the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Apr 2026
Quantifying temperature dependence of Fe reduction in humid tropical soils: a Bayesian model-data integration
Abstract. Humid tropical forests are critical regulators of the global carbon (C) cycle, yet their soil C stocks are increasingly vulnerable to warming. Predicting potential losses requires a mechanistic understanding of the processes that govern soil C stabilization and mineralization, particularly in Fe-rich soils, where iron (Fe) redox cycling plays a dual role in both protecting and decomposing organic matter. However, the temperature dependency of these Fe-mediated processes remains poorly understood. In this study, we quantified the temperature dependence of FeIII reduction by conducting anoxic incubations at 23, 27, and 33 °C and calibrating four kinetic models of increasing complexity to estimate the Q10 and Arrhenius (Ea) using a Markov Chain Monte Carlo (MCMC) framework. Model performance was evaluated using Bayesian information criteria (WAIC, LOO, and LPML) to assess fit, complexity and uncertainty. Short-term warming significantly accelerated Fe-reduction rates, potentially destabilizing mineral-associated organic carbon and enhancing microbial respiration. Estimated Q10 and Ea values ranged from 1.5 to 2.1 and 30.8 to 56.5 kJ mol-1 respectively, comparable to the temperature sensitivity values measured in temperate and tropical biomes. With the available data, Bayesian information criteria preferred the simplest one pool FeII model due to its parsimony. In contrast, the most complex (three pool) model, which includes dissolved organic carbon (DOC) dynamics alongside Fe reduction and oxidation, was generally the least preferred by Bayesian information criteria due to increased uncertainty from unconstrained additional processes. These results underscore the importance of temperature-dependent Fe redox processes in regulating soil C cycle in humid tropical soils and emphasize the need to balance model complexity with data availability when modeling coupled C-Fe interactions.
- Preprint
(3704 KB) - Metadata XML
-
Supplement
(7998 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2111', Anonymous Referee #1, 04 Jun 2026
-
RC2: 'Comment on egusphere-2026-2111', Anonymous Referee #2, 18 Aug 2026
The authors investigated the temperature dependence of microbial iron reduction in a tropical soil. The subject is relevant and fits into the scope of EGUsphere. The major part of the study is devoted to explore the performance of different kinetic models in order to reproduce the experimental findings. In order to compare the performance of the different models, criteria obtained from probabilistic model assessment were used.
I highly appreciate the use of the Bayesian approach to evaluate the performance of kinetic models by considering the increase in uncertainty by making the model more complex with the benefit of improving the quality of the fit. However, the use of this approach for the presented data is misplaced, creates confusion and does not help to answer the main research question (what is the temperature dependence of microbial Fe reduction in this soil).
In my opinion, the presentations of the four different models is misleading and obscures the underlying problems. Based on the experimental data, the rate of microbial Fe reduction can be deduced based on the rates of Fe(II)HCl production or Fe(III)HCl consumption. As Fe(II)HCl production exceeds Fe(III)HCl consumption the authors conclude that the Fe(III)HCl pool only represents a part of the microbially reducible Fe(III)HCl pool. Nevertheless, it might be an interesting question to compare the temperature dependence of these two rates and to discuss whether the temperature dependence of the Fe(III)HCl reduction kinetics is representative for the whole reducible Fe(III) pool.
The first problem in this study are missing stoichiometries and unmatching mass balances.
It does not make any sense to use a kinetic model in which the production of Fe(II)HCl is 1:1 coupled to the consumption of Fe(III)HCl. This model will intrinsically fail to reproduce the experimental data. This is not a question of complexity or whatsoever but simply a mass balance problem. Without any probabilistic model assessment it is obvious that extending a model with a constraint, which violates the mass balance, will lead to a model with lower performance. However, the problem of presumably systematically (this should be presented and discussed in any case) higher rates of Fe(II)HCl production than Fe(III)HCl consumption can be solved by formulating a proper stoichiometry (in any case, kinetic studies should always be based on a proper formulation of reaction stoichiometries). Here a possible approach would be:
X*Fe(III)HCl + (1-x) Fe(III)NonHCl + Reductant = Fe(II)HCl + Oxidized Reductant.
In this stoichiometry, x is the unknown fraction of bio-reducible, non-HCl-extractable Fe(III), which is consumed per produced Fe(II)HCl. The problem in view of formulating a corresponding rate law is that the reactivity of the non-HCl-extractable pool might deviate of that of Fe(III)HCl so that more than one additional variable will be added to the model in order to account for the different reactivity of different Fe(III) phases (in many diagenetic models including Fe(III) reduction more than one Fe(III) pool is included, for example Markelow et al. 2019).
The other problem is the use of Michaelis-Menten kinetics without considering alternatives
In a typical study on the temperature dependence of Fe(III) reduction rates, the initial rates would be determined at various temperatures. Based on the temperature dependency of the initial rates, an apparent activation energy, for example, could be determined. Using initial rates is very convenient and also purposeful when investigating the temperature dependence of Fe reduction is the main goal.
Again, I highly appreciate that the authors want to use kinetic data beyond the initial stage of the reaction. In doing so, the authors, however, should provide arguments why targeting the temperature dependence of Fe reduction beyond the initial stage is relevant. This would then justify to find and parameterize a suitable rate law instead of directly using the measured rates. The authors address this aspect in line 60 but do not get to the core of the matter.
In this study, the authors apply rate laws based on Michaelis-Menten kinetics but do not question the suitability of this rate law or consider other type of rate laws. In view of using the Bayesian approach to find the most appropriate rate law, this is a missed opportunity. For describing the time evolution of Fe(II)HCl production, simple zero or first order kinetics might perform very well. For Fe(II)HCl production this might require that the reducible Fe(III) pool by far exceeds the amount of produced Fe(II)HCl during the experiment. In my opinion, evaluating the performance of different rate laws for the same data set (e.g. the time evolution of Fe(II)HCl) can profit from the Bayesian approach. For this I would use zero order, first order, unknown order (d(Fe(II)HCl)/dt= - k Fe(III)asccit^x ) as well as te M.-M. rate laws.
My suggestion:
The need to determine the temperature dependency of Fe reduction throughout the whole reaction (and not only in the initial state) should be justified. Is it possible that the T dependency of the rates changes throughout the reaction?
The approach to study the T dependency of Fe reduction based on the rates of Fe(III)HCl consumption should also be adequately justified as Fe(III)HCl only represents a fraction of total bio-reducible Fe(III)>
Next to M.-M. kinetics, other rate laws should be included in the study.
Forcing the model to reproduce the time evolution of Fe(II)HCl and Fe(III)HCl concentrations without solving the problem with an unmatching mass balance should be avoided. Instead a two Fe(III) pool model could be tested. This scenario would be similar to the DOC model as it adds a third pool and would make the DOC model obsolete.
In the discussion ,the added value of assessing T dependency based on rate laws and corresponding rate constants in contrast to simply using initial rates should be critically analyzed.
Minor comments
The temperature dependency of the rates of biological processes can be described by the Arrhenius equation in a narrow temperature range. In general, the T dependency of biological processes show an optimum T with decreasing rates towards higher and lower temperatures.
The terms ‘rate’ and ‘rate constant’ are confused a few times, these are not synonyms.
Q10 is a temperature coefficient and not a rate (line 225)
Literature
Markelov, I., Couture, R. M., Fischer, R., Haande, S., & Van Cappellen, P. (2019). Coupling water column and sediment biogeochemical dynamics: Modeling internal phosphorus loading, climate change responses, and mitigation measures in Lake Vansjø, Norway. Journal of Geophysical Research: Biogeosciences, 124(12), 3847-3866.
Citation: https://doi.org/10.5194/egusphere-2026-2111-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 342 | 102 | 33 | 477 | 160 | 30 | 33 |
- HTML: 342
- PDF: 102
- XML: 33
- Total: 477
- Supplement: 160
- BibTeX: 30
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript investigates the temperature dependency of Fe reduction in a soil sample from a humid tropical system. This was assessed experimentally using batch incubation experiments and subsequently interpreted using Bayesian modelling approaches to identify the model with the best performance. To this end, four different models with increasing levels of complexity were evaluated. Overall, I consider the manuscript to be well structured and clearly written, and it presents interesting results that contribute to understanding how increasing temperature may influence Fe reduction processes in this type of system. The application of a Bayesian framework to evaluate model performance in terms of parsimony and uncertainty is also a relevant and valuable contribution.
However, there are several points of concern that, in my opinion, should be addressed before the manuscript can be considered for publication.
A major concern relates to the framing and interpretation of the study in relation to soil C dynamics. The authors state that one of the aims of the manuscript is to enhance the mechanistic understanding of how temperature influences C cycling through its coupling with Fe redox processes. Furthermore, the manuscript claims that the results underscore the importance of temperature-dependent Fe redox processes in regulating soil C cycling. While the coupling between Fe and C cycles is undoubtedly important in these soils, the results and subsequent discussion do not provide direct evidence or substantial new insights into C cycling dynamics. The study is primarily focused on the temperature dependency of Fe redox processes, and no direct measurements related to C destabilization, decomposition, or other C dynamics were performed. Therefore, I recommend that the authors moderate the claims regarding soil C cycling and instead place greater emphasis on discussing Fe redox processes and the modelling framework, which constitute the strongest aspects of the study.
A second major concern relates to methodological aspects. Throughout the manuscript, the authors state that the experiments were conducted under “constant anoxic conditions.” However, no direct evidence or measurements are provided to verify the establishment and maintenance of anoxic conditions (eg., pE, dissolved O2, or O2 saturation). In addition, important methodological details are missing, such as the gas-to-liquid ratio, procedures used to remove O2 (e.g., gas purging), the use of airtight vessels, and how anoxic conditions were maintained throughout repeated sampling events. Without these details, the assumption of constant anoxic conditions is insufficiently supported, which is particularly important given that the interpretation of Fe reduction dynamics strongly depends on this assumption.
In the attached file, I also include several specific comments aimed at improving clarity and interpretation of the results.