Antibiotic Mixtures in Poyang Lake Show Risk Priorities Differ From Concentrations

The research, published in Environmental Geochemistry and Health, combined a large sampling campaign, source-apportionment modeling and probabilistic risk analysis to rank antibiotics by both origin and hazard.

Between 26 and 30 August 2023, during a high-water-period campaign, the team collected 158 samples across the basin: 42 lake-water samples and 116 wastewater samples from sewage treatment plants, hospitals, livestock and aquaculture facilities, and pharmaceutical industries. Ten antibiotics were measured in each sample. Mean total antibiotic concentration in lake water was 73.9 nanograms per liter, with the profile dominated by three tetracycline-class compounds: oxytetracycline, chlortetracycline and tetracycline.

To trace where the lake’s antibiotic load came from, the researchers first ran exploratory socio-economic correlation analyses, which showed category-dependent patterns linking antibiotic occurrence with economic activity, production intensity and sector-related antibiotic use. They then applied positive matrix factorization, a receptor-side statistical technique, to the lake-water data. This resolved three compositional patterns that accounted for 46.0%, 11.8% and 42.1% of the modeled antibiotic mass: an oxytetracycline-rich mixed pattern, a mixed pattern combining sulfamethazine, chlortetracycline and doxycycline, and a chlortetracycline-dominated tetracycline-rich pattern.

The team compared these receptor-side profiles against the four wastewater source categories and used non-negative least squares fitting to estimate how each category could contribute. The result was deliberately cautious: the fits indicated mixed and spatially variable compositional consistency with the wastewater categories, rather than evidence that any single source dominated the lake’s antibiotic profile. In other words, the chemistry of the lake reflects blended inputs whose relative importance shifts across the basin.

Risk assessment told a different story from concentration measurement. In the study’s primary Monte Carlo analysis, which propagates variability and uncertainty through repeated simulation, oxytetracycline and chlortetracycline each had a 16.3% probability of exceeding a risk quotient of 1, and enrofloxacin followed closely at 14.8%. But when the analysis moved from individual compounds to the mixture as a whole, summed risk quotients exceeded 1 in 65.1% of simulations, indicating that combined exposure, not any single drug, drives the aggregate concern.

The prioritization shifted again when the researchers weighted risk quotients by predicted no-effect concentrations, a refinement that accounts for differences in intrinsic toxicity among compounds. This PNEC weighting increased the relative priority of ciprofloxacin in wastewater profiles and, in lake water, moved the top priorities away from oxytetracycline toward enrofloxacin and ofloxacin. The contrast is the study’s central message: a compound that dominates measured concentrations, such as oxytetracycline, is not necessarily the compound that dominates ecological risk once toxicity differences are factored in.

For environmental managers, the implications are twofold. Source control in a basin like Poyang Lake cannot target one wastewater category, because the receptor-side evidence points to mixed, spatially variable contributions from sewage treatment, hospitals, livestock and aquaculture, and pharmaceutical industry effluents. And risk-oriented management should not simply chase the most abundant compounds; the authors argue their findings support source-sensitive, risk-oriented management of antibiotic mixtures in large floodplain lake basins, in which priority setting incorporates both where the drugs come from and how hazardous they are at measured levels.

The study also illustrates the value of pairing compositional source-apportionment methods with probabilistic risk metrics. Positive matrix factorization and non-negative least squares fitting describe the fingerprint of contamination reaching the lake, while Monte Carlo risk quotients and PNEC weighting translate that fingerprint into ranked ecological concerns. Applied together across 158 samples spanning the full range of wastewater sources and open lake water, these tools showed that concentration-based composition and ecological-risk priority diverge in a large floodplain lake system, a divergence the authors say should guide how antibiotic mixtures are monitored and managed.

Subject of Research: Source apportionment and ecological risk prioritization of antibiotics in the Poyang Lake basin

Article Title: Source-related patterns and ecological risk prioritization of antibiotics in a large lake–river system: insights from the Poyang Lake Basin

Article References: Yao, N., Li, J., Ye, N., Yan, S., Zhang, X., Zhang, Y., Liu, Z., & Zhang, M. (2026). Source-related patterns and ecological risk prioritization of antibiotics in a large lake–river system: insights from the Poyang Lake Basin. Environmental Geochemistry and Health, 48(16), Article 655. https://doi.org/10.1007/s10653-026-03548-3

Image Credits: AI Generated

DOI: 10.1007/s10653-026-03548-3

Keywords: antibiotics, Poyang Lake, ecological risk, positive matrix factorization, Monte Carlo analysis, wastewater, risk quotient, tetracyclines, source apportionment, floodplain lake, water quality, China