the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Sep 2026
Temperate agroforestry systems take up more CO2 compared to conventional crop- or grassland, without consuming more water
Abstract. Ecosystem-atmosphere interactions remain poorly quantified in temperate agroforestry systems despite their potential as nature-based solutions against climate change. In this study, we present CO2 and H2O balances measured with eddy covariance over four pairs of agroforestry with open cropland or grassland systems in northern Germany from 2019 to 2024. Annual sums of net ecosystem CO2 uptake were generally 1.2- to 4-fold larger at agroforestry sites (average of -113.9 g C m-2 yr-1) compared to open cropland or grassland (average of 20.4 g C m-2 yr-1), while differences in evapotranspiration between systems were comparatively small (averages of 473.7 mm yr-1 and 423.5 mm yr-1, respectively). Water-use efficiency was thus generally higher at agroforestry, driven primarily by enhanced photosynthetic activity. Site-years with winter crops showed stronger carbon uptake, mostly due to reduced soil emissions in winter. Overall, the inclusion of trees resulted in an extended growing season and reduced sensitivity of these agroecosystems to high air temperature and low soil moisture. Furthermore, although net biome production was positive for most of the site-years, indicating a net carbon release, its magnitude was two or three times higher at open croplands and grassland than agroforestry, indicating stronger carbon losses from the reference systems that always showed a positive balance. These results reveal the potential of agroforestry systems as nature-based solution for climate change mitigation in agriculture, without compromising water availability.
- Preprint
(11165 KB) - Metadata XML
-
Supplement
(1651 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-4973', Anonymous Referee #1, 11 Oct 2026
-
RC2: 'Comment on egusphere-2026-4973', Anonymous Referee #2, 03 Nov 2026
Review of Callejas-Rodelas et al. EGUsphere-2026-4973
The authors compare CO₂ and H₂O fluxes measured with lower-cost eddy covariance (LC-EC) from 2019 to 2024 at four pairs of short-rotation agroforestry (AF) and adjacent open cropland or grassland (OC/OG) in northern Germany. They quantify annual net ecosystem exchange (NEE) and evapotranspiration (ET), ecosystem water-use efficiency (WUE = GPP/ET), responses to drivers (ACE and GAMM analyses), and net biome production (NBP), including carbon exports through harvesting. They conclude that AF systems take up more CO₂ without consuming more water, are less sensitive to heat and drought, and have a better carbon balance, although most site-years remain net carbon sources. The dataset is valuable because it is multi-year and paired observations, and the code and data are openly available.
My main concern is how the authors compute and gap-fill ET, which raises questions about the reliability of the reported WUE values. The reported cumulative ET values seem unusual given the temperate climate and annual precipitation (Table 1 & Callejas-Rodelas et al. 2024, 2026). In addition, the authors claim that the gap-filling procedure for the fluxes (CO₂, H₂O) is described in José Ángel Callejas-Rodelas et al. (2024, 2026); see lines 94-95. However, these publications do not explicitly evaluate the gap-filling approach for the latent heat flux. It is also unclear how the authors use and combine the daytime and nighttime partitioning approaches (see lines 87-89). Futhermore, the manuscript does not provide underlying harvest export dataset, and the cumulative dynamics of NEE and ET (Figures 2 and 3) are difficult to interpret. Other claims in the title and abstract are not adequately supported by the evidence presented, and the manuscript contains internal inconsistencies, including in the NEE sign convention (see lines 150-153). Although the study includes four site pairs, there appears to be no within-site replication or randomisation of treatments. Differences between AF and OC/OG systems may therefore partly reflect soil properties, land-use history, or measurement configurations, including tower heights of 10 m versus 3–3.5 m, rather than from my point of view, the agroforestry type alone.
The data are worth publishing, but the conclusions must be revisited to match what is demonstrated. The claims on climate mitigation, etc.. require either additional analysis or more cautions wording in the title abstract, and conclusions. For these reasons, I recommend a major revision.
Major comments :
1) L170-175 and Figure 3: The maximum amount of the cumulative sums of ET at all studied sites (Figure 3, left) is about 750mm. However, the maximum values in the bar plot, depending on the site and crop system, are close to 2000 mm. How can you explain this difference? Also, what explain the difference between the Wendhausen site compared with the other sites? Please specify how incomplete years were handled to have the annual sum of ET.? I would also suggest that the authors use the more precise term “actual evapotranspiration” (AET), rather than simply “evapotranspiration” (ET), consistently throughout the manuscript.
2) The statement in line 174, “Generally, increases in ET were dampened by crop or grass ripening and harvest”, is not sufficiently supported by the results presented. Showing the temporal evolution of the main meteorological drivers of ET, including radiation, vapour pressure deficit, and wind speed, alongside crop phenology and harvest dates would help substantiate this interpretation. Providing the harvested biomass for each site and year would also add useful context.
3) The abstract reports mean annual ET values of 473.7 mm yr⁻¹ for AF and 423.5 mm yr⁻¹ for OC/OG. This corresponds to approximately 50 mm yr⁻¹,, which does not support the claim of unchanged water consumption without further qualification.
4) Furthermore, footprint effects are discussed only qualitatively, although the AF footprints at Mariensee and Wendhausen visibly extend beyond the agroforestry plots in Figure 1. How might contributions from non-target areas affect the estimated fluxes and the conclusions? A footprint-weighted analysis, or a sensitivity analysis using footprint-based filtering, should be included. Finally, no statistical tests or confidence intervals for the paired AF–OC/OG differences are presented. These are needed to assess the robustness of the reported contrasts
5) Canopy height relative to the AF measurement height of 10 m also changes throughout the rotation cycle, particularly following tree harvesting in February 2021 at Forst, Mariensee, and Wendhausen. How might these changes have affected the flux measurements and the resulting comparisons? The authors should discuss this issue with reference to the relevant literature.
6) The reported WUE uncertainty reflects only the spread across GPP. It does not account for random measurement error or gap-filling uncertainty, whose propagation into GPP and WUE remains unresolved, as acknowledged by the authors (lines 374–375). Lines 144–146: Please clarify which sources of uncertainty are included in these calculations and which are excluded. How are random measurement errors and gap-filling uncertainties propagated into the final estimates? Figure 6: The description of the uncertainty bands as “1.96 times the fitted mean” presumably contains an error. Do the authors mean 1.96 times the standard error? Please clarify and verify the calculation. The captions of Figures 2 and 3 refer to “standard deviation”, whereas the main text refers to “uncertainty”. Please specify exactly what the shaded areas and error bars represent and use consistent terminology throughout.
7) Several definitions of WUE are used in the literature (Baldocchi, 2003; Farquhar & Sharkey, 1982; Beer, 2009; Zhou et al., 2014; etc..). Please explicitly precise the type of WUE evaluated here and justify this choice. Line 119: The statement “GPP_U50 was averaged from daytime and nighttime models” requires clarification. Were the two GPP estimates averaged at each time step before temporal aggregation? Please explain why to combine them and how differences between the partitioning approaches were handled. WUE calculations exclude the period from 1 November to 15 February, whereas annual NEE sums include winter. These do not share the same temporal basis. Please explain the implications for their comparison and provide a sensitivity analysis where appropriate.
Minor comments
- Lines 37–39: The sentence “The eddy covariance (EC) technique provides direct measurements of net ecosystem CO₂ exchange (NEE), H₂O, and energy exchanges between terrestrial ecosystems and the atmosphere” requires greater precision. EC fluxes are calculated from high-frequency measurements of wind velocity and scalar quantities, while estimating NEE may additionally require a storage correction. Please distinguish the measured variables from the derived fluxes.
- Abstract, line 2: Please also revise the statement “we measured CO₂ and H₂O balances”. Annual cumulative CO₂ exchange and AET do not constitute complete ecosystem carbon and water budgets, which require consideration of the relevant inputs and outputs within the system. Where appropriate, use “annual cumulative net CO₂ exchange and actual evapotranspiration” instead, and please correct the use of “balance” throughout the manuscript.
- Line 65: Please specify “LC-EC towers” at the beginning of the sentence, consistent with the equipment used.
- Line 69: Please provide the units of all measured variables, including CO₂ molar density, relative humidity, wind speed, and sonic air temperature. This information is important for checking the subsequent conversions. Similarly, at line 74, please provide the equations and units used to calculate CO₂ and H₂O mole fractions from the raw measurements.
- H₂O mole fractions are derived from relative-humidity measurements (lines 69–74). Please explain how the validation presented in Callejas-Rodelas et al. (2024) constrains potential measurement biases in the AF–OC/OG differences in ET.
- Some annual ET totals at Dornburg and Forst in 2023–2024 appear to reach approximately 700–800 mm (Figure 3), exceeding the long-term mean annual precipitation of 608 and 568 mm, respectively (Table 1). Please evaluate these values against precipitation measured during the corresponding years and discuss why these differences. Mechanisms such as a “longer growing season” and “deeper roots” are proposed but not directly tested. At Forst, the AF carbon sink strengthens after tree harvesting, when the aboveground tree biomass is reduced (lines 154–155). Please explain how this observation relates to the proposed mechanisms and distinguish measured evidence from hypotheses.
- Equation 2: Please specify the units of the response variables and predictors and indicate whether any variables were centred, standardised, or normalised before fitting the model. I have also a huge concern about how was collinearity among the predictors, particularly VPD, air temperature, and soil moisture, was addressed in Equation 2. This requires detailed clarification.
- Soil moisture is not included in the ACE analysis. Furthermore, SPEI and soil moisture are obtained from an external product based on Pohl et al. (2023). Please clarify whether these variables differ between the paired AF and reference plots. If identical values are assigned to both treatments, they cannot resolve the treatment-specific effects on soil water, such as those associated with shading and rooting, invoked in the discussion.
- Please explain how temporal autocorrelation in the weekly GAMM data was assessed and addressed. Also clarify whether the models were fitted separately by site and how site differences and crop identity were accounted for.
- In the ACE analysis, global radiation, air temperature, and VPD are likely to be strongly correlated and to covary with phenology. They are also measured at different heights in AF and reference systems. The conclusions concerning reduced sensitivity to air temperature appear to rely on visual comparisons of second-degree polynomial fits, without confidence intervals or quantitative measures of model performance. Please provide stronger support for these comparisons.
- NEE sign convention, lines 151–152: The manuscript states “net CO₂ emissions (negative NEE) … net CO₂ uptake (positive NEE)”, whereas negative NEE denotes uptake elsewhere (lines 149 and 153–159; Figure 7). Please correct this inconsistency and verify the interpretation of Figure 2.
- The abstract attributes stronger uptake during winter-crop site-years to “reduced soil emissions in winter”, whereas the discussion reports higher winter respiration in AF systems (lines 240–243). Please reconcile these statements. Moreover, soil respiration was not measured directly.
- The statement that CO₂ uptake was “1.2- to 4-fold larger” requires clarification, particularly because the mean NEE of the reference systems indicates a net source (+20.4 g C m⁻² yr⁻¹). Ratios are difficult to interpret when the compared values differ in sign or approach zero.
- Line 148: The statement “All AF sites were stronger CO₂ sinks” should be qualified, given that the sink classification at Wendhausen is described as uncertain in line 162.
- Line 281: The statement “A higher multi-annual WUE at Dornburg OC” contradicts Table 1, which reports 2.8 for AF and 2.7 for OC, as well as the statement in line 265.
Figure 6: Please clarify the soil-moisture units: the axis labels indicate percentages, whereas the axes range from 0 to 1. Also specify the definition and accumulation timescale of SPEI, apparently 90 days, in the methods.
- The title generalises the findings to temperate agroforestry, although the study covers four4 sites with differing and, in some cases, incomplete measurement periods. Please align the scope of the title and conclusions with the systems investigated. The climate-mitigation interpretation should also consider with a lot of caution.
Cited references
1) FARQUHAR, G. D., & SHARKEY, T. D. (1982). Stomatal conductance and photosynthesis.
Annual Review of Plant Biology, 33, 317-345. https://doi.org/10.1146/annurev.pp.
33.060182.0015332) BALDOCCHI, D. D. (2003). Assessing the eddy covariance technique for evaluating car-
bon dioxide exchange rates of ecosystems : past, present and future. Global Change
Biology, 9(4), 479-492. https://doi.org/10.1046/j.1365-2486.2003.00629.x3) BEER, C., CIAIS, P., REICHSTEIN, M., BALDOCCHI, D., LAW, B. E., PAPALE, D., SOUS-
SANA, J.-F., AMMANN, C., BUCHMANN, N., FRANK, D., GIANELLE, D., JANS-
SENS, I. A., KNOHL, A., KO STNER, B., MOORS, E., ROUPSARD, O., VERBEECK,
H., VESALA, T., WILLIAMS, C. A., & G., W. (2009). Temporal and among-site
variability of inherent water use efficiency at the ecosystem level. https://doi.org/
10.1029/2008GB0032334) ZHOU, S., YU, B., HUANG, Y., & WANG, G. (2014). The effect of vapor pressure deficit
on water use efficiency at the subdaily time scale. Geophysical Research Letters,
41(14), 5005-5013. https://doi.org/10.1002/2014GL060741Citation: https://doi.org/10.5194/egusphere-2026-4973-RC2
Data sets
Dataset corresponding to this publication. José Ángel Callejas-Rodelas et al. https://doi.org/10.5281/zenodo.21353133
Model code and software
Code for perform paper's analyses. José Ángel Callejas-Rodelas https://github.com/jangelcrgot/nee_et_wue_agroforestry.git
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 383 | 172 | 64 | 619 | 109 | 65 | 66 |
- HTML: 383
- PDF: 172
- XML: 64
- Total: 619
- Supplement: 109
- BibTeX: 65
- EndNote: 66
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “Temperate agroforestry systems take up more CO2 compared to conventional crop- or grassland, without consuming more water” by Callejas-Rodelas et al., for Biogeosciences
Summary
The manuscript presents findings from eight eddy covariance towers across Germany that represent four pairs of sites: agroforestry adjacent to an open cropland or grassland. As the title indicates, the data show that agroforestry sites indeed take up more CO2 relative to conventional systems and without using more water (via evapotranspiration). This is an impressive dataset (also it’s very nice to see that the data is already published in a peer-reviewed data journal format), spanning several years and different types of production systems. The results should be relevant and of interest to readers of this journal, and to farm practitioners (or the surrounding ‘ecosystem’, like those developing incentive programs). To that end some additional context about relative differences in harvest yield, farm economics or management, labor concerns, etc., as well as the potential limits of extrapolation of the findings, would help these audiences understand the full relevance of the science. The technical writing and construction (and captions) of the figures can be improved, as can some of the analysis.
Major comments
Minor comments
Technical comments