Sewage Overtakes Fertilizer as the Hidden Engine of Nitrate Pollution in a Karst River Basin

In the fractured limestone landscapes of southern China, water moves in ways that defy the tidy assumptions of conventional watershed models. Rainfall does not linger in the soil; it plunges through sinkholes, conduits, and fissures into aquifers that are directly connected to the rivers above. A new open-access study in Environmental Earth Sciences has now traced exactly where the nitrate polluting one such system comes from, and the answer overturns the expectation that fertilizer should dominate. Using dual nitrate isotopes and a Bayesian mixing model, researchers working in the Lipu River Basin of Guilin found that manure and sewage contributed roughly half of the modeled nitrate load, a finding that reshapes how pollution control should be prioritized in karst regions where surface water and groundwater are inseparable.

The Lipu River Basin lies in Lipu City in southern Guilin, in the Guangxi Zhuang Autonomous Region, and drains toward the Lijiang and Guijiang rivers. It is a classic mixed-land-use karst catchment: terraced farmland famous for Lipu taro, dense rural settlements relying in part on dry-toilet systems, and a growing industrial park hosting metalworking, chemical, pharmaceutical, and food-processing facilities. The subtropical climate delivers about 1,441.5 millimeters of rain annually, concentrated between April and August, while winters are comparatively dry and slow-moving. Because the underlying aquifer sits only 0.5 to 3.5 meters below the surface and is riddled with carbonate conduits, whatever enters the water system in one place can emerge, barely degraded, somewhere else entirely.

To untangle the pollution, the team led by Zupeng Wan and Honghu Zeng of Guilin University of Technology ran two sampling campaigns in 2022, one in the wet summer and one in the dry winter. They collected 51 river-water samples from 27 summer and 24 winter sites spanning the mainstream and six tributaries, and 26 groundwater samples from Quaternary porous aquifers, Carboniferous fissure-cave aquifers, and bedrock-fissure aquifers. Nitrate isotopes were measured with the bacterial denitrifier method, converting sample nitrate to nitrous oxide and analyzing it by isotope-ratio mass spectrometry, calibrated against the international reference materials USGS32, USGS34, and USGS35 with analytical precision better than ±0.3 per mil for δ15N and ±0.5 per mil for δ18O.

The concentration data alone told a striking seasonal story. Total nitrogen in river water averaged 7.09 ± 6.59 milligrams per liter in winter, more than double the summer average of 2.95 ± 1.48 milligrams per liter, a difference the authors attribute to reduced dilution when rainfall and runoff diminish. Groundwater, buffered by the aquifer, showed no significant seasonal shift. Nitrate was overwhelmingly the dominant dissolved inorganic nitrogen species, exceeding 77 percent of total nitrogen in river water in both seasons. Twelve percent of all samples exceeded the World Health Organization and Chinese drinking-water guideline of 10 milligrams per liter of nitrate-nitrogen, with exceedances more frequent in groundwater, at 19 percent, than in river water, at 8 percent, and concentrated near the industrial park and in winter river water at tributaries draining grain- and food-processing areas.

The isotopic evidence then did the detective work that chemistry alone could not. Different nitrate sources carry characteristic isotopic fingerprints: manure and sewage typically have high δ15N values, atmospheric deposition has high δ18O values, while mineral fertilizer and soil nitrogen have lower δ15N signatures. Because the ranges of manure and sewage overlap, the researchers combined them into a single end-member and used the Stable Isotope Analysis in R package, SIAR, a Bayesian mixing model fitted by Markov-chain Monte Carlo simulation, to estimate proportional contributions with full uncertainty. A locally collected effluent sample with a δ15N of +8.6 per mil fell squarely within the manure-and-sewage literature range, anchoring the prior without materially shifting the posterior.

The posterior estimates were unambiguous. Across the basin, manure and sewage accounted for 56.7 ± 8.2 percent of modeled nitrate, followed by soil nitrogen at 22.5 ± 6.5 percent, nitrogen fertilizer at 15.3 ± 5.0 percent, and atmospheric deposition at just 5.5 ± 2.5 percent. The contribution was even higher in groundwater, at 61.5 ± 8.5 percent, than in river water, at 50.2 ± 8.0 percent. The authors point out that this dominance exceeds that reported for other Southwest Chinese karst systems, including the Babu karst aquifer in Guizhou, where manure and sewage contributed 37 to 38 percent, and the Erhai Basin, where the figure was 49.6 percent. They attribute the Lipu excess to dense rural settlement, widespread dry-toilet use, manure application on taro fields, and industrial discharge of wastewater rich in dissolved organic nitrogen from the county’s hanger-manufacturing and specialty food-processing sectors.

Equally important is what the isotopes revealed about nitrogen transformations. All sample δ18O values fell within the range produced by microbial nitrification, indicating that the nitrate pool in both river and groundwater is built by bacteria oxidizing reduced nitrogen rather than by direct input of pre-formed nitrate. To test for denitrification, the process that removes nitrate and would partly mask source signals, the team applied a Rayleigh fractionation framework, in which remaining nitrate becomes progressively enriched in both heavy isotopes as its concentration falls, with an expected enrichment-ratio εN/εO of roughly 1.3 to 2.1. Most river samples showed no such coupled enrichment, indicating no resolvable denitrification. Groundwater showed a ratio of about 1.38 with δ15N and δ18O inversely related to nitrate concentration, evidence of some denitrification, but the generally low isotope values and only suboxic conditions, with dissolved oxygen below 2 milligrams per liter, indicated the removal was limited.

Independent tracers reinforced this picture. Chloride behaves conservatively and electrical conductivity integrates total mineralization, so if nitrate rises together with both, the pattern points to mixing rather than in-situ removal. In river water, nitrate correlated positively with both parameters at all discharge types, most strongly at sanitary and industrial outlets, where regression coefficients of determination reached 0.593 and 0.464 respectively for conductivity. In groundwater, however, the nitrate-chloride relationship broke down entirely, a decoupling the authors attribute to vertical karst heterogeneity: in matrix-dominated, oxygen-poor zones, nitrate is removed locally while chloride passes through untouched. A nitrogen enrichment slope analysis added further nuance. At sanitary and industrial outlets, the log of nitrate concentration rose with δ15N, the signature of accumulating waste-derived nitrogen, whereas at other outlets the slope was negative, consistent with excess synthetic fertilizer leaching at low δ15N.

The spatial mapping identified clear management targets. Total nitrogen in groundwater averaged 6.11 milligrams per liter, and nearly 19 percent of groundwater sites exceeded the nitrate guideline. Wells near the metalworking industrial park, sites U10 and U11, recorded total nitrogen two to three times the regional mean, and a nearby karst spring reached 18.3 milligrams per liter of nitrate-nitrogen, likely reflecting inputs from a new-materials facility. Upstream agricultural wells near the taro fields ran 1.65 to 3.57 milligrams per liter above the basin average, driven by heavy organic fertilizer application whose mineralization and nitrification outpace crop uptake. Tributaries such as the Pulu and Dumo rivers, some carrying more than 30 percent of the mainstream flow, emerged as the core nitrogen arteries of the basin, with weakly oxygenated karst reaches suppressing nitrification and allowing ammonium to accumulate.

The study’s practical message is direct: pollution-control strategies designed for porous-media basins translate poorly to karst terrain, where conduits can fast-track contaminants past the soil’s natural attenuation. Because nitrate isotopes were measured only in summer, the source proportions describe the summer basin-scale budget rather than a full seasonal picture, a caveat the authors state explicitly. Still, their recommendations are clear. Regulators should prioritize manure and sewage management across karst-sensitive zones, remediate the localized industrial groundwater hotspots near the metalworking and new-materials parks, and intensify winter monitoring of food-processing tributaries where low flows concentrate point-source signals. In a basin where a dry toilet on a hillside and an electroplating workshop in the valley both drain into the same maze of limestone pipes, protecting the water means treating the aquifer and the river as one system, not two.

Subject of Research: Nitrate source apportionment and nitrogen transformation in the karst river-groundwater system of the Lipu River Basin using dual nitrate isotopes and Bayesian SIAR mixing modeling

Article Title: Identifying nitrate sources and nitrogen transformation in a karst river–groundwater system using dual nitrate isotopes and a Bayesian (SIAR) mixing model

Article References: Wan, Z., Liu, B., Chen, W., Chen, Y., Yan, X., & Zeng, H. (2026). Identifying nitrate sources and nitrogen transformation in a karst river–groundwater system using dual nitrate isotopes and a Bayesian (SIAR) mixing model. Environmental Earth Sciences, 85(16), Article 406. https://doi.org/10.1007/s12665-026-13136-4

Image Credits: AI Generated

DOI: 10.1007/s12665-026-13136-4

Keywords: karst hydrogeology, nitrate pollution, dual nitrate isotopes, SIAR mixing model, groundwater contamination, nitrogen cycle, manure and sewage, denitrification, nitrification, Lipu River Basin, water quality, source apportionment