The study area is located in Renhuai City, Guizhou Province, China (Chishui River Basin). Eight composite surface soil samples (0–20 cm, tilled layer) were collected in August 2024 from representative Hongyingzi sorghum fields across eight towns (S1–S8; Supplementary Table S1; Supplementary Figure S1). At each site, five subsamples were taken within a 10 m × 5 m plot using a stainless-steel auger and composited to approximately 2 kg (Supplementary Method S1). Samples were placed in pre-cleaned polypropylene bags, transported to the laboratory in a cooler with ice packs, and stored at 4 °C prior to pre-treatment. In the laboratory, visible roots and residues were removed; soils were air-dried in the dark, homogenized, sieved (100-mesh), and stored at −20 °C until extraction (within 14 days). Soil physicochemical properties (e.g., pH and organic matter) were not determined in this campaign; their potential influence on PFAS sorption and distribution is discussed as an important limitation.

The concentration of each target PFAS was quantified using external calibration with mixed standards and isotope-labeled internal standards. All calibration curves covered 0.2–200 ng/mL with coefficients of determination (r²) ≥ 0.9978 (Supplementary Table S5). Method detection limits (MDLs) and method quantification limits (MQLs) ranged from 0.005 to 0.014 ng/g and 0.016–0.048 ng/g (dry weight), respectively (Supplementary Table S5). Procedural blanks and solvent blanks were analyzed in each batch, and no target PFAS was detected above the MDL. Matrix-spike recoveries (n = 3) ranged from 80.6% to 104% and were within commonly accepted QA/QC criteria for trace organic analysis (Supplementary Table S5). Because soil matrices can cause ion suppression/enhancement in LC–MS/MS, the use of isotope-labeled surrogates was employed to correct for extraction losses and residual matrix effects; however, we acknowledge that matrix effects remain a potential source of uncertainty.

Data processing and analysis were performed using SPSS version 26.0 (IBM, USA). Differences among sampling sites were examined descriptively owing to the limited sample size (n = 8). Spatial interpolation was conducted using ordinary kriging in ArcGIS 10.2 to visualize regional patterns of ΣPFASs. Given the small number of sampling points, kriging was used as an exploratory tool rather than a definitive predictive model. Leave-one-out cross-validation was applied to evaluate the interpolation performance; the standardized mean error and standardized root-mean-square error were −0.0329 and 0.956, respectively, indicating no substantial systematic bias and acceptable uncertainty for visualization at this scale. We emphasize that higher-resolution mapping will require denser sampling and incorporation of covariates (e.g., land use and soil properties). All figures were prepared using Origin 2019 (OriginLab, USA).

Twenty-four PFASs (11 legacy PFCAs/PFSAs and 13 emerging alternatives) were targeted in the soils from the eight sampling sites. Seven PFASs were detected at least once, and Σ₇PFAS concentrations ranged from 87.8 to 446 ng/kg (dry weight) (Supplementary Table S6; Figure 1). Overall, the PFAS burden in the Hongyingzi sorghum base is lower than that reported for highly industrialized agricultural soils (e.g., ΣPFAS up to 11.2 μg/kg in Shanghai) but is comparable to several background agricultural soils reported internationally (e.g., 10–350 ng/kg in France; 1,200 ng/kg in Canada; 5–400 ng/kg in Sweden). A quantitative comparison with selected studies is summarized in Table 1.

Seventeen target PFASs were not detected in any sample and were therefore reported as non-detects. Their MDLs were 0.005–0.014 ng/g (dry weight) (Supplementary Table S5). The non-detect list includes PFDA, PFUnDA, PFDoDA, PFTrDA, PFTeDA, PFHxDA, PFOS, PFDS, PFNS, PFHxS, PFHpS, PFODA, PFPeS, 4-2FTS, 6-2FTS, 8-2FTS, and PF-3,7-DMOA.

Among the legacy PFASs, PFOA and PFNA were detected at all sites (100% frequency), with Σ₂PFAS concentrations ranging from 87.8 to 204 ng/kg (mean: 145 ng/kg) (Supplementary Table S6). Long-chain PFCAs (PFDA–PFHxDA) and PFSAs (PFOS, PFNS, PFDS) were not detected, which is consistent with decreased use of these legacy compounds following international restrictions (e.g., Stockholm Convention listing of PFOS and PFOA-related substances) and domestic pollution control measures in China.

Five emerging PFASs (PFBA, PFPeA, PFHxA, PFHpA, and PFBS) were detected, and Σ₅PFAS concentrations ranged from 56.9 to 242 ng/kg (mean: 119 ng/kg) (Supplementary Table S6). Emerging PFASs accounted for 41.8% of Σ₇PFASs (Figure 2), indicating that short-chain alternatives can dominate the PFAS profile even in agricultural soils remote from large industrial centers. Among the emerging PFASs, PFBA and PFBS were the most abundant and are further discussed as potential substitutes for PFOA-related uses (Section 3.4).

Σ₇PFAS concentrations varied among sites, with the highest value observed in Shuangjiang Village of Meijiuhe Township (S1; 446 ng/kg) and the lowest in Wuma Township (S2; 87.8 ng/kg) (Supplementary Table S6). The point measurements (Figure 3) indicate a local hotspot in the central–eastern part of the study area, whereas the western sites generally showed lower concentrations. This heterogeneity suggests that PFAS inputs are spatially uneven and may be influenced by proximity to settlements, road networks, and winery-related facilities, in addition to potential atmospheric deposition and hydrologic transport along the Chishui River.

To visualize regional patterns, Σ₇PFAS concentrations were interpolated using ordinary kriging (Figure 4). Because only eight sampling points were available, the interpolated surface should be interpreted as an exploratory map rather than a statistically robust prediction. Leave-one-out cross-validation showed acceptable performance for visualization (standardized mean error = −0.0329; standardized RMSE = 0.956), but uncertainty remains high in areas far from sampling points. Future work should increase sampling density and incorporate environmental covariates (e.g., land use intensity, soil organic carbon, pH, topography, prevailing wind direction, and distance to potential sources) to improve spatial inference and causal interpretation.

Principal Component Analysis (PCA) was applied to explore potential source patterns in the PFAS composition. The seven detected PFASs were standardized prior to PCA, and two principal components (PC1 and PC2) were retained with eigenvalues >1, explaining 68.7% and 19.2% of the variance, respectively (cumulative: 87.9%; Supplementary Table S7). The loading plot and site scores are shown in Figure 5, and the corresponding loading matrix is provided in Supplementary Table S9 to facilitate quantitative interpretation.

PC1 showed high contributions from PFOA and PFNA together with the detected short-chain carboxylates (PFBA, PFPeA, PFHxA, and PFHpA), suggesting a mixed profile that is consistent with diffuse inputs (e.g., atmospheric deposition, runoff from residential areas, and the use of consumer products and materials containing PFAS-related precursors) (Gerardu et al., 2022; Zenobio et al., 2025). PC2 was dominated by PFBS, indicating a comparatively distinct input profile that may reflect localized activities or specific commercial formulations where PFBS-related chemistries are used as replacements for longer-chain sulfonates (O’Rourke et al., 2024).

It should be emphasized that source apportionment based solely on PCA of soil concentration profiles is inherently indirect. Although PFBS has been reported in a range of industrial and commercial applications (e.g., surface treatment agents, coatings, and some fire-fighting formulations) (Simon, 2022), we did not analyze potential source materials (e.g., coatings/paints, packaging, irrigation water, or atmospheric particulates) in this study. Therefore, any linkage between PFBS and specific local activities in Renhuai (including winery-related development) should be regarded as a working hypothesis that requires targeted source sampling and historical usage information for validation.

Regulatory restrictions and market phase-outs of PFOA- and PFOS-related chemistries have accelerated the transition toward alternative PFASs (Li et al., 2024), particularly short-chain PFCAs (e.g., PFBA) and short-chain PFSAs (e.g., PFBS) that can provide similar functional performance with lower bioaccumulation potential (Dewapriya et al., 2023; Elgarahy et al., 2024). PFBA-related chemistries have been used in some water- and grease-repellent treatments, paper/packaging applications, and fluoropolymer processing, whereas PFBS-related chemistries have been used in surface treatment agents and certain industrial formulations. In the Hongyingzi sorghum cultivation base, the dominance of PFBA and PFBS among emerging PFASs (Section 3.1) suggests that local inputs may increasingly reflect this substitution trend (Figure 6). Given the rapid development of winery-related infrastructure and the associated use of treated materials (e.g., packaging, coatings, and construction materials) in Renhuai, these pathways warrant further investigation through targeted source sampling and temporal monitoring.