Peer Review File Manuscript Title: Net greenhouse gas balance of fiber wood plantation on peat in Indonesia Reviewer Comments & Author Rebuttals

This study reports on a wide range of biogeochemical measurements made from an Acacia plantation on peat soils and compares the results with previously published data from degraded peatland soils, and intact peatland. There is a wealth of data reported for each site, including eddy covariance data for CO2 and CH4, a preliminary water balance, and peatland soil characteristics as well as harvest information for the plantation site. These data are extremely valuable. That said, I wondered about the ability to generalize from these results. There is only one replicate of each site, and it appears that the sites may actually lack independence as the degraded site is really an edge location. This may also be true of the intact site; it was unclear from the text. It is also unclear if it is appropriate to assume that the three sites all had the same starting condition and are thus linearly related as the text implies. A broader survey of site characteristics would help confirm if this was the case.

plantation cycle were two times higher than those from the intact forest, but less than the current IPCC Tier 1 emissions factor. Given the challenges and cost of conducting these types of measurements (including N2O fluxes, fluvial carbon export, and biomass carbon loss), this is truly a unique and very rich dataset, particularly in an underrepresented region of the world. The findings of this work are not only applicable to scientists, but also have strong policy implications, and are of relevance to better predicting future climate change.
While the dataset is unique and the findings are of interest to a broad audience, I would encourage the authors to more clearly describe (and consider the assumptions) how uncertainty was calculated. Furthermore, I have found a potential error in the total GHG balance calculations. I would also use RF/ANN gap-filling for methane fluxes. These major comments, along with some more minor comments, are outlined below. However, once these issues are addressed, I believe this manuscript shows strong potential for publication in Nature.
Major comments: 1) Uncertainty estimates: In the paper you describe gap-filling and random uncertainty, and uncertainty due to the friction velocity threshold. You also note that total uncertainty in fluxes was calculated using the law of propagation of errors. Do the total errors represent gap-filling, random, and ustar uncertainty? Or only random and ustar uncertainty (e.g., https://github.com/bgctw/REddyProc/blob/master/vignettes/aggUncertainty.md)? Can you elaborate more on total uncertainty? Furthermore, in Extended Data Table 3, you note that annual estimates are presented with the standard deviation from different gap-filling approaches. Do you mean the spread of the 20 modelled ANN/RF procedures? Or the spread between the ANN/RF/MDS gap-filling values? It doesn't seem like you include random and ustar uncertainty there? It would be helpful to better described how uncertainties in annual sums were calculated.
Furthermore, you note that random uncertainty is estimated using the method of Finklestein and Sims. Why not use the random uncertainty from the MDS standard deviation as described here: https://github.com/bgctw/REddyProc/blob/master/vignettes/aggUncertainty.md? As noted by El madany et al. 2018, "the MDS uncertainties include information on spatial variability (given changes in the footprints between half hours), on temporal variability (given the +/−7 days sampling window around an observation), and to a certain degree also variability in meteorological conditions (which should be relatively small given the narrow bin widths)" while the Finklestein and Sims method "is only based on half hourly turbulent time series (e.g. vertical wind speed and CO2) and their autoand cross-correlation, hence containing no, or only very limited information about spatial, temporal, or meteorological variability." Since you already have the MDS estimate, I would suggest using that as it is more comprehensive.
Related to uncertainty estimates, how did you conduct the ustar filtering. On Line 735 you note "After a set of quality controls69-71,…" does this include ustar filtering? How was ustar filtering done? Also, for the ustar uncertainty, why not run the full ustar distribution for u* uncertainty (e.g., https://github.com/bgctw/REddyProc/blob/master/vignettes/aggUncertainty.md?) 2) CH4 flux gap-filling: Why weren't ANN & RF applied to CH4 flux as well? Recent papers (e.g., Kim et al. [Gap-filling approaches for eddy covariance methane fluxes: A comparison of three machine learning algorithms and a traditional method with principal component analysis] and Irvin et al. [Gap-filling eddy covariance methane fluxes: Comparison of machine learning model predictions and uncertainties at FLUXNET-CH4 wetlands]) advocate for using RF over MDS.
3) Total GHG budgets (e.g., Figure 2 and extended data table 4): Going back to Neubauer and Megonigal, 2015, they engineered the SGWP metric to account for C seq and not C uptake (it's not EC-centric in this way). So instead of a balance between net CO2 uptake and CH4 emitted, it needs to be a balance between C sequestered and CH4 emitted. So for example, for your sites, based on the excerpt from the paper below, you would need to remove the C-CH4 component lost from the net NEE uptake and remove fluvial C export and Biomass C loss (all in units of C). This doesn't change the numbers drastically, but I think it would be important to do if using the SGWP approach. Attached I provide an example calculation for the Intact peatland (hopefully I didn't make any mistakes -but regardless, I would make sure to double check your numbers and consider the points from Neubauer and Megonigal, 2015 below).
Neubauer and Megonigal, 2015 quote: "Any reader who wishes to apply these SGWP and SGCP values to determine whether a site is a source or sink of greenhouse gases should be aware that these ratios are based on rates of ecosystem CO2 sequestration, not net rates of CO2 uptake. This decision was based on practicality, as our impression is that more ecosystem scientists measure C sequestration (for example, by quantifying wood accumulation and/or using 137Cs and 210Pb radiodating techniques in accreting soils) than develop robust annual estimates of net ecosystem CO2 exchange (for example, by using flux chambers or eddy covariance methods)." Minor points: Abstract 1. In the abstract, I would suggest including a first sentence highlighting the importance of tropical peatlands in the global C cycle. 2. In the abstract, in the last sentence, can you elaborate on who/what is impacted by your findings (eg., implications for the climate system? implications for C credits? Implications for national GHG inventories?) 3. Line 72 -can you provide a range of the uncertainty in the text for readers who are less familiar with those estimates? 4. Line 142: I would suggest 'The substantial net CO2 emissions' (rather than The substantial NEE-CO2 during…) 5. Line 154: Usually NEP is defined as -NEE. I know you clarify that you use the same sign convention for NEE and NEP, but that might cause some confusion to readers from different fields. 6. Lines 185-186: I would suggest 'net CO2 emissions' or 'net CO2 uptake' rather than just 'CO2 emissions' or 'CO2 uptake'. And the same elsewhere in the manuscript. 7. Lines 760 -You note "Therefore, we used the GWL as an environmental factor for the look-up table to derive the daytime Reco data using the MDS gap-filling algorithm". Did you include any other variables such as soil temperature or air temperature? 8. For Extended data table 3 (and others) -For 'measurement period', please specify month/year (as to not confuse it with month/day). General This extensive and timely work using Eddy Covariance (EC) method is very much welcome by the scientific community. The authors described the long-term experiments involving natural topical peat swamp forest (PSF), degraded/logged-over forests, and industrial plantation established from PSF conversion. The net GHG exchange results provide useful insights and new understanding of a complex process in tropical environment. However, I have a few concerns that need clarity/ explanation around the flux change approaches in comparing with the IPCC guidelines. Firstly, being the most comprehensive account of the process, it is utmost logical to recognize numerous ground-based/static chamber studies in the background, which were considered to be overestimating by the authors.

Attachment
Secondly, methodological wise, while this is not the first to implement EC approach the description of the experimental setup is lacking necessary background information associated with existing natural condition, given the extent of the flat landscape etc.
Thirdly, it is mentioned throughout the text that the accounting of GHG flux in the plantation system is a full cycle, however, it is not clear which cycle it was. This is just fair as soil C-stock in the second and following cycles would be logically much less event to compare with logged-over PSF. Otherwise the paper almost appears like defending plantation and approve that conversion is not too bad.
Specific I found C-store is used. Probably C-input, C-losses and C-gains may be consistently hyphenated L.49: in terms of simulated C accumulation using modified Holocene Peat  L.168-177: to make the budget complete, this story of GWL effect on emission may be added at the Acacia plantation, in addition to NEE story.
L.178-189: again, it is obscure to jump into this claim from policy perspectives. I would suggest toning down the statement and focus on the science and keep it neutral.
L.198: spell out "LT". Also, evapotranspiration cannot be neglected, especially is dry season. Aerodynamic process (wind removal) can be VERY significant I such a rough surface like forest stands. To make the story complete, I suggest plotting the contrasting daily ETP in wet and dry seasons. At least in the Extended information/data section.
Since ETP represent (potential) atmospheric demand, mifgt be worth plotting the actual evapotranspiration (ETA) and see what the actual effects on GWL and associated fluxes were.
L.209-216: It depends on the phase of IOD. In positive phase, IOD would enhance Southern Oscillation Index (SOI), which means worsening drought. But in negative phase SOI will be diminished or cancelled (see also comment on L.525-530 below L.539 onward: it may be useful to have a background how Acacia plantation was established. I believe there are ample information on story of biodiversity losses, soil compaction, and canalization associated with such a large-scale pristine peat swamp forest conversion. Carbon and nutrient cycling would have been significantly affected in the first place.
L.543: the word 'ensuring" sounds like a special effort has been made to select the site. In fact, it naturally exists, probably throughout the peninsula. Please be factual.

Dear Reviewers,
We thank you for encouraging and positive assessments and for providing thoughtful and constructive suggestions to improve the quality of the paper. We hope that the revised manuscript has addressed all of the concerns raised satisfactorily. Our point-bypoint response is shown below, with the reviewer's comments quoted first (black text), followed by our response (blue text) and blue texts in italics for the actual changes to the paper with the lines in brackets indicate where the corrections have been applied in the revised manuscript.
Referee #1 (Remarks to the Author): This study reports on a wide range of biogeochemical measurements made from an Acacia plantation on peat soils and compares the results with previously published data from degraded peatland soils, and intact peatland. There is a wealth of data reported for each site, including eddy covariance data for CO2 and CH4, a preliminary water balance, and peatland soil characteristics as well as harvest information for the plantation site. These data are extremely valuable. That said, I wondered about the ability to generalize from these results. There is only one replicate of each site, and it appears that the sites may actually lack independence as the degraded site is really an edge location. This may also be true of the intact site; it was unclear from the text. It is also unclear if it is appropriate to assume that the three sites all had the same starting condition and are thus linearly related as the text implies. A broader survey of site characteristics would help confirm if this was the case.
We thank the reviewer for the positive appraisal of the research, particularly in the context of wealth of data in this study and their value to improve knowledge of the GHG balance in tropical peatland ecosystems.
We accept the point made by the reviewer about the challenges of generalizing from our sites, and have sought to acknowledge this in the manuscript (revised text below), and to place our results in the context of comparable data, to the extent that such data exist (e.g. Fig. 3, Extended Data Fig 5). On the other hand, as the reviewer notes, there are few if any similarly comprehensive studies of GHG fluxes using eddy covariance technique in tropical peatlandsso we believe that our study offers unique insights which (with acknowledged caveats) should advance our understanding of these vitally important ecosystems and lead to more measurements from this globally important carbon-rich ecosystem.
We have clarified these points as follows: The point about edge locations is well made and we have addressed this within the site descriptions and footprint analysis for the eddy covariance data. We have clarified the edge and boundary effect points in the degraded and intact peatland in the Methods as follows: "The second EC tower is located on the boundary of the degraded peatland and Acacia plantation (Fig. 1). To represent only the degraded peatland, half-hourly measurements in which the prevailing wind came from the plantation site (90 to 270°) were excluded as is commonly done in EC studies 54 ." (Lines 544 -547) "An integrated climatologic footprint analysis 53 indicated that approximately 80% of the fluxes were derived within 1,000 m in the upwind direction ( Fig. 1), and thus originated from intact forest with neither logging nor canal-construction activity 17 . Some long-term regional effects of hydrological management of surrounding plantations cannot be ruled out, but a previous analysis suggested that the strongest effects occurred within 300 m of the plantation boundary 21 , and recent multivariate analysis indicates that subsidence in the interior forest is independent of distance from plantation canals 22 . The nearest active plantation is 3.5 km from the flux tower and well outside of the flux footprint. Further, to avoid any possible boundary effect and associated bias, measurements from a wind direction between 78 and 191° were excluded in this study (Fig. 1)." (Lines 561 -571) We have clarified "the same starting condition" point as follow: "The relatively close proximity of the Acacia plantation, intact peatland, and degraded peatland sites ( Fig. 1) within the same peat landscape avoids potentially confounding variables such as differences in past natural succession 55 and peat formation 8 ." (Lines 578 -581) I thought it was interesting that the mineral N concentrations, especially nitrate, were so high. This suggests that the sites are well aerated. How frequently and at what depths were these measurements made? I did not see details of the soil biogeochemistry measurements. This is also true for soil C.
Yes, we agree with the referee 1. The measured groundwater levels clearly indicate that the sites were well aerated, including an intact peatland particularly in the dry seasons (Extended Data Table 2).
We have added text about peat sampling for physical and chemical properties in the Methods as follows: " For peat physical and chemical properties of the surface layer (0 -50 cm), four plots in each of the Acacia plantation and degraded site, and three plots in the intact site were randomly selected within the EC flux footprint (200 -1,000 m distance from each tower location, Fig. 1 N2O was measured within 2 km of the flux towers, which seems to be very far away from the rest of the measurements. Were the measurements always in the primary footprint of the towers? How frequently were the N2O measurements made? The statement is made that ~1000 measurements were made, but now how they were distributed across treatments or time. For soil N2O emissions measurement, four plots in each of the Acacia plantation and degraded site, and three plots in the intact were randomly selected within the EC flux footprint (200 -1,000 m distance from tower location, Fig. 1). The measurements were conducted on a bi-monthly basis. As suggested, we have added number of N2O flux chamber measurements and their distribution in time across treatments and in Extended Data Fig. 4 The paper is generally well written although it would benefit from a tighter organization and focus on the primary results.
Thank you for the positive assessment. We have gone through the revised manuscript and tried to tighten it up where possible, in line with the Editorial guidelines.
Referee #2 (Remarks to the Author): Deshmukh et al. present an interesting and unique multiyear dataset of ecosystemscale greenhouse gas (GHG) fluxes from an Acacia crassicarpa plantation, degraded forest and intact forest within the same peat landscape to assess the impacts of land-cover change on GHG fluxes in Sumatra, Indonesia. The measurements represent the first full plantation cycle GHG budget on peatlands worldwide. The primary findings are that the GHG emissions from the Acacia plantation over a full plantation cycle were two times higher than those from the intact forest, but less than the current IPCC Tier 1 emissions factor. Given the challenges and cost of conducting these types of measurements (including N2O fluxes, fluvial carbon export, and biomass carbon loss), this is truly a unique and very rich dataset, particularly in an underrepresented region of the world. The findings of this work are not only applicable to scientists, but also have strong policy implications, and are of relevance to better predicting future climate change.
Thank you for the positive assessment.
While the dataset is unique and the findings are of interest to a broad audience, I would encourage the authors to more clearly describe (and consider the assumptions) how uncertainty was calculated. Furthermore, I have found a potential error in the total GHG balance calculations. I would also use RF/ANN gap-filling for methane fluxes. These major comments, along with some more minor comments, are outlined below. However, once these issues are addressed, I believe this manuscript shows strong potential for publication in Nature.
Thanks for assessing the manuscript positively and providing inputs for further improvement. We have clarified the calculations of uncertainties and total GHG balance and applied Random Forest and ANN gap-filling methods for net ecosystem CH4 exchanges, as discussed further below.
Major comments: 1) Uncertainty estimates: In the paper you describe gap-filling and random uncertainty, and uncertainty due to the friction velocity threshold. You also note that total uncertainty in fluxes was calculated using the law of propagation of errors. Do the total errors represent gap-filling, random, and ustar uncertainty? Or only random and ustar uncertainty (e.g., https://github.com/bgctw/REddyProc/blob/master/vignettes/aggUncertainty.md)? Can you elaborate more on total uncertainty? Furthermore, in Extended Data Table 3, you note that annual estimates are presented with the standard deviation from different gap-filling approaches. Do you mean the spread of the 20 modelled ANN/RF procedures? Or the spread between the ANN/RF/MDS gap-filling values? It doesn't seem like you include random and ustar uncertainty there? It would be helpful to better described how uncertainties in annual sums were calculated.
Furthermore, you note that random uncertainty is estimated using the method of Finklestein and Sims. Why not use the random uncertainty from the MDS standard deviation as described here: https://github.com/bgctw/REddyProc/blob/master/vignettes/aggUncertainty.md? As noted by El madany et al. 2018, "the MDS uncertainties include information on spatial variability (given changes in the footprints between half hours), on temporal variability (given the +/−7 days sampling window around an observation), and to a certain degree also variability in meteorological conditions (which should be relatively small given the narrow bin widths)" while the Finklestein and Sims method "is only based on half hourly turbulent time series (e.g. vertical wind speed and CO2) and their auto-and crosscorrelation, hence containing no, or only very limited information about spatial, temporal, or meteorological variability." Since you already have the MDS estimate, I would suggest using that as it is more comprehensive.
Related to uncertainty estimates, how did you conduct the ustar filtering. On Line 735 you note "After a set of quality controls69-71,…" does this include ustar filtering? How was ustar filtering done? Also, for the ustar uncertainty, why not run the full ustar distribution for u* uncertainty (e.g., https://github.com/bgctw/REddyProc/blob/master/vignettes/aggUncertainty.md?) We apologize for the misunderstanding. We did indeed calculate total uncertainty from random error in measurement, friction velocity quality control criteria, and gap-filling approach in line with the uncertainty approach outlined in Pastorello et al. (2020) used in the global Fluxnet2015 dataset and implemented in the OneFlux software package. We also applied u* threshold filtering as a quality control criterion following Wutzler et al. (2018) as recommended by the reviewer. We have revised the text in Methods for clarity as follows: "The flux random uncertainty was calculated following ref 80  Based on the reviewer's comments, we undertook Random Forest and ANN gap-filling methods for the net ecosystem CH4 exchanges. We are grateful to the reviewer for this suggestion, which is now fully incorporated in the analysis, and has strengthened our findings.
We have added the text as follows: "We gap-filled both low-quality and missing data, as is commonly done in EC studies 16,42,56,[71][72][73][74][75]  3) Total GHG budgets (e.g., Figure 2 and extended data table 4): Going back to Neubauer and Megonigal, 2015, they engineered the SGWP metric to account for C seq and not C uptake (it's not EC-centric in this way). So instead of a balance between net CO2 uptake and CH4 emitted, it needs to be a balance between C sequestered and CH4 emitted. So for example, for your sites, based on the excerpt from the paper below, you would need to remove the C-CH4 component lost from the net NEE uptake and remove fluvial C export and Biomass C loss (all in units of C). This doesn't change the numbers drastically, but I think it would be important to do if using the SGWP approach. Attached I provide an example calculation for the Intact peatland (hopefully I didn't make any mistakesbut regardless, I would make sure to double check your numbers and consider the points from Neubauer and Megonigal, 2015 below).
Neubauer and Megonigal, 2015 quote: "Any reader who wishes to apply these SGWP and SGCP values to determine whether a site is a source or sink of greenhouse gases should be aware that these ratios are based on rates of ecosystem CO2 sequestration, not net rates of CO2 uptake. This decision was based on practicality, as our impression is that more ecosystem scientists measure C sequestration (for example, by quantifying wood accumulation and/or using 137Cs and 210Pb radiodating techniques in accreting soils) than develop robust annual estimates of net ecosystem CO2 exchange (for example, by using flux chambers or eddy covariance methods)." Thank you for the detailed review of the SGWP calculation. In our study, the studied ecosystems were sources of CO2, CH4 and N2O. CO2 emissions from our studied ecosystems have a warming effect on the atmosphere during the measurement period and thus all SGWP values are additive following the outline here. We believe that there no further change is required in our total GHG budgets, but if we have misunderstood the reviewers' point here we would be happy to revisit this.
Minor points: Abstract 1. In the abstract, I would suggest including a first sentence highlighting the importance of tropical peatlands in the global C cycle.
The following sentence has been added: "Tropical peatlands cycle and store globally significant amounts of carbon in their soil and biomass." (Lines 27 -28) 2. In the abstract, in the last sentence, can you elaborate on who/what is impacted by your findings (eg., implications for the climate system? implications for C credits? Implications for national GHG inventories?) We have added the text on implications of our findings as follows:

"Our results should help to reduce the uncertainty in the estimation of GHG emissions from globally important ecosystems, provide estimate of the impact of land-use change on tropical peat, and develop science-based peatland management practices as naturebased climate solutions that help to minimize GHG emissions." (Lines 41 -44)
3. Line 72can you provide a range of the uncertainty in the text for readers who are less familiar with those estimates?
We have added a range in the text as follows: "Existing estimates of GHG emissions from tropical peatlands continue to be debated 23,27 with large observed variability (0.04-2.79 GtCO2-eq yr -1 ) 28 and resulting uncertainty 29 ." (Lines 68 -69) 4. Line 142: I would suggest 'The substantial net CO2 emissions' (rather than The substantial NEE-CO2 during…) We have revised the text as follows: "The substantial net CO2 emissions during the early stage of the plantation …" (Lines 97 -98) 5. Line 154: Usually NEP is defined as -NEE. I know you clarify that you use the same sign convention for NEE and NEP, but that might cause some confusion to readers from different fields.
We do not use NEP term anymore in the revised manuscript to avoid possible confusion to readers from different fields.
We have revised the text as follows: "Net ecosystem exchanges of CO2 (NEE-CO2; net gaseous CO2 exchange between ecosystem and atmosphere) varied with plantation age; it was highest (48.4 ± 4.7 tCO2 ha -1 yr -1 ) in the first year after planting, lowest (-8.8 ± 4.5 tCO2 ha -1 yr -1 ) in the third year with highest tree growth, and then rose again to 11.7 ± 6.0 tCO2 ha -1 yr -1 before harvesting (Extended Data Fig. 1 and Extended Data General This extensive and timely work using Eddy Covariance (EC) method is very much welcome by the scientific community. The authors described the long-term experiments involving natural topical peat swamp forest (PSF), degraded/logged-over forests, and industrial plantation established from PSF conversion. The net GHG exchange results provide useful insights and new understanding of a complex process in tropical environment.
Thank you for the positive assessment.
However, I have a few concerns that need clarity/ explanation around the flux change approaches in comparing with the IPCC guidelines. Firstly, being the most comprehensive account of the process, it is utmost logical to recognize numerous ground-based/static chamber studies in the background, which were considered to be overestimating by the authors.
As the reviewer notes, a substantial number of chamber-and subsidence-based studies that have tried to estimate the carbon and GHG balance of tropical peatlands, we refer to these studies in the context of IPCC Tier 1 emission factors in our introduction, and subsequently compare our results with those obtained using similar (EC) and different (chamber, subsidence) methods in Fig. 3 and associated discussion. Based on this comment we have added some additional text here noting the discrepancy between CO2 fluxes obtained by the different methodswe prefer not to enter into a detailed discussion of the pros and cons of each approach, or to suggest that other methods are intrinsically flawed, but we have noted that the differences we observed here are not unique to tropical peatlands, but does exist for the northern peatlands also as at the following figure. We have added the text as follows: " Secondly, methodological wise, while this is not the first to implement EC approach the description of the experimental setup is lacking necessary background information associated with existing natural condition, given the extent of the flat landscape etc.
Based on this comment and previous comments from Referee 1, we have revised the text for clarity in the Methods section. Specifically, we have added the additional information related to frequency of measurement and the flux footprint area to describe the extent and completeness of the experiments in representing the ecosystem. We have addressed this within the site descriptions and footprint analysis for the eddy covariance datasee response for Referee 1 "Eddy covariance provides half-hourly measurements of turbulent exchanges between an entire ecosystem and the atmosphere above the vegetation canopy 54 . Hence, eddy covariance measurements incorporate all existing sources and uptakes that can vary significantly within an ecosystem in both space and time arising from variation in environmental conditions. Given the flat terrain (slope < 0.05%), using the measured vegetation-canopy height and wind speed, the estimated 80% EC flux footprints represent an area of interest of around 1,000 m radius (Fig. 1)." (Lines 572 -578) Thirdly, it is mentioned throughout the text that the accounting of GHG flux in the plantation system is a full cycle, however, it is not clear which cycle it was. This is just fair as soil C-stock in the second and following cycles would be logically much less event to compare with logged-over PSF. Otherwise the paper almost appears like defending plantation and approve that conversion is not too bad.
This is a good point and we have included in our revision. We have added more information on the establishment of the plantation (see response to comment on Line 539 below), as well as noting that our measurements cover the 4 th rotation of the plantation during 17-22 years after drainage in both the methods and at the start of the results.
"When measurements began in October 2016, the trees were already at the end of the third plantation rotation. All plantation compartments within 2-km radius around the EC tower were harvested between October 2016 and April 2017. Tree height at harvest was in the range from 19 to 24 m, determined from a vegetation survey in permanent sampling plots (20 m × 125 m) around the tower. Replanting for the fourth plantation rotation took place within two weeks after harvesting each individual compartment at a density of 1,667 trees per hectare (3 m × 2 m spacing). Five grams of chelated micronutrients per tree were applied around the seedlings during planting. All compartments within a 2-km radius of the EC tower were harvested between June and August 2021, when the average plantation age was 4.7-years, and replanting for the fifth plantation rotation took place within two weeks after harvesting." (Lines 519 -530) Specific I found C-store is used. Probably C-input, C-losses and C-gains may be consistently hyphenated Thanks and agreed. We have replaced the term "C export" with "C-export", "C input" with "C-input", "C loss" with "C-loss", "C balance" with "C-balance", and "C content" with "C-content" as suggested.

"Other GHG fluxes and C-loss" (Line 181)
"A previous study 36 within the same landscape reported fluvial C-export of 0.3 ± 0.1 and 0.5 ± 0.1 tC ha −1 yr −1 respectively in the intact and degraded peatland. Due to lack of fluvial C-export measurements for the Acacia plantation, we used a value of 0.4 ± 0.1 tC ha −1 yr −1 from a managed oil palm plantation in Southeast Asia 37 . Notably, fluvial Cexports are fairly small compared to direct CO2 emissions. We conservatively assume that all fluvial C-export is ultimately emitted as CO2 (ref 38 ). The increased fluvial Cexport from the plantation and degraded forest may be attributed to enhanced mineralization with deeper GWL 37 ." (Lines 199 -206) "…the measured above-and belowground biomass C-stock was highest…" (Line 207) "…biomass C-losses due to land-use change from intact peatland were…" (Line 213) "Biomass C-loss due to plantation…" (Line 215) "…the average C-balance of high-latitude peatlands with multi-decadal…" (Line 241) "…measured a C-input to the peat of around…" (Line 261)

"This C-input…" (Lines 264-265)
"…soil C-input through better post-harvest residue management. Further research is needed to confirm the potential scale of increase in C-input that could realistically be achieved."  "For the Acacia plantation site, time-integrated NEE-CO2 over the plantation rotation was combined with C-export in the harvested wood. Total C-export in harvested wood and delivered to the mill from the total footprint area of 220 ha over the average plantation age of 4.7-year was calculated using a basic density of 455 kg m -3 and average C-content of 48.2% [56][57][58][59] . The exported wood is converted into pulp products and biomass fuel for bioenergy generation. We applied the conservative assumption that all C in exported wood would be returned to the atmosphere as CO2. Intact and degraded peatlands were considered to have had no biomass C-export during the study period." (Lines 661 -668) "The biomass C-loss due to land-use change was calculated from above-and belowground biomass C-stock differences between the intact peatland and the Acacia …" (Lines 669 -670) "…reported C-balance over a full plantation rotation, and thus they are not in steady state with regard to CO2 uptake and emission, making the underlying peat C-balance uncertain. However, naturally forested peatlands and degraded peatlands were considered to have had no biomass C-export during the study period." (Lines 970 -974) "…and C-export in harvested wood to quantify a conservative estimate of C-input to the peat…" (Lines 1035 -1036) L On revision including all reviews, this statement is no more in the revised manuscript due to space limitations, in line with the Editorial guidelines.
L.87: knowing that IPCC Guidelines are designed for GHG Inventory, I am not sure if it is appropriate to comment of its consistency here. Every new (emission factor) number may be contributed as refinement and used at project based GHG accounting. Having said that, it does not mean that the guidelines fails to proof one of its principles, consistency.
This is a fair point and we have revised the statement to just note our refinement. Following the Nature Editorial guideline on the word limitation, the statement is no more in the revised manuscript.
L.117-119: even so this is not the first EC method applied. It might be prudent to describe the extent and completeness of the experiments in representing the ecosystem while recognizing not being the first Following the Nature Editorial guideline on the word limitation, the statement is no more in the revised manuscript. However, we have provided additional information about EC method in response to the above second comment as: "Eddy covariance provides half-hourly measurements of turbulent exchanges between an entire ecosystem and the atmosphere above the vegetation canopy 54 . Hence, eddy covariance measurements incorporate all existing sources and removals that can vary significantly within an ecosystem in both space and time arising from variation in environmental conditions. Given the flat terrain (slope < 0.05%), using the measured vegetation-canopy height and wind speed, the estimated 80% EC flux footprints represent an area of interest of around 1,000 m radius (Fig. 1)." (Lines 572 -578) L.167: please remove the word "emission" Thanks -we have removed "emission" at the mentioned line.
L.168-177: to make the budget complete, this story of GWL effect on emission may be added at the Acacia plantation, in addition to NEE story.
We have revised as suggested and the text as follows: "The observed net CO2 emissions can be attributed to peat aeration due to a consistently deep GWL, which enhances heterotrophic respiration rates, combined with a higher soil temperature (intact site = 27.5 ± 0.5 °C vs. Acacia site = 29.3 ± 1.0 °C; Extended Data Table 2) due to both canopy-cover loss and GWL drawdown, which further boosts microbial activities and heterotrophic respiration."  L.178-189: again, it is obscure to jump into this claim from policy perspectives. I would suggest toning down the statement and focus on the science and keep it neutral.
We are unsure what the reviewer is referring to here as Line 178-189 in the original submission were "Coarse woody debris from fallen dead trees may also have contributed to the emissions as fallen trees do not decompose instantaneously, providing a lagged but sustained contribution to CO2 emissions." We believe that this sentence does not provide directly any policy aspects.
However, we are grateful to the reviewer for this suggestion and our revised manuscript focuses on the science and tries to present results as neutrally as possible from a policy perspective.
L.198: spell out "LT". Also, evapotranspiration cannot be neglected, especially is dry season. Aerodynamic process (wind removal) can be VERY significant I such a rough surface like forest stands. To make the story complete, I suggest plotting the contrasting daily ETP in wet and dry seasons. At least in the Extended information/data section. Since ETP represent (potential) atmospheric demand, mifgt be worth plotting the actual evapotranspiration (ETA) and see what the actual effects on GWL and associated fluxes were.
We have spelled out "LT" to "local time" throughout the manuscript.
The eddy covariance provided actual evapotranspiration estimates. Actual evapotranspiration measurements clearly show that evapotranspiration seems negligible in the night. The diurnal variation of evapotranspiration in both dry and wet season is added as Extended Data Fig. 3 in the revised manuscript.
To clarify this we have amended the text as follows: "The evapotranspiration measurements clearly indicate that actual daily evapotranspiration (4.2 mm d -1 ) exceeded daily rainfall…" (Lines 148 -149) L.209-216: It depends on the phase of IOD. In positive phase, IOD would enhance Southern Oscillation Index (SOI), which means worsening drought. But in negative phase SOI will be diminished or cancelled (see also comment on L.525-530 below).
We investigated the combined impact of the ENSO and IOD as part of another paper on regional subsidence dynamics (Evans et al. 2022). The two phenomena are closely linked, with strong El Niño consistently followed by a positive phase of the IOD, lagged by a few months, which together drive regional drought. A pattern of positive IOD and La Niña cancelling each other out (or vice versa) does not appear to occur. We have revised the text as follows: "In general, the El Niño 34 and positive IOD 35 occur sequentially, with the positive IOD peaking a few months after the El Niño, exerting a strong combined effect on regional rainfall patterns 22  L.539 onward: it may be useful to have a background how Acacia plantation was established. I believe there are ample information on story of biodiversity losses, soil compaction, and canalization associated with such a large-scale pristine peat swamp forest conversion. Carbon and nutrient cycling would have been significantly affected in the first place.
As suggested, we have provided additional information of water management system in the Study Area to provide all site characteristics including soil physical and chemical properties, which potentially influence GHG dynamics. We amended the text as follows: "This involved clearance of the remaining logged forest, artificial compaction occurred during mechanical land preparation, installation of regularly spaced water management and access canals, and planting of Acacia crassicarpa, which is harvested on a 4-5 year rotation." (Lines 511 -514) "Water management zones comprise navigable canals, typically of 12 m width and 3 m depth, also used for transportation 21 . Branch canals of 5 -8 m width run perpendicular to these canals at a spacing of 500 -800 m to form plantation compartments, which contain 1 m deep field drains at a spacing of 75 m 21 ." (Lines 536 -539)

Reviewer Reports on the First Revision:
Referees' comments: Referee #1 (Remarks to the Author): The authors have addressed the concerns I had with the manuscript and I think it will make a nice contribution to the literature.
Referee #2 (Remarks to the Author): Deshmukh et al. provided clear and comprehensive replies to my previous comments, particularly related to gap-filling and uncertainty estimates.
However, I still have a very minor concern regarding the calculation for the GHG flux estimates. For the net GHG estimates, I realize I made a slight error, although I still believe there is a minor error in the manuscript (similar to an error we made in a paper, which we also had to correct during revisions).
The main issue is related to the NEE term. As I noted, Neubauer and Megonigal engineered the SGWP to account for C sequestration (or C loss as in your case) and not NEE (net CO2 uptake/loss). As such, eddy covariance measured NEE does not represent true C sequestration/loss (i.e., NEE doesn't equal the net ecosystem carbon balance). To fully capture C sequestration/loss at the site, you also need to account for the C lost as CH4 (in addition to the fluvial C export and Biomass C loss which you already account for). As such, C sequestration/loss = NEE + CH4-C + Fluvial C + Biomass C. This will give you the net C sequestration/loss. Below I provide the example for the intact peatland (which doesn't include Biomass C). Then the GHG budget becomes: GHG budget = FCO2-eq + (FCH4 * SGWP) + (N2O * SGWP), where FCO2-eq represents C sequestration/loss (i.e., NEE + CH4-C + Fluvial C, and Biomass-C where applicable). Note: see uploaded xlsx doc for proper formatting of the table.
As such, now you're fully accounting for C sequestration (or in your case loss), which is what should be included in the SGWP calculation. The C contained in CH4 is very minor in this case, but can sometimes be an important component of the C budget in peatlands.
Other than this, I feel that the article is suitable for publication as is.

Attachment:
Net GHG budget from the paper The revised version of the manuscript turns out to be satisfactory. It demonstrates concrete and convincing results regarding net GHG emissions in peat land-use dynamics. The piece of work clearly contributes new scientific findings on those of challenging ecosystem to work and yet very important element in GHG emission reduction efforts.
The vast field data are well handled and thoroughly evaluated using statistically sounds techniques. Uncertainties are clarified. Referencing to the relevant works has been meticulously done that demonstrate the authors comprehension on the existing works.
Overall, I recommend that this piece of important work is accepted for publication.