Geochemical fingerprinting of a radon anomaly: high-resolution PCA–ANOVA case study, Castleisland, SW Ireland

Radon represents a global health risk, and so accurate delineation of radon-prone areas is a prerequisite for evidence-based radon mitigation and public health protection. National probabilistic radon models of Ireland, based on 1:1 M bedrock geological maps that group the Clare Shales with limestones achieve ∼74% accuracy. Despite the high accuracy of Ireland's national radon map, some discrepancies still exist. Using geochemical and geostatistical methods, we investigated a significant and persistent radon anomaly over the Clare Shale Formation at Castleisland, Co. Kerry, Ireland. Fifty-six topsoil samples collected from the A-horizon within a 6 km 2 grid, with samples spaced every 250–500 m apart, were included in the analysis and compared to co-located soil-gas radon measurements. Following centred and isometric log-ratio transformations of inductively coupled plasma mass spectroscopy (ICP-MS)/optical emission spectroscopy (OES) data for 37 elements, soils above the Clare Shales exhibited a median U concentration of 4 mg kg −1 (range 1.4–37 mg kg −1 ). Pearson correlations between log 10 soil-gas radon and individual elements peaked at a Pearson correlation coefficient ( r ) = 0.57 for Sr (coefficient of determination ( R 2 ) = 0.32), with similarly strong associations for V ( r = 0.54), Ag ( r = 0.52), P ( r = 0.47), Au ( r = 0.46), U ( r = 0.45) and Tl ( r = 0.43) (all P < 0.001). One-way ANOVA indicates radon class categories explain a median 43.8% of variance in 14 trace elements, while bedrock geology explains 27.8%. Shared tracers (U, V, P, Ag, Sb) underscore overlapping lithological and radiogenic controls. Principal components 1–3 (PC1–PC3) capture 62.9% of total variance (PC1 = 31.8%, PC2 = 17.2%, PC3 = 13.9%). PC1 is defined by strong positive loadings on Sc, Mg, Fe, Al, Th, Ni and Co, and negative loadings on Sr, Ag, U, P and V, neatly contrasting shale-derived, high-radon soils from carbonate terrains. This study demonstrates that local-scale lithogeochemical proxies, resolved at 250 m and calibrated against a 1:100 k geological framework, can effectively delineate elevated radon sources. This approach offers a systematic means to investigate map anomalies, refine national radon models, enhance spatial accuracy, and support evidence-based radon risk assessment and public health protection initiatives.