the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Feb 2026
Temperature dependence of the contribution of soil moisture to soil respiration and the soil respiration temperature threshold in a temperate deciduous forest
Abstract. Soil respiration (Rs) in forest soils is a key flux governing forest carbon balance and the global carbon cycle. Because this flux is expected to respond rapidly to climate warming, understanding the controls on Rs is essential for predicting changes in forest carbon balance induced by warming. In natural field conditions, soil temperature (Ts) and soil moisture content (SMC) often covary seasonally, which tends to limit our ability to isolate and quantify the independent contribution of SMC and to evaluate how its contribution varies with temperature. Although temperature thresholds in Rs have been reported, few studies have quantitatively identified such thresholds from field observations and interpreted potential shifts in the dominant controls based on how moisture responses differ across the threshold. Here, we used two years of continuous automated chamber measurements in a temperate deciduous forest to estimate a Ts threshold for Rs and to assess how the relative contribution of SMC varies with Ts by comparing models across temperature ranges, with particular attention to changes near the threshold. At the annual scale, the explanatory power of SMC alone was limited, but the relationship between SMC and Rs was significant. In contrast, above 15 °C, the relationship between SMC and Rs strengthened consistently, indicating that the contribution of SMC is constrained at low Ts but increases markedly at high Ts. Piecewise regression of the relationship between Rs and Ts identified a Ts threshold near 17 °C, and models including this threshold improved fit relative to an exponential model. These results show that the relative contribution of SMC can change across a specific temperature range, suggesting that changes in the relative influence of SMC on Rs variability across the threshold may reorganize the dominant controls on Rs. Therefore, projections of forest Rs should jointly consider temperature dependent changes in moisture contribution and the presence of Ts thresholds.
- Preprint
(888 KB) - Metadata XML
-
Supplement
(116 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-605', Anonymous Referee #1, 13 Mar 2026
-
AC1: 'Reply on RC1', Seo Dongmin, 16 Mar 2026
Thank you for your constructive comments.
We have provided detailed answers to all points in the corresponding sections of the attached file “Response_to_Comments.pdf”-
RC3: 'Reply on AC1', Anonymous Referee #1, 19 Mar 2026
There are still things not to clear to me. The authors show this equation: 𝑅𝑠 = 𝑎exp (𝑏𝑇𝑠)(𝑐𝑆𝑀𝐶2 + 𝑑𝑆𝑀𝐶 + 𝑒); which is actually not applied right? at least there is no statistics provided for the fit of this equation, which is pretty much a standard equation used to predict effects of Ts and SWC.
What authors use instead is an equation with this form: 𝑅𝑠 = Trange * (𝑐𝑆𝑀𝐶2 + 𝑑𝑆𝑀𝐶 + 𝑒) (Trange is what you called Tbins, which you actually defined arbitrary...) . Am I right?
IS there differences in the fit of those two equations? I also propose two other equations:
𝑅𝑠 = Season*(𝑐𝑆𝑀𝐶2 + 𝑑𝑆𝑀𝐶 + 𝑒)
Rs= Season* Trange*(𝑐𝑆𝑀𝐶2 + 𝑑𝑆𝑀𝐶 + 𝑒)
and see which one gives the best fit
My feeling is that the different shapes of the relation between SWC and Rs will be related to season, which partially covaries with Ts, but not entirely. This may show that part of the variability in Fig 3 might also be related to phenology
Citation: https://doi.org/10.5194/egusphere-2026-605-RC3 - AC2: 'Reply on RC3', Seo Dongmin, 22 Mar 2026
-
RC3: 'Reply on AC1', Anonymous Referee #1, 19 Mar 2026
-
AC1: 'Reply on RC1', Seo Dongmin, 16 Mar 2026
-
RC2: 'Comment on egusphere-2026-605', Timo Plaçais, 19 Mar 2026
SummaryThis study examines whether the contribution of soil moisture (SMC) to soil respiration (Rs) is temperature-dependent, using two years of continuous automated chamber measurements in a temperate deciduous forest under an Asian monsoon climate. Rs is measured as the total soil CO₂ flux, root and microbial contributions combined, and both Rs and the environmental drivers (Ts, SMC, precipitation) are aggregated to daily means before analysis, so that any sub-daily dynamics fall outside the scope of the study.
Four questions are addressed. Does a significant SMC–Rs relationship exist at the annual scale? Then daily means are considered for the following questions. Does its strength vary across temperature conditions? Does SMC explain additional Rs variability beyond what Ts alone accounts for? And can a formal breakpoint be identified in the Rs–Ts relationship, above which SMC contribution increases markedly? The first three questions are approached by fitting exponential Ts-only and Ts+SMC models within 5°C temperature bins, with ΔAdj. R² used to quantify the added contribution of SMC. The fourth uses segmented regression. A consistent breakpoint near 17°C emerges across both years, above which ΔAdj. R² rises to 0.21–0.62 depending on year and bin, compared to near-zero values below 15°C. The authors conclude that this threshold marks a reorganization of Rs controls, from temperature-dominated regulation at low Ts to a combined Ts–SMC structure at high Ts, and advocate for incorporating this temperature-dependence into carbon cycle models.
Novelty and methodological concerns
The question is well-motivated and the dataset is genuinely valuable, two years of continuous high-frequency chamber data is not trivial to obtain, and the approach of stratifying by temperature bins to isolate the SMC contribution is sensible. The result that SMC contribution to Rs variability is negligible below ~15°C but substantial above it is ecologically meaningful and reasonably well-supported by the data.
My main concern is with how the breakpoint is identified. The authors define it as the temperature above which SMC contribution to Rs variability increases sharply, that is a statement about the explanatory power of moisture. But the breakpoint itself is estimated through segmented regression on the Rs–Ts curve, which identifies where the slope of Rs with respect to Ts changes. A change in slope in Rs~Ts can reflect nothing more than the curvature expected from a standard Arrhenius-type temperature response, and does not in itself say anything about moisture sensitivity. The variable that actually operationalizes the authors' question is ΔAdj. R² as a function of Ts (shown in Fig. 4), and it is on that curve that a breakpoint analysis should be performed. As it stands, the 17°C threshold is borrowed from a different analysis and applied to a question it was not designed to answer. This is a genuine methodological inconsistency, and the conclusions drawn from it are stronger than the framework supports.
Beyond this, the single-site design limits how far the 17°C value can be generalized, and the absence of Rs partitioning makes the mechanistic interpretation difficult to pin down, the breakpoint could reflect a phenological transition in root activity just as well as a moisture threshold for microbial metabolism.
Interpretation
The biological interpretation is largely consistent with established understanding. The invocation of substrate diffusion limitation under dry conditions, oxygen constraint under waterlogged conditions, and temperature-dependent stimulation of microbial and root activity is appropriate and well-referenced. Individual claims do not overreach the data.
Throughout the manuscript, the breakpoint in the Rs–Ts relationship is equated with a threshold in SMC sensitivity, and this equivalence is used to argue for a "reorganization of the dominant control structure governing Rs." This conclusion is repeated in the abstract, results, discussion and conclusions, but the logical bridge between the two analyses is never formally established. A change in slope in Rs~Ts does not imply a change in the variance of Rs explained by SMC, this would require additional assumptions that are neither stated nor tested. The bootstrap test confirms that the segmented model fits Rs~Ts better than an exponential; it says nothing about SMC sensitivity. More cautious phrasing throughout, acknowledging that the two lines of evidence converge suggestively without being formally linked, would substantially strengthen the manuscript.
A further interpretive gap concerns causal attribution. The strengthening of the SMC–Rs relationship above 15–17°C likely coincides with leaf-out and the onset of root activity, both of which covary with Ts and SMC in this monsoon system. Without autotrophic/heterotrophic partitioning, it is not possible to determine whether the breakpoint reflects a moisture threshold for microbial activity, a phenological transition in root respiration, or both. This ambiguity deserves more prominent acknowledgment than it currently receives.
Specific comments
- 27–28 — The authors invoke increasing extreme hydrological events as a motivation for the study, yet daily averaging of SMC and Rs is likely to dampen the very transient responses they allude to — most notably the Birch effect. Could the authors comment on whether rewetting pulses occurred during the study period, and how their representation may have been affected by the temporal aggregation applied?
- 64–65 — The authors acknowledge that SMC effects on Rs depend strongly on timescale, yet daily averaging may itself attenuate the moisture signal they seek to quantify. The manuscript implicitly positions daily resolution as an improvement over annual-scale studies, but does not discuss what temporal resolution would be needed to fully resolve moisture–respiration dynamics near the identified threshold. This point deserves explicit treatment.
- 185 — The SMC thresholds reported (10.8% in 2022, 13.1% in 2023) lack a methodological basis in the text. Were these derived from the fitted quadratic fonction, from a formal changepoint procedure, or from graphical inspection? A clarification is needed for reproducibility.
Figure 5 — A change in slope in Rs~Ts may simply reflect the curvature of an Arrhenius-type temperature response rather than any shift in moisture sensitivity, raising the possibility that the identified breakpoint is an artefact of the functional form rather than an ecological threshold. A more direct test, for instance, examining at what temperature the residual variance of Rs (after Ts removal) begins to increase significantly, would more rigorously operationalize the authors' question. Adding SMC as a color overlay on the Rs–Ts scatterplot would also allow readers to visually assess moisture modulation across the temperature range.
Citation: https://doi.org/10.5194/egusphere-2026-605-RC2 - AC3: 'Reply on RC2', Seo Dongmin, 22 Mar 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 196 | 69 | 25 | 290 | 36 | 39 | 56 |
- HTML: 196
- PDF: 69
- XML: 25
- Total: 290
- Supplement: 36
- BibTeX: 39
- EndNote: 56
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
I believe this study offers a novel perspective on how soil temperature and moisture control soil respiration. Generally, soil temperature is identified as the primary controller of the temporal variability of soil respiration (SR), subsequently modulated by water availability, which limits both autotrophic and heterotrophic respiration when water is scarce. This study demonstrates that, in these temperate forest ecosystems, the opposite view holds true: temperature appears to control the SR response to moisture. This control occurs above a threshold of 15-17°C. In other words, above a certain temperature, the variability in Rs is mostly explained by fluctuations in soil moisture, but with the modulation of Ts. Below this threshold, soil metabolic activity appears to decrease and lose all relationship with moisture, with temperature being the sole factor controlling the temporal variability of Rs.
The result seems interesting to me, as I mentioned above, but some factors and limitations should be considered and discussed in the manuscript. First, the scope of this study needs to be better contextualized, since in this study system water availability is quite high throughout the year, and particularly during the warmer periods. In temperate systems outside the monsoon influence, this is not the case, and soil temperature and moisture tend to have a negative seasonal relationship. That is, in non-monsoon temperate systems, the relationship between soil moisture and Rs at high temperatures is very likely nonexistent because there is insufficient moisture for even minimal autotrophic or heterotrophic metabolism. In other words, in these systems it is humidity that limits the response to temperature, and not the other way around. This should be discussed and contextualized in the discussion.
Another aspect that I believe the authors do not sufficiently discuss in this study are the results obtained in Figure 3. Particularly interesting are the "jumps" in basal respiration rates between the different temperature ranges above 15°C, and the difference in the shape of the relationship between soil moisture and Rs between these ranges. On the one hand, what do these jumps respond to? They aren't controlled by humidity or temperature. They may respond to changes in the biomass of microorganisms and fruit roots in different phenotypic phases during warmer periods. Regarding the form of the relationship, I partially disagree with the interpretation of Fig 3. Made in lines 260. If oxygen limits, should limit Rs either similarly for each Ts ranges or should be higher at the highest temperatures, when O2 demand peaks. Here does not seem consistent with neither of those cases. Again, I think that the phenophase might have played a role here, because in these Ts ranges data from spring and fall are mixed up, and things are generally very different phenologically speaking between spring and fall. Maybe to try to separate also this phenophases with the temperature ranges will help explaining the differences in the slope of the relation between temperature ranges.
Finally, I understand that Figure 5 confirms what we see in Figures 3 and 4. What the authors mean is that below this threshold (which largely coincides with the 15°C shown in Figure 3), the variation in Rs is entirely explained by temperature, while above it, Rs is controlled by the interaction of Ts and SWC. However, the correlation coefficients and the slope that would confirm that Ts's control is more important below the threshold are not shown. It is also possible that below the threshold the roles of Ts and SWC have reversed: Ts is the primary control, modulated by SWC.
Minor comments:
Title: looks like a riddle. Please rephrase
References: Most references are from the last 5 years. Most important references in the study of the role of soil temperature and soil moisture in soil respiration are before 2020. Please make a better bibliographic search and cite other key papers
Methodology: method section could be substantially improved. You learn at the end of the methods section that there were 5 automated chambers installed (what was the criteria where to install them, for instance?) but no information about where the sensor of SWC and Tz where installed with respect to the Rs measures. Were SWC and Ts measured near each automated chamber? Or there were only one measurement point? By the way, why the authors use the acronym SMC instead of the more commonly used in literature SWC (soil water content) to refer to soil moisture? It is just a matter of consistency and reproducibility of results.
Figure 2. Why different symbols if they are presented in two different panels?
Discussion: throughout the discussion I see many paragraphs that looks more results than discussion (e.g. paragraphs 250, 275, 280…). The discussion should be used to discuss results rather than to reporte them again. Please try to improve this too.
Figure 5. Which is the fit for each part of the threshold? It would be nice to see the slope and R2 for the two different sections at both sides of the threshold. Does temperature fit better Rs at colder periods? Below the threshold there is also some variability around the model. Could this be done to fluctuations in SWC?