the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Dec 2025
National-Scale Inventory based Climate Impact Analysis of the Nitrogen Balance, including all Nitrogen Fluxes using Process-Based Modelling with the LandscapeDNDC Model and EURO-CORDEX Ensembles for Greece
Abstract. In this study, we have simulated inventories of arable production and soil carbon and nitrogen cycling on a national scale (0.25 × 0.25-degree) for Greece with the bio-geochemical ecosystem model LandscapeDNDC. Based on observation data, we have aggregated for each grid cell 4 most likely crop rotations, including nitrogen and manure fertilization, tilling and irrigation. The arable management was continuously projected into the future until 2100, while plant phenology was adapted to local conditions, general properties of the arable management were kept constant into the future, such as the selection of crops or the share of irrigated arable land. To understand the impacts of climate change, we used the EURO-CORDEX-11 regional climate ensemble to drive the LandscapeDNDC impact model under scenarios RCP4.5 (16 datasets) and RCP8.5 (32 datasets). The simulation timespan was from 1990 until 2100, using the first 10 years as spin-up to obtain equilibrium in the model's internal carbon and nitrogen pools from the model initialization.
Arable production declines from 2045 onwards by 9.5% or 144 kg C ha-1 yr-1 under RCP4.5 and by 29% or 484 kg C ha-1 yr-1 towards 2100. At present, the ensemble results show an average soil carbon loss of 122.1 kg C ha-1 yr-1 versus 139.7 kg C ha-1 yr-1 in the future. The gaseous outfluxes of the ensemble simulations show N2O emissions of 0.494 to 0.453 kg N2O–N ha−1 yr−1, NO emissions of 0.031 kg NO–N ha−1 yr−1, N2 emissions of 4.806 to 3.377 kg N2–N ha−1 yr−1, NH3 emissions of 24.662 to 35.040 or 34.205 kg NH3–N ha−1 yr−1 and nitrate leaching losses from 54.304 to 58.213 kg NO3-N ha−1 yr−1 comparing present versus future conditions. The overall nitrogen balance of the ensemble simulations reveals a mean nitrogen loss of 4.7 versus 5.7 kg-N ha-1 yr-1 comparing present to future conditions.
- Preprint
(4133 KB) - Metadata XML
-
Supplement
(2296 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-5311', Anonymous Referee #1, 09 Mar 2026
-
AC1: 'Reply on RC1', Edwin Haas, 20 May 2026
In general, we want to point out to the reviewers and the editor the importance and novelty of the work presented in this study. While many models and studies have been addressing the carbon balance of arable systems from site to national to continental to global scale, there are to our knowledge only 3 modelling papers which present the full nitrogen balance including ALL nitrogen fluxes:
- Schroeck et al. 2019 – Estimating nitrogen flows of agricultural soils at a landscape level, https://doi.org/10.1016/j.scitotenv.2019.02.071
- Sifounakis et al 2024 – Regional assessment and uncertainty analysis of carbon and nitrogen balances at cropland scale using the ecosystem model LandscapeDNDC, https://doi.org/10.5194/bg-21-1563-2024, 2024.
- Rahimi et al 2024 – Aggregation of activity data on crop management can induce large uncertainties in estimates of regional nitrogen budgets.
https://doi.org/10.1038/s44264-024-00015-3
All three papers were authored by the corresponding author of this study: Edwin Haas. The reason for the lack of papers reporting the full nitrogen balance was clearly elaborated in a recent discussion paper by Balasz Groscz et al (2023) “Modeling Denitrification: Can We Report What We Don't Know?“ (https://doi.org/10.1029/2023AV000990). This discussion paper resulted from denitrification research project funded by the German Research Foundation (DFG). The conclusion and recommendation are summarized in the following two statements / sections:
- … we argue that including the entire N balance and related parameters should become standard when publishing the results of N model studies.
- We assume that the scarcity of “complete” (i.e. including N2 fluxes and other N pools/pathways) modeled N balances in the soil denitrification literature stems from the reluctance of the scientific community to support the publication of unvalidated modeled output, especially given that the simulation results of these ‘neglected’ N pools may be unrealistic. But this self-censorship of authors has resulted in a missed opportunity to share knowledge and improve our understanding of modeled processes. We recommend that future studies exercise transparency in publishing model outputs. We ask authors to focus on the aspects of their model that were of particular interest (i.e. validated model developments), but, while clearly stating which variables were not validated by measurements, to include all related pools and parameters to the fullest extent possible (e.g. all modeled N pools/pathways, soil aeration and CO2 flux).
Our presented study/manuscript was guided by these recommendations and therefore we combined the presentation of the carbon balance (well studied and less complex) and the nitrogen balance (rarely presented, very complex) for arable land cultivation on the national scale.
Reviewer 1
1.1 Title does not reflect the carbon content of the paper
Reviewer comment: About a third to one half of the paper deals with carbon results, but the word ‘carbon’ does not appear in the title.
Authors’ response: The title has been revised to: “National-Scale Inventory-based Climate Impact Analysis of the Carbon and Nitrogen Balance using Process-Based Modelling with the LandscapeDNDC Model and the EURO-CORDEX Climate Change Ensembles for Greece” The title clearly names the Carbon and Nitrogen Balance. The (now) misleading phrase “including all nitrogen fluxes” has been removed.
1.2 Abstract needs reworking (units, ‘arable production’, gases, two values)
Reviewer comment: The abstract from L31 onwards needs to be reworded to clarify the spatial unit (0.25° vs ha⁻¹ yr⁻¹), the meaning of ‘arable production’, what ‘soil C decrease’ refers to, which gases are emitted, and why two values are given for gaseous emissions.
Authors’ response: The abstract has been revised. We now explicitly differentiate the reported loss of production (carbon in yield) as kg-C ha-1 versus the spatial averages of C and N fluxes all reported as kg-C ha⁻¹ yr⁻¹ or kg-N ha⁻¹ yr⁻¹ on a cropland-area weighted basis. ‘Arable production’ is replaced by ‘crop yield (C-in-yield)’. We added a clarifying sentence: ‘All flux values reported below are ensemble means; ranges of the form “X to Y” refer to present (2005–2034) versus future (2070–2099) conditions’. Soil C is now consistently described as topsoil (0–30 cm) SOC change. The list of gaseous fluxes (N₂O, NO, N₂, NH₃) and aquatic N losses (NO₃⁻ leaching) is given explicitly.
1.3 Highlights require reworking
Reviewer comment: The first two highlights are climate-scenario related, not core results. The 4ᵗʰ point is unclear (overall total C, biomass C, or soil C?).
Authors’ response: The Highlights section has been fully rewritten. The first highlight now states the novelty (first national-scale process-based inventory of the full C and N balance under EURO-CORDEX). The second highlight summarises the propagation of structural uncertainty through the LandscapeDNDC model. The biomass-C highlight now reads ‘Ensemble-mean biomass carbon yield (crop C-in-yield) decreases by 9.5 % under RCP4.5 and by 29 % under RCP8.5 towards 2100’. The SOC highlight is now ‘Ensemble-mean topsoil (0–30 cm) soil-carbon loss …’. The N₂O highlight now explicitly mentions the present-vs-future comparison and the drivers. A new highlight on nitrate leaching dominance has been added.
1.4 Graphical abstract is not stand-alone
Reviewer comment: Parameter names on the y-axis may not be known; ‘5.7 kg N ha⁻¹ yr⁻¹’ requires explanation; inputs/outputs of the N balance should be clearly labelled.
Authors’ response: We acknowledge this limitation. The figure was not appropriate as a highlight as it needs some explanation, especially the cumulative nature of a waterfall diagramm. Therefore we have created a diagram showing the magnitude and the uncertainty of each flux. The figure shows only the magnitude of the flux, independent whether they are influxes of outfluxes. The nitrogen balance (sum of all fluxes has been removed as it needs additional explanation and is therefore not appropriate for the highlight figure.
1.5 Introduction is too broad — focus on national N inventories and modelled C/N impacts under climate change
Reviewer comment: L60-83 is too broad. Add focused review of national N inventories (Fan et al. 2020; Strenge et al. 2023) and a discussion of main sources/sinks of N and C and how they are affected by climate change.
Authors’ response: A new paragraph has been inserted in the Introduction (immediately before the research questions). It (i) cites Fan et al. (2020) and Strenge et al. (2023) as recent regional/national N-budget studies, (ii) reviews modelled climate-change impacts on cropland C and N cycling at site/regional scale (Basche et al. 2016; Carozzi et al. 2022; Deryng et al. 2011; Lugato et al. 2014; Petersen et al. 2021), and (iii) summarises how the main sources (mineral and organic N input, BNF, atmospheric N deposition) and sinks (harvest, gaseous N losses, leaching, SOM turnover) of cropland C and N are affected by shifts in temperature, precipitation, soil moisture and CO₂ (Beaudor et al. 2025; Shen et al. 2020). The novel contribution of the present study (national-scale full C and N balance from process-based ensemble modelling) is then stated explicitly.
1.6 Research questions L101-108 are bullet statements, not questions
Reviewer comment: The research questions on L101-108 are not formulated as research questions, but rather as bullet point statements.
Authors’ response: The four research questions (a–d) have been reformulated as proper questions. They were addressed in the results and discussion sections in this sequence.
- How is agricultural production (yields) on a national scale projected to change under future climate-change scenarios (RCP4.5, RCP8.5)?
- What is the soil carbon balance of the Greek cropland system under present conditions, and how does it evolve under future climate-change projections?
- What is the full nitrogen balance of the Greek cropland system under present conditions, as relevant for UN FCCC reporting?
- How are the coupled carbon and nitrogen cycles and their associated fluxes projected to respond to climate change towards the end of the 21st century?
1.7 LandscapeDNDC description: add spatial/temporal resolution, daily vs hourly
Reviewer comment: The description (L112-138) must include the spatial and temporal resolution; L128 states ‘daily or hourly climate variables’ but which one was used in the study?
Authors’ response: The description here refers to the general model description. The description of the spatial and temporal resolution for the study is provided in section 2.2 in details.
1.8 L151-162: more detail on management input data
Reviewer comment: Briefly describe the level of spatial resolution of the input data. What management data was used (tillage, plant residues, seeding/harvest dates, fertiliser application/type)? Was the Haas et al. 2022 crop rotation transposed to Greece, and does this make sense?
Authors’ response: The present study used an existing and published model input dataset (Haas et al. 2022) in order to avoid performing redundant work and to avoid the explicit repetition of a lenthy description of data sources and data processing. We have added an extensive description of the used agricultural-management in the supplementary data section now explicitly lists the management variables (crop types and rotations, sowing and harvest dates, mineral and organic fertiliser amounts and timing, tillage operations, residue management and irrigation amount and timing). The inputs are previously aggregated to the same 0.25° grid as the soil and (CORDEX) climate inputs.
1.9 L153 vs L31 contradiction (15 vs 10 years warm-up)
Reviewer comment: L153 states 15 years (1990–2005); the abstract (L31) says 10 years.
Authors’ response: The abstract has been corrected from ‘10 years’ to ‘15 years (1990–2005)’ to be consistent with the methods section. Both are now harmonised at 15 years of spin-up.
1.10 L187-199 unclear: 2099 vs 2100; ‘spatial and temporal means’; ‘these statistics’
Reviewer comment: Section is unclear. 2099 contradicts other parts that state simulations to 2100. ‘Spatial and temporal means are assumed’ and ‘these statistics’ are vague.
Authors’ response: We have rewritten this paragraph. We now state explicitly: ‘“present” and “future” refer to the 30-year intervals 2005–2034 and 2070–2099, respectively, throughout all figures and tables (unless explicitly stated otherwise).’ The terms ‘spatial and temporal means’ are now defined: first calculatin of cropland-area-weighted spatial means and second, temporal means of the special means. ‘These statistics’ has been replaced by an explicit description (mean ± SD plus IQR with Q25, median, Q75; see also 1.11).
1.11 L208: describe statistical analyses
Reviewer comment: Describe which statistical analyses were carried out.
Authors’ response: An ensemble impact analysis results in a distribution of results instead of discrete numbers. These distributions were condensed into statistical measures such as mean, median and quantiles. A new sentence has been added describing the statistics: ‘Ensemble statistics are reported as mean ± SD together with the interquartile range (IQR; Q25, median, Q75).
1.12 Missing calibration & validation section at start of Results
Reviewer comment: Add a section on calibration and validation of LandscapeDNDC (observation data, spatial resolution, time period, method, objective function).
Authors’ response: The Mediterranean agriculture lacks the availability of detailed measurements of e.g. N fluxes such as N2O emissions or NO3 leaching. Therefore, we cannot calibrate and validate the model explicitly for this study. To prove the capabilities of the model, we have simulated the LandscapeDNDC validation tests reported by Molina et al. (2016) with the model version used in this study and we will present them in the in the supplementary material section for the readers interested in this. We are aware that this study does not have a validation part and we refere to the statement in the introduction to this “respose to reviewers” that we should publish results without validation not to miss an opportunity to gain knowledge.
We have added a new subsection at the start of the Results:
The Mediterranean agriculture lacks the availability of detailed measurements of N fluxes such as N2O emissions or NO3 leaching. There fore the study does not provide an explicite calibratin and validation section. The results of the LandscapeDNDC validation reported by Molina et al. (2016) was repeated with the model version of this study and results are reported in the supplementary material section. The LandscapeDNDC setup used in this study is the same as used by Sifounakis et al. (2024) for the Thessaly region (Greece). Simulated crop yields and N uptake in harvest were shown to be consistent with regional statistics (Sifounakis et al., 2024). Calibrated crop parameters and simulated fluxes are therefore considered representative for Greek cropland and are transferred to the national-scale application presented here. No additional calibration has been performed for the present study.1.13 Q25/Q75 should be introduced
Reviewer comment: Is it standard to use Q25/Q75 in the text? Introduce its meaning (presumably quantile).
Authors’ response: We now introduce Q25, median and Q75 explicitly as the lower quartile, median and upper quartile of the ensemble (i.e. the interquartile range). See also response 1.11.
1.14 L227-8 pre-empts results (‘systematic decrease expected’)
Reviewer comment: Why is ‘a systematic decrease expected under both RCPs’? This seems to pre-empt the results.
Authors’ response: Please be aware that this result analysis relates to the Euro Cordex climate data and not to the ecosystem simulation results. The wording has been changed from ‘systematic decrease expected’ to a neutral description of what the ensemble shows for present and future conditions, without anticipating the result.
The section reads now as follows (line 403 ff):
For the precipitation dynamics, shown on Figure 1 (b, d), a very moderate decrease is shown under both RCPs, with stronger declines in RCP8.5.1.15 Description of main crops and management practices missing
Reviewer comment: Which crops are examined? What are typical management practices?
Authors’ response: A full despriction of the arable management is added to the supplementary material section describing the used arable crops (…) and the associated typical management (sowing/harvest windows, synthetic-N rates, manure use, irrigation share, tillage). The selection of the 4 most-likely crop rotations per grid cell is also re-stated here.
1.16 Section 3.2 unclear (response, ‘arable production’, figure)
Reviewer comment: Section 3.2 needs reworking. Which response is meant? Where is the figure? What is ‘arable production’ — is it C content of biomass?
Authors’ response: Section 3.2 has been edited so that ‘arable production’ is defined as ‘carbon in crop C yield’.
The text reads as follows:
Arable production (expressed by carbon in crop yield) is considered the averaged biomass carbon yield across all various crops within the rotations for the various inventory simulations. The ensemble response to the climate change scenarios shows different responses for the RCP4.5 and RCP8.5, especially in the second half of the century (see Figure 2).1.17 Why does C decrease in the future? Reflection in Fig 2/3?
Reviewer comment: It is not clear why C decreases over the future period (Fig 2). Presumably yield decrease — why? How is the decrease reflected in Fig 3?
Authors’ response: Please be aware that LandscapeDNDC is a carbon and nitrogen model. Therefore all crop yield results are expressed by carbon in yields instead of yield biomass or dry weight/dry matter. Figure 2 refers to arable production as crop yields exressed in C in yield. Yield response to climate change itself is (in the model) driven by heat stress and water stress due to reduced precipitation under RCP8.5. The corresponding visual signal in Fig. 2 (declining C in yield after ~2050 under RCP8.5) is therefore described explicitly in the text.
This comment was resolved by adding the explanation that arable production is expressed in carbon in yields in the text and figure 2 and the figure should become clear.1.18 Suggest moving Section 3.3 before Section 3.2
Reviewer comment: Move Section 3.3 (C processes) before Section 3.2 (biomass C yield).
Authors’ response: After internal discussion we decided to keep the current ordering, because Section 3.2 (biomass C yield) introduces the main driver of the C cycle which is then decomposed in 3.3. This is for us the logical order of presenting the results.
We have however added cross-references between Sections 3.2 and 3.3 to make the logical flow clearer.1.19 Discussion repetition L395-401; modelled vs measured N₂O
Reviewer comment: L395-401 contains repetition. Are the studies on L501 measured or modelled N₂O emissions?
Authors’ response: The potential repetition in L395-401 is the essence of this study. Readers who scan the paper than rather reading it in details need to capture this message as it makes this study unique! After an internal discussion we will not remove the paragraph or shorten it.
The study of (Sifounakis et al., 2024) was performed by the same team than this study and is a pure modelling study. It has focused on one mayor arable production area in Greece and on a regional uncertainty assessment.
In the discussion we have added more comparisons to observations reported in recent literature. The section reads now as follows:
Analysing the N2O emission fluxes, our study reports rather low N2O emission fluxes compared to Sifounakis (2024), reporting modelled N2O emissions for the intensively cultivated region of Thessaly (Greece) of 2.6 kg N ha-1 yr-1 under much higher N influxes of up to 220 kg N ha-1 yr-1. N2O emission measurements for Greece have not been reported in the literature. For the Mediterranean region, Volpi et al (2019) (https://doi.org/10.1007/s10705-019-10032-1) reported N2O emissions only in the growing season of 0.344 to 0.656 kg N2O-N ha-1 yr-1, Guardina et al (2024) (https://www.sciencedirect.com/science/article/pii/S0959652624026350) reported averaged post harvest N2O emissions of 0.241 kg N2O-N ha-1 yr-1 and Carbonell-Bojollo et al. (2022) (https://www.mdpi.com/2073-4395/12/6/1349) reported cumulative N2O emissions of 0.4 to 0.6 kg N2O-N ha-1 yr-1 within 180 days for 3 seasons which agree very well with the N2O emissions of our study.1.20 L466: Greece lumped into one basket
Reviewer comment: Greece is lumped into one basket when discussing soil C losses. Land-use specific results would be more useful. How comparable is Greece with Haas et al. 2022 if climate and soils differ?
Authors’ response: We are aware of this. The study of Haas et al. 2022 has even lumped all of Europe into one basket to make an assessment on SOC change versus N2O emissions. Please be aware that the objective of this study was to make coarse assessments of the carbon and nitrogen balances of Greece on the national scale due to data availability and resources available to process at the time the project started. It was never intended to make a detailed analysis of the presponsiveness of various drivers such as soil properties to various N fluxes of the full balance. Note that no other country in Europe has performed and reported the carbon and nitrogen balance in such a detail on the national scale before. The comparison to the study of Haas et al. 2022 shows clear differences (Mediterranean vs continental, lower SOC stocks in Greece) and concludes that the Greek values are at the lower end of the EU range, consistent with the climatic gradient. After some consultation, we will not add any additional discussion about this comment to the manuscript.
1.21 Missing uncertainty section
Reviewer comment: A section on uncertainties is missing (input data, spatial resolution, calibration and validation, interpretation).
Authors’ response: A new subsection ‘Uncertainty assessment and limitations’ has been added at the end of the Discussion. It summarizes and concludes many of both reviewers comments. It groups uncertainty into five categories: (i) input data, (ii) spatial resolution, (iii) model parameter and calibration uncertainty, (iv) the fixed land-use/land-cover assumption, and (v) the fixed irrigation management. Each category is discussed individually and quantified where possible (e.g. ± SD of N₂O and SOC fluxes). A concluding paragraph is added on heat-stress and vernalisation representation. The section now reads as follows:
4.1 Uncertainty assessment and limitationsThe predictions presented in this study are subject to a number of uncertainties that can be grouped into five categories: (i) input data uncertainty, (ii) spatial-resolution uncertainty, (iii) model-parameter and calibration uncertainty, (iv) uncertainty from fixed land-use and land-cover assumptions, and (v) uncertainty from fixed irrigation management. In the following we summarise the implications of each for the reported national-scale C and N budgets.
(i) Input data. National totals of synthetic fertiliser and manure use (FAO, national statistics) were spatially redistributed per grid cell using the arable-land share derived from Corine Landcover. Because the aggregated arable-land class includes unfertilised land components (orchards, vineyards, fallow, gardens), per-hectare fertiliser rates on the truly fertilised land may be somewhat underestimated. Atmospheric N deposition fields are taken from EMEP/EDGAR products whose reported uncertainty for the Mediterranean is of the order of 20–30 %, which translates into a comparable uncertainty on the N input side of the balance.
(ii) Spatial resolution. The simulations are performed on a 0.25×0.25° grid (~25 km at these latitudes), which is coarse relative to the field scale where biogeochemical processes operate. Sub-grid heterogeneity in soil texture, organic-matter content, terrain and management is represented only through the four most likely crop rotations per grid cell. This aggregation necessarily smooths out extreme local values (e.g. hotspot N₂O emissions on wet organic soils or high leaching on sandy soils) and contributes to the observed compression of the IQR in the ensemble statistics relative to site-scale studies such as Sifounakis et al. (2024).
(iii) Model parameters and calibration. The LandscapeDNDC parameter set used here is identical to the one calibrated and evaluated for Thessaly by Sifounakis et al. (2024) (see Section 3.1). While parameter distributions were not re-propagated in the present national-scale application for computational reasons, the reported fluxes ± SD ranges for the main fluxes (e.g. N₂O: 2.6 ±0.8 kg N2O-N ha⁻¹ yr⁻¹; SOC change: 0.5 ±0.3 t C ha⁻¹ yr⁻¹) can be taken as a proxy for parameter uncertainty. Be aware, Sifounakis et al. (2024) reported N fertilization up to 220 kg-N ha-1 and the crop rotation included a legume perennial feed crop frequently incorporated into the soil. Extrapolating a regional calibration to the full Greek territory introduces additional structural uncertainty for regions whose soils or climate fall outside the calibration envelope.
(iv) Land-use and land-cover (LULC) assumption. The spatial distribution of arable land is kept constant throughout the simulation horizon. Climate change is, however, likely to trigger LULC redistribution (e.g. abandonment of drought-marginal cropland, northward or upward migration of crops, expansion of irrigated land where water remains available). Such dynamics would affect both the denominator (area) and the numerator (fluxes per hectare) of all reported national budgets and are not represented here. This constitutes a first-order source of uncertainty for projections beyond the mid-21ˢᵗ century, in particular under RCP8.5.
(v) Irrigation management. The 52 / 48 % rainfed/irrigated split and the seasonal irrigation volumes were held constant into the future, as no consistent Greek-specific irrigation projections were available at the start of the study. Under RCP8.5 in particular, projected precipitation declines in parts of Greece may reduce the practical availability of irrigation water; our assumption therefore probably leads to an optimistic view of future crop yields and an underestimate of drought-driven reductions in crop C input to the soil. Conversely, if irrigated area were to expand to buffer heat stress, soil moisture-dependent emissions (N₂O, N₂) and nitrate leaching could be higher than projected here. We view this as the dominant management-related uncertainty for the RCP8.5 projections.
In addition, crop phenology under strongly warmed climates – in particular winter vernalisation of cereals – is represented with a level of mechanistic detail that may become insufficient under RCP8.5 end-of-century conditions, as already noted above. Taken together, these uncertainties do not invalidate the qualitative conclusions of this study (direction of change in yield, SOC, N₂O, NO₃ leaching) but they justify caution with respect to the exact magnitudes reported for the late-century time slice.
1.22 Avoid bullet points in the conclusion
Reviewer comment: Avoid writing bullet points in the conclusion.
Authors’ response: The bullet list (a–g) in the Conclusion has been deleted and rewritten as a single flowing paragraph (‘First… Second… Finally…’). The conclusion now reads as follows:
To increase prediction reliability and reduce uncertainty, several aspects should be considered in future assessments. First, the EURO-CORDEX ensembles for Greece exhibit large variance for the historical period, underlining the need for a general bias correction of the climate change projections before they are propagated through biogeochemical models. Second, for detailed climate-change impact analyses the same set of GCM/RCM downscaled projections should be used across RCP scenarios, in order to avoid introducing model-structural uncertainty that is only present in one scenario. Third, projections of irrigation management (spatial and temporal availability of irrigation water) should be derived consistently from each climate-change scenario in advance of the simulations, rather than assumed to be constant. Fourth, climate-change projections offer new perspectives for deriving future crop-cultivation strategies and timelines, including shifts in crop calendars, double-cropping potentials, and the replacement of winter crops whose vernalisation may fail under end-of-century warming. Fifth, nutrient supply and fertilisation should be dynamically calibrated towards the future to match evolving crop demand while respecting soil nutrient availability. Sixth, the process representation of severe heat-stress impacts (e.g. anthesis stress) needs to be improved in the model to reliably assess climate-change impacts on cereal yields. Finally, rather than refining outdated input datasets (e.g. the arable-area aggregation used here), the recently available high-resolution datasets for soil, climate and crop management should be adopted, which would enable high-resolution, field-scale national inventory simulations in subsequent studies.1.23 English proof-reading; unclear wording (L320-323; L309-311)
Reviewer comment: The manuscript could greatly benefit from English proof-reading. Specific examples of unclear sentences are given.
Authors’ response: All specific sentences flagged by the reviewer (L320-323, L309-311) have been rewritten in the revised manuscript with the missing qualifier ‘average’ and explicit attribution to mean/median and to the corresponding RCP. The full manuscript has been re-read for clarity and the use of ambiguous ‘or’ between two values has been replaced by clear ‘mean = X (median = Y)’ formulations. A professional language edit will be performed prior to final submission.
1.24 ‘C balance’ / ‘N balance’ terminology is misleading
Reviewer comment: The use of ‘C balance’ and ‘N balance’ throughout the document is misleading because the results show selected components of the cycles, not the balance.
Authors’ response: We respectfully retain the ‘balance’ terminology, since the manuscript does report all major input fluxes (mineral/organic fertilization, BiologicalNitrogenFixation, atmospheric N deposition, photosynthetic C fixation) and all major output fluxes (harvest C and N, gaseous N losses N₂O/NO/N₂/NH₃, NO₃⁻ leaching, ecosystem respiration), so that the algebraic sum constitutes a true balance, and this of the national-scale. To address the reviewer’s concern, the abstract and introduction now state explicitly which input and output components constitute the C and N balances reported here, and the figure captions of all balance figures have been extended to list the components shown (see comment below).
1.25 All figure captions need to be stand-alone
Reviewer comment: All captions of the figures should be reworded to reflect the contents of the figures so that they are stand-alone.
Authors’ response: All figure captions have been extended to be self-contained. Each caption now states (i) what is plotted (variable, units), (ii) the time slice and RCP scenario, (iii) the spatial aggregation (e.g. cropland-area-weighted national mean), (iv) the meaning of any uncertainty bands (mean ± SD, IQR with Q25–median–Q75), and (v) the data source/figure-construction method (e.g. waterfall diagram, with a brief explanation of how to read it, in response to the comment on waterfall plots).
Citation: https://doi.org/10.5194/egusphere-2025-5311-AC1
-
AC1: 'Reply on RC1', Edwin Haas, 20 May 2026
-
RC2: 'Comment on egusphere-2025-5311', Anonymous Referee #2, 18 Mar 2026
General evaluation:
This paper is of potential interest to the journal. The model landscapeDNDC was used to simulate future crop yield and carbon and nitrogen balances/fluxes over Greece, with input from a large number of GCM-RCM combinations for two different time slices and two different concentration pathways. While the study is interesting, it has some important limitations, which are discussed later in the paper but not when the model setup is introduced or in the results section. In my opinion, assumptions and limitations should be highlighted when introducing the model setup. Additionally, the paper does not provide all the necessary details about the model experiments. A few additional model simulations (e.g. for different irrigation scenarios) would significantly strengthen the paper. Finally, the introduction should be improved by providing a clearer explanation of the novel contribution of the work. The language used in the paper, especially in the introduction, should be improved. Overall, I recommend a major revision of the paper.
Main comments:
Introduction: The motivation for this work is not clear in the introduction. The review of the modelled impact of climate change on carbon and nitrogen budgets and fluxes in agro-ecosystems is rather limited. This should be extended. In fact, the discussion provides more background on this, and some of this material would be better placed in the introduction.
Section 2.2.1: How was management considered in the model spin-up? Management will have an important impact on the initial conditions, calling into question the assumption of equilibrium conditions at the beginning of the simulation.
The assumption that irrigation will remain constant in the future is unrealistic. Either irrigation will increase in response to lower precipitation and higher potential ET, or it will not increase due to limitations in irrigation. The timing of irrigation would also change. One possible solution is to make irrigation a model-internal function of simulated soil moisture content, for example by initiating irrigation when soil moisture decreases below the critical threshold. While the authors acknowledge the limitations of their approach to irrigation, their proposed solution is not convincing.
Section 2.3: It seems that land use and land cover change is constant for all time slices. This decision should be discussed directly in the text. The reasons for this assumption and its limitations should be made clear.
Section 3 contains mainly tables and time series plots summarising the results. It would be useful to include some maps showing the simulation results. For example, ammonia emissions, nitrate leaching and changes in carbon storage for different time slices.
Results: It would be good to discuss the results in more detail. For example, why is there a carbon loss under current conditions, despite the model spin-up having simulated an equilibrium state? In this context, it would be important to have detailed information on the model spin-up setup (management, time period, transient climate and CO₂ concentration, etc.).
Given the significant impact of farmer nitrogen input on the system, it would be important to simulate different scenarios. For example, under conditions of lower precipitation and lower yields, the input of organic and mineral nitrogen by farmers could also be reduced. While it is challenging to anticipate future changes, simulating at least two realistic scenarios would be helpful.
Detailed comments:
L30–L32: These years are inconsistent with those provided in the paper.
L129: Is this a complete meteorological input dataset? Does the model not require wind speed and humidity?
L172–L173: The relevance of these should be explained more clearly.
L224: It seems that Q25 and Q75 have been mixed up here.
Table 2: Remove the column showing percentage changes in temperature.
Figures 2, 7 and 8: increase the font size. The lines in the figures (e.g. yellow on a red background) are not very clear.
L262–L263: This description lacks information on the removal of carbon by surface water and erosion. Please mention all components.
L326: A brief explanation of the results would be helpful here, such as why NH₃ and N₂O emissions change.
L340: Were fluxes to open waters not mentioned before, or do you mean leaching to groundwater?
Figure 7 appears before Figure 6 in the paper.
L395–L401: Please state explicitly that these are modelled results.
L405–L406: This is incorrect. More irrigation is likely needed, but this has not been considered. One simulation scenario could have been to investigate the impact on yield if irrigation amounts do not increase. Another simulation scenario could have considered the impact on yield, carbon and nitrogen if irrigation is adapted.
L435: What is the role of increasing potential ET and other stressors, such as heat stress? It would be helpful if you could provide more information on the factors contributing to reduced yield, such as reduced precipitation and increased heat stress.
L480–L481: Do you mean leaching into groundwater?
Conclusions: see my general comments regarding the treatment of irrigation and the possibility of performing different simulation experiments for different fertilisation scenarios.
Language (not exhaustive):
L35-L36: rephrase sentence
Reformulate all highlights
L63: 2 x effects
L66: “etc.” Remove or add further examples.
L71-L72: reformulate sentence
L81: systematic model errors?
L87-L89: rephrase sentence
L90: reformulate sentence
L137: error message in text
L143-L144: rephrase sentence
L191-L192: clarify sentence
L226-L227: rephrase: “Data is shown in Figure (…)”
L249-L251: rephrase sentence
L369: rephrase sentence
L421-L422: sentence unclear to me.
L441: error message in text
L447: process description
L493: due to the
Citation: https://doi.org/10.5194/egusphere-2025-5311-RC2 -
AC2: 'Reply on RC2', Edwin Haas, 20 May 2026
In general, we want to point out to the reviewers and the editor the importance and novelty of the work presented in this study. While many models and studies have been addressing the carbon balance of arable systems from site to national to continental to global scale, there are to our knowledge only 3 modelling papers which present the full nitrogen balance including ALL nitrogen fluxes:
- Schroeck et al. 2019 – Estimating nitrogen flows of agricultural soils at a landscape level, https://doi.org/10.1016/j.scitotenv.2019.02.071
- Sifounakis et al 2024 – Regional assessment and uncertainty analysis of carbon and nitrogen balances at cropland scale using the ecosystem model LandscapeDNDC, https://doi.org/10.5194/bg-21-1563-2024, 2024.
- Rahimi et al 2024 – Aggregation of activity data on crop management can induce large uncertainties in estimates of regional nitrogen budgets.
https://doi.org/10.1038/s44264-024-00015-3
All three papers were authored by the corresponding author of this study: Edwin Haas. The reason for the lack of papers reporting the full nitrogen balance was clearly elaborated in a recent discussion paper by Balasz Groscz et al (2023) “Modeling Denitrification: Can We Report What We Don't Know?“ (https://doi.org/10.1029/2023AV000990). This discussion paper resulted from denitrification research project funded by the German Research Foundation (DFG). The conclusion and recommendation are summarized in the following two statements / sections:
- … we argue that including the entire N balance and related parameters should become standard when publishing the results of N model studies.
- We assume that the scarcity of “complete” (i.e. including N2 fluxes and other N pools/pathways) modeled N balances in the soil denitrification literature stems from the reluctance of the scientific community to support the publication of unvalidated modeled output, especially given that the simulation results of these ‘neglected’ N pools may be unrealistic. But this self-censorship of authors has resulted in a missed opportunity to share knowledge and improve our understanding of modeled processes. We recommend that future studies exercise transparency in publishing model outputs. We ask authors to focus on the aspects of their model that were of particular interest (i.e. validated model developments), but, while clearly stating which variables were not validated by measurements, to include all related pools and parameters to the fullest extent possible (e.g. all modeled N pools/pathways, soil aeration and CO2 flux).
Our presented study/manuscript was guided by these recommendations and therefore we combined the presentation of the carbon balance (well studied and less complex) and the nitrogen balance (rarely presented, very complex) for arable land cultivation on the national scale.
Reviewer 2
2.1 Introduction: improve motivation and review of climate-change impacts on C/N budgets
Reviewer comment: Motivation is not clear; review of modelled climate-change impact on C and N budgets in agro-ecosystems is rather limited. Some material from the Discussion would fit better in the Introduction.
Authors’ response: See also comment 1.5 of reviewer 1. We have inserted a new paragraph in the Introduction integrating both reviewers’ requests: it cites Fan et al. (2020) and Strenge et al. (2023) on national N inventories, and Basche et al. (2016), Carozzi et al. (2022), Deryng et al. (2011), Lugato et al. (2014), Petersen et al. (2021), Beaudor et al. (2025) and Shen et al. (2020) on modelled climate-change impacts on cropland C and N cycling. Material previously in the Discussion that provided background was moved to this new Introduction paragraph.
2.2 Section 2.2.1: Management during model spin-up
Reviewer comment: How was management considered in the model spin-up? Management will affect the initial conditions and the assumption of equilibrium.
Authors’ response: We now state explicitly that during the 15-year spin-up (1990–2005) the agricultural management of each grid cell — i.e. the 4 most likely crop rotations including fertilization, tillage and irrigation regimes used for the impact period — was applied repeatedly so that soil C and N pools approached a management-consistent dynamic equilibrium under historical climate. The remaining initial-condition uncertainty (the spin-up does not constitute a strict equilibrium because management changed over the historical period) is now discussed in the Uncertainty section (1.21). In the supplementary material document, the agricultural management is presented in full detail.
2.3 Constant irrigation in the future is unrealistic
Reviewer comment: Either irrigation will increase or it cannot increase due to limitations. Suggest making irrigation a model-internal function of soil moisture; or simulate at least two realistic irrigation scenarios.
Authors’ response: We fully agree that the constant-irrigation assumption is a major limitation of the study. To our knowledge, there is no study available which describes a prototype on how to perform such a projection of irrigation capacity under a climate change ensemble into the future. We know, that the approach we use does not consider the soil water availability for irrigation within each ensemble simulation, but we use a rather low irrigation regime. We understand the criticism but we need to point out:
- Programming a model-internal irrigation regime into LandscapeDNDC
It would be possible to trigger irrigation on a soil water content threshold. This is the GGCMI approach (Jaegermeyr et al, 2021, nature climate change). But this would pose two major concerns: i) Each ensemble simulation would end up having a different irrigation regime due to the differences in the precipitation levels and timings of each climate ensemble member. To our understanding, this would pose high uncertainties and would cause heterogeneity, which would make it difficult to treat the ensemble as a homogenous set of simulations. Ii) This would not fulfill any constrain of potential water mass balances of the ensemble members.
We do not think that this is mor realistic than our selected scenario. - Defining and deploying realistic irrigation scenarios
The literature does not report any projection of irrigation into the future under climate change conditions. There is no e.g. MIRCA2000 projection until 2100 e.g. under CMIP6 equivalent. This does not exist so far. End of 2024 a dataset on “Projected Global Area Equipped for Irrigation Datasets during 2020-2100 under SSP scenarios” (https://zenodo.org/records/14177960) has become available in combination with ISIMIP. This is still not compatible with the EURO Cordex Ensembles and was not available at the beginning of the study when the input data was collected and the management was prepared.
Under the given circumstances we decided not to follow the reviewer’s advice and to use automatic irrigation or “define at least two realistic irrigation scenarios” and rerun the full set of ensemble simulation. In our opinion, doing this would pose a much larger uncertainty to the study. We would like to point out, that a number of studies in the past have used the assumption of projecting an arable management into the future as we did, therefore keeping this approach makes our study comparable.
Re-running the full ensemble with two new irrigation scenarios is unfortunately not feasible within the revision time-frame. To address this concern we have (i) explicitly stated the limitation already in Section 2.3 (formerly only in the Discussion), (ii) added a dedicated paragraph in the new Uncertainty subsection that quantifies the likely direction and order-of-magnitude effect of fixed irrigation on yield, soil C, N₂O and NO₃⁻ leaching under RCP8.5, and (iii) added to the Conclusion stating that future studies should derive irrigation projections consistently from each climate-change scenario in advance of the simulations and explore demand-driven (soil-moisture-based) irrigation as suggested by the reviewer.
2.4 Section 2.3: LULC kept constant — discuss reasons and limitations
Reviewer comment: It seems that land use/land cover change is constant for all time slices. The reasons and limitations of this assumption should be discussed in the text.
Authors’ response: A short paragraph has been added at the end of Section 2.3 stating that the spatial distribution of arable land was held constant over the simulation horizon, that no LULC scenarios were prescribed (because no internally consistent Greek-specific LULC projection ensemble was available), and that potential future LULC shifts (abandonment of drought-marginal cropland, crop-type migration, irrigation expansion) are therefore not represented. The implications are discussed in the new Uncertainty subsection and addressed in the conclustion.
2.5 Include some maps of simulation results
Reviewer comment: Section 3 contains mainly tables and time-series plots. Include maps for ammonia emissions, nitrate leaching, changes in C storage, etc.
Authors’ response: We appreciate this comment. The approach to assess and present the full carbon and nitrogen balances und climate change offers various potential maps to be displayed. The study focuses on a national-scale aggregation and this is already a challenge to present and discuss. Presenting maps in the main manuscript will rise the need of analysis and discussion on spatial patterns, correlations to soil etc. and would therefore distract the focus of the study, limit the presentation and discussion as provided by the manuscript.
We have committed to provide a Supplementary Material section containing all mentioned spatially resolved maps for the present (2005–2034) and future (2070–2099) time slices under RCP4.5 and RCP8.5 for: carbon in yield, soil C change, N₂O emissions, N₂ emissions, NO₃⁻ leaching, NH₃ volatilization and total N balance.
A pointer sentence has been added at the start of the Results explaining that the main text focuses on national-scale aggregates and that spatial maps are in the supplementary materials document for completeness. In the discussion, we mentioned the issue of spatial resolution as uncertainty source. In the conclustion we added:
Finally, rather than refining outdated input datasets (e.g. the arable-area aggregation used here), the recently available high-resolution datasets for soil, climate and crop management should be adopted, which would enable high-resolution, field-scale national inventory simulations in subsequent studies.2.6 Why is there a C loss under current conditions despite spin-up equilibrium?
Reviewer comment: Why is there a C loss under current conditions, despite the model spin-up having simulated an equilibrium state?
Authors’ response: LandscapeDNDC performs a soil carbon initialization into various carbon pools represent the initial given SOC mass and ensure stability. Typically, observed SOC data and agricultural management fulfills these equilibrium assumptions. The 15-year spin-up (1990–2005) brings the system additional into a dynamic equilibrium relative to the provided management and climate of that period. Potentially, this could mean, that the amount of crop residues plus below ground crop litter does not sustain the initial soil carbon stocks in long term. The assumption of residue management is e.g. 50% residues (straw) remain on the field and 50% get removed for cerials as derived from FAO data. The ‘present’ time slice (2005–2034) already includes ~20 years of the projected EURO-CORDEX climate, which is on average warmer and with shifted precipitation compared to 1990–2005. The simulated soil C loss for ‘present’ therefore reflects the response of the spin-up state to the ensemble climates of 2005–2034 and the management, not a disequilibrium of the spin-up itself. Any other ecosystem model would respond similarly. We have added an explicit explanation in Section 3.3 of the Results:
This indicates, that the amount of residues left on the field (50%) plus the below ground crop litter (roots) does not fully sustain soil carbon decomposition.2.7 Different fertilization scenarios should be simulated
Reviewer comment: Given the significant impact of farmer N input, simulate different fertilization scenarios (e.g. lower N input under lower yield conditions).
Authors’ response: We agree that fertilization scenario analysis is an important follow-up. As for irrigation (2.3), running the full EURO-CORDEX ensemble with several N-fertilization scenarios is beyond the scope of this revision.
We have projected a so-called baseline scenario into the future with all constraints and criticism. As stated in the Diskussion of Haas et al. (2022), this projection should be recalibrate on e.g. a decadal basis for the fertilization demand based on plant growth and residues return and being adapt it into the future. Potentially, a climate change projection could lead to a yield decline and then the historic fertilization levels would pose a potential overfertilization. But for a study based on ensembles, this would lead to heterogeneity and creating difficulties in the ensemble analysis.
We have therefore (i) added a sentence to the Discussion acknowledging that the constant-fertilization assumption is a limitation of the study, and (ii) listed the dynamic calibration of N fertilization to evolving crop demand and soil nutrient availability as one of the key recommendations for follow-up studies in the revised Conclusion.2.8 L30-L32: time-slice years inconsistent with the rest of the paper
Reviewer comment: The years given in the abstract are inconsistent with those elsewhere in the paper.
Authors’ response: The abstract now explicitly defines the time slices as present (2005–2034) and future (2070–2099), and these intervals are now used consistently throughout the manuscript (methods, figure captions, tables). A clarifying sentence has been added in the methods stating that figures and tables follow this convention unless otherwise stated.
2.9 L129: Is the meteorological input dataset complete? Wind speed, humidity?
Reviewer comment: Does the model not require wind speed and humidity?
Authors’ response: LandscapeDNDC does provide several modules to model plant growth and soil biogeochemistry. The configuration used in this study is based on the DNDC plant growth module in combination with the DNDC soil bio-geochemistry module. They calculate photosynthesis without . The more advanced plant growth module plamox computes photosynthesis based on Farquhar plus the Leuning stomatal model taking wind speed and humidity into account. But these modules need hourly time steps cnreasing the computational effort by a factor of at least 20. Additionally, the availability of climate change ensemble members were already limited by the availability of input files in the repository for the complete time slice and for all input variables, which was often not the case leading to the heterogenic sets of ensemble members between RCP4.5 and RCP8.5. Limiting the needed input variables helped to maintain the ensemble sets to their large extend.
Therefore, we have not considered these options in this study. The DNDC modules compute potential evapotranspiration with the Thornthwaite/Hamon scheme (driven by T and radiation) and plant growth with an empirical plant growth description based on growing degree days and water and nutrient limitations, which does not require wind speed and humidity. We have added a sentence in Section 2.2.1 to explain this and to note that wind speed and humidity are therefore not required as input.2.10 L172-L173: relevance unclear
Reviewer comment: The relevance of these two lines should be explained more clearly.
Authors’ response: As described in the answer above, the selection of the sets of ensemble members for RCP4.5 and RCP8.5 were not by choice but by the limitation of data availability at the beginning of the project. In the ideal case, both ensembles should have climate change projections representing all GCMs and RCMs to a large extent. Unfortunately, the Euro Cordex data was distributed across several European Datacenters and accessibility was limited. Additionally, some datacenters did not offer all data variables or time slices needed for the climate input, some datasets could not being downloaded or some files were not consistent or corrupted and could not being reprojected or processed. After a large effort to download, reproject and process the climate data, we ended up with the available ensemble members as stated in table 1. The attribute common / unique refers to whether this GCM/RCM combination is in both (common) RCP4.5 and RCP8.5 ensembles or only in one of them (unique) represented.
The section has been rewritten to make explicit why the two ensembles to not contain the ideal selection of ensemble members.
The section states as follows:
In the ideal case, both ensembles should have climate change projections representing all GCMs and RCMs to the largest extent, but due to constraints in data availability for download from the various data centers/repositories, the completeness for all necessary variables and time slices, as well as data consistency to repoject them onto the simulation grid, we had to limit the study to the climate change projections summarized in Table 1 at the start of the project. Note that 14 out of 32 GCM/RCM combinations are common in both RCP ensembles, while the RCP4.5 ensemble contains 2 unique combinations not present in the RCP8.5, and the RCP8.5 ensemble contains 18 unique combinations not present in the RCP4.5 ensemble. This is summarized with the common/unique attribute in Table 1.
The table 1 caption states:
14 GCM-RCM combinations are common in both RCPs, while RCP4.5 ensemble contains 2 unique combinations, and the RCP8.5 ensemble contains 18 unique combinations. 1) global circulation model, 2) regional climate model (for details, see Olschewski et al., 2024).2.11 L224: Q25 and Q75 mixed up
Reviewer comment: Q25 and Q75 appear to be swapped at L224.
Authors’ response: We confirm the swap. The sentence reporting the temperature increase under RCP4.5 has been corrected to ‘Q25 of 1.31 and Q75 of 1.43 °C’ (instead of the previous Q25 1.43 and Q75 1.31). All other Q25/Q75 entries have been re-checked.
2.12 Table 2: remove the percentage-temperature column
Reviewer comment: Remove the column showing percentage changes in temperature.
Authors’ response: After consultation, we want to keep the percentage-change column for temperature as we have it for precipitations. Considering only absolute temperature changes (°C) would only make sense if we need save space. We would like to keep the terminology to present absolute and and relative (%) changes.
2.13 Figures 2, 7 and 8: increase font size and improve line contrast
Reviewer comment: Increase font size; lines (e.g. yellow on red background) are not very clear.
Authors’ response: a colour-blind safe palette (Wong, 2011). ????
2.14 L262-L263: missing C removal by surface water and erosion
Reviewer comment: Description lacks information on the removal of carbon by surface water and erosion.
Authors’ response: We have added a sentence to clarify that horizontal C export by erosion and surface runoff is not represented in the current LandscapeDNDC setup. We have added the following statement at the definition of carbon outfluxes:
Other carbon losses such as erosion or lateral transport via surface runoff is not considered in this setup.2.15 L326: brief explanation of NH₃ and N₂O changes
Reviewer comment: A brief explanation of the results would be helpful here, e.g. why NH₃ and N₂O emissions change.
Authors’ response: We have added a short mechanistic explanation in the results section: NH₃ volatilisation increases with temperature (because the NH₃/NH₄⁺ partition shifts to gas phase under warmer conditions) and partly compensates the reduction in N₂O and N₂ losses, which are themselves driven down by reduced soil moisture and a shorter window of denitrification activity in the drier future climate.
2.16 L340: ‘open waters’ vs leaching to groundwater
Reviewer comment: Were fluxes to open waters not mentioned before, or do you mean leaching to groundwater?
Authors’ response: We now use the term ‘NO₃⁻ leaching to groundwater’ consistently. This was a wording inconsistency and vertical nitrate transport by percolation water fluxes is the process.
2.17 Figure 7 appears before Figure 6
Reviewer comment: Figure 7 appears before Figure 6 in the paper.
Authors’ response: The figure 6 has been removed and will be added to the supplementary material document..
2.18 L395-L401: state explicitly that these are modelled results
Reviewer comment: Please state explicitly that these are modelled results.
Authors’ response: Throughout L395-401 (and in equivalent passages elsewhere) we have inserted ‘simulated’ or ‘modelled’ qualifiers wherever the values originate from the LandscapeDNDC ensemble.
We added the statement: The mentioned three assessments were all based on modelling studies.2.19 L405-L406: irrigation statement incorrect
Reviewer comment: ‘More irrigation is likely needed’ has not been considered. Suggest two simulation scenarios: (a) yield with constant irrigation; (b) yield, C and N with adapted irrigation.
Authors’ response: The statement was misleading and has been removed from the manuscript.
2.20 L435: role of potential ET, heat stress and reduced precipitation
Reviewer comment: What is the role of increasing potential ET and other stressors such as heat stress?
Authors’ response: We have extended the Discussion at L435 to describe explicitly the three main drivers of yield decline under RCP8.5: (i) reduced precipitation and shortened wet seasons, (ii) increased potential evapotranspiration leading to earlier soil-moisture stress and shorter grain-filling windows, and (iii) episodic heat stress around anthesis for wheat and around silking for maize. The limitations of the current heat-stress representation in LandscapeDNDC are noted in the Discussion and Uncertainty section.
2.21 L480-L481: leaching into groundwater?
Reviewer comment: Do you mean leaching into groundwater?
Authors’ response: Yes — the wording has been clarified to ‘NO₃⁻ leaching into groundwater’, consistent with 2.16.
2.22 Conclusions: irrigation and fertilisation simulation experiments
Reviewer comment: See general comments regarding irrigation and the possibility of performing simulations for different fertilization scenarios.
Authors’ response: The Conclusion paragraph has been rewritten as flowing prose (in response to 1.22) and now explicitly recommends, as follow-up steps: (i) deriving irrigation projections consistently from each climate-change scenario in advance of the simulations, (ii) implementing demand-driven irrigation, and (iii) dynamic decadal based calibration of future fertilization rates to evolving crop demand and soil nutrient availability under future climate. Adaptations to the manuscript have been made.
2.23 Language list (L35-36, highlights, L63, L66, L71-72, L81, L87-89, L90, L137, L143-144, L191-192, L226-227, L249-251, L369, L421-422, L441, L447, L493)
Reviewer comment: A list of specific lines requiring rephrasing is provided.
Authors’ response: Each of the items in the language list has been individually addressed in the tracked-changes manuscript: duplicate words removed (L63 ‘effects effects’), ‘etc.’ replaced or removed (L66), error messages cleaned (L137, L441), unclear sentences (L35-36, L71-72, L87-89, L90, L143-144, L191-192, L226-227, L249-251, L369, L421-422) reformulated, ‘systematic model errors’ clarified (L81), ‘process description’ added at L447, and ‘due to the’ corrected at L493. Highlights have been fully reformulated (1.3). A final professional language edit will be performed prior to final submission.
We thank both reviewers once more for the time and effort spent on this manuscript. We hope that the changes summarised above and implemented in the revised manuscript address all major concerns and are sufficient for the paper to be considered for publication in Biogeosciences.
Citation: https://doi.org/10.5194/egusphere-2025-5311-AC2
-
AC2: 'Reply on RC2', Edwin Haas, 20 May 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,242 | 941 | 115 | 2,298 | 312 | 100 | 104 |
- HTML: 1,242
- PDF: 941
- XML: 115
- Total: 2,298
- Supplement: 312
- BibTeX: 100
- EndNote: 104
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The paper addresses a novel and relevant scientific topic of carbon and nitrogen fluctuations under climate change in Greece. However, the scientific methods and assumptions are not clearly outlined, there are several shortcomings, such as inadequacies in the presentation of the methodology; the description of experiments and calculations are insufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results). The results are not sufficient to support the interpretations and conclusions, so that the conclusions cannot be retraced. Therefore, manuscript is not ready for publication.
The manuscript struggles to convey in clear language the scientific methodology and discuss robust results. The study is difficult to read primarily due to unclear methodological descriptions and a lack of several important details. A scientific manuscript also has the purpose to document results so that they are reproducible, however the inconsistencies and lack of attention to details leaves me wondering if all of the co-authors have proof-read the paper before submission.
The manuscript deals with C and N balances. As such I would expect a clear and prominent description of both formulae providing the inputs and outputs of the respective balances that the model deals with, and which are furthermore presented and discussed in the manuscript. Any indication of the inputs and outputs are currently lacking.
As well, not all readers are familiar with a waterfall diagram. An explanation of how to interpret such a diagram would be very useful. E.g. what are the start and end points? Do the gaps or lines correspond to anything?
As a reviewer, it is not my job to correct the wording or English syntax, so I shall omit undertaking these corrections and focus more on the content.
The title does not clearly reflect the contents of the paper, since about a third to one half of the paper deals with carbon results, but the word carbon does not appear in the title.
The abstract (particularly from L 31 onwards) needs to be reworded to reflect the results and clarify several issues mentioned below. Several sentences are not clearly phrased. What is the spatial unit examined in the study, because 0.25 degree is stated, yet the results are presented in ha-1 yr-1, why is this so? Be more specific when stating “arable production” as I believe this refers to the crop yields in terms of C in the biomass. What does soil C decrease? Which gases are emitted? Why are 2 values given for the gaseous emissions, what do these refer to? Etc.
The highlights require reworking. The first two points are related to climate scenarios, not to the core results of the study. The third point is not clear (see above comment pertaining to abstract). Does the 4th point refer to overall total C, or just biomass C, or just soil C?
The graphical abstract is not a “stand alone” figure, as the parameter names (y axis) may not be known to all readers. The number in the top left 5.7 kg N ha-1 yr-1 requires explanation. The inputs and the outputs of the N balance should be clearly marked or labelled.
In the introduction, the first part L 60-83 is too broad and does not add much new information. I would expect a more focused look at national studies of national N inventories, such as by Fan et al. 2020 and Strenge et al., 2023. A look at the main sources and sinks of N and C, and how these are affected by climate change is relevant to this study.
The research questions on L 101-108 are not formulated as research questions, but rather as bullet point statements.
The description of the Landscape DNDC model L 112-138 must include the spatial and temporal resolution at which it was set up. What was the input data used to set up the model including the time periods. L 128 states “daily or hourly climate variables”, but which was used in the study? All of the above is important information, as it lends credibility and rigor to the methodology. Thus, referring to Hass et al., 2013 is not sufficient, in my opinion.
L 151- 162: briefly describe the level of spatial resolution of the input data. What management data specifically was used as input? Tillage, plant residues, seeding and harvesting dates, fertilizer application and type? Was the same crop rotation in Hass et al., 2022 used for Greece and does this make sense to transpose the cropping systems to Greece?
L 153 states that 15 years were used for the warm up period (1990-2005). This contradicts L 31 of the abstract (10 years warm up period).
L 187-199 this section is very unclear. E.g. the year 2099 is stated, which contradicts several other parts of the study that state the simulations were carried out to the year 2100. L 195 mentions “spatial and temporal means are assumed” but what does this mean? Also, “these statistics” are mentioned, but which statistics are meant?
-L 208: describe which statistical analysis were carried out
I am missing a section at the beginning of the Results section on the calibration and validation of the model Landscape DNDC. How robust is this model? What was the observed data used to calibrate and validate the model (SOC, N), and which spatial resolution and time period was examined? How was the model calibrated and validated, with which method? What was the objective function used? etc…
- L 223-4: is it standard to use Q25 and Q75 in the text? Would be nice to introduce the meaning first, presumably it refers to quantile?
L 227-8: why is “a systematic decrease expected under both RCPs”? this seems to pre-empt the results
Furthermore, generally in the manuscript a description of the main crops grown in Greece are missing. Which crops are being examined in the study mainly? And what are typical management practices that are being examined and discussed in this manuscript?
Section 3.2 (L 248 -) needs to be reworked. It is not clear which response is meant. Is there is figure of these results? What is meant by “arable production”? Is it the C content of the biomass specifically?
It is not clear why the C decreases over the future period (also shown in Fig 2). This is presumably a decrease in yields. Why do they decrease? Looking at Fig 3 it is also not clear how the decrease is reflected in this figure 3?
I also suggest to move Section 3.3 before Section 3.2, so that the authors state the main C processes and then focus on the results of C in the biomass.
The discussion has repetition which can be omitted, see L395-401. The studies mentioned e.g. L501, are these measured or modelled N2O emissions?
L 466 - The whole country of Greece is lumped into one basket when discussing soil C loss rates. How useful is this? It would be more appropriate to have nuanced and land-use specific results discussed here. How comparable is Greece with the study from Hass et al., 2022, if the climate and soils are different between the studies?
I am missing a section in the uncertainties of the study, e.g. input data, spatial resolution, calibration and validation process and data. Interpretation and comparability with measured data.
Avoid writing bullet points in the conclusion.
The manuscript could greatly benefit from an English proof-reading service. Here are some examples of unclear scientific findings: L 320-323: „For the future time slice, we notice stronger differences in gaseous out-fluxes as N2O emissions under RCP4.5 of 0.476 (mean and median) versus 0.453 or 0.457 kg N2O–N ha−1 yr−1 under RCP8.5. N2 emissions show a decline in the future, comparing under RCP4.5 of 4.252 or 4.354 kg N2–N ha−1 yr−1 versus 3.377 or 3.171 kg N2–N ha−1 yr−1 under RCP8.5.” The values presented are not clearly attributable to what they are referring to; the mean or the median, or both? What about the RCP scenario) Also, the use of the preposition “or” is very confusing.
Another example: L 309-311, “Nitrogen deposition from atmospheric sources contributed by 5.68 kg N ha−1 yr−1 on croplands (see Table 5 and Figure 5), while synthetic nitrogen fertilization was 100.88 kg N ha-1 yr-1 and organic nitrogen fertilization was 19.89 kg N ha-1 yr-1 across both scenarios and all years and rotations.” I would expect to read the word “average” at least once in this description.
The use of the “C balance” and “N balance” throughout the document (e.g. in figure titles) is misleading. These results do not show the “balance” but rather selected components of the cycles.
All the captions of the figures should be reworded to reflect in detail the contents of the figures and explain them, so that they are “stand alone” figures.
The journal has high standards, which in my opinion are simply not met in this manuscript.
References
Fan, X., Worrall, F., Baldini, L.M., Burt, T.P., 2020. A spatial total nitrogen budget for Great Britain. Sci. Total Environ. 728, 138864 https://doi.org/10.1016/j. scitotenv.2020.138864.
Strenge E, Zoboli O, Mehdi-Schulz B, Parajka J, Schönhart M, Krampe J and Zessner M 2023 Regional nitrogen budgets of agricultural production systems in Austria constrained by natural boundary conditions J. Environ. Manage. 347 119023 https://doi.org/10.1016/j.jenvman.2023.119023