Where Does the Nitrogen Go? Model Intercomparison and Benchmarking of the Global Terrestrial Nitrogen Cycle and Carbon–Nitrogen Interactions

Nitrogen (N) is an essential element for life and a fundamental regulator of terrestrial ecosystem productivity, carbon sequestration, and climate feedbacks, yet it remains one of the most weakly constrained and uncertain components of Earth system models. Despite major advances in terrestrial biosphere modeling, large discrepancies persist in how models represent nitrogen inputs, internal cycling, losses, and their coupling to the carbon and water cycles. Here we present the first comprehensive, process-resolved benchmark of the global terrestrial nitrogen cycle based on coordinated simulations from the Nitrogen Model Intercomparison Project Phase 3 (NMIP3) combined with multiple independent observational constraints.We trace the full nitrogen cascade across land ecosystems—from natural and anthropogenic inputs (biological nitrogen fixation, atmospheric deposition, and fertilizer and manure application), through plant uptake and soil–microbial transformations, to hydrological export to inland waters and gaseous losses to the atmosphere (N₂O, NH₃, NO, and N₂). This integrated framework allows us to evaluate not only individual fluxes and pools, but also the internal consistency of regional and global nitrogen budgets and their emergent coupling with carbon cycling.Across models, we find broad agreement in the magnitude of total nitrogen inputs and in first-order global spatial patterns. However, models diverge strongly in how nitrogen is partitioned among vegetation, soils, and loss pathways. The largest spreads occur in biological nitrogen fixation, soil nitrogen turnover, nitrate leaching, and gaseous emissions, producing substantial inconsistencies in regional budget closure and large uncertainty in carbon–nitrogen feedback strength. These discrepancies are especially pronounced in intensively managed agricultural regions and climate-sensitive ecosystems, including the tropics and high latitudes, and they propagate directly into uncertainty in the magnitude and spatial distribution of the terrestrial carbon sink.By systematically comparing model structures and process representations, we diagnose the dominant sources of these divergences and show that a small number of key processes control most of the uncertainty. Our analysis demonstrates that improving the representation and observational constraint of biological nitrogen fixation, soil organic matter turnover, and coupled nitrification–denitrification pathways can substantially reduce uncertainties in nitrogen budgets, N₂O and NH₃ emissions, and land carbon sink estimates.More broadly, this work establishes a community framework for nitrogen cycle benchmarking that moves the field from qualitative model intercomparison toward quantitative, process-level accountability. Our results show that coordinated benchmarking can transform nitrogen–carbon–climate projections into more robust and policy-relevant tools, with direct implications for climate mitigation, air and water quality management, and integrated carbon–nitrogen stewardship.