Unveiling spatial nitrogen transformations in agricultural lacustrine soils and their contribution to groundwater ammonium contamination: An integration of isotopic tracers and multivariate statistics

Nitrogen contamination in groundwater threatens water security in agricultural lake basins. However, the mechanisms controlling the transformation and transport of nitrogen from the lacustrine soil surface to the aquifer remain unclear. This study integrates isotopic fingerprinting (δ¹⁵N, δ¹⁸O, δ²H) with multivariate statistical analyses (SOM–K-means and PCA) to investigate nitrogen sources and transformation along the soil–groundwater continuum in the Poyang Lake Basin, using 53 soil samples (0–6 m, six profiles). Three distinct biogeochemical zones were identified: (1) surface soils (0-30 cm) with high agricultural nitrogen (NO₃⁻–N of 25.57 mg/kg, NH₄⁺–N of 38.52 mg/kg), (2) an intermediate zone (30-200 cm) of nitrogen depletion (NH 4 + –N < 4 mg/kg, NO 3 - –N < 2 mg / kg) from denitrification and uptake, and (3) deep soils (>300 cm) exhibiting NH 4 + –N enrichment (22.18 mg/kg). This deep ammonium pool has a dual origin: primarily from in-situ mineralization of legacy organic nitrogen, but also from direct infiltration of agricultural sources in specific areas. Elevated NH₄⁺–N concentrations in groundwater directly connected to high-ammonium soils (ZGW) (5.05 ± 3.45 mg/L) and in domestic wells (MGW) (8.97 ± 1.81 mg/L) further confirm this linkage. Multivariate analysis revealed dominant controls: agricultural input (PC1), leaching (PC2), and reducing-condition retention (PC3). Isotopic signatures confirmed subsurface transfer, with mineralized deep-soil nitrogen (δ¹⁵N-NH₄⁺: +3.35 to +7.56‰) contributing to ZGW (+4.27 to +5.39‰) and MGW (+2.36 to +6.90‰). Key mechanisms include clay-facilitated retention, nitrification, and depth-stratified mineralization. These insights establish a framework for identifying high-risk zones and designing strategies to mitigate both legacy and current nitrogen sources for groundwater protection. • Multi-isotope and machine learning reveal depth-dependent N transformation mechanisms • Agricultural inputs govern surface soils; mineralization dominates in deep layers • Nitrification-denitrification coupling controls N loss in intermediate vadose zone • Soil texture and redox conditions jointly regulate N retention and conversion • Deep soil NH₄⁺-N from mineralization and shallow sources contributes to groundwater contamination