POXC and SOC in tropical paddy soils — Northern Thailand dataset

This dataset accompanies the manuscript "An elevated POXC/SOC ratio in tropical paddy soils of Northern Thailand: pedological and analytical limits on its interpretation" by Suphathida Aumtong (2026). The study examines permanganate-oxidizable carbon (POXC) and total soil organic carbon (SOC) across four rice management systems — conventional rice (RICE), rice–legume rotation (RICE_BEAN), organic rice (RICE_OR) and rice–vegetable rotation (RICE_VEG) — distributed across six provinces of Northern Thailand (Chiang Mai, Chiang Rai, Lampang, Lamphun, Mae Hong Son, Phayao). Sampling was carried out between January and March 2023, after the main wet-season rice harvest at every site, using an identical protocol at every plot in all systems and provinces. Topsoil samples were analysed for eleven chemical parameters: pH and electrical conductivity (1:1 and 1:5 water suspensions), available P (Bray II) and K (NH4OAc), exchangeable Ca and Mg (NH4OAc), soil organic matter (loss on ignition), SOC (Walkley–Black), POXC (modified Weil method: 3 g soil, 0.5 mm sieve, 0.02 M KMnO4, 30 min at 120 rpm, 550 nm), and NH4+ and NO3− (KCl extraction). A 50-sample subset carries paired bulk-density measurements for carbon-stock estimation. Sampling design 28 geographic waypoints → 61 field profiles → 421 depth-layer records (0–10 cm down to 90–100 cm). Each profile contributes one record per depth layer. Sampling depth varies by profile: some profiles were sampled to 30 cm only, the remainder to 100 cm. Reproducing the analytical dataset Keep rows where Depth_cm is 0–10, 10–20 or 20–30 → 183 topsoil records from 61 profiles. Keep rows where SOC_gkg / SOM_gkg falls between 0.30 and 1.10 inclusive → 133 records from 58 profiles (RICE 83, RICE_BEAN 16, RICE_OR 22, RICE_VEG 12). For the profile-level analysis, average the retained depth layers within each Profile_ID → 58 observations (RICE 36, RICE_BEAN 7, RICE_OR 10, RICE_VEG 5). POXC was not determined for one Alfisol profile, so tests involving POXC and the POXC/SOC ratio rest on 50 profiles and 114 depth-layer observations rather than 51 and 117. All values reported in the manuscript reproduce from this dataset when these steps are applied. Note on units of analysis The three topsoil layers sampled at one profile are not independent observations. Between-system tests are therefore reported at both the depth-layer level (n = 133) and the profile level (n = 58). Carbon pools are non-significant at both levels; the number of significant fertility parameters falls from seven to four at the profile level. Key findings The pooled POXC/SOC median was 10.00%, or 6.0–9.5% after adjustment for Walkley–Black recovery factors of 60–95%, against the 1–4% range reported for temperate agricultural soils. Within the dataset the ratio varied with soil order — 9.78% under Alfisols against 15.48% under Entisols (p = 0.0119 at both the depth-layer and the profile level) — but not with management. Restricted to Alfisol profiles, available P, Mg, Ca and EC continued to separate the cropping systems while SOC, POXC, SOM and the ratio did not, and a paired contrast between ten conventional and ten certified-organic fields at one site gave the same answer. The soil-order contrast should be read with care. Soil order is completely confounded with cropping system in this design: all seven Entisol profiles are rice–legume and all fifty-one Alfisol profiles belong to the other three systems, so the two comparisons are arithmetically the same one. The Entisol group comes from two localities. Excluding the two profiles with unresolved provenance leaves p = 0.0314 and excluding the Chiang Rai site leaves p = 0.0219, but restricting the group to Mae Hong Son alone (n = 4) gives p = 0.0620. No pedological property beyond the taxonomic class was measured. Because the temperate benchmarks derive from dry-combustion SOC, cross-study comparison of POXC/SOC is confounded unless SOC protocols are matched. Changes in version 2.0 Corrected cropping-system labels. Eleven profiles carried an incorrect label in v1: ten profiles at Ban Don Jieng coded as rice–legume rotation are conventional rice, and one Chiang Rai profile recorded as rice carries an off-season soybean crop and is rice–legume. Rice_system_code now holds the corrected value and Rice_system_original preserves the v1 label, so the change is fully traceable. Every result in the accompanying article uses the corrected labels. Profile_ID, Site_group, Province and District added, so the profile-level analyses can be reproduced directly. The manuscript file has been removed. This deposit now contains data and documentation only, and no farmer names or contact details. Description updated to match the current manuscript title and findings. Funding Agricultural Research Development Agency (Public Organization), Thailand (ARDA), Project Code PRP6605030280.