Arsenic-driven agricultural soil risk exceeds industrial contamination in Punjab, India: a replicated EDXRF and 16S rRNA assessment across three land-use types and three agro-ecological districts.

Arsenic loading from irrigated agriculture poses a greater composite ecological risk than industrial heavy-metal contamination in Punjab, India, a finding that directly challenges the primacy of metal-load-based soil hazard indices across South Asian agro-industrial regions. Using a formally replicated 3 × 3 factorial design (three land-use types × three agro-ecological districts; N = 36 energy-dispersive X-ray fluorescence (EDXRF) samples; n = 4 field replicates per sampling unit) combined with 16S ribosomal RNA (rRNA) V3-V4 amplicon sequencing at three representative sites, we demonstrate that land-use type is the primary geochemical driver across 15 of 32 quantified elements (partial eta-squared, η2p, up to 0.862), with significant land-use × district interaction effects detectable only through factorial design. The agricultural Amritsar sampling unit recorded the highest composite ecological risk (Potential Ecological Risk Index, PERI = 723.54), exceeding the primary industrial hotspot (PERI = 624.17; Pollution Load Index, PLI = 20.11), driven by extreme arsenic enrichment from chronic flood irrigation with arsenic-bearing groundwater (As = 80.8 ± 10.7 mg/kg; Enrichment Factor, EF = 45.19; geo-accumulation index, Igeo, Class 6), a risk pathway invisible to PLI-based ranking. All nine sampling units simultaneously exceeded Central Pollution Control Board (CPCB) guideline values for zinc, lead, arsenic, chromium, nickel, and copper (PLI range 8.46-20.11), confirming region-wide multi-element soil pollution. Descriptive 16S rRNA community profiles revealed a diversity gradient consistent with Pollution-Induced Community Tolerance (PICT) theory (Shannon entropy H': 3.906 → 3.007), with Pseudomonadota enrichment (60.1% → 83.0%), Actinomycetota depletion (20.0% → 3.5%), and eight genera forming a candidate biomonitoring panel. Procrustes alignment (m12 = 0.170) provides exploratory evidence of trace element-microbiome co-structure. Three co-existing contamination pathways are identified, and an integrated EDXRF-16S rRNA framework for land-use-stratified soil health monitoring is proposed for South Asian agro-industrial regions.