the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Apr 2026
Simulation of long-term peat accumulation dynamics & vulnerability: Insights from a Pole Forest and Palm Swamp in Amazonia
Abstract. Peruvian peatlands represent one of the largest reservoirs of carbon in Amazonia. This heterogeneous landscape exhibits several types of ecosystems, including pole forest (PF), palm swamp (PS), open peatlands (OP), and seasonal flooding forest (SFF). We apply the HPMTrop_EcoTy model, a novel development that represents ecological succession via bespoke parametrisations of ecohydrological mechanisms, to gain insights into long-term peat accumulation dynamics across these ecosystems and to assess their vulnerability to carbon gain and loss. Model results suggest that carbon accumulation rates in Amazonian peatlands are similar to or greater than those reported for the Congo Basin and Southeast Asia. Peat and carbon accumulation in Amazonia are particularly sensitive to local-scale changes, especially those driven by ecosystem succession. Amazonian peatlands appear less sensitive to precipitation changes, likely due to the extremely high rainfall across the Peruvian Amazon. However, reducing rainfall to levels similar to those of the present-day Congo Basin (45 % reduction) produces an exponential decline in peat and carbon accumulation, suggesting a critical tipping point. Sensitivity analysis shows that PF, the most carbon-dense ecosystem, is the most sensitive, likely because it is rain-fed and therefore more vulnerable to ecohydrological changes, whereas SFF, the least carbon-dense, is the least sensitive. Considering the exceptionally high precipitation in the region, peatland formation appears mainly controlled by local processes such as river migration, which drives vegetation succession linked to peatland development.
- Preprint
(17945 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2272', Anonymous Referee #1, 22 May 2026
-
AC1: 'Reply on RC1', Yarin Tatiana Puerta Quintana, 05 Sep 2026
We want to extend our big thanks to the reviewer for all the constructive comments, which have greatly helped us improve the manuscript. The comments encouraged us to address several points and seek additional information to support our work further, and we believe that substantial improvements have been made to the manuscript. We greatly appreciate the reviewer’s time, expertise, and thoughtful comments.
Citation: https://doi.org/10.5194/egusphere-2026-2272-AC1 -
AC2: 'Reply on RC1', Yarin Tatiana Puerta Quintana, 06 Sep 2026
Major comments: The model is validated against peat presence/depth over time using a single peat core from each site. I am particularly concerned about the lack of validation data. At this stage, it is not possible to say whether the model can reproduce peatland accumulation dynamics. The depth and presence of peat at a particular time point is a great first step to validating the model. However, the depth and presence of peat is not indicative of carbon accumulation rates or carbon storage.
Response: We agree with the reviewer that the original validation was limited and that peat presence and depth alone do not provide sufficient evidence for evaluating carbon accumulation rates or carbon storage.
The HPMTrop model is a mechanistic peat accumulation model. For each modelling site, the peat-core depth-age curve (radiocarbon dating) was used as an independent reference for calibrating the model, rather than as a direct model input. Calibration was performed separately for each site using site-specific parameters derived from available in situ and historical information. Where direct observations were unavailable, parameters were tuned within ecologically plausible ranges. Thus, the model calibration was conducted independently for each site while maintaining the ecological constraints of the model parameterisation.
Once a satisfactory calibration was obtained between the reference curve (peat core-age: radiocarbon dating curve) and the model curve, the model was used to simulate peat accumulation and carbon dynamics through time. In particular, the model's peat-height variable represents the annual accumulation of peat mass and therefore provides information on peat accumulation dynamics that is distinct from the cumulative peat-depth variable. The definitions and distinction between peat height and peat depth are now clarified in the revised manuscript (Lines 128–131 and Appendix A 3.9).
Importantly, following the reviewer's suggestion, we sought additional independent observations with which to evaluate model outputs. We identified laboratory measurements of bulk density (BD) and carbon concentration from the peat cores at the NYO-03 and VEN-02 sites. These measurements provide an independent basis for evaluating two fundamental quantities represented by the model: bulk density and carbon stock.
We calculated carbon stock from the core measurements as using the laboratory information as following:
C_stock (Mg C ha^-1) = BD x Carbon_con x Thickness
where (BD) is bulk density, (Carbon_con) is carbon concentration, and thickness is the thickness of the corresponding peat layer. The resulting core-based carbon-stock estimates were compared with the corresponding model estimates, where the model assumed that carbon is 0.5 of the organic mass (please note that the model does not simulate carbon concentration). The values were 1676.9 Mg C ha ^-1 (core) versus 1799.4 Mg C ha^-1 (model) for NYO-03, and 558.8 Mg C ha ^-1 (core) versus 563.1 Mg C Mg C ha ^-1 (model) for VEN-02.
In addition, we compared the vertical patterns of laboratory-measured bulk density with the corresponding model simulations. The modelled bulk-density values fall within the range of the measured values at both sites. These independent comparisons provide additional support that the model can reproduce key properties related to peat accumulation and carbon storage, although we acknowledge that the available independent observations remain limited to two sites.
We have added a new subsection, “Model comparison with field data” (Appendix C), to provide an independent evaluation of model outputs using laboratory measurements from the NYO-03 and VEN-02 peat cores.
Specifically, we added:
- Bulk-density comparisons: Two new figures compare simulated bulk density with laboratory-derived bulk-density measurements for NYO-03 and VEN-02 (Figs. C1 and C2).
- Carbon-stock comparison: A new table compares carbon stock estimated from the peat-core measurements with the corresponding model estimates (Table C1).
Major comments: The manuscript refers to in situ NPP measurements from other studies. Can this be used to validate the model in the present day?
Response: The in situ NPP measurements reported in previous studies cannot be considered fully independent validation data for our model framework because observations from these studies were used as initial values to parameterise the model at the corresponding sites. This data is the same as that summarised in Appendix Table A1.
The HPMTrop_Ecoty model estimates NPP dynamically throughout the simulation based on the simulated water-table conditions and ecosystem-specific relationships between water table and productivity. NPP is calculated separately for above-ground and below-ground components, including leaves, stems, and roots, and is subsequently aggregated to obtain ecosystem- or site-level NPP (we have added these relationships in Appendix B2). Thus, although observed NPP values were used to constrain the initial model conditions, NPP throughout the simulation is determined by the modelled ecosystem dynamics and water-table conditions rather than being prescribed directly through time.
As a consistency check, we compared the long-term modelled NPP among ecosystem types with patterns reported in the literature. The model produces the highest NPP for the seasonal flooded forest (SFF), followed by the palm swamp (PS), whereas the pole forest (PF) has lower productivity. This relative ranking is consistent with observations reported by Aguila-Pasquel et al. 2013, SFF and Dargie et al. 2024, PF and PS. However, because these studies contributed to the initial parameterisation of the model, we do not consider this comparison an independent validation. Instead, it provides evidence that the simplified ecosystem-specific relationships implemented in HPMTrop_EcoTy produce ecologically plausible differences in productivity among ecosystem types.
Major comments: WTD (estimated from precipitation) is the primary driver of NPP in the model. Can this be validated against present day site data? I encourage the authors to validate at least one other model output.
Response: In HPMTrop_EcoTy, water-table (WT) is estimated dynamically from precipitation through ecosystem-specific relationships. These relationships determine the monthly WT throughout the simulation and, in turn, drive the estimation of NPP for each ecosystem. The NPP formulation includes separate aboveground and belowground components (leaves, stems, and roots), which are subsequently aggregated to obtain ecosystem- or site-level NPP. We have added these relationships in Appendix B1: Water table and B2: NPP.
As the NPP values, the present-day WT observations were used to derive the relationships between precipitation, WT, and NPP, which were incorporated into the model parameterisation and therefore cannot be considered independent validation data.
Major comments: In terms of the structure and readability of the manuscript, much more can be done to improve the flow and readability.
Response: We agree with the reviewer that the structure and readability of the manuscript can be further improved. We have carefully revised the manuscript throughout to improve the overall flow, clarity, and readability.
Major comments: There are only five figures in the main text, despite an extensive supplementary material and appendix. The number of figures and tables is relatively low for a manuscript of this length.
Response: We agree that additional tables and figures in the main text can improve the interpretation of the model structure and results. We considered the balance between the main text and the supplementary material, while also taking into account the journal's formatting and length requirements. In particular, we have added a table summarising the main model parameters to the main text (Table 1, Lines 132). This table provides the reader with an overview of the key parameters used in the HPMTrop_EcoTy model. And also we added the figure with the graphical representation of the model structure and the main processes (Figure 2 Line 198).We have retained the more detailed equations, parameter descriptions, and supporting analyses in Appendices A and B and in the Supplementary Information (in the repository), including parameter descriptions and code availability.
Major comments: Many details are omitted from the main text, including a comprehensive description of relevant model processes and I think including more figures and tables in the main text would help clarify several details
Response: Yes, we agree that a clearer presentation of the model processes and key model components would improve the interpretation of the study. Therefore, we have included a graphical representation of the model structure and the main processes (Figure 2 Line 198), together with a table summarising the principal model parameters (Table 1. Line 132) and the relationships among precipitation, water-table dynamics, NPP, litter production, peat decomposition, peat mass, peat thickness, and carbon storage. The main equations required to understand the model formulation are now provided in Appendix B, where they are accompanied by definitions of the variables and parameters.
Major comments: It remains unclear which parameters were changed. This is a very exciting study and a wonderful application of the model.
Response: To address this point, we have added the model parameterisation used in this study to the main text (Table 1, Line 132). This table now provides the relevant parameter values used in the simulations and identifies the parameters that were modified for the study sites. A more detailed description of all model parameters, including their definitions, values, and the corresponding code used in the simulations, is provided in the Supplementary Information and in the associated repository.
Small comments:
L18. Peatland formation is the result of a very slow process spanning thousands of years, during which the material from dead vegetation accumulates a process regulated by climatic and hydrological conditions.
Response: Updated with the suggestion. Lines 17-18
L22. I assume you mean unique or endemic? Can you give an example?
Response: No, the intention of this phrase was to introduce the heterogeneity of the peatlands found in the Amazon, particularly those located in the western Amazonia, where environmental factors such as precipitation and river migration are more intense in terms of magnitude than in other parts of the basin. I apologise for the unclear redaction. I have improved this paragraph; hopefully, it will be clearer. See Lines 22-27.
L25. Remove ‘their history and’.
Response: Done
L37. Important how? Carbon-rich? Biodiversity? Oldest?
Response: We speculate -because other peatlands within the Amazon basin remain unquantified- that these western peatlands are the largest peatland area of the Amazon, and they are likely the most carbon-dense and also the most studied. We have added that they are probably the most extensive and carbon-dense within the region. Lines 39-41.
L40. Here or in the methods? Could you explain how/why these sites have developed to be a pole forest and a palm swap?
Response: We have added a brief explanation to the Introduction (Lines 44–47) based on the pollen records from the two study sites. The pollen records indicate that the NYO site underwent several vegetation changes before developing into a pole forest, including phases characterised by palm swamp and seasonally flooded forest vegetation (Åkesson et al. 2026). Similarly, the VEN site experienced changes from open swamp to palm swamp vegetation. These shifts in vegetation are potentially associated with changes in hydrological conditions, including river migration, distance from the river, and changes in flooding dynamics (Lawson et al., 2026).
L43- L48. Net primary production? How was this measured? Can this be used as validation data in this study?
Response: The Net Primary Production data for each site were derived from field measurements collected over one year and compiled by Dargie et al. (2024). Where Total NPP (Mg C ha⁻¹ year⁻¹) was calculated as the sum of the mean NPP contributions from the different components, including litter, stems, and roots. These NPP data were used as input data to parameterise the model and therefore form part of the model parameterisation. For this, they cannot be considered independent validation data for the model outputs.
L67. Large? Do you mean tall? How tall? How thin? What are the dominant aboveground species? What are the dominant peat-forming species?
Response: Thanks for pointing out this error. We have revised the description to clarify that the trees at this site are taller and have smaller DBH than those at the other site. The dominant aboveground species include Pachira nitida and Hevea guianensis, with average tree heights of approximately 24 m and DBHs of approximately 17 cm. Further details on structural traits and species composition are described in Dargie et al. (2024) and Honorio Coronado et al. (2021). This description was included in Lines 71- 74.
L70. What is a permanent vegetation plot?
Response: By permanent, we mean a plot that was established for repeated monitoring of an extended period, rather than a plot surveyed during a single field visit. In this case, the plot was monitored for 1 year (2018-2019), during which biomass measurements were collected and decomposition experiments were conducted. We simplified the phrase to avoid confusion. Line 77.
L71. What method was used to collect the core?
Response: Peat cores were collected using a Russian-type peat corer. We have added this information to the methods description. Line 78.
L72. Re Åkesson et al. data? Do you use this? Why is it referenced?
Response: Åkesson et al., in prep (in preparation) corresponds to an ongoing manuscript describing the palaeoproxies data obtained from the peat core, including radiocarbon dating, pollen analysis, bulk density, LOI and others. These data were provided by the authors for use in our modelling purposes. However, this reference was updated because the manuscript was recently published; it is now Åkesson et al. (2026).
L73 – L76. What time periods does ‘initial formation’, ‘middle to later stage’, and ‘most recent stage’ span? What were the drivers of these transitions?
Response: This is an important point. We have added the approximate timing of the different stages in parentheses for clarification (Lines 80-85). The initial stage of peat formation occurred at approximately 6,000 cal. yr BP, when the site was characterised by palm swamp vegetation. The middle-to-later stage spans approximately 3,700–200 cal. Your BP and was characterised by seasonal flooded forest and palm swamp vegetation. The most recent stage occurred at approximately 100 cal. yr BP, when the site developed into a pole forest.
Regarding the drivers of these transitions, the available pollen record does not provide sufficient evidence to identify specific environmental or climatic drivers. Rather, the pollen data indicate changes in the dominant vegetation through time. We therefore defined the vegetation stages and transition periods based on changes in dominant vegetation identified in the pollen record, without attributing these transitions to a specific driver. The vegetation periods identified for the VEN and NYO sites, together with the calculations used to define the transition boundaries and the corresponding model timing, are provided in Appendix A2.
Comment: When giving the length of the ‘peat core’, was this the length of the core or the peat section? Did both cores include basal peat? It’s unclear.
Response: The reported core length refers to the length of the peat section, rather than the total length of the core. As the model is only capable of representing peat accumulation. We have clarified this in the text (Line 77 and Line 89).
L85. Again, what were the drivers of the transitions? What is the dominant vegetation today? And the dominant peat forming vegetation?
Response: This comment is related to the clarification provided in response to Comment L73–L76. The present-day dominant vegetation and the dominant peat-forming vegetation were identified from the pollen record and field observations, respectively, from studies such as Dargie et al. (2024), Honorio Coronado et al. (2021), and Åkesson et al. (2026). The dominant vegetation at NYO and VEN is pole forest (swamp hardwood forest) and palm swamp, respectively. Likewise, the dominant peat-forming vegetation is hardwood forest and palm trees (assuming you are also referring to the present day).
L95. Here an elsewhere. I don’t see the need to abbreviate ecosystem, palm swamp, and pole forest. It reduces the readability of the manuscript.
Response: Thanks for this suggestion. We understand that the use of abbreviations sometimes limits readability; however, to maintain a balance in wording length, we use abbreviations for the names of ecosystem types, as they are repeated many times throughout the manuscript. We change the EcoTy abbreviation to the corresponding word or phrase, and readability improves.
L98. Are you calling the new parameterisations, a new model? Is it not just a new application of an existing model? Was any model code changed? If only new parameters, I wouldn’t change the name of the model.
Response: No, we are not calling a new parameterisation a new model. The conceptual framework, governing equations, and core processes of HPMTrop remain unchanged. However, the model code was substantially modified to enable the use of different parameter sets for different ecosystem types and successive periods of ecosystem development. In the original HPMTrop implementation, a single parameter set is applied throughout the simulation, effectively representing one ecosystem type. In the extended implementation, the model can dynamically switch between ecosystem-specific parameter sets according to the ecosystem type and its corresponding period identified from the palaeoecological record.
Thus, HPMTrop_EcoTy is not a fundamentally new model or a new set of governing equations. Rather, it is an extended implementation of HPMTrop that incorporates ecosystem-type- and period-dependent parameterisation. We use the name HPMTrop_EcoTy to distinguish this extended implementation from the original HPMTrop code and to make explicit its main new functionality.
The code modifications go beyond simply changing parameter values. In particular, we introduced new configuration and parameter-management files, including hpmT2_params_config.m, hpmT2_main_config.m, hpmT2_global_figures_init.m, and hpmT2_local_figures_init.m. Most importantly, the parameter-management structure was redesigned. In the original implementation, parameter values were entered directly into the MATLAB code, whereas HPMTrop_EcoTy uses an external Excel file containing the parameter values for the different ecosystem types. The new parameter-management code reads these values and assigns the appropriate parameter set according to the ecosystem and simulation period.
We have therefore added an illustrative figure to point out the differences between HPMTrop and HPMtrop_EcoTy in Appendix Figure B8. And in the repository there is the corresponding HPMTrop_Ecoty code.
Comment: Table 2. Where does this information come from? You’ve derived this from the sampled cores? Does the information become model input? Or validation data?
Response: The information presented in Table 2, now Table 3 in the revised manuscript, was derived from two sources. Column 2 (periods, model input) was derived from the interpretation of the pollen records obtained from the sampled peat cores at each study site and was used as input to define the vegetation periods represented in the model. Column 6 (vegetation dominance) was also based on the interpretation of the pollen records, but is provided only as a description of the dominant vegetation during each period and is not used as a model input.
The remaining columns present model outputs from the simulations for each vegetation period or ecosystem type. These include peat accumulation rate, peat height, carbon accumulation rate, NPP, decomposition, and water table. Therefore, Table 3 provides a summary of the model simulations for each ecosystem type and period at the two study sites, distinguishing between the pollen-derived information used to parameterise the temporal vegetation sequence and the variables produced by the model.
Comment: Figure A1. Simulation results should be in the results section.
Response: We thank the reviewer for this recommendation. Figure A1 presents the results of the model calibration, rather than the final simulation results used for the analysis. The purpose of this figure is to show how the simulated peat accumulation developed during model calibration after incorporating the vegetation periods into the model code, and to demonstrate the resulting agreement between simulated and observed peat-depth–age relationships. The analysis of peat accumulation dynamics was conducted only after the calibration was completed and the model adequately reproduced the observed peat-depth–age relationship. Therefore, we consider Figure A1 to document the model calibration and development process rather than a main result of the study. For this reason, we have retained the figure in the Appendix.
Figure 1. Please change labels, core VEN-O2 and NYO-03 to site names. Can you show a photo of the core for both sites? Perhaps this figure can be sperated into two: First figure: The two Maps. The second figure: photo of each site (are these photos of these specific sites or ecosystem type in general. Specific site would be preferable), Photo of each core, radiocarbon dates of each core. It would be good to perhaps use dots in the figure so it is clear, at what ages, the samples were taken .
Response: We have requested a photo of each core from the dataset manager. The photos representing the Palm Swamp and Pole Forest ecosystems are from the same area where the cores were taken and are intended to show differences in aboveground vegetation between these ecosystems; however, we have requested specific photos of the sites of interest. We have revised and added the dots to the core reference curves to clearly indicate the ages at which samples were collected. We maintain a single figure to avoid increasing the manuscript length. The labels VEN-O2 and NYO-03 correspond to the specific core, not just the site; we considered it more appropriate to keep the core label rather than the site name.
L105. Can you include equations in SI?
Response: We have included all of the equations in Appendix B.
L116. Where can I find the codedataavailability or Table HPMTrop_EcoTy_Parameters.xlsx? I looked throughout the manuscript but could not find it.
Response: The file was originally provided in the Code Availability section (L440) of the manuscript. In the revised manuscript, this information has been moved to Lines 469–470. The repository containing the code and the HPMTrop_EcoTy_Parameters.xlsx file is available at DOI: 10.5281/zenodo.18713887 (Zenodo repository). Link: https://shorturl.at/dgBoZ
Comment: Table 1. How does the pollen record vs the peat core agree/disagree? Are the dates radiocarbon dates? How were the ecosystem types determined? Using pollen or peat core visual analysis? Figure A3 and A4 are useful to show you developed the ecosystem types and transitions. It would be great if these were included in the main manuscript alongside the core photos.
Response: The dates reported in Table 1 (now Table 2, Line 152) are directly derived from the pollen diagrams, following the criteria described in Appendix A2. Importantly, the base of the calculation of these periods was taken from the core age associated with the length of the peat section.
They are therefore not individual radiocarbon dates. The pollen records and the radiocarbon dating agree in the total period of analysis, but the timing resolution is different.
The temporal framework of each peat sequence was established from radiocarbon dating of the peat section (1129 yBP for VEN-02 and 6175 yBP for NYO-03). Whereas the pollen record provides the vegetation composition associated with different periods. All pollen-derived ages used here were provided by Åkesson et al., 2026 and converted to the model's internal time coordinate transforming age from pollen record to model time using as base the radiocarbon date of the core corresponding to the peat section.
The periods used for model parameterisation were extracted from the pollen records because these provide the resolution required to identify changes in ecosystem state. However, the radiocarbon dating curve from the core was used as a reference curve to calibrate the model (peat-depth age curve). The ecosystem types were therefore determined primarily from the pollen assemblage composition and its changes.
Following the reviewer's suggestion, we have requested core photographs to include alongside Figures A3 and A4. These figures remain in the supporting information as they are not the main focus of discussion of the paper, they were useful to define ecosystem types and the corresponding stable and transitional periods identified from the pollen records and provide a link between ecosystem parameterisation and periods used in the model.
L140. Precipitation product was compared to meteorological station data and bias corrected. Can you show this in a figure?
Response: We have added a new Figure A1 to the Appendix, together with Appendix A1, “Bias correction of precipitation forcing.” Figure A1 shows three monthly precipitation time series: the original CCSM3 TraCE-21ka data, ground-station observations from SENAMHI, and the bias-corrected TraCE-21ka precipitation data after applying the monthly bias-correction factors.
L152. Was this WTD assumption validated against in situ site observations? This is critical to determine whether the simulated changes in NPP and peat accumulation are related to representative changes in WTD.
Response: Water table (WT) was obtained from dataloggers measurements in the two sites (NYO-03 and VEN-02; Darguie et al., 2024 and Flores Llampazo et al., 2022) and the water deficit (WD) is estimated from a linear relationship developed between water table and water deficit. This relationship was developed by ecosystem:
- Pole forests (data from NYO-03; Dargie et al., 2024),
- Palm samp (data from VEN-02; Dargie et al., 2024), and
- Seasonal flooded forest (data from Flores Llampazo et al., 2022),
The relationships between WT and WD have now been added to Appendix B1.
The assumption of evapotranspiration fixed in 100 mm is supported by Aragão et al., 2007; Griffis et al., 2020 and follows the same approach as Kurnianto et al 2014.
The same approach was used for NPP. NPP was parameterized using non-linear relationships between NPP and WT based on field observations for each ecosystem type: pole forest (NYO-03; Dargie et al., 2024), palm swamp (VEN-02; Dargie et al., 2024), and seasonal flooded forest (del Aguila-Pasquel et al., 2022). Because field NPP data were not available for open peatland, a general peatland NPP dataset from Hirano et al. (2012) was used. The NPP–WT relationships have now been added to Appendix B2.
The reliability of these outputs from the simulation (WT and NPP time series) relies on the range of the time series output that falls in an ecological plausible range for each type of ecosystem.
L155. Please include the equations for NPP, litter, peat depth, peat height, and carbon storage here in the main text. These equations are critical for interpreting the results of this study.
Response: We have added a new section in the supplementary material (Appendix B) presenting the equations in a logical flow used in HPMTrop, including those for NPP, litter production and decomposition, peat mass, peat depth and height, and carbon storage, together with a description of the model formulation. The parameters used in these equations are also described in detail in the HMTrop_EcoTy_Parameters table, which is available in the Zenodo repository (DOI: 10.5281/zenodo.18713887) https://shorturl.at/dgBoZ.
L200. Peat height relative to what?
Response: In the model, peat depth and peat height are treated as very different concepts. As shown in the equations in the previous comment. Peat depth is the cumulative peat during the entire simulation, which means the final peat depth at the end of the simulation, therefore, corresponds to the core peat depth curve. On the other hand, peat height is relative to the year, which means the peat accumulated during that year. In other words, we can say that peat depth is the final photo of the peat accumulation, and peat height is the photo by year tracking the peat accumulation dynamic. So peat height allows us to understand which year peat was increased or decreased during the entire simulation. These concepts were defined and added to the main text, Lines 129-132, and Appendix B3.
Comment: Fig 4. Core/observed data should be in black. Please then use a solid line for the base simulation with shading to indicate ±25%. Which parameters were changed? Why does some figures have a grid and others not? Please be consistent. Why does the shaded grey patch span across all time period but the caption reads that it represents the Holocene hiatus? Shouldn’t this be vertical? Can you also include lines or discuss over which time periods you expect, based on literature, the Holocene hiatus at these sites?
Response: We have revised Figure 4 (now Figure 5) regarding the sensitivity simulations, and the parameters varied in each panel are explicitly identified in the figure caption. Panels (a) and (b) show changes in NPP, panels (c) and (d) show changes in decomposition, and panels (e) and (f) show changes in precipitation. These simulations were conducted to assess the sensitivity of the modelled peat accumulation to these key processes.
The peat height plot was presented with a grid, as this represented the change by year, while the peat depth is only interesting to see at the end of the simulation, so the grid is not needed. However, for homogenisation, we’ll remove the grid in both plots.
The shaded path in grey only tries to highlight the period during which peat accumulation was minimal. We follow other similar studies such as Garcin et al., 2022 and Young et al., 2021.
Yes, we discussed the “hiatus” in Lines 307-320 with the reported “ghost interval" from Congo (Garcin et al 2022 and Young et al 2021) and it was added to the hiatus reported by Swindles et al. (2017), for another site in the same basin in Peruvian Amazonia (PMFB). However, as one hypothesis from this study, we do not attribute the hiatus to the same causes in the Congo Basin. This reduction in peat accumulation following the simulation results is associated with the ecosystem dominance which corresponds to a non-peat forming ecosystem represented by a deeper and dynamic WT and faster decomposition.
Discussion: Perhaps, I missed it but a limitation of this study seems to be the reliance of pollen analysis to infer vegetation presence. Pollen records only capture pollen producing plants. Can the authors discuss to what extent non-pollen producing plants likely to have been present at these sites?
Response: It is a great point. We agree that reliance on pollen analysis is a limitation of this study because pollen records primarily capture pollen-producing plants. In this study, however, pollen analysis was used to distinguish broad ecosystem categories based on the dominant vegetation types represented in the pollen record, rather than to reconstruct the complete plant community at each site. Therefore, non-pollen produced species present in the past are not included in the category, as we don't have information on them. However, we consider that absence of these non-pollen produced species from the pollen record is unlikely to affect the broad ecosystem classification, as the most conspicuous species in present-day (used to classify the same type of ecosystem) are pollen produced species and were registered in the pollen record. At least for our modelling purposes, where broad periods of ecosystem types are needed, we assume that the pollen record is a good source of these classification periods.
Comment: Table A1. It's a bit unclear but it seems that these are model input parameters. Without presenting model equations, it's difficult to understand how these parameters are relevant to the study or used by the model. Please provide model equations in the methods of the main text.
Response: Thanks for the comment. The tables A1 and A2 provide the initial values for the simulation parameters for NPP and decomposition, which are included in the main table of parameters and the code. These values trigger the start of the simulation. For instance, instead of 0 in the NPP, the model uses the initial value provided in this table to initialise the NPP variable, and then the equation (in Appendix B2) associated with NPP and WT is used to recalculate this value within the model on a monthly time step.
Citation: https://doi.org/10.5194/egusphere-2026-2272-AC2
-
AC1: 'Reply on RC1', Yarin Tatiana Puerta Quintana, 05 Sep 2026
-
RC2: 'Comment on egusphere-2026-2272', Anonymous Referee #2, 15 Aug 2026
First, I apologize for the delay in providing this review, as August is a common vacation period.
This manuscript uses a relatively simple one-dimensional model to simulate long-term peat and carbon accumulation at two peatland sites in the Peruvian Amazon. This is an important research topic, as our understanding of the long-term development of tropical peatlands—and peatlands more generally—remains limited. I find the research question valuable and clearly defined. Although the model is relatively simple, it appears broadly capable of reproducing long-term peat accumulation at the two sites. Nevertheless, its simplicity introduces several assumptions and uncertainties that require clarification.
I hope the following comments will help improve the manuscript.
-
Use of abbreviations. The frequent use of abbreviations somewhat disrupts readability. Examples include pole forest (PF), palm swamp (PS), open peatland (OP), seasonally flooded forest (SFF), and ecosystem type (EcoTy). Because these are not commonly used abbreviations and the full terms are relatively short, I often had to return to earlier sections to recall their meanings. I suggest reducing the number of abbreviations, particularly “EcoTy”.
-
Precipitation tipping point. How likely is a precipitation reduction of 30–45% at these particular sites under future climate and land-use scenarios? If such a reduction is unlikely, I suggest presenting the result primarily as precipitation sensitivity rather than as evidence of a tipping point. The use of “tipping point” may require stronger evidence than a sharp response among several prescribed precipitation scenarios.
-
Constant evapotranspiration. The assumption of constant evapotranspiration at 100 mm month⁻¹ seems highly simplified, particularly because the simulated water table depends directly on this value. Are evapotranspiration data available from the palaeoclimate simulation? Alternatively, could evapotranspiration be reconstructed using temperature, precipitation, or another simple method? At minimum, the uncertainty introduced by this assumption should be discussed or evaluated through sensitivity analysis.
-
NPP–water-table relationship. The empirical relationship between NPP and water-table position is an important component of the model. Can this relationship be evaluated against independent observations or through cross-validation? The manuscript should provide information about the amount of data used, goodness of fit, and uncertainty in the fitted relationship. Could past temperature variability also have affected NPP? Although temperature may not be a major present-day limitation on tropical peatland productivity, the authors should discuss whether temperature changes over the last approximately 6000 years may be relevant.
-
Peat depth versus peat height. Around L203, please clearly explain the difference between “peat depth” and “peat height”. Figures 2 and 3 appear to show similar temporal patterns, with the main visual difference being the direction of the y-axis. Please clarify their physical meanings, how the model calculates each variable, and why both need to be presented.
-
Human influence. Around L312, were these peatlands affected by human activities during their developmental histories? If anthropogenic disturbance, burning, vegetation modification, or hydrological alteration is plausible, its omission should be acknowledged as a source of uncertainty.
-
Sign convention in Figure 5. Net ecosystem carbon uptake is often presented as negative, whereas carbon loss to the atmosphere is positive. Figure 5 appears to use the opposite convention. Please either adopt the more common convention or clearly state the sign convention in the axis label and caption.
-
Causes of ecosystem transitions. Section 4.4 would benefit from a clearer explanation of why the ecosystem types changed over time. Were these transitions related to precipitation, river dynamics, local hydrology, autogenic succession, or a combination of these processes? If precipitation is considered a possible driver, showing the reconstructed precipitation history together with the prescribed ecosystem transitions would help readers understand their relationship.
-
Figure B1. Please add a clearly labelled scale for water-table position, apparently on the right-hand y-axis. The units, sign convention, and reference level should also be specified.
-
L4. Typo in “parameterisations”.
-
River migration. River migration is not explicitly represented in the model. Therefore, the conclusion that it controls peatland development should be moderated or better supported.
-
Model transferability. Please discuss whether the calibrated model and its parameters can be applied to other peatland sites, or whether substantial site-specific calibration would be required.
Overall, I consider this modelling framework useful as an exploratory and hypothesis-generating tool. However, the manuscript should more clearly distinguish between model calibration and independent validation, and between processes represented explicitly by the model and interpretations inferred from external evidence. With these clarifications and more cautious interpretation of the tipping-point and river-migration results, the study could make a useful contribution to understanding long-term tropical peatland development.
Citation: https://doi.org/10.5194/egusphere-2026-2272-RC2 -
AC3: 'Reply on RC2', Yarin Tatiana Puerta Quintana, 09 Sep 2026
First, we would like to sincerely thank the reviewer for the thoughtful and constructive comments. They encouraged us to reflect more deeply on our work and to consider several aspects from different perspectives, which has helped us improve the manuscript. We greatly appreciate the time and effort dedicated to reviewing our work and the valuable suggestions provided. Below, we respond to each comment in detail.
Citation: https://doi.org/10.5194/egusphere-2026-2272-AC3
-
-
AC4: 'Reply on RC2', Yarin Tatiana Puerta Quintana, 09 Sep 2026
- Use of abbreviations. The frequent use of abbreviations somewhat disrupts readability. Examples include pole forest (PF), palm swamp (PS), open peatland (OP), seasonally flooded forest (SFF), and ecosystem type (EcoTy). Because these are not commonly used abbreviations and the full terms are relatively short, I often had to return to earlier sections to recall their meanings. I suggest reducing the number of abbreviations, particularly “EcoTy”.
Response: We agree with the reviewer that the frequent use of abbreviations can affect readability. In response, we have removed the abbreviation “EcoTy” throughout the manuscript and replaced it with the term “ecosystem type” (or “type of ecosystem”, or “ecosystems” where appropriate). We have retained the abbreviations PF, PS, OP, and SFF for the individual ecosystem types because these terms occur frequently throughout the manuscript. Retaining these abbreviations helps avoid repetitive use of these terms and keeps the text concise.
- Precipitation tipping point. How likely is a precipitation reduction of 30–45% at these particular sites under future climate and land-use scenarios? If such a reduction is unlikely, I suggest presenting the result primarily as precipitation sensitivity rather than as evidence of a tipping point. The use of “tipping point” may require stronger evidence than a sharp response among several prescribed precipitation scenarios.
Response: We agree with the reviewer that the term “tipping point” may imply stronger empirical evidence than is available from our prescribed precipitation scenarios. Even with our sensitivity analysis, we do not have sufficient evidence to conclude that a 45% reduction in precipitation is likely at our study sites under future climate and land-use scenarios. Although a 45% reduction has been identified as a threshold for Amazonian peatlands in previous work (Flores et al., 2024).
Attending to this recommendation and to avoid exaggerating the implications of our results, we have therefore replaced “tipping point” with “threshold” throughout the manuscript. We frame these sensitivity results primarily as evidence that peatlands are sensitive to precipitation reduction, identifying a precipitation threshold associated with a strong peatland ecosystem response.
- Constant evapotranspiration. The assumption of constant evapotranspiration at 100 mm month⁻¹ seems highly simplified, particularly because the simulated water table depends directly on this value. Are evapotranspiration data available from the palaeoclimate simulation? Alternatively, could evapotranspiration be reconstructed using temperature, precipitation, or another simple method? At minimum, the uncertainty introduced by this assumption should be discussed or evaluated through sensitivity analysis.
Response: We agree that the assumption of a constant evapotranspiration (ET) rate of 100 mm month⁻¹ is a simplification and that, because ET directly influences the simulated water table, its potential effect on model outcomes should be evaluated. We explored whether ET could be obtained from the palaeoclimate simulations or reconstructed using meteorological variables. However, no suitable palaeoclimate ET dataset was found at an appropriate spatial and temporal resolution for our study sites. We also evaluated several approaches for estimating potential evapotranspiration, including the Priestley–Taylor (Priestley & Taylor, 1972), Penman–Monteith, and Oudin equations. These approaches would require additional assumptions regarding parameters that are uncertain for the Holocene, while the available temperature reconstructions (e.g., Temperature 12k) also contain substantial uncertainties when downscaled to individual sites. We therefore retained the constant ET to maintain model parsimony.
The value of 100 mm month⁻¹ was selected based on previous ecological studies reporting comparable evapotranspiration fluxes for tropical forests in the Amazon (Aragão et al., 2007; Griffis et al., 2020) and following the same approach as Kurinanto et al. (2014), who used HPMTrop in other tropical peatlands. As an additional check, we estimated present-day potential evapotranspiration for the study region using the Thornthwaite equation (Thornthwaite & Mather, 1957) with temperature data from the Quistococha flux tower in Peru, obtaining an average value of approximately 116 mm month⁻¹.
Importantly, thanks to this suggestion, we conducted a sensitivity analysis by increasing the constant ET rate from 100 to 110, 116, 125, 135 and 145 mm month⁻¹, corresponding to increases of 10%, 16%, 25%, 35% and 45%, respectively. We found that increasing ET resulted in a progressively deeper water table and reduced final peat depth. A 45% increase in ET reduced the average simulated WT level by approximately 31% and reduced final peat depth by approximately 24%. In comparison, 10%, 16%, 25% and 35% increases in ET reduced final peat depth by approximately 2%, 5%, 8% and 19%, respectively. Reducing ET below the reference value of 100 mm month⁻¹ resulted in positive changes in final peat depth of less than 2%.
These results indicate that ET is an important control on simulated WT dynamics and, consequently, on peat accumulation. However, they also show that moderate increases in ET produce relatively limited changes in final peat depth, whereas larger increases have a more substantial effect. We have therefore added the sensitivity analysis in the manuscript (lines 436-437).
Finally, although river flow is not explicitly represented in the model, we find a relatively strong correlation between the observed present-day WT and river-level time series (R = 0.72), suggesting that river dynamics may partly buffer the influence of atmospheric water losses on local water-table conditions. This provides additional context for interpreting the sensitivity of the simulated water table to ET, although we acknowledge that this buffering mechanism cannot be directly evaluated within the current model structure.
We have retained 100 mm month⁻¹ as the reference ET value because it is supported by previous ecological studies and provides a parsimonious representation in the absence of reliable site-specific palaeo-ET data. We have nevertheless made the associated uncertainty explicit and now identify the representation of ET and its relationship with WT dynamics as an important area for future model development.
- NPP–water-table relationship. The empirical relationship between NPP and water-table position is an important component of the model. Can this relationship be evaluated against independent observations or through cross-validation? The manuscript should provide information about the amount of data used, goodness of fit, and uncertainty in the fitted relationship. Could past temperature variability also have affected NPP? Although temperature may not be a major present-day limitation on tropical peatland productivity, the authors should discuss whether temperature changes over the last approximately 6000 years may be relevant.
Response: We developed an empirical quadratic relationship between NPP and water-table (WT) position separately for each ecosystem type, using available field-based observations. The datasets used to parameterise these relationships are as follows: (1) pole forest, using NPP observations from site NYO-03 reported by Dargie et al. (2024); (2) palm swamp, using NPP observations from site VEN-02 reported by Dargie et al. (2024); (3) seasonal flooded forest, using data from del Aguila-Pasquel et al. (2022); and (4) open peatland, for which ecosystem-specific observations were unavailable, so we used the peatland NPP data reported by Hirano et al. (2012). The number of observations, fitted quadratic relationships, and corresponding coefficients of determination (R²) are now provided in Appendix B2.
We acknowledge that an independent validation dataset or formal cross-validation could not be performed due to the limited available observations. We therefore do not present the NPP–WT relationships as independently validated relationships. Instead, they represent empirical parameterisations constrained by the available field observations. As an additional check, we evaluated whether the resulting simulated NPP values remained within ecologically plausible ranges for the corresponding ecosystem types.
Regarding temperature, we agree that past temperature variability could potentially influence NPP and have added a discussion of this issue to the manuscript (Lines 444-446). Temperature is not explicitly included as a driver of NPP in HPMTrop_EcoTy. Instead, the model represents changes in ecosystem-specific productivity through the temporal shifts in vegetation types inferred from the pollen records. Thus, when the reconstructed vegetation state changes from one ecosystem type to another, the associated ecosystem-specific NPP parameterisation also changes. This approach indirectly incorporates the integrated ecological response to environmental changes that are reflected in vegetation turnover, while avoiding the introduction of poorly constrained temperature–NPP relationships.
For the approximately 6000-year simulation period considered here, we consider this simplification reasonable for western Amazonia because available palaeotemperature studies generally indicate relatively limited Holocene temperature variability compared with the pronounced temperature changes associated with the transition from the Last Glacial Maximum to the Holocene. For example, Bendle et al. (2010) reconstructed relatively stable temperatures of approximately 26 °C over the last ~9500 years in eastern Amazonia. More recently, a brGDGT-based reconstruction from western Amazonia (study in progress) has suggested a low-amplitude, gradual warming trend during the Holocene (Hällberg et al., 2026). These reconstructions suggest that large temperature-driven changes in NPP are unlikely to be the dominant control on the long-term peatland dynamics simulated here, although we acknowledge that the available palaeotemperature data are spatially limited and that smaller-scale temperature variability cannot be excluded.
- Peat depth versus peat height. Around L203, please clearly explain the difference between “peat depth” and “peat height”. Figures 2 and 3 appear to show similar temporal patterns, with the main visual difference being the direction of the y-axis. Please clarify their physical meanings, how the model calculates each variable, and why both need to be presented.
Response: Thanks a lot for pointing out the need to clarify the distinction between peat depth and peat height. Although these variables are related, they represent different quantities in the model, and we have added this to the manuscript to make this distinction explicit. Let me explain better here:
Peat depth is the cumulative peat over the entire simulation. It tracks the vertical position within the peat profile, and at the end of the simulation, the resulting peat-depth profile represents the present-day vertical distribution of peat and can therefore be compared with the depth profile reconstructed from a peat core. The age associated with each depth in Figure 2 represents the age of the peat material at that position in the core, based on the corresponding dating information.
Peat height represents the total thickness of peat accumulated at each model time step. It therefore describes the temporal trajectory of peat accumulation, from the initial state with no accumulated peat to the final peat thickness at the end of the simulation. For example, in Figure 3, the coordinate (~6000 yr BP, 0 m) represents the initial model state with no accumulated peat, whereas (0 yr BP, ~4.5 m) represents the final simulated peat thickness at the present day. Changes in peat height between successive time steps indicate periods of greater or lower peat accumulation.
Thus, although Figures 2 and 3 show related temporal patterns, they answer different questions. Peat depth describes the vertical position and age-depth structure of the accumulated peat profile, whereas peat height describes the temporal development of the total peat thickness. Both variables are therefore presented because the depth allows comparison with the present-day peat-core profile, while the height allows us to examine the dynamics of peat accumulation throughout the simulation. To avoid confusion with this two concepts, we have made the following changes to the manuscript:
- Added the equations defining peat depth and peat height in Appendix B. - Added “Age (years BP)” to the x-axis of Figure 2 (now Figure 3) to clarify that this axis represents the age of peat at different depths in the present-day core. - Added explicit definitions of peat depth and peat height in Section 2.2, Model application description (Lines 129-132).
- Human influence. Around L312, were these peatlands affected by human activities during their developmental histories? If anthropogenic disturbance, burning, vegetation modification, or hydrological alteration is plausible, its omission should be acknowledged as a source of uncertainty.
Response: We agree that anthropogenic influences could be an important source of uncertainties as human activities have not been taken into account in the model. The objective of this study was to develop a baseline representation of peatland development under the environmental conditions, while incorporating anthropogenic disturbance scenarios, including land-use change and fire, is planned as a follow-up work of the model application. For the Peruvian Amazon peatlands considered in this study, available paleoecological evidence does not indicate a clear or substantial anthropogenic influence on long-term vegetation development (Roucoux et al., 2013; Åkesson et al., 2026). Human use of Peruvian Amazonian peatlands suggests that the main activities, such as harvesting fruits, hunting, and timber, have been developed sustainably (Schulz et al., 2019), different from central and eastern Amazonia, where some proxies indicate the significant legacy of past human activities (McMichael et al., 2025).
We have now explicitly acknowledged this limitation in the manuscript adding a few lines of discussion in section 4.4 of limitations (Lines 447-452).
- Sign convention in Figure 5. Net ecosystem carbon uptake is often presented as negative, whereas carbon loss to the atmosphere is positive. Figure 5 appears to use the opposite convention. Please either adopt the more common convention or clearly state the sign convention in the axis label and caption.
Response: Thanks for raising this point. Yes, we agree that the sign convention in Figure 5 could cause confusion if it is not explicitly defined. The variable shown in Figure 5 is the Net Carbon Balance (NCB), also referred to as Net Ecosystem Carbon Balance (NECB). In our simulation positive NCB values indicate net carbon gain by the peatland ecosystem, representing net carbon uptake and accumulation within the ecosystem, whereas negative values indicate net carbon loss from the ecosystem (ecosystem perspective).
This differs from the convention commonly used for Net Ecosystem Exchange (NEE), which is typically expressed from the atmospheric perspective, where negative values represent net ecosystem carbon uptake (a sink) and positive values represent carbon release to the atmosphere (a source). Because our analysis focuses on changes in carbon stored within the peatland ecosystem, we use the ecosystem perspective for NCB. To avoid ambiguity, we have explicitly defined this sign convention in the caption of Figure 5 (now Figure 6) in the manuscript.
- Causes of ecosystem transitions. Section 4.4 would benefit from a clearer explanation of why the ecosystem types changed over time. Were these transitions related to precipitation, river dynamics, local hydrology, autogenic succession, or a combination of these processes? If precipitation is considered a possible driver, showing the reconstructed precipitation history together with the prescribed ecosystem transitions would help readers understand their relationship.
Response: Thank you for this important comment. We have discussed more about the mechanisms underlying the observed ecosystem transitions in Section 4.4 to clarify both the evidence available from the pollen record and the potential environmental processes that may have contributed to these transitions.
The pollen record identifies changes in dominant vegetation, but it does not provide sufficient evidence to attribute individual transitions to a specific climatic or hydrological driver. We therefore interpret the transitions as potentially resulting from a combination of interacting processes, including local hydrological conditions, fluvial dynamics, and autogenic vegetation development. In particular, previous paleoecological studies suggest that river migration and associated fluvial dynamics may have been an important driver of ecosystem change in the Peruvian Amazon. Conversely, the available evidence suggests relatively stable Holocene climate conditions in this region, providing limited support for major precipitation-driven shifts (Swindles et al., 2017; Lawson et al., 2026). We have added further discussion of these potential drivers and their interactions in Section 4.4 (Lines 412–415).
The timing of ecosystem transitions in the simulations was prescribed as periods identified from the pollen records (Appendix A2). Thus, the model does not mechanistically simulate the environmental processes responsible for initiating these historical ecosystem transitions. Instead, when a transition occurs, the corresponding ecosystem-specific parameterisation is applied, which alters processes such as water table dynamics and productivity.
The reconstructed water-table dynamics along with the prescribed ecosystem transitions are shown Appendix Figure A8. This figure illustrates how the ecosystem-specific parameterisations influence hydrological conditions during the simulated periods. Although precipitation is an important model driver, the reconstructed TraCE-21ka precipitation series is relatively stationary over the period considered, with no clear long-term trend or major climatic events that would provide strong evidence for precipitation as the primary driver of the observed ecosystem transitions. In general better precipitation climate reconstruction is needed for the Amazonian.
- L4. Typo in “parameterisations”.
Response: done.
- River migration. River migration is not explicitly represented in the model. Therefore, the conclusion that it controls peatland development should be moderated or better supported.
Response: We agree that, because river migration is not explicitly represented in our model, our conclusions should not imply that river migration has been demonstrated to control peatland development.
Our simulations, together with available paleoecological evidence, suggest that large-scale precipitation changes are unlikely to be the primary explanation for the observed ecosystem transitions and peat accumulation in the Peruvian Amazon. The TraCE-21ka precipitation reconstruction used in our simulations shows relatively stable conditions during the Holocene. Previous studies have also suggested that relatively stable Holocene climate conditions in the region may have limited the role of large-scale precipitation variability, while fluvial dynamics may have played an important role in shaping local environmental conditions.
However, because river migration and other local processes are not explicitly represented in the model, we cannot directly test their effects or establish river migration as the causal mechanism underlying the observed transitions. We have therefore stated that our results, in combination with previous paleoecological evidence, suggest that local factors, including fluvial and hydrological processes, may have been more important for peat accumulation and ecosystem development than large-scale precipitation variability in the Peruvian Amazon peatlands (Lines 465–470).
- Model transferability. Please discuss whether the calibrated model and its parameters can be applied to other peatland sites, or whether substantial site-specific calibration would be required.
Response: The HPMTrop_EcoTy parameterisation includes four ecosystem types: (1) pole forest (PF), representing hardwood, thin-stemmed peatland forest with palms; (2) palm swamp (PS), characterised by palm-dominated vegetation such as Maruitia; (3) seasonal flooded forest (SFF), representing hardwood forests that are generally not peat-forming; and (4) open peatland (OP), representing open-canopy peatlands dominated by grasses. If the evolutionary history of another site can be represented by one or more of these ecosystem types, the corresponding ecosystem-specific parameters can potentially be transferred without substantial recalibration.
However, site-specific information remains necessary, particularly for the precipitation forcing and the timing and sequence of ecosystem transitions. In addition, the relationships between water-table depth and water deficit (WT–WD) and between water-table depth and NPP (WT–NPP), described in Appendices B1 and B2, may require site-specific parameterisation when sufficient observational data are available. Likewise, productivity values should not necessarily be assumed to be transferable among ecosystems or sites with different environmental conditions.
Thanks to this comment, we applied an initial test of transferability. We used the same ecosystem-specific parameterisation on another core in the same region of Veinte de Enero, an independent site for which peat-depth information (158 cm and 1399 years) and pollen-derived ecosystem periods are available. At this new site, two major ecosystem periods were identified: open and palm swamp (similar to our VEN-02 site). We ran the model using the same parameter values as for the original site VEN-02, changing only the site-specific precipitation time series and the timing of the ecosystem periods. The model produced a good representation of the observed peat-depth trajectory, reporting a peat depth of 150 cm. Although this represents only one additional site and therefore does not constitute a comprehensive validation of transferability, it provides an initial indication that the ecosystem-specific parameterisation can be applied beyond the calibration site when the site has a comparable ecological trajectory.
Thanks to this comment we have added a discussion of model transferability to the manuscript, clarifying which parameters and inputs can be transferred among sites and which require site-specific adjustment (Lines 428-430).
Citation: https://doi.org/10.5194/egusphere-2026-2272-AC4
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 452 | 128 | 38 | 618 | 39 | 36 |
- HTML: 452
- PDF: 128
- XML: 38
- Total: 618
- BibTeX: 39
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Major comments
This is an interesting site-specific modelling study investigating peat development in two Peruvian peatlands. Research on tropical peatlands is critically underrepresented. Therefore, more modelling studies on tropical peatlands are urgently needed. The model is validated against peat presence/depth over time using a single peat core from each site. I am particularly concerned about the lack of validation data. At this stage, it is not possible to say whether the model can reproduce peatland accumulation dynamics. The depth and presence of peat at a particular time point is a great first step to validating the model. However, the depth and presence of peat is not indicative of carbon accumulation rates or carbon storage. The manuscript refers to in situ NPP measurements from other studies. Can this be used to validate the model in the present day? WTD (estimated from precipitation) is the primary driver of NPP in the model. Can this be validated against present day site data? I encourage the authors to validate at least one other model output.
In terms of the structure and readability of the manuscript, much more can be done to improve the flow and readability. There are only five figures in the main text, despite an extensive supplementary material and appendix. The number of figures and tables is relatively low for a manuscript of this length. Many details are omitted from the main text, including a comprehensive description of relevant model processes and I think including more figures and tables in the main text would help clarify several details. It remains unclear which parameters were changed. This is a very exciting study and a wonderful application of the model. With a bit more work, this paper can become a very useful contribution.
Small comments:
L18. Peatland formation is the result of a very slow process spanning thousands of years, during which the material from dead vegetation accumulates a process regulated by climatic and hydrological conditions.
L22. I assume you mean unique or endemic? Can you give an example?
L25. Remove ‘their history and’.
L37. Important how? Carbon-rich? Biodiversity? Oldest?
L40. Here or in the methods? Could you explain how/why these sites have developed to be a pole forest and a palm swap?
L43- L48. Net primary production? How was this measured? Can this be used as validation data in this study?
L67. Large? Do you mean tall? How tall? How thin? What are the dominant aboveground species? What are the dominant peat-forming species?
L70. What is a permanent vegetation plot?
L71. What method was used to collect the core?
L72. Re Åkesson et al. data? Do you use this? Why is it referenced?
L73 – L76. What time periods does ‘initial formation’, ‘middle to later stage’, and ‘most recent stage’ span? What were the drivers of these transitions?
When giving the length of the ‘peat core’, was this the length of the core or the peat section? Did both cores include basal peat? It’s unclear.
L83. Again, pollen, DBD, LOI etc. Are you analysing this data? Why is it referenced?
L85. Again, what were the drivers of the transitions? What is the dominant vegetation today? And the dominant peat forming vegetation?
L95. Here an elsewhere. I don’t see the need to abbreviate ecosystem, palm swamp, and pole forest. It reduces the readability of the manuscript.
L98. Are you calling the new parameterisations, a new model? Is it not just a new application of an existing model? Was any model code changed? If only new parameters, I wouldn’t change the name of the model.
Table 2. Where does this information come from? You’ve derived this from the sampled cores? Does the information become model input? Or validation data?
Figure A1. Simulation results should be in the results section.
Figure 1. Please change labels, core VEN-O2 and NYO-03 to site names. Can you show a photo of the core for both sites? Perhaps this figure can be sperated into two: First figure: The two Maps. The second figure: photo of each site (are these photos of these specific sites or ecosystem type in general. Specific site would be preferable), Photo of each core, radiocarbon dates of each core. It would be good to perhaps use dots in the figure so it is clear, at what ages, the samples were taken .
L105. Can you include equations in SI?
L116. Where can I find the codedataavailability or Table HPMTrop_EcoTy_Parameters.xlsx? I looked throughout the manuscript but could not find it.
Table 1. How does the pollen record vs the peat core agree/disagree? Are the dates radiocarbon dates? How were the ecosystem types determined? Using pollen or peat core visual analysis? Figure A3 and A4 are useful to show you developed the ecosystem types and transitions. It would be great if these were included in the main manuscript alongside the core photos.
L140. Precipitation product was compared to meteorological station data and bias corrected. Can you show this in a figure?
L152. Was this WTD assumption validated against in situ site observations? This is critical to determine whether the simulated changes in NPP and peat accumulation are related to representative changes in WTD.
L155. Please include the equations for NPP, litter, peat depth, peat height, and carbon storage here in the main text. These equations are critical for interpreting the results of this study.
L200. Peat height relative to what?
Fig 4. Core/observed data should be in black. Please then use a solid line for the base simulation with shading to indicate ±25%. Which parameters were changed? Why does some figures have a grid and others not? Please be consistent. Why does the shaded grey patch span across all time period but the caption reads that it represents the Holocene hiatus? Shouldn’t this be vertical? Can you also include lines or discuss over which time periods you expect, based on literature, the Holocene hiatus at these sites?
Discussion: Perhaps, I missed it but a limitation of this study seems to be the reliance of pollen analysis to infer vegetation presence. Pollen records only capture pollen producing plants. Can the authors discuss to what extent non-pollen producing plants likely to have been present at these sites?
Table A1. It's a bit unclear but it seems that these are model input parameters. Without presenting model equations, it's difficult to understand how these parameters are relevant to the study or used by the model. Please provide model equations in the methods of the main text.