Search

Abstract Background We analyze the forest carbon stock development following the recent historically unprecedented dieback of coniferous stands in the Czech Republic. The drought-induced bark-beetle infestation resulted in record-high sanitary logging and total harvest more than doubled from the previous period. It turned Czech forestry from a long-term carbon sink offsetting about 6% of the country's greenhouse gas emissions since 1990 to a significant source of CO2 emissions in recent years (2018–2021). In 2020, the forestry sector contributed nearly 10% to the country's overall GHG emissions. Using the nationally calibrated Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3) at a regional (NUTS3) spatial resolution, we analyzed four scenarios of forest carbon stock development until 2070. Two critical points arise: the short-term prognosis for reducing current emissions from forestry and the implementation of adaptive forest management focused on tree species change and sustained carbon accumulation. Results This study used four different spruce forest dieback scenarios to assess the impact of adaptive forest management on the forest carbon stock change and CO2 emissions, tree species composition, harvest possibilities, and forest structure in response to the recent unprecedented calamitous dieback in the Czech Republic. The model analysis indicates that Czech forestry may stabilize by 2025 Subsequently, it may become a sustained sink of about 3 Mt CO2 eq./year (excluding the contribution of harvested wood products), while enhancing forest resilience by the gradual implementation of adaptation measures. The speed of adaptation is linked to harvest intensity and severity of the current calamity. Under the pessimistic Black scenario, the proportion of spruce stands declines from the current 43–20% by 2070, in favor of more suited tree species such as fir and broadleaves. These species would also constitute over 50% of the harvest potential, increasingly contributing to harvest levels like those generated by Czech forestry prior to the current calamity. The standing stock would only be recovered in 50 years under the optimistic Green scenario. Conclusion The results show progress of adaptive management by implementing tree species change and quantify the expected harvest and mitigation potential in Czech forestry until 2070.

Mapping and assessing the carbon stock change (CSC) in urban areas can support the allocation of Nature-based Solutions (NbS) to mitigate climate change and advance urban sustainability. However, an effective framework concerning historical CSC and future simulation to support the allocation of NbS implementation is lacking. To fill this gap, we proposed a framework and applied it in the Zhejiang coastal region based on the assessment of historical (from 1990 to 2020) and predicted future (2030) CSC and local context analysis of urban and ecosystem challenges. Over the past three decades, the Zhejiang coastal region has experienced a considerable C stock loss of 20.34 Tg, predominantly owing to fast urbanization. The severest C stock reduction occurred from 2000 to 2010, with a slowdown in the following decade. Even so, more effective spatial management policies are urgent to mitigate further C stock depletion. Our framework identified 50.51% of the study area as the allocation area for NbS implementations where current and future C sequestration demand existed. Within the allocation area, six NbS types identified from literature were allocated or co-allocated, leading to eight tailored NbS implementations to tackle specific urban and ecosystem challenges of each location. The most widely allocated NbS implementations were "NbS1 × NbS2 × NbS3" and "NbS2 × NbS4", covering 42.86% and 34.69% of the allocation area. NbS2 covered nearly the entire allocation area (98.80%), with its primary role of habitat preservation and to control urban expansion. The proposed framework can be adapted to support various planning decisions regarding the prioritization and spatial allocation of NbS., Competing Interests: Declaration of competing interest There is no conflict of interest between the authors., (Copyright © 2024 Elsevier Ltd. All rights reserved.)

The reducing emissions from deforestation and forest degradation (REDD+) framework has been implemented over the past decade, and has led to a restructuring of forest governance systems in host countries. In the case of Lao People’s Democratic Republic, which is promoting REDD+, activities have been implemented at project, sub-national, and national scales. Project-scale REDD+ is assumed to be compatible with small-scale forestry, and usually targets local people to enhance participatory forest management through technology transfer. Such projects were also supported by foreign governments under bilateral cooperation or by private funding. In the case of sub-national- or national-scale REDD+, the Lao People’s Democratic Republic government aims to develop a system of forest monitoring, as well as related structures required by international REDD+ entities. These activities are supported by substantial funding from multilateral organizations. Lessons learned from project-scale REDD+ in northern Lao People’s Democratic Republic showed a gap in expectations among different donors and recipients regarding how to implement REDD+, in particular how to reduce dependency on forest resources in rural areas, and how to estimate and account for greenhouse gas emissions reductions with consistent methodologies at different scales. Such differences are related to the attitudes of local people toward participation, and those of the private entities that fund projects and ground-based activities. In future REDD+ schemes, the structural network or structural social capital among project-, sub- national-, and national-scale activities should be reconsidered to enhance the continued participation of stakeholders and make use of their accumulated experience and knowledge of small-scale forestry management.

Abstract Short rotation plantations are often considered as holding vast potentials for future global bioenergy supply. In contrast to raising biomass harvests in forests, purpose‐grown biomass does not interfere with forest carbon (C) stocks. Provided that agricultural land can be diverted from food and feed production without impairing food security, energy plantations on current agricultural land appear as a beneficial option in terms of renewable, climate‐friendly energy supply. However, instead of supporting energy plantations, land could also be devoted to natural succession. It then acts as a long‐term C sink which also results in C benefits. We here compare the sink strength of natural succession on arable land with the C saving effects of bioenergy from plantations. Using geographically explicit data on global cropland distribution among climate and ecological zones, regionally specific C accumulation rates are calculated with IPCC default methods and values. C savings from bioenergy are given for a range of displacement factors (DFs), acknowledging the varying efficiency of bioenergy routes and technologies in fossil fuel displacement. A uniform spatial pattern is assumed for succession and bioenergy plantations, and the considered timeframes range from 20 to 100 years. For many parameter settings—in particular, longer timeframes and high DFs—bioenergy yields higher cumulative C savings than natural succession. Still, if woody biomass displaces liquid transport fuels or natural gas‐based electricity generation, natural succession is competitive or even superior for timeframes of 20–50 years. This finding has strong implications with climate and environmental policies: Freeing land for natural succession is a worthwhile low‐cost natural climate solution that has many co‐benefits for biodiversity and other ecosystem services. A considerable risk, however, is C stock losses (i.e., emissions) due to disturbances or land conversion at a later time.

A comprehensive assessment of European forest-based biomass harvest potentials, their future utilization and implications on international wood product markets and forest carbon dynamics requires the capability to model forest resource development as well as global markets for wood-based commodities with sufficient geographical and product detail and, most importantly, their interactions. To this aim, we apply a model framework fully integrating a European forest resource model and a global economic forest sector model. In a business-as-usual (BaU) scenario, European Union harvests increase seven percent by 2030 compared to past levels (485 million m3 on 2000-2012 average and 517 million m3 in 2030). The subsequent annual carbon stock change is a ten percent reduction by 2030 compared to 2000-2012 average (equal to 119.3 Tg C yr-1), corresponding to decreasing carbon-dioxide removal by the European forests. A second, high mobilization scenario (HM), characterized by the full utilization of the potential wood supply and a doubling of EU wood pellets consumption, was designed to explore potential impacts on forest carbon dynamics and international wood product markets under intensive exploitation of biomass resources. In the HM scenario, harvest increases by 55% (754 million m3 in 2030) compared to the BaU scenario. Fuelwood accounts for this increase in harvest levels as overall competition effects from increased wood pellets consumption outweighs synergies for material uses of wood, resulting in slightly reduced harvests of industrial roundwood. As expected, this increasing harvest level would significantly impair carbon-dioxide forest sequestration from the atmosphere in the medium term (-83% in 2030, compared to 2000-2012 average).

Robust estimates of CO2 budget, CO2 exchanged between the atmosphere and terrestrial biosphere, are necessary to better understand the role of the terrestrial biosphere in mitigating anthropogenic CO2 emissions. Over the past decade, this field of research has advanced through understanding of the differences and similarities of two fundamentally different approaches: "top‐down" atmospheric inversions and "bottom‐up" biosphere models. Since the first studies were undertaken, these approaches have shown an increasing level of agreement, but disagreements in some regions still persist, in part because they do not estimate the same quantity of atmosphere–biosphere CO2 exchange. Here, we conducted a thorough comparison of CO2 budgets at multiple scales and from multiple methods to assess the current state of the science in estimating CO2 budgets. Our set of atmospheric inversions and biosphere models, which were adjusted for a consistent flux definition, showed a high level of agreement for global and hemispheric CO2 budgets in the 2000s. Regionally, improved agreement in CO2 budgets was notable for North America and Southeast Asia. However, large gaps between the two methods remained in East Asia and South America. In other regions, Europe, boreal Asia, Africa, South Asia, and Oceania, it was difficult to determine whether those regions act as a net sink or source because of the large spread in estimates from atmospheric inversions. These results highlight two research directions to improve the robustness of CO2 budgets: (a) to increase representation of processes in biosphere models that could contribute to fill the budget gaps, such as forest regrowth and forest degradation; and (b) to reduce sink–source compensation between regions (dipoles) in atmospheric inversion so that their estimates become more comparable. Advancements on both research areas will increase the level of agreement between the top‐down and bottom‐up approaches and yield more robust knowledge of regional CO2 budgets. [ABSTRACT FROM AUTHOR]

Short rotation plantations are often considered as holding vast potentials for future global bioenergy supply. In contrast to raising biomass harvests in forests, purpose‐grown biomass does not interfere with forest carbon (C) stocks. Provided that agricultural land can be diverted from food and feed production without impairing food security, energy plantations on current agricultural land appear as a beneficial option in terms of renewable, climate‐friendly energy supply. However, instead of supporting energy plantations, land could also be devoted to natural succession. It then acts as a long‐term C sink which also results in C benefits. We here compare the sink strength of natural succession on arable land with the C saving effects of bioenergy from plantations. Using geographically explicit data on global cropland distribution among climate and ecological zones, regionally specific C accumulation rates are calculated with IPCC default methods and values. C savings from bioenergy are given for a range of displacement factors (DFs), acknowledging the varying efficiency of bioenergy routes and technologies in fossil fuel displacement. A uniform spatial pattern is assumed for succession and bioenergy plantations, and the considered timeframes range from 20 to 100 years. For many parameter settings—in particular, longer timeframes and high DFs—bioenergy yields higher cumulative C savings than natural succession. Still, if woody biomass displaces liquid transport fuels or natural gas‐based electricity generation, natural succession is competitive or even superior for timeframes of 20–50 years. This finding has strong implications with climate and environmental policies: Freeing land for natural succession is a worthwhile low‐cost natural climate solution that has many co‐benefits for biodiversity and other ecosystem services. A considerable risk, however, is C stock losses (i.e., emissions) due to disturbances or land conversion at a later time. [ABSTRACT FROM AUTHOR]

Cassava-based ethanol has been promoted in China and Southeast Asia as an effective means to reduce greenhouse gas (GHG) emissions and promote energy security. However, existing life cycle assessments of the environmental impacts of cassava ethanol have used highly-aggregated empirical methods to estimate ecosystem C stock changes, which do not capture finer-scale characteristics of different cassava growing regions within a country. We investigated carbon debts, GHG emissions, and payback periods for cassava-based ethanol in Vietnam using a life cycle assessment approach coupled with the widely-used ecosystem biogeochemical CENTURY model. The model simulated regionally-specific carbon stock changes associated with cassava cultivation for biofuel feedstock under different land use change, cassava yield and fertilization scenarios. We found that switching land use to cassava production for biofuel substantially reduced soil organic carbon in all major cassava growing regions in Vietnam. GHG emissions, carbon debts, and payback periods of Vietnam's cassava ethanol were strongly dependent on cassava yield. The mean carbon debt due to direct land use change to cassava production for ethanol ranged from 66 to 97 Mg of CO 2 per hectare, and the net carbon dioxide equivalent emission of cassava-based ethanol ranged from 36 to 95 g MJ −1 , depending on the range of cassava fresh weight yield (from 18 to 60 Mg ha −1 ). To repay a carbon debt from direct land use change within 25 years, the average fresh weight yield of cassava used as feedstock for ethanol production must be above 33 Mg ha −1 . [ABSTRACT FROM AUTHOR]

The purposes of this study were 1) to estimate the biomass in the mangrove forests using satellite imagery and airborne LiDAR data, and 2) to estimate the amount of carbon stock changes using biomass estimated. The study area is located in the coastal area of the South Sumatra state, Indonesia. This area is approximately 66,500 ha with mostly flat land features. In this study, the following procedures were carried out: (1) Classification of types of tree species using Satellite imagery in the study area, (2) Development of correlation equations between spatial volume based on LiDAR data and biomass stock based on field survey for each types of tree species, and estimation of total biomass stock and carbon stock using the equation, and (3) Estimation of carbon stock change using Chronological Satellite Imageries. The result showed the biomass and the amount of carbon stock changes can be estimated with high accuracy, by combining the spatial volume based on airborne LiDAR data with the tree species classification based on satellite imagery. Quantitative biomass monitoring is in demand for projects related to REDD+ in developing countries, and this study showed that combining airborne LiDAR data with satellite imagery is one of the effective methods of monitoring for REDD+ projects. [ABSTRACT FROM AUTHOR]

Up-to-date and accurate information of ecosystem state and functioning becomes ever more critical for decision-making and policy. For complex ecosystems such as forests, these demands can in some cases not be met by field observations only, especially at larger scales. Additionally, methodological requirements include comparability and transparency. To satisfy these needs, models can provide an important supplement or alternative. We examined the validity of the litter decomposition and soil carbon model Yasso07 in Swiss forests based on data on observed decomposition of (i) foliage and fine root litter from sites along a climatic and altitudinal gradient and (ii) of 588 dead trees from 394 plots of the Swiss National Forest Inventory. Our objectives were to (i) examine the effect of the application of three different published Yasso07 parameter sets on simulated decay rate; (ii) analyze the accuracy of Yasso07 for reproducing observed decomposition of litter and dead wood in Swiss forests; and (iii) evaluate the suitability of Yasso07 for regional and national scale applications in Swiss forests. From the three examined parameter sets, the set was identified which resulted in the best agreement between Yasso07 results and observed decomposition. No significant differences were found between simulated and observed remaining C in foliage and fine root litter after 10 years and in lying dead trees after 14–21 years. The model overestimated the decomposition of standing dead trees. We concluded that Yasso07 can provide accurate information on temporal changes in C stocks in litter and deadwood in Swiss forests in a transparent manner that is valid for, e.g., reporting purposes under the UNFCCC and the Kyoto Protocol. [ABSTRACT FROM AUTHOR]

Highlights: [•] A comprehensive evaluation of alternative LUC and fertilization schemes. [•] The GHG intensity of palm oil greatly depends on the LUC scenario. [•] Colombian palm area expansion resulted in negative or low palm oil GHG intensity. [•] GHG emissions from plantation vary significantly with N2O emission parameters. [ABSTRACT FROM AUTHOR]

International agreements require countries to annually report on greenhouse gas emissions and removals. For the land-use sector, this includes estimating stock changes in various carbon pools. For carbon pools like mineral forest soil where a country-level statistical inventory based on measurements is very difficult, models are usually applied together with data from case studies. In this paper, we present a country-specific model together with case studies that aim at capturing major soil processes due to forestry activity. These processes include “hot moments”, e.g., disturbances that occur rarely but might result in relatively high emissions. The model only aims at developing a conservative estimate, rather than a central one, of net country-level carbon stock change with emissions overestimated and removals underestimated. The model is partially parameterised using paired sampling of soil organic carbon in the uppermost 30-cm layer, applying standard methods including those suggested by IPCC, in afforestations on former croplands and in artificial regenerations. Results show that soils of afforested croplands act as a sink, and carbon stock after regeneration might decrease due to disturbance by forest operations, but might also increase due to transfer of carbon from dead roots to soil depending on disturbance levels. The estimation at the country level, which involves additional considerations and data from the literature, suggests that overall, forest soils are a net sink in Hungary, but also that artificially limiting soil organic carbon changes estimation to the uppermost 30-cm layer as applied in the IPCC methodology might lead to artefacts. [ABSTRACT FROM AUTHOR]

We assessed the environmental impacts of producing bread, as delivered to the consumer, assuming the use of Norwegian ingredients only. Ten impact categories, including global warming potential (GWP), were quantified by mixed modelling and life cycle assessment. Firstly, we quantified the impacts of growing barley, oats, winter and spring wheat on 93 farms that were representative of the main cereal production regions in Norway. We used wide system boundaries, which included all relevant processes occurring both pre-farm and on-farm. Secondly, we assessed a representative production chain for bread, including transport, milling, baking and packing processes. On-farm processes accounted for a large share of the environmental impacts attributable to the production of bread (e.g. 50% for GWP). There is thus considerable potential for environmental improvements through changes in farm management. In total, the GWP per kg of bread (freshweight) was 0.95 kg CO2-equivalent. The environmental footprint of transport was small. [ABSTRACT FROM PUBLISHER]

In recent days, climate change is a main global problem which concerns the scientists. Climate change is defined as the change of global mean temperature. While the reasons for global climate change were investigated, the effect of CO2, which is the most important greenhouse gas, was found. In the world the oceans and forests are the biggest carbon sinks. So the forests are very important for climate change and carbon storage. 75% of carbon was stocked by plants in the forests. In this study, we aimed to determine the carbon stock change between 1970 and 2009 in Saraycık Forest Management Chiefdom using the carbon conversion coefficient calculated for Turkey's forests by Asan (2002). The stand type maps and forest management plans produced in 1970 and 2009 were digitized for calculating the carbon stock. In addition the carbon stock maps were produced with ArcGIS 10.0 software. [ABSTRACT FROM AUTHOR]

There is a need to assess climate change mitigation opportunities in forest sector in India in the context of methodological issues such as additionality, permanence, leakage and baseline development in formulating forestry mitigation projects. A case study of forestry mitigation project in semi-arid community grazing lands and farmlands in Kolar district of Karnataka, was undertaken with regard to baseline and project scenario development, estimation of carbon stock change in the project, leakage estimation and assessment of cost-effectiveness of mitigation projects. Further, the transaction costs to develop project, and environmental and socio-economic impact of mitigation project was assessed. The study shows the feasibility of establishing baselines and project C-stock changes. Since the area has low or insignificant biomass, leakage is not an issue. The overall mitigation potential in Kolar for a total area of 14,000 ha under various mitigation options is 278,380 t C at a rate of 20 t C/ha for the period 2005-2035, which is approximately 0.67 t C/ha/year inclusive of harvest regimes under short rotation and long rotation mitigation options. The transaction cost for baseline establishment is less than a rupee/t C and for project scenario development is about Rs. 1.5-3.75/t C. The project enhances biodiversity and the socio-economic impact is also significant. [ABSTRACT FROM AUTHOR]

Chinese fir is an important timber species in Southern China. Millions of hectares of Chinese fir plantations have been established during the past decades, and increasing areas are re-planted as second or later rotations. Since the 1980s, scientists and forest managers have reported apparent yield decline and soil fertility degradation over successive rotations. If this yield decline is accompanied by a reduction in carbon stocks, and if it is caused by management, it has to be considered as a form of forest degradation and will become important from a carbon-accounting point of view.In this paper we have collected and compiled published data relevant to growth and soil properties over successive rotations of Chinese fir, calculated the stocks of stand biomass and soil organic carbon and nitrogen, and have analyzed the impact on carbon stocks of growing Chinese fir over successive rotations.We found that on an average, stand biomass increments were reduced by 24% from the first to the second rotation, and by a further 40% from the second to the third rotation. Soil organic carbon was reduced by 10% and 15% between the first and second, and second and third rotations, respectively. Soil carbon losses were usually accompanied by nitrogen losses but carbon losses were generally larger than nitrogen losses. The reduction of carbon stocks in successive rotations appears to be related to increases in soil bulk density and nutrient losses caused by burning of residues during site preparation.Potential afforestation/reforestation projects under the clean development mechanism (CDM) of the Kyoto Protocol may also need to consider the possible yield decline and soil degradation over successive rotations of plantations. [Copyright &y& Elsevier]

Abstract Data from 57 permanent monitoring sites are used to document the growth in woody vegetation and estimate the carbon sink in 27 M ha of eucalypt woodlands (savannas), contained within c. 60 M ha of grazed woodlands in Queensland (northeast Australia). The study sites are shown to be representative of the environment and structure of the eucalypt woodlands in the defined study area. Mean basal area increment for all live woody plants in 30 long-term sites, with an average initial basal area of 11.86 ± 1.38 (SE) m2 ha-1 , was 1.06 m2 ha-1 over a mean 14 years timeframe. The majority of the measurement period, commencing between 1982 and 1988, was characterized by below-average rainfall. The increase in live tree basal area was due primarily to growth of existing trees (3.12 m2 ha-1 ) rather than establishment of new plants (0.25 m2 ha-1 ) and was partly offset by death (2.31 m2 ha-1 ). A simple but robust relationship between stand basal area and stand biomass of all woody species was developed for the eucalypt dominant woodlands. Analysis of above-ground carbon stocks in live and standing dead woody plants gave a mean net above-ground annual carbon increment for all 57 sites of 0.53 t C ha-1 y-1 , similar to values estimated elsewhere in world savannas. Published root : shoot ratios were used to infer C flux in woody root systems on these sites. This results in an estimated sink in above- and below-ground biomass of 18 Mt C y-1 over the eucalypt woodlands studied, and potentially up to 35 Mt C y-1 if extended to all grazed woodlands in Queensland. It is suggested that introduction of livestock grazing and altered fire regimes have triggered the change in tree-grass dominance in these woodlands. Thus, change in carbon stocks in the grazed woodlands of Queensland is identified as an important... [ABSTRACT FROM AUTHOR]

The expected increased share of renewables due to the ongoing energy transition may reduce the estimated potential mitigation effect of wood. Here, we estimated the climate change mitigation effect for five scenarios of wood products use in Europe applying dynamic substitution factors embracing a future energy mix with an increasing share of renewables in accordance with the emission reductions necessary to achieve the Paris Agreement targets. Our innovative modelling approach also included the elimination of eternal recycling loops, the inclusion of more realistic wood use cascading scenarios, and adoption of a more realistic marginal (ceteris paribus) substitution approach. Results show that the mitigation effect derived from material substitution is 33% lower in 2030 than previously predicted, and even 96% lower in 2100, showing its expiry date by the end of the century. Nevertheless, the mitigation effect of wood product use, in addition to mitigation by forests, may represent 3.3% of the European emission reduction targets by 2030. [Display omitted] • A dynamic modelling approach was used to simulate carbon balance in wood products. • Policies should promote the use of long-lived wood products and cascade chains. • Mitigation effect from substitution 33% lower in 2030 than previously predicted. • Mitigation effect from substitution 96% lower in 2100 than previously predicted. • Mitigation by substitution has an expiry date by the end of the century. [ABSTRACT FROM AUTHOR]

• Change in land use has changed the carbon and nitrogen stocks in the soil. • The permanence time of the cover crops affected the carbon dynamics in the soil. • The stable fraction of organic matter contributed to soil carbon accumulation. Soil organic carbon (SOC) is the main terrestrial carbon reservoir. The spatial distribution of SOC is mainly regulated by environmental factors and anthropogenic activity. Historically, land-use change has been responsible for much of the world's greenhouse gas emissions. However, carbon storage can be improved through agricultural management, either by increasing carbon inputs through higher crop yields or by delaying the release of carbon into the atmosphere. Therefore, it is essential to correctly choose which cover crops to use and how long these crops will remain in the crop rotation system. In order to clarify these issues, the objective of this work was to evaluate the chemical quality and changes in soil carbon stock between no-tillage and soybean/corn crop succession systems and integrated crop-livestock systems. The study was carried out in Querência (MT), Brazil, located in the Cerrado/Amazon ecotone. The integrated systems were studied between 2007 and 2014. These systems followed annual crop rotation and forage rotation, and were laid out as a consortium or were isolated, with the aim of producing grain and meat. In order to determine soil chemical properties, 170 sampling plots were randomized among the treatments. The highest carbon values were observed in the integrated crop-livestock systems (ICL), probably due to higher inputs of organic residues. Crop rotation with plants of different families allowed better land use and soil exploration. However, integrated production systems comprising longer pasture periods, no-tillage, and soybean/corn crop succession decreased soil carbon. When comparing carbon stocks from 2010 to 2014, only integrated production systems ICL2 and ICL3 were able to increase carbon stocks. Our results highlight the importance of length of growing and correct choice of cover crops in maintaining SOC stocks, and thus provide insights to improve SOC stocks in agricultural land in Brazil's main producing region. [ABSTRACT FROM AUTHOR]

Robust estimates of CO 2 budget, CO 2 exchanged between the atmosphere and terrestrial biosphere, are necessary to better understand the role of the terrestrial biosphere in mitigating anthropogenic CO 2 emissions. Over the past decade, this field of research has advanced through understanding of the differences and similarities of two fundamentally different approaches: "top-down" atmospheric inversions and "bottom-up" biosphere models. Since the first studies were undertaken, these approaches have shown an increasing level of agreement, but disagreements in some regions still persist, in part because they do not estimate the same quantity of atmosphere-biosphere CO 2 exchange. Here, we conducted a thorough comparison of CO 2 budgets at multiple scales and from multiple methods to assess the current state of the science in estimating CO 2 budgets. Our set of atmospheric inversions and biosphere models, which were adjusted for a consistent flux definition, showed a high level of agreement for global and hemispheric CO 2 budgets in the 2000s. Regionally, improved agreement in CO 2 budgets was notable for North America and Southeast Asia. However, large gaps between the two methods remained in East Asia and South America. In other regions, Europe, boreal Asia, Africa, South Asia, and Oceania, it was difficult to determine whether those regions act as a net sink or source because of the large spread in estimates from atmospheric inversions. These results highlight two research directions to improve the robustness of CO 2 budgets: (a) to increase representation of processes in biosphere models that could contribute to fill the budget gaps, such as forest regrowth and forest degradation; and (b) to reduce sink-source compensation between regions (dipoles) in atmospheric inversion so that their estimates become more comparable. Advancements on both research areas will increase the level of agreement between the top-down and bottom-up approaches and yield more robust knowledge of regional CO 2 budgets., (© 2019 John Wiley & Sons Ltd.)