General comments

The study by Farooq et al. presents a novel work that explicitly includes termites within Microbial Explicit Soil Carbon (MES-C), a process-based SOC model. The relevance of such work is well justified, particularly due to the key role of termites in regulating SOC sequestration and CO₂ and CH₄ emissions, which is missing in global SOC models.

The main technical weaknesses of the presented work (e.g. uncertainty in some assumptions, simplification of key soil processes, homogeneous spatial representation of termites despite their highly heterogeneous distribution) are reasonable taking into account issues in the availability of data, the novelty of the approach, and common limitations in model structure. Moreover, they have been also properly addressed by the authors.

Nevertheless, there is one issue with the perspective of a detailed representation of termites into global models. The authors have divided termites in three feeding guilds, with their own parameters and specialized interactions with litter and SOC pools. Such level of detail is very reasonable for ecosystem-scale models, but it might be excessive for global models, which is the claimed goal of the authors. Despite it can be done, as shown by the authors, a question remains: should we really add all that complexity in global models? There is no doubt that termites are an important biota group with high impact, but the same can be said for several other groups, like ants, earthworms, nematodes, etc. Global models normally do not represent any of them, and if they were to include them, it would not be feasible to add each taxon separately (let alone with several guilds or subgroups per taxon). While the specific framework developed by the authors can be interesting for specific simulations related to termites, or as reference for future model developments, its future integration into global models seeking to improve the representation of soil fauna will probably need a simplification. After reading this study, I see right now two possibilities (for future developments):

1) Justify (if that is the case) that termites must remain separately from any other biota due to unique roles, but perhaps combining all guilds into one pool, and parameterize the different roles of each guild within that common pool. For example, equations S7-S9 could be modified using one single TER pool, but adding in each case a multiplier with the fraction of the TER pool that corresponds to the fraction of biomass of each feeding guild (at each time step). Some processes could retain the specialized effects of each guild that way, while other processes should be simplified with an average effect of the entire termite biomass across guilds, which would reduce the number of parameters (together with reduced interactions and fluxes) and therefore reduce complexity.

2) Discuss if termites could be mixed with other fauna groups, splitting if needed their guilds into different functional groups of soil fauna, in the same way that for example in some soil food web models nematodes are divided into different functional groups depending on their body size and/or feeding roles (e.g. microbivores, herbivores, predators). Maybe the way to go could be to modify generic functional groups (once they are added to global models) to account for the specific roles played by termites, as for example including the role of TER_SF within microbivores, and the roles of TER_FG and TER_XP within detritivores, taking into account that termites will only be a fraction of such pools.

I don’t think there is any need to modify the study for that reason, the presented perspectives (and all equations, results, etc.) can stay as they are, but I would suggest to add to the discussion also those challenges related to adding termites to global models while keeping complexity as low as possible. Despite this work has merit on its own and can certainly help to improve models, I would like the authors to address, from their expertise on modelling termites, what would be the best option to deal with future global modelling frameworks including termites along with other soil fauna groups.

Overall, the manuscript is interesting, well written, it represents a first step towards a potentially relevant improvement of global SOC models, and it should be suitable for publication after a minor revision.

Specific comments

Table 2 – The total Area was apparently calculated as the sum of all the areas by vegetation type. If I understand the represented concept correctly, termites occupy 17.3 m/km2 of tropical evergreen, 5.9 m/km2 of tropical deciduous, etc. Then what you are suggesting with that total is that, in general over all vegetation types, termites occupy 72.3 m/km2, right? I don’t think that is correct. You should calculate an average instead of a sum for such variable, or even a weighted average considering the different total areas of each vegetation type. If my interpretation is not correct, please explain better what exactly are you showing there.

You explain in the manuscript (L210) that GPP was smoothed using a 7-day running mean, while in the supplementary (L58) you also mention that you obtained GPP from MODIS 8-day productivity product, and that you constructed daily climatological time series. Can you explain in more detail how did you process the 8-day data? In particular, for generating the daily data; the use of a 7-day running mean from daily data is more obvious and should not require more details, but it would be better to explain the method used to go from 8-day to daily.

Supplementary:

L44 – the symbols ω and δ determine the fractions of decomposed carbon entering the LMWC pool for frass and necromass, respectively, with their complementary fractions determining what goes to MAOC in each case; the text wrongly describes them as if they refer to the different SOC pools, when they refer to different output fluxes from termite biomass.

Table S3 – the values shown for fractions of total termite biomass for Asia and Africa not always add up to 1. In particular, the first two rows for tropical vegetation (evergreen and deciduous) sum in total 0.91, and savanna and dense shrubland only have 0.33. Please explain why is that. Conversely, for grassland, there is already 1 for soil-feeding termites (SF), but also shows a zero for xylophagous termites (XP), while other cases in which one guild has 1 the others show a hyphen instead of a zero. Is that because the shown values for grassland are rounded, but SF guild is actually lower than 1, and XP higher than 0?

Equation S14 – add an explanation of why the denominator is the square root of 12.

Technical corrections

The manuscript title is not very fluent to read, with the double colon. I would suggest to remove “Ideas and perspectives:” entirely, or if not, then at least remove the first colon, or rewrite the title in a more fluent way (not a correction, just a suggestion).

L17 – “their contributions … remain” (instead of “remains”).

L198 – “carbon biomass” instead of “biomass carbon”.

L256 – In Fig. 2 caption, better make a clearer reference to the latitudinal mean representation, e.g. “The subset in each plot (blue line) shows…”. The same applies for Fig. S2.

L407 – “is available” instead of “are available”

Supplementary:

L21 – missing space in “times(Wang et al., 2010)”.

L26 – missing space in “underwentmicrobial”.

Equations S13 – k_B should be k_n

Table S3 – “from Sanderson (1996)”, the author name is explicitly referenced, so it should not be inside the parenthesis.

L68 – “carbon” instead of “substrates”, within the full name of LMWC.

General comments

I had the pleasure of reading Farooq et al.’s paper on termite effects on the global C cycle. It is an incredibly well-written article presenting a mechanistic framework for how termites can be incorporated in soil carbon models and eventually into earth system models. They also provide global estimates of termite-mediated C-fluxes, both into the atmosphere as GHG gasses, and into the soil as labile and stabilised C-pools. Impressively, the emergent estimate of global termite-driven CH4 emission is very consistent with estimates from the Global Carbon Budget.

The framework is clearly presented and logical. Proper and thorough sensitivity analyses are conducted and presented clearly. Relevant recommendations for future research are outlined.

I have no severe concerns with this manuscript, and recommend it for rapid publication in Biogeosciences. Below I suggest some minor comments that can hopefully help improve the manuscript further.

Specific comments

• Title: consider getting rid of the double colon.
• 77-78: may be worth relating this to the total annual CH4 emission, to make the estimate more immediately interpretable. If I remember correctly, it can be around 5% of total methane emissions, no?
• 114-115: It is not clear exactly how you distinguish between these three litter pools. Can you give a short description of what each of them comprise?
• 158: Is the pronounced aggregate formation through soil particle ingestion and regurgitation during mound construction for Macrotermes contributing to the MAOC formation? I suppose they would feature as ‘fungus-farmers’ in your framework, as they do not really ingest soils for nutritional purposes?
• 186-193: These simplifications seem fair to me, but I would still like a rough description of what the structural, metabolic and CWD pools comprise, perhaps just with a few examples at first mention - you do give examples for CWD, so that is less critical.
• 252-253: I agree, but you could make this clearer by relating some of your numbers to total global annual fluxes or similar, i.e. X% of total annual CO2 emissions.
• Table 2: I see the termite-driven turnover of structural litter in grasslands is 0, which again makes me curious what exactly this litter pool is exactly. If structural tissue is defined by molecular composition, there would be structural compounds in grasslands (e.g. cellulose, hemi-cellulose and lignin), but perhaps there are no termites eating those pools in grassland biomes?
• 273-274: Overall the sensitivity analysis seems quite robust to me. Yet, it may be helpful to suggest here towards the end what this means in terms of future research to constrain and reduce this uncertainty. It is not completely clear what research we need to reduce King uncertainty. It is better termite biomass (stocks and fluctuations) estimates, is it better grip around field metabolic requirements, feeding preferences or what is it?
• 283-288: This is very interesting, and makes sense due to their role as physical ecosystem engineers able to optimise the environmental conditions of their surroundings (built structures) to their advantage. Reminds me of the ecosystem resilience arguments brought forward for drylands in the 2015 Science paper by Bonachela et al.: https://www.science.org/doi/10.1126/science.1261487 and for tropical rainforests in the 2019 Science paper by Ashton et al. https://doi.org/10.1126/science.aau9565 . Perhaps worth citing.
• 290-295: There is perhaps even more to it than just translocation into mounds. Macrotermes typically ingest and chew CWD together with spores from Termitomyces as a form of inoculation before they excrete both to form the fungus comb.
• 311: Impressively consistent!
• 361: and as mentioned, it is not clear to me whether the ingested-regurgitated soil aggregates for mound construction is included, which has been reported to account for up to 30% of the soil matrix in areas with high termite activity.
• Discussion: One aspect that hasn’t really been touched on, but which could be an interesting aspect for future research to mention in the discussion is the interactive effects of termites and larger animals on C-processes. Some places to begin are:
  • https://doi.org/10.1017/S0266467411000125
  • https://doi.org/10.1016/j.tree.2021.09.006
  • https://doi.org/10.1002/ece3.7445