Nitrogen and phosphorus trends in lake sediments of China may diverge

The study addresses how nitrogen (N) and phosphorus (P) levels in Chinese lakes have evolved since the mid-19th century and how they are likely to change through 2100. The motivation stems from limited long-term monitoring data for Chinese lake waters (systematic nutrient records begin only around 2003), which hampers understanding of eutrophication drivers and temporal trends, and constrains effective water quality management. Nutrients enter lakes via soil erosion, runoff, and atmospheric deposition (notably N). Because direct long-term water-column records are lacking, sediment archives provide a viable alternative to reconstruct nutrient trajectories. The purpose is to compile and analyze sedimentary TN and TP records across China’s six lake districts to (1) reconstruct historical trends and identify structural transition periods, (2) assess the combined roles of anthropogenic pressures and climate variability, and (3) project future district-level trajectories under different socioeconomic and climate scenarios, thereby informing targeted nutrient management. The work is important given extensive eutrophication issues, large investments in remediation, and uneven water resource distribution affecting millions in China.

Prior research shows humans have substantially altered aquatic environments through increased nutrient emissions and climate change. China has invested heavily in water restoration, yet eutrophication persists in iconic lakes (e.g., Taihu, Fuxian). Lake nutrient levels are commonly assessed using water TN and TP, but long-term observational records in China are sparse, with systematic data only since ~2003. Sediment cores are established proxies to reconstruct long-term nutrient histories, but prior Chinese sediment records were limited, restricting regional synthesis and causal attribution. Studies indicate strong agricultural contributions (fertilizer use, livestock), sewage discharge, and atmospheric N deposition to nutrient loads, with evidence of regional differences linked to population density, land use, and industrialization. Global and Chinese studies also highlight climate influences on erosion/runoff and biogeochemical processes, and suggest many lakes are P-limited, implying P control is crucial for eutrophication management. This study addresses gaps by assembling a broader Chinese lake sediment dataset and linking it with climatic and anthropogenic drivers to produce historical reconstructions and future projections.

Data synthesis and sites: The authors compiled 61 published lake sediment core records with reliable chronologies (average temporal resolution <5 years) and added 8 new cores, totaling 69 lake records across China’s six lake districts. Of these, 62 include total nitrogen (TN) and 49 include total phosphorus (TP). New cores were dated using the CRS 210Pb model; TN was measured by Kjeldahl and TP by colorimetric methods. Records were interpolated to annual resolution (R 4.2.0) to build time series from 1850 onward.

Historical change-point detection: Structural transition nodes (change points) in TN and TP time series were detected using the R package trend, with visualization via ggplot2/ggpubr. Mean ages of transition nodes were determined and district-level normalized trends produced.

Drivers and datasets: Climatic variables (temperature, precipitation) used CRU TS v4 for correlations; for projections, precipitation from MRI-ESM2-0 and temperature from GFDL-ESM4 (CMIP6, ESGF). Anthropogenic drivers included fertilizer consumption (global gridded datasets mapped to lake grids), population (provincial/gridded projections for China), and atmospheric N deposition estimated from EDGAR emissions harmonized with CMIP6 (input4MIPS) and calibrated against China-specific N deposition simulations (CHND). Fertilizer consumption (1967–2013) was regressed against global trends (1960–2100) to extend 2013–2100 by district (r2: N=0.95, P=0.93). N deposition models were built linking EDGAR emissions to CHND (1981–2015, r2=0.92) and extended with ESGF trends (1990–2015, r2=0.84) to 2100 at district scale. Population density (1955–2020) was basin-extracted and aligned with projections.

Model construction and selection: Multiple linear regression models were built per district for TP and TN using combinations of independent variables while checking factor independence and redundancy (AIC/BIC). The chosen Pop_M model balanced fit and parsimony, using climate (precipitation, temperature), fertilizer consumption (P for TP; N for TN), population, and N deposition (for TN). Model forms: Y_TP = A(Prep, Temp, Pfert, Pop) + δ; Y_TN = B(Prep, Temp, Nfert, Pop, Ndep) + δ'. Annual series for 1901–2022 (no gaps) were used for fitting; climate series were smoothed (10-year window) to match sediment trends. Models were significant at p<0.05; performance assessed via r2, SSE, RMSE, AIC, BIC. An example indicated adding GDP led to overfitting; LUCC and lake temperature were excluded due to counterintuitive correlation or redundancy with included variables.

Projections: Using SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenario inputs (CMIP6 climate, extended fertilizer use, population, and N deposition), district-level TN and TP trajectories were projected for 2030–2100, with uncertainty summarized via RMSE shading. Analyses also examined lake morphometric context (volume) and regional characteristics (e.g., Yunnan–Guizhou Plateau tectonic basins) to interpret spatial differences.

Uncertainty: Data limitations (factor availability, harmonization of EDGAR/ESGF, smoothing, spatial resampling), model simplicity (linear form, limited factors), and weaker fit in specific cases (e.g., TP model for Inner Mongolia r2=0.05) were acknowledged. Sensitivity of predictions to future policy changes and potential ecological tipping points was noted.

• Historical increases: Across China, sediment TN and TP concentrations increased by ~267% and ~202%, respectively, since 1850.
• Transition timing: Structural transition nodes cluster around ~1950–1965 (mean TN 1956; mean TP 1957), aligning with the onset of the Anthropocene and rapid growth in population, agriculture, and industry.
• Regional differences: Highest post-transition variation ratios for TN and TP occur in Northeast Plain and Mountains (NEPM, IV), Eastern Plain (EP, V), and Yunnan–Guizhou Plateau (YGP, VI), regions east of the Hu Huanyong Line with intensive agriculture, industry, and higher atmospheric N deposition (dry 2.25–44.69 and wet 2.06–37.56 kg/ha/yr for 2006–2015). YGP lakes show anomalously high TN and TP (3–4× other districts) due to tectonic basins with steep catchments, fewer outlets, and long residence times.
• Lake size effect: Lakes with volumes ~1–100×10^6 m^3 exhibit larger ΔTN and ΔTP and higher variation ratios; very small or very large lakes show lower susceptibility to rapid eutrophication.
• Driver correlations: TN/TP correlate more strongly with fertilizer consumption and population than with climate variables, indicating dominant recent anthropogenic influence; atmospheric N deposition is a substantial N source (2006–2014 average deposition ~0.74× cropland N emissions; deposition ≈8.13% of total watershed N emissions).
• Future TN trends (2030–2100): TN generally decreases in five districts; stronger declines in southern districts (QTP I, EP V, YGP VI) and slower declines or stabilization at high levels in northern districts (XJ II, IMP III, NEPM IV). Average projected decreases: southern lakes ~87%, northern lakes ~19% (aggregate figures). Decreases in EP and YGP are driven by reductions in N deposition and population; QTP decreases reflect low human activity.
• Future TP trends (2030–2100): TP continues increasing in several districts, notably XJ (II), EP (V), and YGP (VI), with average increases ~25%. Only QTP (I) shows significant TP decreases; elsewhere, increases slow or level off after ~2060 in some scenarios.
• Management targets: To align northern TN declines with southern trends by 2100, N emissions reductions of at least ~38% (XJ II) and ~97% (NEPM IV) are required. To curb TP accumulation by 2100, P emissions need reductions of at least ~43% (XJ II), ~58% (EP V), and ~89% (YGP VI).

Reconstructing 1850–2020 sedimentary TN and TP across 69 lakes reveals that nutrient enrichment accelerated around the mid-20th century in step with intensified human activities, validating sediment proxies for long-term nutrient trajectories in the absence of historical water-column monitoring. The identification of consistent transition nodes across districts supports a nationwide shift toward eutrophication during the Anthropocene. Strong associations with fertilizer use and population underscore the primacy of anthropogenic drivers over climate in recent decades, while acknowledging climate’s ongoing and future influence via altered hydrology and biogeochemistry.

Model projections indicate divergent future trajectories: TN tends to decrease in most districts under scenarios featuring declining atmospheric N emissions and slower population growth, reflecting existing and anticipated policy measures (e.g., water pollution controls, carbon neutrality goals). In contrast, TP is projected to continue increasing in several populous and agriculturally intensive districts (XJ, EP, YGP), consistent with persistent phosphorus legacies and P limitation of primary production. This divergence implies that eutrophication risk may persist or even heighten via P-driven productivity and algal blooms despite progress on N controls. The findings argue for district-specific management, with intensified P monitoring and reduction strategies where TP is projected to rise, while maintaining and strengthening N controls, especially in northern districts with high sensitivity to agricultural N inputs. Quantified reduction targets provide actionable guidance for aligning regional outcomes with national goals.

The study compiles a China-wide sedimentary dataset (69 lakes) to reconstruct long-term TN and TP trajectories and identifies a mid-20th-century shift to accelerated nutrient accumulation. It links nutrient trends primarily to anthropogenic pressures (fertilizer use, population, atmospheric N deposition) and projects that, from 2030 to 2100, TN will generally decline while TP will continue to rise in several districts, leading to divergent nutrient futures. These insights support district-level, element-specific management strategies, emphasizing strengthened P controls alongside sustained N reduction.

Future research should include: high-frequency monitoring of nutrient carriers within watersheds; improved quantification of lake material cycling and internal loading; incorporation of additional drivers (e.g., sewage, livestock, land use) as robust scenario datasets emerge; and assessment of ecological tipping points. Policy directions include optimizing agricultural fertilizer usage and promoting dietary shifts that reduce feed-related fertilizer demand and livestock emissions.

• Data constraints and harmonization: Limited long-term monitoring; reliance on sediment proxies; harmonization between EDGAR and ESGF emissions and deposition datasets introduces uncertainty. Spatial resampling and gridding may add small errors.
• Model structure: Multivariate linear models with a limited factor set (climate, fertilizer, population, N deposition) balance fit and parsimony but may omit relevant processes; adding more variables (e.g., GDP) caused overfitting in tests. The TP model for Inner Mongolia has weak fit (r2≈0.05), reducing confidence there.
• Temporal smoothing and resolution: Climate series smoothed (10-year) to match sediment trends, potentially damping variability.
• Scenario and policy dependence: Projections assume current management paradigms; future policy changes could alter trajectories. Lack of future scenario data for some factors (e.g., sewage, livestock) required proxies.
• Ecological complexity: Potential tipping points and internal loading dynamics are not explicitly modeled, which may affect nutrient trajectory realism in high-nutrient systems.