Supplementary Table S3 compiles the REY and HFSE data for high-Ti and low-Ti Emeishan basalts, which provide a comparative geochemical reference, from Lai et al.43, Li et al.44, Song et al.47, and Xiao et al.45,46. The main REY geochemical indicators in soils and potential source rocks of the study area. The sources of the REY and high-field-strength element (HFSE; Nb, Ta, Zr, Hf) data compiled for this study are detailed in Supplementary Tables S2 and S3. As presented in Supplementary Table S2, data for coal seam roof-and-floor materials (Coal R&F), coal seams, and tuff interlayers, serving as proxies for coal gangue, were sourced from published literature (Dai et al.31,40,41; Zou et al.30,42). Furthermore, Supplementary Table S3 compiles the REY and HFSE data for high-Ti and low-Ti Emeishan basalts, which provide a comparative geochemical reference, from Lai et al.43, Li et al.44, Song et al.47, and Xiao et al.45,46. (a) Boxplots of ΣREYs in soils at different depths (0–10, 10–20, 20–30 cm); (b) Comparison of average ΣREY concentrations in soils with potential source rocks. HT: high-Ti basalts, LT: low-Ti basalts, Coal R&F: coal seam roof-and-floor materials. Distribution parameters of REYs in the soils The distribution of REYs in the study soils was characterized using the LREY/HREY ratio (mass ratio of light to heavy REYs), together with Ce and Eu anomalies (Table 3). Across the 0–30 cm horizons, the relative abundances of individual REYs exhibited consistent patterns without systematic depth-dependent variation. The REY abundance sequence was identical across the topsoil, midsoil, and subsoil horizons, following the order Ce > La > Nd > Y > Pr > Sm > Gd > Dy > Er > Yb > Eu > Ho > Tb > Tm > Lu. The leading triad (Ce, La, Nd) accounted for 67–72% of ΣREYs at all depths, confirming the consistent predominance of LREYs in the soils. To quantify the degree of fractionation between LREYs and HREYs, the LREY/HREY ratio was used instead of (La/Yb)N, as group-sum ratios are less affected by individual elemental behavior and provide a more robust comparison in soils with complex mineral hosts. Across all three depth intervals, the LREY/HREY ratios ranged from 3.1 to 3.9 (mean 3.4), closely approximating the PAAS ratio of 3.7, indicating consistent enrichment of LREYs over HREYs throughout the soil profile. In the PAAS-normalized REY patterns (Fig. 3), the soil curves were generally flat, with a gentle convexity in the Sm – Gd interval, and nearly parallel to those of Coal R&F and tuff, suggesting similar levels of inter-group differentiation (Fig. 3; Table 3). The predominance of LREYs in the soils is broadly consistent with distributions reported for Chinese agricultural soils70. Even in ion-adsorption REY deposits of southern China, known for HREY production, soils remain LREY-enriched, though with much lower LREY/HREY ratios averaging 1.9811. Across sampling locations (Fig. 1b), LREY/HREY remains within a relatively narrow range (3.1–3.9; Table 6), indicating broadly comparable LREE–HREE partitioning among sites despite differences in enrichment levels. Comparative PAAS-normalized REY patterns for soils, high-Ti basalts (HT), low-Ti basalts (LT), coals, limestones, coal gangue, and tuff. AV_SA, AV_SB, and AV_SC represent the average REY concentrations of the 0–10 cm (topsoil), 10–20 cm (midsoil), and 20–30 cm (subsoil) horizons, respectively. Previous studies have shown that process-oriented descriptors based on MREE curvature and fractionation patterns (e.g., MREE-curvature indices) can complement concentration-based metrics by quantifying subtle convexity in shale-normalized REY patterns and thereby informing interpretations of REY mobility controls in mining-impacted systems71,72. In this study, the PAAS-normalized soil patterns exhibit a gentle convexity in the Sm – Gd interval (Fig. 3), suggesting that fractionation accompanies bulk enrichment. Because MREE curvature can arise from multiple, non-unique controls, future work should integrate coupled soil–water REY patterns (including dissolved–colloidal partitioning where relevant), the hydrochemical context governing REY speciation and transport (notably pH, salinity/electrical conductivity, and major-ion chemistry, particularly sulfate), and constraints on alternative mechanisms and reactive carrier phases (especially phosphate-related controls and Fe oxyhydroxides) to better resolve lithogenic inputs versus process-driven fractionation in coal-gangue–impacted soils. δCe and δEu were used to quantify Ce and Eu anomalies73, which are widely applied as indicators of specific geochemical processes and sources of REYs11,74. In the studied soils, δCe values range from 0.88 to 1.24 (mean 1.02), showing weak anomaly development and no systematic vertical variation (Table 3). The Eu anomaly is similarly subdued, with δEu values ranging from 0.83 to 1.15 (mean 1.00), and likewise showing no discernible depth-dependent trend across the 0–10, 10–20, and 20–30 cm horizons. Within the PAAS-normalized framework used in this study, the near-neutral Ce and Eu anomalies likely reflect the combined influence of mixing between gangue-derived materials and local soils, along with redistribution during weathering and cultivation-related disturbance. Overall, the subdued and vertically uniform behavior of Ce and Eu aligns with the broadly similar REY patterns observed across the three sampled depths (Fig. 3). Sources of REY contamination in agricultural soils The provenance of REYs in the soils was investigated using a multi-proxy geochemical approach. PAAS-normalized distribution patterns, immobile-element ratios, and principal component analysis (PCA) provided complementary lines of evidence, which were further interpreted within the regional stratigraphic and paleogeographic framework to evaluate potential source contributions. The standardized partition model of REYs reflects average source patterns, which can indicate the characteristics of the sediment source area of basins75,76. The PAAS-normalized REY patterns of the soils (AV_SA, AV_SB, AV_SC) are nearly parallel across depths, consistently characterized by LREY enrichment, subdued HREY slopes, and a gentle convexity in the Sm – Gd interval (Fig. 3). The close similarity of REY patterns among the three sampled depths indicates that vertical variations within the 0–30 cm profile are limited, likely reflecting the combined influence of weathering, mixing of gangue-derived materials with local soils, and cultivation-related disturbance. The soils display PAAS-normalized REY patterns that are highly comparable to those of Coal R&F tuff, and broadly similar to limestone (Fig. 3). In terms of abundance, the mean ΣREYs in the soils (365.6 mg kg−1) are lower than those of Coal R&F (432.8 mg kg−1) and tuff (787.8 mg kg−1), but exceed those of limestone (14.9 mg kg−1). The primary abundance of REYs in soil-forming materials is a fundamental prerequisite for the extent of REY enrichment observed in weathered crust77. In South China, this principle is well illustrated by weathered crust elution–deposited REE ores, which preferentially develop on REY-rich parent materials77,78. Accordingly, given that the sampled agricultural soils are developed on limestone bedrock, in situ weathering of this inherently REY-poor lithology is unlikely to account for the observed soil REY inventory alone. In this context, the parallel curves and relatively higher REY abundances of Coal R&F and tuff suggest that these lithologies represent the more plausible contributors to the soil REY budget. Coal seams also exhibit REY distribution patterns broadly comparable to those of the soils, showing LREY enrichment and subdued HREY slopes (Fig. 3). However, a distinct difference occurs in the Eu anomaly, which is strongly negative in coal seams but close to unity in soils (Table 3; Fig. 3). Any contribution from coal seams would therefore have been diluted or modified by admixture with materials lacking negative Eu anomalies or showing positive ones, such as basalt-derived weathering products and tuffaceous material. As a result, coal seams are unlikely to have served as the primary source of REYs in the agricultural soils. Within the Emeishan Large Igneous Province (ELIP), basalts are classified into high-Ti and low-Ti types79. The low-Ti basalts are mainly distributed in the western ELIP, whereas the high-Ti basalts occur predominantly in the eastern ELIP79,80. Both types display PAAS-normalized REY patterns dominated by LREY enrichment and Sm – Gd convexity, with high-Ti basalts containing notably higher ΣREYs. In contrast, the soils lack the stronger Sm – Gd enrichment and positive Eu anomalies typical of the basalts, indicating that basalts were not direct contributors to the soil REY inventory. This interpretation is further supported by the absence of basaltic clasts in the soils and the lack of basalt outcrops within the study area. Stratigraphic and geochemical studies have shown that weathering products of high-Ti basalts contributed substantially to the REY budget of the Xuanwei Formation77, which is coeval with the marine Longtan Formation underlying the study area81. The presence of tonsteins and tuffaceous interbeds within the Longtan Formation provides evidence for volcanic inputs from the ELIP31,82–85. During the Late Permian, the study coalfield was located at the eastern margin of the ELIP, in proximity to the main distribution area of the high-Ti basalts. Volcanic materials, including high-Ti basalts and associated volcanic ash, were likely transported into peat-forming environments, leading to the deposition of tonsteins and tuffaceous layers within the coal seams. Thus, the ELIP served as the primary source of volcanic material, while the contemporaneous paleo-peatlands that developed into the coal-bearing strata acted as sedimentary sinks. The subsequent weathering of exposed coal gangue has facilitated the release of REYs into modern agricultural soils. Immobile high-field-strength elements (HFSEs, such as Zr, Hf, Nb, and Ta) are predominantly hosted in resistant heavy minerals, particularly zircon and rutile, and thus retain their primary geochemical signatures during weathering and sedimentary recycling 1. The Zr / Hf and Nb / Ta ratios are widely applied as robust tracers for constraining parent lithologies in soils and regolith77,86. In the Zr – Hf and Nb – Ta diagrams (Fig. 4), soil samples define tightly constrained linear trends (R2 > 0.96). Data points for limestone, high-Ti basalt, low-Ti basalt, and Coal R&F plot close to these trends, whereas tuff and coal lie farther away. This distribution pattern indicates geochemical linkages between the soils and the limestone, basalts and Coal R&F assemblage. As the local agricultural soils are developed on limestone, their HFSE concentrations being approximately 25–31 times higher than those of the limestone (Table S4) suggest additional inputs from geochemically related lithologies. The average Zr / Hf and Nb / Ta ratios (Table 4) in the soils (≈ 38.18 and 13.86, respectively) are most similar to those of Coal R&F (39.61 and 15.16), and the absolute contents of Zr, Hf, Nb, and Ta in Coal R&F are overall slightly higher than in the soils (Fig. 4, Tables S2 and S4). Taken together, these features suggest that Coal R&F materials represent a plausible contributor to the HFSE inventories of the soils, supporting the broader geochemical inference that weathered coal gangue may have played a role in modifying the soil’s elemental composition, including REYs. Bivariate plots of (a) Zr vs. Hf and (b) Nb vs. Ta for soil samples and potential sources. Soil data are from this study, whereas geochemical data for high-Ti basalts (HT), low-Ti basalts (LT), and coal seam roof-and-floor materials (Coal R&F) were compiled from published sources as detailed in “Sample analysis” Section. Table 4Immobile element ratios in soils and potential source lithologies.RatioLimestonesHTLTCoalsCoal R&FTuffSoilsZr/Hf44.0243.3841.5250.2339.6144.0238.18Nb/Ta15.3415.3816.95319.8815.1616.0813.86 Immobile element ratios in soils and potential source lithologies. The Pearson correlation matrix (Supplementary Table S5) shows strong positive correlations among most REYs (r > 0.90, p < 0.001), reflecting coherent geochemical behavior under common controls. Compared with the other REYs, Ce and Eu exhibit slightly lower correlation coefficients, which is consistent with their weak anomalies and reflects the additional geochemical controls governing their behavior relative to the remaining elements. PCA of REY concentrations yielded a single dominant factor (PC1), accounting for 92.3% of the total variance (Table 5; Fig. 5). Most REYs exhibited high loadings on PC1, accompanied by low uniqueness values, indicating strong common variance. By contrast, Ce and Eu showed lower loadings and higher uniqueness, consistent with their known anomalies linked to redox-sensitive fractionation of Ce3+/Ce4+ and feldspar-related partitioning of Eu during magmatic or sedimentary differentiation50,57. Table 5Principal component loadings and uniqueness values of REYs in agricultural soils.PC1UniquenessDy0.9950.011Tb0.9950.011Ho0.9920.016Er0.9860.028La0.9850.03Nd0.9840.031Pr0.9840.032Gd0.9830.035Tm0.980.039Y0.9780.043Sm0.9740.051Yb0.9720.054Lu0.9670.065Ce0.8770.231Eu0.7250.474Applied rotation method is promax. Principal component loadings and uniqueness values of REYs in agricultural soils. Applied rotation method is promax. Scree plot of principal component analysis (PCA) for REYs in agricultural soils. Stratigraphic and geochemical evidence indicates that PC1 captures a composite signal from several lithologies that share a common volcanogenic imprint related to ELIP activity. Coal roof-and-floor materials (Coal R&F), tuff, and limestone, which constitute the main lithologies associated with the Longtan coal-bearing strata and coal gangue, exhibit geochemical affinities with the soils and are linked to ELIP volcanism through basalt-derived weathering products and tuffaceous (volcanic-ash) inputs. When combined with regional stratigraphic and paleogeographic constraints, these features suggest that coal gangue has likely acted as an important pathway for transferring volcanogenic REYs from the ELIP into the surrounding agricultural soils. In addition, the location-based summary (Table 6) provides complementary spatial context, as higher enrichment levels are generally observed at sites closer to the gangue pile (e.g., L01 and L21) compared with the more distal site L03 (Fig. 1b), consistent with the proposed downslope influence of weathered gangue materials. The correlation matrix and PCA loadings highlight the usefulness of multivariate statistics for detecting common source signals among REYs, but they also emphasize the intrinsic limitations of such approaches. The strong coherence of REY concentrations may reflect integrated contributions from several lithologies with similar volcanogenic signatures, rather than the dominance of a single end-member source. Consequently, robust provenance interpretation requires that statistical results be evaluated in conjunction with independent stratigraphic, mineralogical, and geochemical evidence. Soil pollution by REYs of the study area Pollution load index evaluation The Pollution Load Index (PLI) was employed to evaluate the comprehensive contamination status of REYs in the soils. Across all samples, PLI values ranged from 1.72 to 3.30, with a mean of 2.19. All values exceeded the threshold of 1, confirming site-wide multi-element contamination throughout the study area (Fig. 6). The pollution load index (PLI) of the REYs at different soil depths in the study area. Depth-resolved analysis revealed that the highest degree of contamination occurred in the 10–20 cm horizon, which had a mean PLI of 2.26 (range: 1.72–2.63). This value was slightly higher than those in the overlying 0–10 cm (mean: 2.19; range: 1.74–2.37) and underlying 20–30 cm horizons (mean: 2.13; range: 1.73–3.30). Overall, the PLI values in the 0–20 cm horizons tended to be higher than those in the 20–30 cm horizon, although some individual samples showed the opposite pattern. This general tendency suggests relatively elevated contamination levels in the upper part of the profile. A location-based comparison further indicates that PLI differs among sites across the farmland block, with lower PLI values at L03 (1.7–1.8) and the highest value observed at L21C (PLI = 3.3) (Table 6), consistent with the downslope influence of the gangue pile. Geo-accumulation index evaluation The Igeo values for REYs across the three soil horizons were predominantly within the range 0 < Igeo ≤ 1 (96% of cases), which corresponds to “uncontaminated to moderately contaminated” (Fig. 7). Among all elements, La showed the lowest Igeo values at each depth, ranging from – 0.23 to 0.75 (mean 0.14). By contrast, Tb, Sm, Dy, and Gd displayed the highest mean Igeo values, with Tb (0.43–1.47, mean 0.84) consistently ranking first across horizons. Boxplot of Geoaccumulation index (Igeo) of each REY in soil. (a) 0–10 cm (topsoil), (b) 10–20 cm (midsoil), and (c) 20–30 cm (subsoil). In the topsoil (0–10 cm), the Igeo values of all elements except La remained below 1, indicating contamination within the uncontaminated to moderately contaminated range (Fig. 7a). In the midsoil (10–20 cm), most elements also remained within this range, although Igeo values of Tb occasionally exceeded 1 (0.52–1.13, mean 0.88), placing a subset of samples in the moderately contaminated class (Fig. 7b). In the subsoil (20–30 cm), exceedances above 1 occurred for most elements except La, Ce, and Eu; however, these cases were confined to a single site (Fig. 7c). Overall, the limited vertical variability in the Igeo values of REYs across soil horizons indicates that exceedances are due to site-specific conditions, not a systematic depth trend. In summary, the combined application of PLI and Igeo consistently indicates that agricultural soils in the study area are affected by REY contamination. The PLI results classify the entire profile as polluted (PLI > 1), whereas Igeo values place most elements in the uncontaminated to moderately contaminated category, with only localized exceedances reaching moderate levels. The general agreement between the two indices, coupled with their limited vertical variability, supports the robustness of the assessment and suggests that REY enrichment is not a transient anomaly but reflects a persistent geochemical signal in the soils. Ecological risk assessment The ecological risk of REYs in the soils was evaluated using single-element ecological risk factors () and the potential ecological risk index (RI), which integrate both the concentration and the toxic-response factor of each element. Across all samples, values range from 1.28 to 63.43. For La, Ce, Pr, Nd, Sm, Gd, Dy, Y, Er, and Yb, values at all three soil depths remain below 20, indicating uniformly mild ecological risk (Fig. 8). As the two most abundant REYs in the studied soils, Ce and La also fall within this mild ecological risk class, consistent with findings from a typical ion-adsorption REE mining area in South China, where values for Ce and La (0.37–3.79 and 0.18–6.15, respectively) were likewise well below 20 and thus indicative of mild ecological risk for these LREEs87. In contrast, Tb consistently falls within the moderate-risk class (20 ≤ < 40), and Eu is likewise dominated by moderate ecological risk values, with only one mild risk case occurring in the subsoil. Ho and Tm are predominantly moderate, whereas Lu exhibits the highest values (63.43), with more than half of the samples in each horizon classified as moderate and a substantial fraction reaching the strength ecological risk class (40 ≤ < 80). The ranking of mean values was consistent across the three horizons, with Lu, Tb, Eu, and Ho forming the leading group. For all samples combined, mean values were highest for Lu (40.0), followed by Tb (27.2), Eu (23.6), and Ho (22.4) (Fig. 8). These patterns indicate that single-element ecological risk is dominated by Eu and the HREYs (Tb, Ho, Tm, and especially Lu), whereas the remaining REYs contribute only mild ecological risk. Potential ecological risk factors for single REY () at different soil depths in the study area. (a) 0–10 cm (topsoil), (b) 10–20 cm (midsoil), and (c) 20–30 cm (subsoil). RI values range from 168 to 326 (Fig. 9). At the location scale (Fig. 1b), RI also shows site-to-site differences (Table 6), with consistently lower values at L03 (168.0–179.1 across depths) and higher values at locations closer to the gangue pile (e.g., L01 and L21), reaching a maximum at L21C (RI = 325.9). Based on the thresholds of Chen et al.68, the soils are mainly characterized by moderate (110 ≤ RI < 220) and strength (220 ≤ RI < 440) potential ecological risk, with no sites falling into the mild or strong categories. In the topsoil, approximately 57% of RI values lie in the moderate-risk class and 43% in the strength risk class, whereas in the midsoil the proportion of strength risk sites increases to about 57%. In the subsoil, RI is dominated by moderate risk conditions (about 86% of sites), with strength risk values occurring only locally. The broadly similar risk levels in the 0–10 cm and 10–20 cm horizons likely reflect soil mixing within the cultivated layer due to agricultural practices, resulting in a relatively uniform ecological risk pattern in the upper 0–20 cm of the profile, while the 20–30 cm horizon shows slightly lower overall risk (Fig. 9). RI indices at different soil depths in the study area. Elemental contributions to RI (Fig. 10) showed that, across all soil depths, the five dominant contributors were consistently Lu, Tb, Eu, Ho, and Tm, together accounting for 62.3–62.7% of the total RI. Within this group, the HREY subset (Lu, Tb, Ho, Tm) contributed 51.1–51.7%, indicating that RI is shaped not only by total REY concentrations but also by the toxic-response coefficients () assigned to individual elements. In line with the framework of67, elements assigned higher values exert disproportionate influence on RI even at moderate concentrations. A similar dominance of HREYs in ecological risk calculations has been reported for REE-enriched soils in ion-adsorption rare earth mining areas of South China, where Eu, Ho, Lu, Dy, and Gd were identified as primary contributors to RI11. This tendency reflects the inherently stronger complexation affinity and higher biological toxicity potential of HREYs, in addition to the weighting structure embedded in RI calculations, which together amplify the ecological risk contributions of Lu, Tb, Ho, and Tm11,68,88. Proportional contributions of individual REYs to the comprehensive ecological risk index (RI) at different soil depths. (a) 0–10 cm (topsoil), (b) 10–20 cm (midsoil), and (c) 20–30 cm (subsoil). Overall, these results emphasize the dominant role of HREYs in shaping the ecological risk profile, while also indicating that individual LREYs such as Eu may exert comparable influence under specific geochemical conditions. The vertical patterns of ΣREY concentrations and contamination indices show that the 0–20 cm horizon, which corresponds to the cultivated soil layer, exhibits consistently higher and more uniform REY enrichment. This pattern is likely related to soil mixing associated with agricultural activities, and it indicates that this layer is particularly important for evaluating potential environmental exposure. Accordingly, environmental monitoring should focus on the 0–20 cm soil horizon as well as on agricultural land located downslope of coal gangue piles, where REYs are most likely to accumulate or migrate. Risk management may also involve regular soil quality assessments and targeted mitigation measures to limit REY release from weathered gangue materials, particularly after periods of intense rainfall that promote leaching and downslope transport. Taken together, these findings provide a consistent characterization of the magnitude and depth distribution of REY-related ecological risk in soils surrounding a representative coal gangue pile in Chongqing. The present assessment is intentionally focused on REYs in soils and does not include paired crop measurements or a formal human health risk assessment. Given that many abandoned coal mines in Chongqing exploit Late Permian coal-bearing strata that are naturally enriched in REYs, coal gangue piles derived from these strata may act as broader sources of REY inputs to surrounding agricultural soils and, potentially, to local food chains. From a regional perspective, future investigations should expand toward integrated, city-scale assessments of REY contamination, ecological risk, and potential human exposure in agricultural soils adjacent to coal gangue piles derived from Late Permian Longtan Formation coal seams. Such assessments would provide a scientific basis for distinguishing areas of elevated concern from those of lower risk, thereby supporting risk-based zoning and the development of tiered management approaches. Within this framework, areas identified as higher-risk may warrant strengthened soil control or remediation efforts, whereas lower-risk areas could be addressed through routine monitoring, early-warning measures, and appropriate land-use management. In parallel, further research into environmentally sound beneficial use and management of coal gangue is needed to reduce long-term sources of REY release to surrounding soils. Conclusions ΣREY concentrations in agricultural soils ranged from 289.8 to 530.5 mg kg−1, consistently exceeding both the global and Chinese soil background values. Geochemical evidence suggest that coal gangue has likely acted as a major pathway transferring REYs ultimately derived from ELIP-related volcanic materials (and their weathering products) into adjacent agricultural soils. Pollution assessments indicate overall REY contamination, with PLI classifying the soils as polluted, whereas element-specific Igeo values placed most cases in the uncontaminated to moderately contaminated range. Ecological risk results show that RI values fall exclusively within the moderate and strength classes, with elevated contributions from HREYs (Lu, Tb, Ho, Tm) and Eu, which collectively account for ~ 63% of the total index. The upper 0–20 cm horizon exhibits slightly higher ecological risk than the 20–30 cm horizon, likely reflecting soil mixing within the cultivated layer together with additional inputs from weathered gangue materials. These characteristics underscore the importance of prioritizing the 0–20 cm horizon in environmental exposure assessment and risk management for agricultural soils affected by coal-gangue weathering. Collectively, these findings highlight the need for integrated, city-scale evaluations of REY contamination, ecological risk, and potential human exposure across agricultural soils in Chongqing that may be influenced by coal gangue piles derived from Late Permian Longtan Formation coal seams. Such region-scale assessments would provide a scientific basis for risk-based zoning and differentiated management strategies, while further research on environmentally sound utilization and management of coal gangue may help reduce long-term sources of REY release. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Acknowledgements We sincerely thank the anonymous reviewers for their insightful and constructive comments, which greatly improved the clarity and scientific rigor of this manuscript. We also extend our gratitude to Prof. Hu Jing of the Institute of Geochemistry, Chinese Academy of Sciences, for her valuable support with laboratory analyses and instrumental guidance throughout this study. Author contributions The contribution of the authors is men-tioned as follows: (1) Maochao Zhang: Conceptualization, Methodology, Software, Investigation, Formal Analysis, Writing - Original Draft, Writing - Review & Editing; (2) Lihua Peng: Investigation, Data Curation, Writing - Original Draft; (3) Hong Li: Software, Visualization, Investigation; (4) Liusan Cheng: Validation, Resources; (5) Mingqiu Wang: Software, Visualization. Funding This study was funded by the Characteristic Innovation Projects in Ordinary Universities in Guangdong Province (Natural Science) (Grant No. 2023KTSCX387), the Project supported by the Scientific Research Starting Foundation for Qingyuan Polytechnic and the Science and the Technology Research Program of Chongqing Municipal Education Commission (Grant No. KJQN202203406; KJQN202503406). Data availability The data used to support the findings of this study are available from the corresponding author upon request. Declarations Competing interests The authors declare no competing interests. Ethical approval All authors have read, understood, and have complied as applicable with the statement on “Ethical responsibilities of Authors”. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Data Availability Statement The data used to support the findings of this study are available from the corresponding author upon request.