| 18 Dec 2025
CO2 and H2O isotope exchange and flux partitioning in Amazonia
Abstract. Understanding the coupled exchange of H2O and CO2 between ecosystems and the atmosphere remains limited due to our inability to partition net fluxes into their individual source and sink components. For the Amazon rainforest, which plays an important role in the global balance of water and carbon, investigating these individual fluxes is critical given the environmental changes in recent years. Here, we apply a stable isotope-based approach to partition ecosystem-scale gas exchange from simultaneous eddy covariance measurements of H2O and CO2 isotopologues. During the 2022 CloudRoots-Amazon campaign at the Amazon Tall Tower Observatory, high-frequency isotope flux measurements from 57 m were used to derive multi-day composite diurnal cycles of δ fluxes and ecosystem source compositions. A steady-state midday interval, constrained with independent leaf and soil isotopic observations, allowed us to coherently link the H2O and CO2 isotopic states throughout the ecosystem (soil, canopy, leaf, atmosphere) using δ18O.
Isotopic flux partitioning indicates that transpiration accounts for 95.5 % of the net evapotranspiration (ET) of water at 14:00, with soil evaporation being responsible for 4.5 %. For CO2, δ18O-based partitioning indicates that the respiration flux from the soil equals −44 % of the net ecosystem exchange (NEE), where the photosynthetic assimilation flux in turn is 144 % of NEE. The partitioning of NEE was found to be strongly dependent on the leaf intercellular-to-atmospheric CO2 ratio (ci/ca) which determines the (apparent) isotopic composition associated with photosynthetic assimilation (δP). This underlines how important detailed leaf and soil level measurements of isotopic compositions and leaf characteristics are for ecosystem-scale flux partitioning.
The manuscript by Moonen et al. contributes to recent body of literature developing novel approaches to improve methods to partition net ET and NEE fluxes in its gross components. This is paramount to better understand the role of terrestrial ecosystems for regional water and carbon cycles and to allow better upscaling of these fluxes using remote sensing or modelling tools.
This partitioning is for various reasons not trivial for tropical forest and recently novel neural network-based methods have been developed that try to improve the “traditional” so-called daytime and nighttime methods. I think that besides the limitations, which are clearly recognized by the authors, adding isotopes exchanges to the ET and NEE partitioning “toolbox” disserves to be discussed in the scientific literature allowing for future experiments to improve also this method. Afterall, this isotope flux method It is simply a more direct and physical based method, i.e. we add one isotope flux equation to solve two unknowns.
This paper assesses this method on a very relevant site (ATTO, Brazil) and was made possible using novel eddy covariance measurements of isotopologues of H2O and CO2. The authors address in a very detailed and (often) clear way the followed methodology and potential caveats for an “average day” of a 13-day period in the dry season in the Amazonian tropical rainforest. Despite the limitations, but this is the case for all current existing partitioning methods, this paper disserves to be published to allow further refinement by other interested research groups, this is the only way how the potential of this method can be tested and further developed. Below I give one major comment mainly related to the steady state assumption made and several minor comments that are meant to further increase the clarity of this complex and integrated methodology. Finally, the paper also correctly advocates for detailed isotopic measurements of leaf and soil endmembers, which is paramount further refine and test this methodology.
Major comments
I have some concerns on what the authors describe as a midday steady state (line 68). This is mostly linked to the “isotopic composition of transpiration” (line 125). Can you really consider that mass of water that is entering the leave matches the mass of water evaporating from the leaf? During photosynthesis light splits H2O and is “consumed” to generate glucose. Furthermore, H2O is also used to incorporate H into leaf waxes (alkanes). These processes have a (large) biosynthetic fractionation factor and hence also isotopically enrich leaf water. Maybe there is a misunderstanding in the (too short) description of the steady state here? Furthermore, I agree, in general root water uptake is assumed not to be fractionating, though this is increasingly debated. In addition to that (and see comment on Table A1) root water uptake seems to come from deeper than 10 cm soil depth. This is also something that needs more attention in the text.
Minor comments
I believe also “non-isotope” experts would be highly interested in this methodology. Hence a consistent and unfirm terminology is important, i.e. what is a “delta flux” (line 7, 48, 90, etc.) or (H2O and CO2) isotopic fluxes (line 88, 68, etc.), explain the relation between delta and isotope ratio notation (e.g., Eq; 6, 7 vs. Eq. 4, as both are used in the various equations.
Line 81: I don’t understand this sentence: … “describe the composition of a negative flux”?
Line 120, Eq. 6, 10: explain all parameters in the text. Note here you use isotope ratio’s (R) and not a delta notation (make clear to a non-expert public). Please check this also consistently for all other equations in the text.
Line 210: is it correct that “d18c” is calculated from Eq. 6?
Line 243: what is the meaning of 10^5 here?
Line 255: It is unclear how the outlier filtering was done.
Line 259: You are sure these uncertainties are correct?
Figure 1 and related text: please add a bit more technical details on the equipment and which isotope standards that have been used.
Line 269: Here you refer to the “steady state” (see major comment), we need a bit more explanation on this I believe.
Line 280: Explain why this background isotopic composition was unstable.
Line 301: “This finding was confirmed by,…” this “co-spectra” notation is not clear.
Fig. 2: improve legend and caption for a better understanding of the figure. For a) also add it is a “30 min water flux”.
Fig. 3: idem for legend in this figure.
Line 311-312: “stable conditions lead to horizontally heterogeneous conditions”; make clear please.
Line 322: correct terminology please: write (maybe) CO2-isotopologue flux,…
Line 343-345: this section is unclear to me. Can you try to rephrase?
Line 348: where we see this -15 per mill lower values?
Fig. 5: explain the y-axis titles well in the caption.
Line 379: Make clear in the captions that all data are for 14:00 hour and a 13-day composite diurnal flux - > independent reading of Tables and Figures.
Table 1: please not that Tair and Tleaf can be different. So not sure what the 33°C represents.
Line 394: “isotopic composition of the topsoil”. Please note root water uptake is not from the topsoil alone and, depending on tree species, distributed over different soil layers. Please make this and its complications clear here. This is also a bit misleading at the bottom of figure 6, as root water uptake is water with (for dD) a value between -25.8 and -14.4 per mil. Furthermore see also my comment for table A1 and in the major comment.
Line 399-400. Here is the concept of mass conservation again. See my comments above where I argue that is not really the case. Can this be adjusted?
Line 420: explain clearly to which “reservoir” you refer.
Fig. 6: I can’t recall from the text how/where the “canopy” values were derived? Maybe this can be further clarified?
Line 423, 433 (and elsewhere): I think you need to better clarify in the text (e.g. M&M) what you mean with kinetic, physical, apparent and equilibrium fractionation. This is now used a bit randomly/confusing. Please clarify this throughout the text.
Line 439-443: unclear please revise.
Line 447-448 this equality (for dE) needs to be better explained.
Line 458: can you better explain that equation?
Line 462: “Uptake fractionation”? See my earlier comments on fractionation
Line 498: you might consider discussing the possibility to derive integrated ci/ca ratios via 13C analyses in cellulose of leaves.
Line 507: the ci/ca of 0.7 seems to me on the high end? I would expect values of 0.5-0.6. Can you back this up with literature data?
Line 522, 542-544: if the isotope steady state is so crucial, I think you need to guide/explain the (new) reader a bit better on this here and in the Materials and Methods.
Line 560 and beyond: I agree you need better description of the diurnal dynamics of the isotope reservoirs. But I think to use this isotope-based partitioning method on a longer time scales than just days we need also seasonal (e.g. dry-wet season) information on the isotope reservoirs. Second, it would be good, if feasible, to compare your isotope-based partitioning with other (debated) flux partitioning methods such as the (modified day-time method) or more recent neural network-based approaches.
Figure A1: check the title of the y-axis. Is this notation correct?
Table A1/A2: mention in the caption this data is at 14h00 and 13-day composite average. Make sure your parameters have all the same notation as in the text (e.g. for the fractionation factors this seems different. If I look at the dD and d18O data for topsoil and xyleme water, it is clear (assuming no fractionation) that trees take up water from below 10 cm depth. I have added an earlier comment on root water uptake depth. You might need to (re)consider this aspect in the text.