The regional variability in tundra and boreal carbon dioxide (CO₂) fluxes can be high, complicating efforts to quantify sink‐source patterns across the entire region. Statistical models are increasingly used to predict (i.e., upscale) CO₂ fluxes across large spatial domains, but the reliability of different modeling techniques, each with different specifications and assumptions, has not been assessed in detail. Here, we compile eddy covariance and chamber measurements of annual and growing season CO₂ fluxes of gross primary productivity (GPP), ecosystem respiration (ER), and net ecosystem exchange (NEE) during 1990–2015 from 148 terrestrial high‐latitude (i.e., tundra and boreal) sites to analyze the spatial patterns and drivers of CO₂ fluxes and test the accuracy and uncertainty of different statistical models. CO₂ fluxes were upscaled at relatively high spatial resolution (1 km²) across the high‐latitude region using five commonly used statistical models and their ensemble, that is, the median of all five models, using climatic, vegetation, and soil predictors. We found the performance of machine learning and ensemble predictions to outperform traditional regression methods. We also found the predictive performance of NEE‐focused models to be low, relative to models predicting GPP and ER. Our data compilation and ensemble predictions showed that CO₂ sink strength was larger in the boreal biome (observed and predicted average annual NEE −46 and −29 g C m⁻² yr⁻¹, respectively) compared to tundra (average annual NEE +10 and −2 g C m⁻² yr⁻¹). This pattern was associated with large spatial variability, reflecting local heterogeneity in soil organic carbon stocks, climate, and vegetation productivity. The terrestrial ecosystem CO₂ budget, estimated using the annual NEE ensemble prediction, suggests the high‐latitude region was on average an annual CO₂ sink during 1990–2015, although uncertainty remains high.

The response of evapotranspiration (ET) to warming is of critical importance to the water and carbon cycle of the boreal biome, a mosaic of land cover types dominated by forests and peatlands. The effect of warming-induced vapour pressure deficit (VPD) increases on boreal ET remains poorly understood because peatlands are not specifically represented as plant functional types in Earth system models. Here we show that peatland ET increases more than forest ET with increasing VPD using observations from 95 eddy covariance tower sites. At high VPD of more than 2 kPa, peatland ET exceeds forest ET by up to 30%. Future (2091–2100) mid-growing season peatland ET is estimated to exceed forest ET by over 20% in about one-third of the boreal biome for RCP4.5 and about two-thirds for RCP8.5. Peatland-specific ET responses to VPD should therefore be included in Earth system models to avoid biases in water and carbon cycle projections.

Arctic wetlands are known methane (CH₄) emitters but recent studies suggest that the Arctic CH₄ sink strength may be underestimated. Here we explore the capacity of well-drained Arctic soils to consume atmospheric CH₄ using >40,000 hourly flux observations and spatially distributed flux measurements from 4 sites and 14 surface types. While consumption of atmospheric CH₄ occurred at all sites at rates of 0.092 ± 0.011 mgCH₄ m⁻² h⁻¹ (mean ± s.e.), CH₄ uptake displayed distinct diel and seasonal patterns reflecting ecosystem respiration. Combining in situ flux data with laboratory investigations and a machine learning approach, we find biotic drivers to be highly important. Soil moisture outweighed temperature as an abiotic control and higher CH₄ uptake was linked to increased availability of labile carbon. Our findings imply that soil drying and enhanced nutrient supply will promote CH₄ uptake by Arctic soils, providing a negative feedback to global climate change.

Wetland methane (CH₄) emissions ( $${F}_{{{CH}}_{4}}$$F C H 4 ) are important in global carbon budgets and climate change assessments. Currently, $${F}_{{{CH}}_{4}}$$F C H 4 projections rely on prescribed static temperature sensitivity that varies among biogeochemical models. Meta-analyses have proposed a consistent $${F}_{{{CH}}_{4}}$$F C H 4 temperature dependence across spatial scales for use in models; however, site-level studies demonstrate that $${F}_{{{CH}}_{4}}$$F C H 4 are often controlled by factors beyond temperature. Here, we evaluate the relationship between $${F}_{{{CH}}_{4}}$$F C H 4 and temperature using observations from the FLUXNET-CH₄ database. Measurements collected across the globe show substantial seasonal hysteresis between $${F}_{{{CH}}_{4}}$$F C H 4 and temperature, suggesting larger $${F}_{{{CH}}_{4}}$$F C H 4 sensitivity to temperature later in the frost-free season (about 77% of site-years). Results derived from a machine-learning model and several regression models highlight the importance of representing the large spatial and temporal variability within site-years and ecosystem types. Mechanistic advancements in biogeochemical model parameterization and detailed measurements in factors modulating CH₄ production are thus needed to improve global CH₄ budget assessments.