Unexpectedly minor nitrous oxide emissions from fluvial networks draining permafrost catchments of the East Qinghai-Tibet Plateau

The study investigates how streams and rivers draining permafrost catchments contribute to atmospheric N2O emissions, a potent greenhouse gas and ozone-depleting substance. Prior work on fluvial N2O has focused on human-impacted lowland systems where anthropogenic nitrogen elevates emissions, leaving large uncertainty in natural, high-altitude cryospheric regions. Permafrost stores vast carbon and nitrogen pools, and thaw can deliver inorganic N to rivers, potentially enhancing N2O production. The central questions are: (1) What are the magnitudes and seasonal patterns of N2O concentrations and fluxes from permafrost-affected rivers on the East Qinghai-Tibet Plateau (EQTP)? (2) What terrestrial, biogeochemical, microbial, and physicochemical processes regulate these dynamics? (3) How do these emissions scale regionally and compare globally, and how might they change under future warming? Addressing these gaps is important for constraining global N2O budgets and understanding potential non-carbon climate feedbacks from permafrost regions.

Global fluvial N2O emissions are estimated at ~291 ± 59 Gg N2O-N yr−1 but are poorly constrained, particularly for pristine and cryospheric regions. Most studies emphasize anthropogenically enriched lowland rivers with high DIN and N2O. Permafrost soils contain large nitrogen stocks and are evident N2O sources; Alaskan streams exhibit elevated inorganic N and often N2O supersaturation. However, direct N2O flux measurements from permafrost-affected fluvial networks are scarce, limiting extrapolation. Additionally, riverine N2O production is modulated by oxygen, nitrate availability, and denitrification/anammox pathways, with stream order, hydrogeomorphology, and suspended sediments influencing scaling of emissions. These insights frame expectations but highlight the lack of high-altitude permafrost river data.

Study area: Four headwater catchments (Yangtze, Yellow, Lancang-Mekong, Nu-Salween) across the East Qinghai-Tibet Plateau (EQTP), spanning elevations 1650–7000 m and covering ~7.36×10^5 km². Sites were categorized by permafrost extent (continuous vs. non-continuous: discontinuous, sporadic, isolated). Stream orders sampled were 3–7. Sampling design: Daytime sampling in spring (May–June), summer (July–August), and fall (September–October) from 2016–2018; Yellow sampled 7 times, Yangtze 4, Lancang and Nu 3 each. Total of 114 site visits and 342 water samples for N2O concentration; 436 flux measurements. N2O concentration measurements: Triplicate 120 mL serum bottles filled at wrist depth; preserved with saturated ZnCl2, sealed, stored dark. Local ambient air sampled to compute equilibrium concentrations. Headspace equilibration with GC-μECD (Agilent 7890B) for N2O; pCO2 measured per prior work. Flux measurements: Four floating chambers per transect spanning bank-to-mid-channel, 1 h deployments with headspace sampling at 0, 5, 10, 20, 40, 60 min; GC-μECD analysis. Chambers had streamlined collars to minimize turbulence and were foil-covered to limit heating. Partitioning diffusion vs. ebullition: Assumed CO2 flux is purely diffusive. Computed FCO2 from linear pCO2 increase; derived kCO2 using Fick’s law and Henry’s constant; converted to kN2O using Schmidt number scaling (n=1/2 for turbulent/rippled, n=2/3 for smooth). Diffusive FN2O = k(Cwater − Ceq). Ebullition = total FN2O − diffusive fraction. Ancillary measurements: In situ DO, pH, ORP, conductivity, water temperature; meteorology (air temp, pressure, wind). Water chemistry: NH4+, NO2−, NO3−, DOC, total phosphorus. Sediment N2O production (Supplementary Methods). Microbial analyses: 26 sediment samples (triplicate). DNA extraction from ~0.5 g sediment. qPCR quantified dissimilatory nitrite reductase genes (nirS, nirK) and nitrous oxide reductase genes (nosZ, nosZn) to obtain (nirS+nirK)/nosZ ratios. GIS and vegetation: Vegetation coverage mapped within 5 km radius buffers around sites using published datasets; NDVI from 1×1 km monthly data (2016–2018) matched to sampling dates. Hydrological analysis assigned Strahler stream orders and computed stream lengths and widths to estimate surface areas (including additional Google Earth measurements). Upscaling: Monte Carlo (1000 iterations) for 3rd–7th orders combining random flux measurements and normally distributed surface areas; summed over ice-free season (210 d). For 1st–2nd orders, extrapolated fluxes from the observed stream-order relationship and included corresponding areas. Annualized by dividing ice-free emission by 0.85 to account for winter outgassing (~15%). Statistics: Pearson correlations; stepwise multiple regression for FN2O vs. %O2, pH, temperature, TP, NO3−; regression tree for N2O concentrations using %O2 and NO3− with 10-fold cross-validation and 1-SE pruning; significance at P<0.05.

- Supersaturation and concentrations: All waters were N2O-supersaturated (117.9–242.5%, n=342). Dissolved N2O ranged 10.2–18.9 nmol L−1, mean 12.4 ± 1.7 nmol L−1, ~one-third of global average (37.5 nmol L−1). Highest concentrations in spring (P<0.001); no significant differences among the four river systems. - Fluxes: Diffusive N2O fluxes mostly positive, −14.0 to 40.6 μmol m−2 d−1, mean 9.4 ± 6.2 μmol m−2 d−1 (n=436), ~10× lower than global average (94.3 μmol m−2 d−1). Summer and fall fluxes > spring (P<0.05). Ebullition averaged 0.74 ± 2.47 μmol m−2 d−1, contributing 4.1 ± 11.9% of total N2O flux; total flux (diffusion+ebullition) averaged 10.2 ± 7.1 μmol m−2 d−1. - Terrestrial controls: Riverine DIN decreased with greater vegetation coverage across sites; in non-continuous permafrost zones, higher NDVI corresponded to lower DIN, consistent with plant uptake. In continuous permafrost, high NDVI coincided with higher DIN, implying surplus terrestrial N supply exceeding plant demand. Overall DIN was low (0.54 ± 0.30 mgN L−1), at the lower end of global ranges, consistent with low N2O. - Biogeochemical controls: Simple linear models poorly predicted N2O using individual variables (e.g., DO R2=0.004; NH4+ R2=0.1). Under DO undersaturation (%O2<100%), N2O correlated positively with NO3−, indicating denitrification influence. Regression tree identified %O2 as the primary splitter for N2O concentration. - Microbial gene ratios: Sediment (nirS+nirK)/nosZ ratio averaged 1.96, far below other lotic settings (range 2.16–3.24×10^6; average 19.8), indicating low N2O yield (greater capacity for N2O reduction to N2). The ratio negatively correlated with %O2, suggesting enhanced N2O reduction under higher oxygen conditions. - Physicochemical and geomorphic controls on FN2O: Stepwise regression found %O2, pH, temperature, TP, and NO3− significant but weak predictors (overall R2=0.14). Longitudinally, FN2O highest in 3rd-order streams, declined in 4th–5th orders, and slightly increased in 6th–7th orders, likely due to decreasing hyporheic exchange and perimeter-to-area ratios with stream size, and increased DIN and suspended sediments in larger rivers. - Regional upscaling (EQTP): Estimated 0.206 Gg N2O-N yr−1 (5–95%: 0.129–0.291) from 3rd–7th orders (2603 km² area). Including extrapolated 1st–2nd orders: 0.275 Gg N2O-N yr−1 (0.162–0.400) over 3049 km². As CO2-equivalent share of total GHG (CO2+CH4+N2O), N2O comprised 1.0% for 3rd–7th orders and 0.4% for 1st–7th orders, within pristine river ranges (0.2–1.2%) and much lower than human-impacted systems (2.8–13.9%). Per unit stream/river area and basin area: 0.08 t N2O-N km−2 yr−1 and 0.32 kg N2O-N km−2 yr−1, respectively, ~10× lower than global lotic averages. - Plateau-wide estimate (QTP): Applying EQTP rates across QTP 1st–7th orders yields 0.432–0.463 Gg N2O-N yr−1, a minor ~0.15% of global fluvial N2O despite accounting for ~0.7% of global stream/river area. These may be overestimates because winter ice-melt outgassing (~15% annual) is likely limited in permafrost systems with minimal winter N inputs. - Overall conclusion: Despite high CH4 emissions from these rivers, N2O emissions are unexpectedly small, indicating decoupled CH4 and N2O dynamics in EQTP rivers.

The findings demonstrate that permafrost-affected streams and rivers on the EQTP are minor N2O sources relative to global fluvial systems. This addresses the initial hypothesis by showing that, contrary to expectations from thawing permafrost soils, aquatic N2O emissions remain low due to: (1) limited DIN reaching rivers after substantial terrestrial plant uptake; (2) environmental conditions that do not strongly favor N2O accumulation via denitrification except under localized low-oxygen conditions; and (3) microbial communities with low (nirS+nirK)/nosZ ratios, enhancing N2O reduction to N2 and yielding low N2O. Weak relationships between fluxes and standard water chemistry variables and the prominence of %O2 highlight the role of oxygen status and hydrogeomorphology in shaping N2O dynamics. Longitudinal patterns indicate headwaters as relatively higher emitters per area, with attenuation and modest resurgence downstream linked to geomorphic and suspended sediment controls. Regionally and globally, including these cryospheric rivers reduces overestimation biases in N2O inventories. However, the study also identifies mechanisms by which future warming may increase N2O: deeper permafrost thaw increasing DIN export beyond plant uptake, warmer waters lowering gas solubility and promoting hypoxia and denitrification, longer ice-free seasons, and added anthropogenic N, all of which could shift these systems from minor to significant N2O sources, contributing to a positive non-carbon climate feedback.

This study provides the first cross-regional, seasonally resolved measurements of N2O concentrations and fluxes from permafrost-affected rivers of the EQTP, revealing unexpectedly low N2O emissions relative to global averages. Low riverine DIN due to terrestrial plant uptake, environmental conditions limiting N2O buildup, and microbial communities favoring N2O reduction underpin these low emissions. Upscaled estimates indicate that QTP fluvial N2O is a small fraction of global riverine N2O. The work refines global N2O budgets by incorporating a major cryospheric region and identifies oxygen saturation as a key correlate of aquatic N2O in high-altitude streams. Looking forward, accelerated warming, deeper thaw, extended ice-free seasons, and increased anthropogenic N could substantially enhance fluvial N2O emissions, implying a potential non-carbon permafrost feedback. Future research should target low-order streams, expand spatial and temporal coverage across cryospheric basins, integrate detailed nitrogen pathway measurements (including denitrification/anammox partitioning and microbial functional genes), and improve upscaling with better hydrologic and geomorphic constraints.

- Limited observations for 1st–2nd order streams required extrapolation from stream-order relationships, introducing uncertainty in headwater contributions. - High cross-validated relative error for the regression tree model indicates limited predictive skill; findings are more suggestive than predictive. - Potential overestimation of annual emissions due to inclusion of winter ice-melt outgassing (~15%) that may be minimal in permafrost systems with frozen soils and low winter N inputs. - Spatial heterogeneity and limited sampling frequency may miss extreme events or hot moments; data restricted to daytime and ice-free seasons. - Stepwise regression explained a small fraction of flux variance (R2=0.14), indicating unmeasured drivers (e.g., fine-scale hyporheic exchange, microzones, hydrodynamics) may be important.