A redistribution of nitrogen fertiliser across global croplands can help achieve food security within environmental boundaries

The study addresses how to reconcile global food security with reducing reactive nitrogen (Nx) pollution to within planetary boundaries. Industrial nitrogen fixation for fertiliser has increased reactive nitrogen inputs from ~100 Tg N/yr in 1860 to 286 Tg N/yr today, with 110 Tg N/yr from synthetic fertiliser. Accumulation of reactive nitrogen contributes to climate change (via N2O), ozone depletion, biodiversity loss and eutrophication. To remain within a “safe operating space,” anthropogenic Nx creation should be limited to roughly 60–100 Tg N/yr (potentially up to 130 Tg N/yr). Maize, wheat and rice use over 60 Tg N/yr of synthetic fertiliser (~60% of global use) and supply ~50% of human calories. The research question is whether spatial redistribution of nitrogen fertiliser can maintain or increase cereal yields while reducing environmental nitrogen losses, particularly N2O emissions and nitrate leaching, under near-term (to 2030) constraints.

Prior work has established planetary boundaries for nitrogen and highlighted the need to limit anthropogenic Nx creation while ensuring food security. Studies have examined global food demand growth, yield gaps, and nitrogen management strategies, including trade-off frontiers for nitrogen use efficiency and cereal production. Previous analyses often either (a) focused on yield-gap closing without explicit environmental nitrogen boundaries, or (b) enforced nitrogen boundaries using assumed future efficiency gains. This paper builds on these by quantifying a spatial redistribution strategy that directly links fertiliser allocation to yield responses and environmental losses using a process-based model, thereby providing an explicit, data-driven scheme to improve global nitrogen use efficiency in cereals within environmental constraints.

The authors use the LandscapeDNDC (LDNDC) process-based biogeochemical model to simulate nitrogen, carbon, and water flows in cereal systems (maize, wheat, rice) at 0.5° spatial resolution. Simulations target 2015-like conditions through 2030, assuming constant climate and CO2 impacts on yields (validated as small before 2030), and constant irrigation and management except fertiliser rates. Nutrients other than nitrogen (P, K, etc.) are assumed non-limiting. Baseline fertiliser application rates combine subnational crop-specific data for major producers with national rates elsewhere and are aligned with the LUH2 dataset. Model outputs are averaged over 2006–2015 climate to represent typical 2015 conditions. Harvested areas are based on MIRCA2000, scaled to FAOSTAT 2015 totals; simulations cover 566.4 Mha (~99% of global maize, wheat, rice area). For each grid cell, crop, and water management (rainfed/irrigated), the team ran 21 fertiliser levels (0–600 kg N/ha per cropping season) and interpolated fertiliser response curves for yields, N2O emissions (direct and indirect), and nitrate (NO3−) leaching. NO3− concentration in soil-water leachate is derived by dividing NO3− leaching by total water outflow (percolation + runoff). Indirect N2O emissions are calculated using IPCC 2019 emission factors (1.1% of leached NO3− and 1% of volatilised NH3 converted to N2O). A stochastic optimisation redistributes N fertiliser globally to minimise N use for a fixed global production level, subject to a constraint that NO3− concentration in leachate is below 2.5 mg N/l in every grid cell (unless already exceeded at zero synthetic N). Scenarios are optimised across a frontier of production vs. N use, maintaining the maize:wheat:rice production ratio up to a 9% increase (beyond which maize is constrained by yield saturation and leaching limits). Targets for 2030 are based on projected crop demand increases (7–15% over 2015–2030) and IPCC 1.5 °C-compatible N2O trajectories (−29% to +1% over 2015–2030). LDNDC calibration/validation draws on extensive field, regional, and global studies across climates and management intensities, with particular attention to simulating N2O and NO3− responses to variable N inputs. The model is validated against FAO yields and N contents, tier-1 country-scale N2O and NO3− estimates, and GAEZ nutrient deficiency yield gaps.

- Baseline (circa 2015): 62 Tg N/yr synthetic fertiliser to cereals (22 Tg maize, 21 Tg wheat, 19 Tg rice) plus 8 Tg N/yr manure, yielding 2580 Tg/yr cereals. N2O emissions total ~1.02 Tg N/yr (0.76 direct, 0.25 indirect). NO3− leaching is 14.5 Tg N/yr (21% of applied N). 40% of harvested area (48% of production) exceeds 2.5 mg N/l NO3− leachate; 7% exceeds 11 mg N/l (unsafe for drinking water). Production is concentrated in North America, East Asia, and Europe (53% of production on 38% of area); Sub-Saharan Africa (SSA) has 9% of area but 4% of production, reflecting fertiliser inequality (East Asia 236 kg N/ha/season; North America 143; Europe 94; SSA 11). - Low emission scenario: Maintains and slightly increases production to 2640 Tg/yr (+2%) with a 32% cut in synthetic N to 42 Tg N/yr. N2O emissions drop by 29% (maize −29%, wheat −40%, rice −13%); NO3− leaching falls by 57% overall (maize −62%, wheat −71%, rice −24%). NO3− concentration is below 2.5 mg N/l on 96% of harvested area. Regional shifts: East Asia N use decreases from 236 to 88 kg N/ha/season (production −18%), North America 143→107 (−6% production); NO3− leachate concentration drops in East Asia from 7.1 to 1.2 mg N/l (−81%) and in North America from 2.8 to 1.7 (−40%); N2O emissions decline 53% in East Asia and 25% in North America. Modest N increases in SSA (11→54 kg N/ha) and Eurasia (11→59) yield large production gains (+67% and +123%). By crop: Wheat shows highest potential to reduce N use and N losses without yield loss (−45% N, −40% N2O, −71% NO3−); maize shows substantial potential (−33% N, −29% N2O, −62% NO3−); rice has lower potential due to paddy system dynamics. - High yield scenario: Increases global cereal production by 15% to 2960 Tg/yr with essentially unchanged synthetic N use (63 Tg N/yr, +2%). N2O emissions change by −5% overall (maize −8%, wheat −6%, rice +4%); NO3− leaching declines by 46% (maize −50%, wheat −59%, rice −14%), with >96% of area below 2.5 mg N/l. Regionally, East Asia production decreases by 11% with a 51% N use cut (to 117 kg N/ha/season), lowering N2O by 44% and NO3− leaching by 77%. Large production increases occur in SSA (+93%, N to 94 kg N/ha) and Eurasia (+159%, N to 114 kg N/ha). - Maintain regional production scenario: Meets or exceeds 2015 production in all regions, yielding 3050 Tg/yr (+18%) with 70 Tg N/yr (+13%). NO3− leaching decreases by 28%, while N2O increases by 4%. - Trade-off frontier: Mapping the frontier shows diminishing efficiency of converting N to yields at higher production, with N losses rising faster than yields. Small reductions in demand (e.g., diet shifts, reduced waste) would significantly reduce required N use and N losses at the margin. - East Asia maintenance: Keeping East Asia’s current production with internal redistribution requires 22% less synthetic N, but NO3− pollution remains problematic: >50% of harvested area still exceeds 2.5 mg N/l; maize leaching averages 57 kg N/ha over 45.4 Mha versus rice 14 kg N/ha over 33.7 Mha.

The findings demonstrate that spatially redistributing nitrogen fertiliser to locations with the greatest marginal yield response can significantly reduce environmental nitrogen losses while maintaining or increasing global cereal production by 2030. This approach mitigates the nitrogen challenge by lowering N2O emissions consistent with 1.5 °C pathways (low emission scenario) and keeps NO3− leaching below ecological thresholds across most cropland. Redistribution reduces reliance on current breadbasket regions (e.g., US Midwest, Eastern China), spreading risk and potentially improving resilience to climate shocks and geopolitical disruptions. It enables under-fertilised regions, notably Sub-Saharan Africa, to approach or achieve cereal self-sufficiency at current consumption levels, while curbing excessive nitrogen losses in heavily fertilised regions such as East Asia. The trade-off frontier indicates that as production targets rise, nitrogen-use efficiency declines and environmental costs grow disproportionately, underscoring the value of complementary demand-side measures (diet shifts, reducing food waste). Compared to prior studies, this work explicitly quantifies a redistribution mechanism that satisfies environmental constraints without assuming exogenous efficiency gains, thereby offering a practical pathway to align food security and nitrogen boundary targets.

Spatial redistribution of synthetic nitrogen fertiliser across global cereal croplands can, by 2030, either maintain current production with one-third less N or increase production by 15% with similar N use, all while substantially reducing nitrate leaching and cutting N2O emissions. This strategy diffuses production away from concentrated breadbaskets, advances food self-sufficiency in regions like Sub-Saharan Africa, and alleviates nitrogen pollution in highly fertilised regions, notably East Asia. The study provides a quantified framework and trade-off frontier to guide policy and management, highlighting that even partial implementation yields benefits. Future research should refine model processes (particularly for maize and rice), incorporate crop rotations and multi-cropping interactions, integrate pests/diseases, explore region-specific nitrate critical loads, and evaluate socio-economic feasibility and governance mechanisms for implementing redistribution alongside complementary agronomic practices (4R fertiliser stewardship, inhibitors, residue management) and improved nutrient recycling from human and animal waste.

- Temporal scope: Assumes 2015-like climate, CO2 fertilisation, irrigation, and management through 2030; does not capture near-term climate variability impacts beyond the 2006–2015 averaging. - Nutrient assumptions: Phosphorus, potassium, and other nutrients are assumed non-limiting; in practice, co-limitation may constrain yield responses. - Cropping systems: Single season per year; crop rotations, multi-cropping interactions, and biological N fixation in rotations are not explicitly simulated. - Environmental constraint: Applies a globally homogeneous NO3− critical load of 2.5 mg N/l; real landscapes vary (e.g., wetlands may increase retention). If the threshold is exceeded even at zero synthetic N, no further enforcement (e.g., manure reduction) is applied. - Model biases: Global calibration indicates maize yields are under-predicted (~10%) and rice over-predicted (~20%), potentially biasing N surplus and loss estimates by crop. - Socio-political feasibility: Implementing yield reductions in high-yield regions and reallocating inputs faces political, economic, and logistical challenges. - Scope: Focuses on cereals; other crops and non-agricultural Nx sources must also be addressed to meet planetary nitrogen limits.