Around one third of current Arctic Ocean primary production sustained by rivers and coastal erosion

The study addresses how terrigenous nutrient inputs from rivers and coastal erosion influence Arctic Ocean net primary production, a system previously thought to be dominated by nutrient supply from upwelling and lateral oceanic inflows. Given rapid Arctic climate change, increased river discharge, and accelerating permafrost coastal erosion, the authors aim to quantify the magnitude, seasonality, and spatial distribution of land-derived carbon and nutrient fluxes to the Arctic Ocean and to assess their net impact on marine NPP. Prior work focused primarily on physical drivers and often neglected terrigenous nutrients; this gap is critical because the Arctic Ocean receives disproportionately large river discharge relative to its volume, and its eroding permafrost coasts may supply significant nutrients. The study proposes that terrigenous nutrients are a key control on present and future Arctic Ocean productivity.

Observed Arctic Ocean NPP increased by about 57% from 1998 to 2018, largely attributed to warming, sea-ice decline, and circulation changes. Model projections of current and future Arctic NPP diverge widely due to complex light-nutrient interactions and physical biases. Previous estimates of riverine nutrient contributions to Arctic NPP ranged from about 1–4% using fixed recycling assumptions to 5–13% using explicit nitrogen cycling, often neglecting coastal erosion. Pan-Arctic riverine flux extrapolations primarily used data from the six largest rivers, while nutrient fluxes from coastal erosion lacked pan-Arctic quantification due to sparse soil nutrient measurements. Terrigenous particulate effects on turbidity and delta processes have also been underrepresented in observational and modeling studies.

The authors developed two gridded, seasonally varying pan-Arctic forcing datasets north of 60°N: (1) riverine dissolved carbon and nutrients and (2) coastal erosion total (particulate plus dissolved) carbon and nutrients. Riverine fluxes of alkalinity, dissolved inorganic and organic carbon, dissolved inorganic and organic nitrogen, total dissolved phosphorus, and silicate were extrapolated from monthly observations for the six largest Arctic rivers (ArcticGRO) to all Arctic watersheds using multiple linear regression with watershed predictors (permafrost extent, lake proportion, topsoil clay and organic carbon content, glacier cover, runoff). Annual fluxes were first predicted to minimize uncertainty; monthly fluxes were derived by scaling with monthly to annual discharge ratios with fitted exponents capturing dilution or flushing behaviors, and a sinusoidal seasonal term for DIN. Concentrations were bounded by observed ArcticGRO ranges. The final river forcing was provided on a 1° grid and remapped to the ocean model grid proportional to runoff, applied in the upper 13 m. For coastal erosion, total organic carbon fluxes were computed by combining spatially resolved erosion rates (1950–2010) with soil carbon content. Total nitrogen fluxes were inferred using coast-specific soil C:N ratios (North America 15.1:1, Eurasia 10.5:1), and total phosphorus using a global soil N:P of 13:1. Monthly seasonality was prescribed (2% May, 5% June, 15% July, 29% August–September, 15% October, 5% November). Fluxes were applied along coastal grid cells proportional to area. Ocean-biogeochemical simulations used NEMO-LIM-PISCES (ORCA025) at eddy-admitting resolution (~14 km in the Arctic). Because PISCES enforces marine Redfield stoichiometry and does not separately track terrigenous DOC, DON, and DOP, all terrigenous organic inputs were added as inorganic nutrients at the land-ocean interface (nitrate for N), acknowledging an overestimate of immediately bioavailable nutrients. Particulate matter effects on turbidity were not simulated. Three simulations (1990–2010) were run with identical physical forcing: Baseline (rivers + coastal erosion), NoCoast (rivers only), and NoTerr (no terrigenous inputs). Analysis focused on 2005–2010 after model spin-up. Uncertainties were quantified separately for forcing construction and organic matter lability. Riverine flux uncertainties combined temporal (4%) and spatial extrapolation (±10%) to ±11%. Coastal erosion flux uncertainties were larger due to erosion rate and soil stoichiometry variability, assuming −40% to +57% for N and −41% to +57% for P. Reactivity adjustments considered that only 20–80% of riverine DON and 70–90% of coastal erosion organic N are remineralised fast enough to fuel NPP, with additional sources from riverine particulate N escaping deltas and subsea coastal erosion accounted for in sensitivity scaling. Remote-sensing-based NPP (2005–2010) provided observational constraints for regional comparisons.

• Riverine dissolved nitrogen input to the Arctic Ocean is estimated at 1.0 (0.9–1.1) Tg N yr−1, composed of ~32% inorganic and ~68% organic forms. Sixty percent is delivered by five rivers (Mackenzie, Pechora, Ob, Yenisei, Lena). Strong seasonality peaks in June at ~0.33 Tg N month−1.
• Coastal erosion supplies 1.6 (1.0–2.5) Tg N yr−1 of total nitrogen, predominantly in particulate organic form, with highest inputs along the Eurasian coast and seasonal peaks in August–September (~0.45 Tg N month−1).
• Consistency with previous carbon flux estimates from coastal erosion: 15.4 (9.2–24.2) Tg C yr−1, slightly above recent upper-end literature values, and within regional ranges.
• Baseline simulated Arctic Ocean NPP (as defined by basin boundaries) averages 380 Tg C yr−1 for 2005–2010. North of 66°N, simulated NPP is 551 ± 89 Tg C yr−1, agreeing with remote-sensing-based 540 ± 25 Tg C yr−1. Regional agreement is generally good, with noted mismatches in Barents (model lower), Chukchi and CAA (model higher) likely due to biases in lateral oceanic inflows.
• Removing terrigenous inputs (NoTerr) reduces Arctic Ocean NPP by 138 Tg C yr−1 (−36%) relative to Baseline. Partitioning indicates coastal erosion sustains ~79 Tg C yr−1 (21%) and rivers ~58 Tg C yr−1 (15%) of Arctic Ocean NPP.
• Terrigenous contribution is highest on Siberian shelves: Kara 59%, Laptev 80%, East Siberian 57%. Regions strongly influenced by ocean inflows show smaller relative impacts.
• Terrigenous impact increases through summer from 22% (May) to 47% (September).
• Recycling: The observed NPP enhancement implies terrigenous N is recycled on average about seven times before export, burial, or lateral export. Most remineralisation occurs in shallow shelf sediments and near-surface waters; 88% of remineralisation happens above 55 m depth; 52% in sediments, 26% in water column, 17% via zooplankton excretion, with low burial (~3%).
• Accounting for organic matter reactivity and missing sources, adjusted estimates suggest terrigenous nitrogen sustains 28–51% of Arctic Ocean NPP. Rivers account for 9–11% and coastal erosion for 19–41% of total NPP.
• Comparison to prior studies: riverine contributions align with 9 (5–13)% when explicitly modeling N cycling, higher than earlier 1–4% fixed-recycling estimates. Coastal erosion contributions are newly quantified and found to exceed riverine contributions.

The findings demonstrate that terrigenous nitrogen from both rivers and especially coastal erosion is a dominant control on Arctic Ocean NPP, challenging the traditional view that emphasizes only oceanic upwelling and lateral inflow. Efficient nutrient recycling on extensive shallow shelves, driven by strong benthic-pelagic coupling and rapid remineralisation in sediments and the water column, amplifies the effect of relatively modest external nitrogen inputs. Spatial patterns show the largest terrigenous influence on Siberian shelves with limited oceanic nutrient supply, while regions with strong Atlantic or Pacific inflows display smaller relative effects. Temporal increases in terrigenous sustenance through summer reflect seasonal delivery and enhanced recycling. The good agreement between modeled and satellite-derived NPP when terrigenous inputs are included supports the model representation of N limitation and the critical role of land-derived nutrients. These results imply that recent NPP increases attributed solely to physical changes likely also reflect rising terrigenous inputs and that future NPP trajectories will depend on evolving river discharge and coastal erosion under climate change. Integrating terrigenous nutrient fluxes into Earth System Models is therefore essential for more reliable projections of Arctic productivity and ecosystem services.

This study provides the first combined, seasonally and spatially resolved pan-Arctic estimates of terrigenous carbon and nutrient inputs from rivers and coastal erosion and quantifies their impact on Arctic Ocean NPP using an eddy-admitting global ocean-biogeochemical model. It shows that terrigenous nitrogen sustains roughly one third, and possibly up to one half, of current Arctic Ocean NPP, with coastal erosion contributing more than rivers due to efficient recycling on shallow shelves. These insights highlight coastal erosion as a major driver of Arctic productivity and underscore the need to better constrain land-sea nutrient fluxes and to consistently represent them in Earth System Models. Future research should reduce uncertainties in coastal erosion rates and soil stoichiometry, quantify the lability and burial of terrigenous organic matter, resolve nearshore turbidity and delta processes, and incorporate additional nitrogen cycle pathways to improve projections of Arctic NPP and fisheries potential under ongoing climate change.

Key limitations include: (1) Model constraint requiring addition of all terrigenous organic matter as inorganic nutrients, likely overestimating immediate bioavailability; adjusted analyses suggest only 20–80% of riverine DON and 70–90% of coastal erosion organic N are rapidly remineralised. (2) Omission of terrigenous particulate matter effects on turbidity and light, which may delay and localize NPP responses in nearshore zones. (3) Exclusion or simplified treatment of additional terrigenous nitrogen sources, such as riverine particulate N escaping deltas and subsea coastal erosion, and limited accounting for delta transformation processes. (4) Large uncertainties in coastal erosion rates, soil carbon and nutrient contents, and stoichiometric ratios, leading to wide error bounds for flux estimates. (5) The model does not explicitly simulate certain nitrogen cycle processes (e.g., benthic denitrification, diazotrophy, atmospheric deposition nuances, Greenland glacier inputs) which could partly offset each other but remain spatially and temporally decoupled. (6) Potential biases in simulated lateral oceanic inflows affecting regional NPP comparisons (e.g., Barents, Chukchi, CAA).