Transforming UK agriculture: the impact of regenerative farming in Yorkshire under a changing climate

Dataset overview This record provides Version 2 of the FixOurFood regenerative agriculture modelling dataset. It contains outputs from site-specific APSIM Next Generation simulations developed to evaluate how individual and combined regenerative agriculture (RA) practices affect crop productivity and soil organic carbon (SOC) dynamics in a rainfed arable system near Tadcaster, North Yorkshire, United Kingdom. The modelling represents a commercial cereal–oilseed production system under historical (1993–2023) and projected future (2030–2060) climates. Fixed and time-varying atmospheric CO₂ treatments were included to distinguish the effects of projected climate change from the potential effects of increasing atmospheric CO₂. The simulations were conducted using APSIM Next Generation, version 2026.07.8076.0. The conventional Baseline comprised two three-year rotations: winter wheat–winter barley–winter oilseed rape; and winter wheat–spring barley–winter oilseed rape. Five RA practice categories were evaluated individually: crop diversification, simple cover cropping, reduced synthetic nitrogen fertilizer, perennial herbal leys and livestock integration. Because mineral nitrogen fertilizer was reduced by 10%, 30% and 50% in separate simulations, the five practice categories produced seven individual scenario variants. Three combined RA systems were also simulated: Ideal farming system: crop diversification, simple cover cropping, a three-year herbal ley and livestock integration; Most productive system: crop diversification, a three-year herbal ley and livestock integration, excluding the simple cover crop; and Least productive system: simple cover cropping combined with a 50% reduction in mineral nitrogen fertilizer. These combined systems are modelling archetypes constructed to compare contrasting combinations of regenerative practices. Their names describe the scenarios used in the study and should not be interpreted as universal recommendations or performance rankings. Dataset contents The dataset package contains five principal Excel workbooks: Individual RA crop yield, flowering periods and SOC: calculation worksheets and publication-style summaries for the conventional Baseline and individual RA scenarios. Individual RA historical daily-report data: daily APSIM outputs for the historical period, 1993–2023. Individual RA future daily-report data: daily APSIM outputs for the projected future period, 2030–2060. Combined RA systems yield, flowering periods and SOC: calculation worksheets and publication-style summaries for the three combined RA systems. Combined RA systems historical and future daily-report data: daily APSIM outputs for the combined systems under both climate periods. The package also includes a reader-oriented README in Word and PDF formats and a SHA-256 checksum manifest that can be used to verify file integrity. The daily-report workbooks retain variables describing crop yield and development, above-ground biomass, soil-water dynamics, nitrogen cycling, crop stress and SOC. These data support recalculation of the reported outcomes and further investigation of the biophysical processes underlying differences among management systems and climate treatments. The principal SOC variable used in the accompanying manuscript is SOC_Accrual_30cm_tonnes_C_ha, representing SOC in the upper 0–300 mm soil layer. The additional SOC_Accrual_60cm field, representing the upper 0–600 mm profile, is retained in the individual-practice daily-report workbooks for supplementary analysis with reference to the APSIM soil-profile configuration. Data processing and derived outcomes Crop-yield records were summarized by simulation, climate period, rotation, zone, crop and calendar date. A 15-day running average was calculated for each simulation–zone–crop combination. Reported crop yield corresponds to the maximum running-average yield, converted from kg ha⁻¹ to t ha⁻¹. Optimal flowering periods were defined as the date window during which the running-average yield was at least 95% of its maximum. This procedure was applied consistently across the Baseline, individual RA practices and combined RA systems. SOC was reported primarily as stock and change in stock in the upper 0–300 mm soil layer. Absolute SOC accrual was calculated as the difference between SOC stocks at the end and beginning of each 31-year simulation period. Average annual SOC change was derived from the change in SOC stock over the corresponding simulation period. The revised manuscript also used the dataset to: · compare the observed responses of combined RA systems with the expected sum of their component-practice responses; · classify combined-system responses as additive, super-additive or sub-additive; · partition the contributions of management, climate and their interactions to long-term and annual SOC dynamics; and · identify the principal ecological pathways associated with crop productivity and SOC change using SHAP-based explainable machine learning. In the additivity analysis, a combined system could provide the greatest absolute productivity or SOC benefit while remaining sub-additive if its observed improvement was smaller than the sum of the improvements produced by its component practices when simulated individually. Scientific contribution and principal findings The revised analyses provide new insight into both the performance of regenerative systems and the ecological processes underlying their responses. Regenerative management generally increased SOC accrual and often enhanced or maintained crop productivity relative to the conventional Baseline. Perennial herbal leys, crop diversification and livestock integration produced the greatest individual benefits, whereas reducing synthetic fertilizer without equivalent biological nutrient replacement created stronger productivity and SOC trade-offs. The combined RA systems integrating crop diversification, perennial herbal leys and livestock produced the greatest absolute improvements in crop productivity and SOC. The Ideal farming system achieved the highest overall SOC accrual, reaching approximately 27.6 t C ha⁻¹ under the historical climate and 24.4 t C ha⁻¹ under the projected future climate. However, additivity analysis showed that the responses of the combined systems were predominantly sub-additive rather than synergistic. Their absolute benefits were greater than those of the Baseline and individual practices, but the observed improvements were generally smaller than the sum of the effects produced by their component practices when simulated separately. This indicates that several practices influenced overlapping or increasingly limiting ecological processes, resulting in diminishing proportional gains as additional practices were stacked. Projected future climate reduced pooled crop yield and absolute SOC accrual by approximately 15.9% and 13.6%, respectively. Higher atmospheric CO₂ increased the corresponding outcomes by approximately 6.5% and 1.9%, but these gains were insufficient to offset the projected climate-driven losses. Regenerative systems therefore reduced the magnitude of some climate impacts but did not eliminate them. Variance decomposition showed that management explained approximately 94.2% of the variation in long-term SOC accrual. Annual SOC dynamics were more strongly climate-dependent: the interaction between management and climate year explained approximately 47.9% of annual SOC variation, while climate year independently explained a further 34.1%. These results indicate that management established the long-term capacity of the systems to accumulate SOC, whereas annual climate variability influenced the rate at which that capacity was realised. The explainable machine-learning analyses identified soil evaporation and surface-residue carbon as leading predictors of annual SOC change. Maximum crop biomass was the principal predictor of crop-yield outcomes, with additional crop-specific effects associated with water stress and harvest timing. Collectively, the analyses indicate that the benefits of regenerative management arose principally from restoring complementary ecosystem functions governing biomass production, biological nitrogen supply, carbon inputs, soil-water regulation and nutrient cycling, rather than from strong synergistic interactions among stacked practices. Intended uses The dataset can be used to support: · independent recalculation of crop-yield, optimal-flowering-period and SOC summaries; · verification and reproduction of manuscript figures and tables; · comparison of conventional, individual and combined regenerative-management scenarios; · evaluation of additive, super-additive and sub-additive system responses; · investigation of soil-water, nitrogen, biomass, residue and crop-stress processes; · further analysis and refinement of APSIM-based management scenarios; · development of process-based and explainable machine-learning analyses; and · research on climate-resilient arable-system design. Interpretation and limitations All outputs are model-derived scenario results rather than direct field measurements. Their interpretation should therefore consider the site-specific soil profile, climate data, crop-management assumptions, atmospheric CO₂ treatments and APSIM configuration used in the study. The simulations represent a case-study arable system in North Yorkshire and should not be interpreted as universal predictions for all farms, soils or climatic regions. The scenario names assigned to the combined systems are study-specific and do not constitute general rankings or prescriptive recommendations. SOC accrual represents a change in modelled soil carbon stocks. It should not be interpreted as a complete net whole-farm greenhouse-gas balance because the simulations reported here do not integrate all methane, nitrous oxide, livestock and external-input emissions. Users should consult the introductory worksheets, calculation worksheets and raw daily-report workbooks when reproducing results or conducting additional analyses. Related manuscript This dataset accompanies the manuscript: “Transforming UK agriculture: impact of regenerative farming in Yorkshire under a changing climate.” Publication or preprint details will be added to the Zenodo record when available. Version 2 update This release is Version 2 of the dataset. Relative to Version 1, the principal data revisions concern updated SOC outputs and the corresponding SOC calculation worksheets. The individual-practice historical and future daily-report workbooks contain revised values for: · SOC_Accrual_30cm_tonnes_C_ha; and · SOC_Accrual_60cm. The associated SOC calculation and summary worksheets have been updated to use these revised values. The introductory worksheets and README have also been expanded to document SOC variables, calculations and interpretation and to align the scenario descriptions and terminology with the revised manuscript. Version 2 additionally documents the expanded analyses undertaken during manuscript revision, including combined-system additivity, variance decomposition and SHAP-based process interpretation. These analyses produced novel findings that were not captured in the preliminary Version 1 description. In particular, the earlier preliminary characterization of stronger combined-system “synergies” has been refined: the combined systems achieved the greatest absolute productivity and SOC benefits, but their responses were predominantly sub-additive relative to the summed effects of their component practices. Crop-yield records, flowering-period results and other model outputs are retained from Version 1, except where an associated worksheet or description required updating to maintain consistency with the revised SOC calculations and manuscript.