the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Assessing terrestrial carbon fluxes and stocks in South America and its major biomes using CMIP6 Earth System Models
Abstract. South America plays a significant role in the global carbon cycle, with its ecosystems storing substantial amounts of carbon in vegetation and soils. This study analyses the behaviour of components of the carbon cycle and stocks in the whole continent and two of its major biomes, the Amazon and the Savannas, using a set of 18 Earth System Models (ESMs) from the Coupled Model Intercomparison Project Phase 6 (CMIP6). We discuss the variability of the model simulations throughout the 20th and first 20 years of the 21st centuries. Results show that South America accounts for 25–30 % of the global Gross Primary Productivity (GPP), 21–28 % of the global Net Primary Productivity (NPP), 17–50 % of the global Net Ecosystem Productivity (NEP), and 15–30 % of the global Net Biome Productivity (NBP), and also contributes significantly to global autotrophic (Ra) and heterotrophic (Rh) respiration. The temporal evolution of NBP in South America indicates a combination of the values estimated for the Amazon and the Savannas, with most models showing a small decreasing trend in the 20th century, likely dominated by emissions from land use change, and shifting to positive values after 1990, likely driven by an increasing productivity in response of atmospheric CO2 fertilization. Comparing the Amazon and Savannas, apart from the magnitude of the fluxes, we see similarities in both ecosystems responses when widespread dry years occur, with higher NBP and GPP in wet years and higher Rh and disturbances in dry years. These results highlight the vulnerability of South America to climate change, with the potential for parts of the continent to shift from carbon sinks to sources under widespread droughts.
- Preprint
(2041 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-942', Anonymous Referee #1, 08 May 2025
Comments:
The author provides a brief comparison of 18 Earth System Models (ESMs) from CMIP6 in simulating carbon cycle variables over South America, including global Gross Primary Productivity (GPP), Net Primary Productivity (NPP), Net Ecosystem Productivity (NEP), Net Biome Productivity (NBP), as well as autotrophic (Ra) and heterotrophic respiration (Rh). The study highlights both the consistencies and discrepancies among the model simulations. However, this work cannot be considered a proper assessment or evaluation. Firstly, it lacks validation against observational data. Secondly, it does not employ rigorous statistical analyses to support the inter-model comparisons. For example, the use of a Taylor diagram would be an appropriate and widely accepted approach for evaluating model performance. Given these shortcomings, the manuscript is not currently suitable for publication. I recommend rejecting the submission.
Detailed comments:
Line number: Please use consecutive line numbers.
Line 15: Please also show the fraction of global autotrophic (Ra) and heterotrophic (Rh) respiration.
Line 25: “Carbon is a critical element in the Earth system.”, please give the reference.
Line 26-28: This sentence also need corresponding reference.
Line 35: Suggest to introduce both biogeophysical and biogeochemical paths, such as changing the energy balance on land surface.
Line 43: It's typo? Was it should be 0.076 PgC?
Line 60-64: Please provide more information about the model experiments, especially the MIPs (maybe DECK, historical and CORDEX) that related to this study.
Line 65: “discuss”, it will be better to use evaluate or assess.
Line 72: Suggesting to put figure 1 into supplementary materials.
Line 73: “120 PgC yr-1”, please give the reference.
Line 82: The authors might indicated the carbon loss by hydrological process such as leaching, “water fluxes” makes people misunderstanding.
Line 84: It was used as “cVeg” instead of “cVEg”.
Figure 1: Before explain detail of this figure, please give a general figure description.
Table 1: Please move this table to supplementary materials, for there are too many figures and tables in the manuscript.
Line 101-108: This paragraph seems like discussion, please move it to discussion or just delete it.
Line 134: “km2”, please use superscript.
Line 138-140: This sentence is hard to understand.
Figure 2: The author did not clearly specify in the manuscript how the Amazonian biome area and the South American Savannas biome areas were defined, which may cause confusion for readers. It is recommended to add a corresponding legend in Figure 2 to clarify the biome boundaries and to improve the image resolution to enhance readability.
Line 153: Where is the results? If the authors mentioned some result, please show it in main text or supplementary materials.
Line 165: Is this result supported by any statistical analysis?
Line 173: Also need statistical evidence.
Figure 3: what is the yellow dash line and horizontal line represent respectively?
Citation: https://doi.org/10.5194/egusphere-2025-942-RC1 -
RC2: 'Comment on egusphere-2025-942', Anonymous Referee #2, 16 May 2025
General Comments:
This manuscript assesses key components of the terrestrial carbon cycle—GPP, NPP, NEP, Ra, and Rh—over South America, with a focus on the Amazon and savanna biomes. Using output from 18 Earth System Models, the authors compare carbon fluxes across models and against literature benchmarks (e.g., CARDAMOM), and also examine differences between ‘dry’ and ‘wet’ years.
The topic is relevant and the multi-model perspective is potentially valuable, particularly for carbon cycle assessments in tropical ecosystems. However, the analysis and presentation are currently too limited and underdeveloped. Significant restructuring and clarification would be required to reach publishable quality. While I hope the comments provided are useful for future revisions, I unfortunately recommend rejection in the current form.
Specific comments:
Structure and framing of the manuscript:
- It’s not entirely clear what the main ‘results’ of this paper are and often the framing changes throughout. For instance, I would suggest the novelty of this paper comes from the attempted assessment of how well models are able to simulate carbon components in South America & its biomes, and how the models compare against one another for the region and in wet/dry years. Yet the abstract doesn’t mention the outcome of such an assessment but rather stating the patterns of the temporal evolution (which could be argued, isn’t really anything new).
- On a similar note, the introduction lacks a justification for the research presented, what are the research gaps and why did the authors feel that this work should be done? What is the novelty? The paragraph starting at L101 under Earth System Models is trying to argue this I think, but this should be earlier on in the manuscript and argumentation improved. The several lines on the 2024 burning seems somewhat unnecessary, more context should be added (eg. This shows vulnerability and fast changes that can occur in terrestrial ecosystems and their carbon cycles etc) or it should be removed.
- Results/Discussions and Conclusions could be better organised. By the conclusion, several points have been repeated numerous times.
Models:
More information is needed on the models presented throughout this study and likely some further analysis is required on the following points:
- 18 ESMs are used according to data availability. However, the models selected greatly vary in their set up and functionality, especially related to carbon dynamics and terrestrial processes, which will be influencing the results you show and discuss. For example, the EC-Earth3-Veg model used is a configuration that does not include ocean biogeochemistry and therefore does not have a fully coupled carbon model activated whilst other models such as CESM2 does have a fully coupled carbon model on. On the other hand, I believe that in the French IPSL-CM6A-LR model, ORCHIDEE does not have the dynamic vegetation scheme activated, which limits PFT responses and hence carbon dynamics. Whilst it is completely understandable to not have descriptions of each of the ESMs, Table 1 should be expanded to have more details of differences in the models or something similar added to SI. I also think there needs to be a discussion on how these differences impact the results added.
- More details should be added about what CARDAMOM is and how it works. Looking at late figures in the manuscript (e.g. Fig.8), it looks like CARDAMOM disagrees with other estimates in literature. As a reader I therefore have no indication about how ‘good’ CARAMOMs outputs are. Therefore, a couple of sentences about CARDAMOMS performance or sources for CARDAMOM evaluations would help.
- Simulation details: More details on the simulations that generated the ESM output in this study would be beneficial under the Earth System Models/Methods section. For example, what data was used to drive land surface change in the models? This would help later in the discussions of land use change impacts on NBP – do the model simulations include the land use changes discussed from L396 onwards? How?
Biome definition:
- How are the biomes presented in Figure 2 defined? Did the authors select the regions based on observations/literature/models? Figure 2 looks like perhaps it is from CARDAMOM - this should be added to the methods and figure caption.
- The next related question is, how are these regions implemented in the analysis. Presumably each model will have slightly different simulated regions for the 2 biomes and also different PFTs defined within them which then influences the carbon fluxes and stocks calculated in the results. Some spatial plots of the models either in the results or as an SI figure are needed to see how spatially different the vegetation in the models are and how well they align with Fig 2. For instance, dominant PFT in each gridcell or vegetation biomass.
NEP: The sentence defining NEP/NEE (L78) is worded in a way that suggests NEP is always related to forests, when perhaps the authors meant just the net flux of CO2 from a given ecosystem? It should also be made clear how NEP was calculated in the analysis, i.e was it calculated by looking at the net ecosystem flux for given spatial definitions (e.g masks such as in Fig.2) or is it calculated only from gridcells that contain ‘forests’ in the model outputs (for instance where tree cover >60% or similar). How is NEP calculated for the whole of south America (Fig.3)?
L152: In your selection of Wet/Dry years, the author writes that comparing observational years with simulation meteorological conditions, the results were mostly inconsistent. However, this is what is expected with ESMs. Models have their own internal variability (natural, chaotic fluctuations in the climate system that occur even under the same external forcings) and therefore timing of specific events, such as wet and dry years is often don’t align with observations. It is therefore good practice not to compare observations with specific years in most ESMs. I would change the lines 152-155 to explain this and frame your approach accordingly.
Figures: In general, the size of figures should be increased relative the legend text size. The horizontal line around zero (yellow dashed line) should be explained in the captions, changed colour or removed.
- Fig 6: Where does the big range in carbon pools come from? Again, this could maybe be explained by spatial vegetation plots.
Wet/Dry years: Why did the authors choose not to do the wet/dry years analysis also for the whole of South America as they did with the temporal fluxes?
Section 4.3: One could argue that the model results do not match well with the reference data in all cases. For example, in Figure 10. GPP and Rh reference data for the Amazonian do not match with the ensemble of models, yet the authors indicate otherwise. Perhaps a better explanation is needed here.
Technical comments:
- L45 lacks a comma after factors. I would also avoid the use of ‘probably’ here.
- L393: EC-Earth3-Veg is listed as a model whose coupled surface model is CLM which is not the case.
Citation: https://doi.org/10.5194/egusphere-2025-942-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
163 | 47 | 9 | 219 | 13 | 14 |
- HTML: 163
- PDF: 47
- XML: 9
- Total: 219
- BibTeX: 13
- EndNote: 14
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1