Process-based reconstruction of peatland carbon dynamics in Switzerland
Abstract. In Switzerland, over 90 % of peatlands have been degraded and transitioned from long-term carbon sinks to persistent carbon sources. Assessing peat carbon vulnerability and restoration opportunities requires understanding not only the present-day distribution of peat carbon stocks, but also the historical trajectories that shaped them. Existing assessments are largely based on inventories and emission-factor approaches, limited in representation of the interactive effects of climate, peatland processes, and land-use change over time. By combining the LPX-Bern Dynamic Global Vegetation Model with a historical record-informed peat-derived area and carbon accounting framework, we reconstructed peatland conversion pathways and tracked the fate of peat carbon through successive land-use transitions from the pre-industrial period to the present. Modelled peatland and peat-derived areas in Switzerland declined from approximately 724 km2 in the pre-industrial period (1701–1710) to 376 km2 at present (2011–2020), while only 5 % remained near-natural, in broad agreement with historical reconstructions and recent synthesis products. Over the same period, peat carbon stocks decreased from 121.9 Mt C under near-natural conditions to 30.4–56.4 Mt C. Although only a small fraction of the original peatlands remained near-natural, 86–92 % of the remaining peat carbon was stored in agricultural, forested, and unmanaged drained land-cover classes. By explicitly tracking peat carbon through historical land-use transitions and attributing remaining peat carbon to present-day land-cover classes, this framework provides a spatially explicit basis for assessing peat carbon vulnerability, identifying restoration opportunities, and informing peatland management and climate mitigation strategies. This approach is transferable to other regions where information on peatland extent, land-use history, and peat-specific disturbances is available.
Summary:
This study applies LPX-Bern, a Dynamic Global Vegetation Model (DGVM), in conjunction with land-change forcing from the Land-use Harmonization 2 (LUH2) dataset and dynamic peat extraction and unmanaged drained peat areas from Wüst-Galley et al. (2019), to characterise evolution of peatlands in Switzerland from 1700 to 2020. The study tracks peatlands (and their carbon stocks) through land-use transitions over this interval. The results show not only the complete loss of about half of Switzerland's peatlands / peat-derived areas (more on that later), but also the almost complete dominance of anthropogenic peatlands in the modern day, with some 95% of peatland area now classified as cropland, pasture, forested or unmanaged, or urban and built. Loss of peat carbon stocks is also estimated, at about 25% to 46% loss. The Abstract claims that "this framework provides a spatially explicit basis for assessing peat carbon vulnerability, identifying restoration opportunities, and informing peatland management and climate mitigation strategies" and that the "approach is transferable to other regions where information on peatland extent, land-use history, and peat-specific disturbances is available." The Summary additionally states that "Most peat carbon is currently stored outside intact peatlands, showing the opportunities to reduce future carbon losses extend beyond the small fraction of remaining near-natural peatlands."
Contribution:
Peatlands drained for conversion to agriculture or peat extraction become persistent sources of CO2 to the atmosphere and their continued use is incompatible with achieving net zero in the land use sector. This creates a practical, as well as a scientific, motivation to better account for existing peatland carbon stocks and how they have been and will be transformed by land use change. In the simulation of peatland evolution in the Holocene using DGVMs / TBMs, it can be fairly said that LPX-Bern is near or at the forefront. A focused study applying LPX-Bern to the trajectory of Switzerland's peatlands and their carbon stocks is therefore a welcome contribution to the literature.
Assessment:
This is a good paper, but in its current form it is rather difficult to work out what exactly was done and why it is important. If this aspect is fixed, I feel it could be a very good paper. This will require a combination of adding missing information and some reorganization. The premise is important, the methods are appropriate, and the data support the conclusions. I find the Discussion informed and thoughtful; one rarely finds papers that really engage with the literature these days, and this paper does. The main challenge is to make clearer what is contributed by LPX-Bern, what by the LUH2 dataset, and what by the Wüst-Galley et al. (2019) reconstruction in this study. For further explanation, please see the general comments below. I got confused at a few points, so some comments may be misguided; but perhaps my comments can still be useful by indicating the aspects of the manuscript that were confusing.
General comments:
1. The motivation for the study should be made clearer and kept as a thread through the Abstract and Introduction. The last part of the Discussion has a heavy conservation / restoration focus, but this framing could be clearer in the Introduction.
2. A revision could aim to highlight in the Introduction and start of the Methods, and possibly briefly in the Abstract, the essence of the approach: How LPX-Bern, the LUH2 dataset, and the combined LUH2 and (interpolated) Wüst-Galley et al. (2019) datasets were used together, and the attribution of land-use change to peatlands in proportion to their area. A simple diagram might help.
3. Good plain-language summary: it is dense with information yet very clear.
4. Terminology is important and tricky ('peat-derived areas') and could use some work in the revision. It would be good to establish precise terminology in the Introduction. For example, I initially had the impression that any area that had been peatland could be considered peat-derived, and only later learned that loss of extracted and drained areas results in "permanent loss of peat-derived area". Another example: the 19 km2 in 3.1 seems to conflict with the 375 km2 in the Abstract; I believe this is related to the distinction between "peatland" and "peat-derived area"? A table could help (possibly an expanded and revised version of Table 1?), or a simple diagram showing the admissible transitions between land classes.
5. Reaching the end of the Introduction, I ask myself: Is the argument here that we can best trust peat volume / carbon stock estimates that come from a dynamic model that is consistent with historical data? Consider making this argument more explicitly?
6. It was not clear to me to what extent the application of LPX-Bern to managed peatlands here involved new code or calculations. The conclusion speaks of a "framework", but what is the framework? Should I then infer that the accounting in the Methods section is an extension of what has been done with LPX-Bern previously?
7. Exponential decay of peat carbon is a slightly surprising choice and would benefit from some explanation / justification.
8. The "Data analysis" section does not include enough detail to reconstruct methods; the processed outputs and plotting scripts are available from Zenodo (great!), but a reader needs more information in the manuscript about what was done. Code can become unreadable because of software changes, but prose is forever.
Specific comments (P for page, L for line):
9. P1 L13-14, "combining the LPX-Bern Dynamic Global Vegetation Model with a historical record-informed peat-derived area and carbon accounting framework,":
How exactly were they combined? I was not sure at the end of the first read. Does the "historical record" here refer to the combination of LUH2 and Wüst-Galley et al. (2019)?
10. P2 L43, "agriculturally":
Does this include silviculture?
11. P2 L46-47, "the mitigation potential of peatland management is an irreversibly depleting resource":
Is the mitigation potential a resource? Consider expanding (if this is what was meant) or rephrasing (if not)?
12. P2 L66, "peat carbon stocks":
Consider specifying that this refers to Switzerland?
13. P3 L88, "Sphagnum":
Italicise (if used as a Latin genus), or make lowercase (if used as an English common name)
14. P3 L90-92, "Peatland processes are developed from the LPX-WHyMe framework (Wania et al., 2009) with dynamic peatland area (Stocker et al., 2014). In this study, peatland expansion was constrained using the GIEMS-MC wetland extent product (Bernard et al., 2025":
Brief descriptions (a phrase each) of what was contributed by the LPX-WHyMe framework and the GIEMS-MC wetland extent product would be helpful.
15. P3 L102-103, "Land-use change is implemented as net area changes, without explicit gross transitions as detailed in Müller and Joos, 2021.":
Are these net area changes in each grid cell? Mention this?
16. P3 L104, "proportionally":
Consider citing Humpenöder et al (2020: https://doi.org/10.1088/1748-9326/abae2a), which applied a similar approach to land change disaggregation, when the approach is briefly explained in the Introduction or Discussion?
17. P3 L104, "is converted to land-use classes":
In proportion to the fraction of the cell occupied by peatland?
18. P3 L105, "when total land-use area increases":
Where does the overall land change forcing come from? Is this provided by LUH2?
19. P4 L116, "Periods outside of the coverage of other forcing datasets were extended using the first year or last year of data.":
Which forcing datasets were extended and how long? An indicative statement is needed to understand the extent to which this might affect results.
20. P4 L121-122, "𝐴(𝑡) under other land-cover classes other than peatland was diagnosed during post-processing using a conserved tracer.":
What does this mean (what is the "conserved tracer" here)?
21. P4 127-128, "prioritises conversion to anthropogenic land-use classes":
Prioritises conversion to anthropogenic land-use classes over what? Conversion to natural land is prevented; besides anthropogenic land-use classes and natural land, what else is there? In this context, is "forested area" considered non-anthropogenic? To avoid such questions, consider stating more explicitly how the prioritisation was done (perhaps as a list or a very brief step-by-step procedure)?
22. P4 L132-135, "Since the LUH2 dataset does not account for peatland-related activities such as peat extraction and peatland drainage (unlike land-use conversion which does not distinguish unmanaged drained peatlands as a separate class), these transitions were incorporated in post-processing using reconstructed fractions from Wüst-Galley et al. (2019), available for year 1710, 1890–1900, 1943–1951, and 2015, and linearly interpolated in time":
See general comments; it would be helpful to highlight this important aspect of the design in the Introduction and possibly hint at it in the Abstract.
23. P4 L136, "active peatland class,":
Is "active peatland class" defined? Does this mean areas that are currently classified as natural peatlands?
24. P4 L137-139, "extracted and drained areas are treated as terminal states, such that any subsequent reduction in these areas results in a permanent loss of peat- derived area from the system.":
Interesting choice; I think it would benefit from further justification. If this is done elsewhere, can we be directed to that section?
25. P5 Table 1, "Tracer accounting":
It is not yet clear to me what this means.
26. P5 L143-144, "For comparison with contemporary inventory-based estimates, forested peat-derived areas and unmanaged drained peatlands are grouped into a single category, termed forested/unmanaged drained peatlands.":
Where exactly is this different classification used? If only for one section of the Discussion (i.e., if unused in the Results), consider stating this there instead?
27. P5 L148-149, "This land-cover classification was used as an interpretive framework rather than as a formal restoration suitability standard.":
Are these last two sentences directly linked to this slightly different classification (with a forested/unmanaged drained peatlands category), or do they pertain to the base classification with separate "forested" and "unmanaged drained" categories? If the latter, consider, for clarity, making a separate paragraph and / or moving before the discussion of the modified classification?
28. P5 L155-156, "Carbon densities in peatlands are output from the model and are applied as boundary conditions at each timestep without being altered during redistribution.":
Does this notation fully capture what was done - was there really a uniform carbon density in each land cover class? Or is this a mean, based on per-pixel accounting or an underlying modeled distribution? If this text implicitly describes accounting within a single pixel, consider stating this at the beginning of the subsection, or adjusting the subsection title?
29. P5 L163-166, "a deep drainage scenario assumes that, in addition to the complete loss of acrotelm carbon, 50 % of the peat carbon below 0.3 m is decomposed and released during the drainage initiation year. These selected thresholds provide lower and upper bounds spanning the range of plausible peat carbon loss associated with drainage.":
I am unclear on to what extent the calculations here are or are not extensions of LPX-Bern?
30. P5 L167, "Peat carbon remaining in all peat-derived non-peatland categories is subjected to first-order decay:":
Exponential C loss in each land class---surprising to me, are there precedents?
31. P6 L169, "Comparison with national inventory and peatland reconstructions":
At this point in the manuscript I remember that in the Abstract it was said that LPX-Bern would be "combined" with a "historical record-informed peat-derived area", and become confused about whether that might be referring to the Wüst-Galley et al. (2019) reconstruction or perhaps the LUH2 dataset.
32. P6 L187, "19 km2":
376 km2 in Abstract; does this relate to what a "peatland" is?
33. P7 L204, "Figure 1":
It seems to me that the way the colors are stacked makes the differences between the left and right panels seem even larger than they are. Could the stacking be altered to place lost peat extraction area (transition to "non-peat-derived", white) at the bottom of the plot? This could make the similarities between the left and right panels more obvious, and the other differences easier to discern.
34. P7 L204-205, "Results based solely on model output":
Is the result in panel (a) not more a matter of using solely the land change forcing from LUH2 (rather than any role of applying the forcing to LPX-Bern)? In either case, clarify?
35. P7 L205-206, "Results after incorporating reconstructed peat extraction and unmanaged drained peat from Wüst-Galley et al. (2019)":
See general comments; possible to briefly but explicitly convey how the LUH2 and Wüst-Galley et a datasets were combined?
36. P7 L208-209, "historical reconstructions":
Tell us a little more about the basis of the historical reconstructions?
37. P8 L227, "record-informed process-based modelling":
At this point in the manuscript it was not clear to me, even in broad outline, how this was done; I had not yet read the Methods carefully, but many readers won't have, either. Consider concisely explaining how LUH2, LPX-Bern and Wüst-Galley et al. (2019) were used together?
38. P8 L246, "Figure 3":
How many of these pixels are there in all of Switzerland? The blockiness seems problematic in an area with such complex terrain, so it might be advantageous to briefly mention any subgrid aspects of the simulation and / or datasets.
39. P8 L250-251, "These land-cover classes contained 26.0–52.0 Mt C of peat carbon,":
How confident can we be in how much C is left in these land uses? Some information on the extent to which estimates were calibrated / bias-corrected would be helpful for interpretation.
40. P9 L262, "Carbon stored in peat-derived agricultural land increased throughout the nineteenth and twentieth centuries":
Could be misinterpreted---Consider rephrasing to make it clearer that peat carbon in agricultural land increased due to conversion of peatlands to agriculture, and not due to increases in peat carbon stock in already-converted peatlands?
41. P10 L301, "The LPX-Bern simulation":
See general comments re. contributions of LPX-Bern per se vs. forcing
42. P11 L310, "GIMES-MS constraint":
What is this, exactly? Explain in a phrase?
43. P13 L402-403, "The reconstructed distribution of peat-derived area and carbon therefore provides a spatial basis for assessing the potential vulnerability, rather than a direct measure of restoration potential.":
Interesting; I have the impression that this statement represents several logical links that have not been spelled out. Consider unpacking into two sentences?
44. P14 L421-423, "Moreover, the effectiveness of water-table management is likely to change under future climate conditions with temperature and water-table dynamics jointly control CO2 and other greenhouse gas emissions from drained peat":
This paragraph is dense with good and interesting points.
Technical corrections:
45. P4 L140, "terms area":
missing "of"
46. P7 L205, "rea":
area
47. P10 L294, "sock":
stock