the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Sep 2023
Elevated atmospheric CO2 and vegetation structural changes contributed to GPP increase more than climate and forest cover changes in subtropical forests of China
Félicien Meunier
Marc Peaucelle
Guoping Tang
Ye Yuan
Hans Verbeeck
Abstract. Abstract: The subtropical forest gross primary productivity (GPP) plays a pivotal role in the global carbon cycle and in regulating the global climate. Quantifying the individual and combined effects of forest cover change (FCC), vegetation structural change (VSC, i.e., leaf area index (LAI)), CO2 fertilization, and climate change (CC) on annual GPP dynamics of various subtropical forest types are essential for mitigating carbon emissions and predicting climate changes, but these impacts remain unclear. In this study, we used a processed-based model to comprehensively investigate the impacts of these factors on GPP variations with a series of model experiments in China’s subtropical forests during 2001–2018. Simulated actual GPP showed a significant increasing trend (26.72 TgC year−1, p < 0.001) under the interaction effects of these factors. The CO2 fertilization (8.23 TgC year−1, p < 0.001) and VSC (4.55 TgC year−1, p = 0.005) were the two dominant drivers of total subtropical forest GPP increase, followed by the effect of FCC (1.35 TgC year−1, p < 0.001) and CC (1.11 TgC year−1, p = 0.08). We observed different responses to drivers depending on forest types. The evergreen broadleaved forests have a high carbon sink potential due to the positive effects of all drivers. Both the FCC (1.29 TgC year−1, p < 0.001) and CC (0.53 TgC year−1, p < 0.05) significantly decreased evergreen needleleaved forest GPP, while their negative effects were almost offset by the positive impact of VSC. Our results indicated that forest structural change outweighed the forest cover change in promoting GPP, which is an overlooked driver that needs to be accounted for in studies, as well as ecological and management programs. Overall, our study offers a novel perspective on different drivers of subtropical forest GPP changes, which provides valuable information for policy makers in forest management to mitigate climate change.
- Preprint
(7629 KB) - Metadata XML
-
Supplement
(6510 KB) - BibTeX
- EndNote
Tao Chen et al.
Status: final response (author comments only)
-
RC1: 'Comment on bg-2023-140', Anonymous Referee #1, 29 Oct 2023
In their study, the authors investigate the influence of different drivers on changes in GPP in subtropical forests in China. The considered drivers were climate change, forest cover change, change in vegetation structure, and changes in CO2 concentrations.
The authors use the BEPS model and run multiple simulations to disentangle the impact of the different drivers and find that atmospheric CO2 and vegetation structure play the most important roles.This is an interesting and well-conducted study and the manuscript is decently written. In my opinion, this study can be published in Biogeosciences after it went through some major revisions.
I mainly find that there needs to be some more model evaluation. Furthermore, some results need to be explained better. Also, the discussion has some points that need to be made clearer or added (see details below).
I'd further suggest some streamlining of results, figures, and text. There are 10 Figures, often with 6 panels. I believe this could be made more concise.
Further detailed comments follow below.
Abstract:Why call it VSC and not just LAI?
Introduction
l. 39: the statement about the 30% is not a result of the cited study and is also not cited there... Please find a better reference
l. 55: should be 0.82 billion I guess.
l. 59: is this compared to global surface temp or temp over land?
Methods:
l. 103: what about the spread of temperature as you mention for precipitation?
l. 115: NEP was not introduced. Generally, a glossary with abbreviations would be helpful.
l. 116: some more text on the model is necessary to allow the reader to get a basic understanding of it. It may go into the supplements.
l. 148: flux partitioning is not quality control.
l. 151: ER not introduced
l. 170: I am not an expert on this. Any reason why GOSIF was not used? I thought this would be the state-of-the-art GPP product.
l. 210: this reads strange. In S1 the land cover is fixed. But then you write that "in this scenario, LCC may lead to changes..."
l. 212: this is confusing. You talk about the conversion of forest to non-forest, and then about forest cover change. Is that not the same thing?
Improve Table S3, explain more. What is remote sensing, what is modeled, etc.
Results:
The model performance section is very good.
But only GPP is evaluated. What about other model outputs?
Also, Fig 3 does not really convince me. Can you discuss why the GPPs are so different?l. 242: typo: "203-2010"
l. 240-245: any explanation as to why some of the sites are performing much better? R2 as low as 0.43 in one site, up to 0.85 in another
Fig 2: do you have any explanation about the small bias in DHS at low observed values? This is also visible in all years in the supplements.
Fig 2: The caption misses that the dots are observations
l. 266: This is an issue: obviously the increase in GPP is similar in a study with the same model. The next data product has a much lower increase, 0.017, compared to this study's 0.026.
BEPS simulates a higher GPP compared to all the other products, and a higher trend, too. This needs to be discussed further.l. 276: what do you mean by simulated actual GPP?
l. 280: grammar
l. 290s: streamline this section to make clear that the change in GPP comes from the increasing/decreasing areas
l. 305: In section 3.3.2, the point needs to be better explained that although climate change contributes to a 1.11 TgC/year most of the area has a decreasing trend. This increase seems to stem from a small region in the west. What is happening in this region? E.g. Fig 6b
l. 346: why is LAI increasing at all?
l. 349 and in general: The wording "Especially, the positive effect of VSC on EBF" is strange. I mean, the VSC change inside the EBF and that led to a change in GPP in those forests.
Fig S10: There is a rapid increase in trend around 2011. Why is that? Also, how does LAI look in the model pre-2000?
l. 355: You write:
"results showed that most GPP increases in China’s subtropical forests due to the increase of LAI, which also offset the negative effects of VSC on GPP, thus allowing VSC to play a key driving factor in promoting GPP increases throughout the forest area.""This is confusing. Did you mean FSC maybe instead of VSC at the first mention? LAI is the same as VSC, right? So how does the effect of change in LAI on GPP offset the effect of change in LAI on GPP? They are the same thing? Or do you mean, there is more positive change that heavily offsets the negative changes?
l. 361: verb is missing
Fig. 9: This is a nice figure that shows the main results.Fig 9: I am puzzled that, e.g., in b) CC-ALL is nowhere near the sum of the three. I understand that there will be interactions, but I find it quite strange that the interactions are quite strongly positive but each of the components is almost 0. Maybe these cancel each other out over the entire region.
Results: when you describe the changes for each of the forest types, I believe the results stem solely from the changing areas. It would be better to show the changes on a per-area basis or in the simulations even keep the forest cover stable...Discussion:
l. 416: "which is mainly converted from cropland". You need to elaborate here. Croplands can be highly productive. A few models even indicate that in some regions in China, cropland could potentially be more productive than forests in terms of GPP (Fig. 3 in https://doi.org/10.1038/s41598-022-23120-0). To back your claim, can you provide some numbers here on GPP values of the crops that have been reforested?
l. 419: what do you mean by the negative effect of a specific forest type on forest GPP variations? That the planting of a certain forest type may result in a lower GPP than the previous land cover? Or something else?
Generally in this section, you need to be careful with the wording as you refer to "forest GPP" most of the time, but sometimes you mean the GPP of the entire area.
l. 447ff: citations for the claim? Also, drought relates more to precipitation, maybe you can instead mention increased VPD as a result of a high temp increase.
l. 450 mention again the magnitudes. They should explain that the smaller area of increase outweighs the larger area of decrease
l. 461: why is that?
l. 488: Forest protection has greater carbon uptake potential than what? This also relates to my comment on l. 416. Also, you only refer to GPP. Can you make any claims on NPP?
section 4.1.4: here I also find that some discussion on the relation of GPP to carbon sequestration is missing.
Conclusionl. 560ff: I am not sure about this last concluding statement. You basically show that changes in the vegetation structure have a strong impact on GPP. You don't show anything about NPP or NEE. I would doubt that the growth of an entire new forest would have a lower impact on the carbon balance than improving the current ones. At least this claim cannot be made based on your work.
Citation: https://doi.org/10.5194/bg-2023-140-RC1 - AC2: 'Reply on RC1', Tao Chen, 25 Nov 2023
-
RC2: 'Comment on bg-2023-140', Anonymous Referee #2, 30 Oct 2023
Review of “Elevated atmospheric CO2 and vegetation structural changes contributed to GPP increase more than climate and forest cover changes in subtropical forests of China” by Chen et al.
The manuscript by Chen et al. investigates drivers of subtropical forest GPP trends in China using a process-based model that runs to provide causal attribution. The study concludes that the primary drivers of GPP change are the CO2 fertilization effect and increased LAI. While the study conducts comprehensive model experiments and maintains a well-organized structure, it lacks a convincing theoretical framework for designing the experiments and conducting the analysis, which is essential for consideration in publication. Additionally, the manuscript requires careful revision for the English language and logical syntax. Please refer to my comments for further details.General comments:
- Introduction: In the second paragraph, several relevant drivers are listed, followed by the research question “the relative contributions of these factors…not clear” in the next paragraph. It does not adequately explain to the reader why these factors are crucial to GPP or provide mechanistic expectations. For instance, in Line 60, rather than stating the increased temperature “has also influenced the forest carbon uptake”, it would be beneficial to summarize the specific mechanisms and reasons behind this influence. Is the influence positive or negative? Some clarifications would be helpful.
- Experiment design: I have two main concerns concerning the experiment design in Table 1. A) When assessing the effect of climate variables on GPP, one of the climate variables (e.g., precipitation) is fixed as the value in 2001 in the forcing for the S2 scenario. As I understand it, that means in the S2 scenario there is no climatological cycle at all. The difference in GPP between S2 and the control run should include the effect of both the long-term trend and short-term variabilities of climate. This means, by design, the trend of GPP driven by climate is overshadowed by the shorter-term variabilities (Figure 6). However, when designing the CO2 and LAI scenarios, the difference of CO2 or LAI forcings are less variable (Figure S10, S11), thus a “clear” trend of GPP can be observed in both Figure 7 (a) and 8 (a). There is no surprise when the authors find that CO2 and LAI are the most prominent drivers, when they are comparing the effect of “trend” (e.g., CO2) and “trend + variabilities” (e.g., precipitation). One may need to test to which extent the way of prescribing climate forcings influences the conclusion, e.g., by removing the trend of climate variables but keeping variabilities. B) Is the GLASS LAI also sensitive to climate change and increasing CO2? With an increased carbon uptake due to increasing CO2, more carbon can be allocated to leaf growth. I wonder if the authors have some thoughts about the causal link when discussing the effect of LAI on GPP.
- Results: This study compares the contribution of different drivers to GPP in the unit of TgC/year (e.g., Figure 9). It is not introduced in the method section how the total GPP is calculated. If I assume GPP in TgC/year is the sum of GPP from all regions or the sum for each PFT, then it is highly related to the specific regions. Figure 1 shows the study region is mostly occupied by EBF and ENF, there is no wonder GPP is higher in TgC/year in EBF. In addition to that, the title indicates that CO2 and LAI contribute more to GPP than forest cover changes. However, only very few regions are affected by forest cover change (Figure 5c), by contrast, all of the regions are under increasing CO2 in the model experiment. It is unfair to compare the relative impact between these two drivers when looking at the total GPP. Or one has to make it clear in the beginning, that only total GPP in this specific region is considered.
- Discussion: I like they discuss the model uncertainties. Most of the model discussion is about the input data, though it is important, the inherent model structure and underlying assumptions and how would these possibly affect the attribution is not so well discussed. For instance, it is not clear how the model simulates plants’ response to CO2. It would greatly enhance the understanding of the contribution results if the authors included more discussion on these elements.
Specific comments:
- L16: If you only use LAI to represent vegetation structural change, it might not be necessary to mention "VSC" explicitly.
- L29: Please be consistent with abbreviations.
- L30: What do you mean by “overlooked”?
- L32: How might these findings inform climate change mitigation efforts or forest management strategies?
- L37: Carbon emissions?
- L66-68: Which regions are they looking at? The major drivers on GPP vary a lot depending on regions and even seasons. Please be precise here.
- L70-71: The term “CO2 fertilization” has not been introduced. Do you mean the CO2 fertilization effect is stronger in China than in other regions, or the CO2 effect is stronger in forest ecosystems than in other ecosystems?
- L73-74: “…most of the current studies…”, really? At least different PFTs are represented in land surface or earth system models.
- L86: How “better-performed” is BEPS? It seems unusual to encounter the conclusion without having reviewed the results, where the performance of the BEPS model has been tested.
- L93: Do you mean different GPP products?
- L95-96: I find this statement not specific. Also, see my comment before.
- L139: What are “the other parameters”?
- L147-149: How is the “nighttime flux correction” done? Gap filling and flux partitioning are not data quality control.
- L150: Which u* is used for each site?
- L167: Vague statement. What does “robust enough” mean?
- L195: You mean “original vegetation classes”?
- L210-213: The sentence is not clear.
- L244: “reasonably well” is not an accurate phrasing, notably considering that all R2 values are below 0.5. Why is NEP only used for testing model performance? Why is NEP exclusively used for testing the model's performance? There seems to be a lack of additional results or discussion regarding NEP thereafter.
- What do the green lines and circles represent in Figure 2?
- L254-255: It is not clear how the spatial correlation is calculated.
- L261-264: The number does not align within the range of all five GPP products as mentioned. Additionally, the reference to 'another BEPS' requires clarification. How to interpret the difference between “another BEPS” and “this BEPS” in Figure S7d?
- L268-269: Rather than a simple conclusion that BEPS-GPP aligns well with other GPP products, it would be more informative to delineate areas of agreement and disagreement between the models.
- L277: Please explain what is the “interactive effect”.
- L281: “…of the forest GPP”, do you mean forest areas showed increased and decreased GPP?
- L297: What is “stable state”? No forest cover change? Or no significant effect of forest cover change?
- In Figure 5 (b), the time series of GPP in MXF seems to be very symmetric with GPP in ENF, any explanations for that?
- L307: Is the increasing trend significant?
- L334: “…58.2% of the…”, but quite a lot of white spaces are shown up on the map. How is the 58.2% derived? Are you referring to Fig. 6h in this statement?
- In Figure 6a, most of the variabilities are from EBF, any explanations?
- L381-383: Where does the conclusion “…EBF…has the highest carbon uptake potential” come from?
- L423-424: But in Table S6, the majority of the ENF has been observed to transition into MXF (19040 km2).
- L450: Could you explain how climate warming negatively influences GPP in your study?
- L460-462: Why do you observe different behaviors between EBF and ENF? Any hypothesis for that?
- L486-L488: How much increase in LAI is related to the forest protection projects?
- L495: Chen et al. attribute drivers to GPP in gC/m2/year, which is not comparable with the GPP attribution in this study because of different regions and units as I mentioned in the general comments. The results in Zhan et al. stem from a land surface model instead of eddy covariance records.
- L515-517: “…still in the early developing stage…” Could you specify the limitations of using this Vcmax25 product? Is the limitation about the theory or data quality?
- Kindly utilize diverging color schemes with the midpoint at 0 for clarity.
- I suggest minimizing the use of abbreviations in the conclusion for better clarity. If necessary, they can be reintroduced.
Technical corrections:
- L164: “yearly” means “from year to year”.
- L470: “increase” instead of “improve”.
Citation: https://doi.org/10.5194/bg-2023-140-RC2 - AC1: 'Reply on RC2', Tao Chen, 25 Nov 2023
Tao Chen et al.
Tao Chen et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 256 | 78 | 14 | 348 | 39 | 7 | 11 |
- HTML: 256
- PDF: 78
- XML: 14
- Total: 348
- Supplement: 39
- BibTeX: 7
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1