Continental-scale evidence of farm management impacts on soil carbon

Continental-scale evidence of farm 1 management impacts on soil carbon 2 3 Julian Helfenstein1, Nick van Dijk1, Anna Edlinger2, Gabriel Y.K. Moinet3, Sophie van Rijssel1, 4 Alexandre M.J.-C. Wadoux4, Rachel Creamer3, Carmen Vazquez3, Vera L. Mulder1 5 1 Soil Geography and Landscape Group, Wageningen University & Research, Wageningen, the 6 Netherlands 7 2 Wageningen Environmental Research, Wageningen University & Research, Wageningen, the 8 Netherlands 9 3 Soil Biology Group, Wageningen University & Research, Wageningen, the Netherlands 10 4 College of Science and Engineering, James Cook University, Cairns, Australia 11 12 Keywords: carbon farming, climate-smart farming, agricultural management, soil carbon 13 dynamics, soil organic carbon stocks 14 15 16 ABSTRACT 17 There are high expectations that agricultural practices can mitigate climate change and improve 18 soil health by increasing soil organic carbon (SOC). However, existing large scale SOC monitoring 19 treats agricultural management as a black box, meaning that observed patterns and trends cannot 20 inform on sustainable practices. Here, we for the first time combine management data from 21 systematic farm surveys (n =81,688 farms) and representative soil monitoring data (n = 8,834 22 locations) to quantify the impact of agricultural practices on three SOC metrics across Europe: 23 stocks, stocks relative to pedoclimatic benchmarks, and yearly change in SOC concentration. Our 24 findings show that management intensity is a significant contributor to SOC loss across Europe, 25 with varying impact by soil and climate region. In turn, several practices (e.g. high share of 26 manure, organic management, and a high proportion of leys in crop rotation) demonstrated 27 potential for increasing SOC in different agricultural systems. This research opens new avenues 28 for understanding leverage points to enhance soil health, and provides unprecedented evidence for 29 informing policies promoting sustainable farming practices. 30 31 MAIN 32 Maintaining soil organic carbon (SOC) stocks in agricultural systems is critical for sustaining soil 33 health, and avoiding CO₂ emissions from degradation1–4. Protecting existing SOC helps preserve 34 nutrient cycling, soil structure, biodiversity, and water regulation, all of which underpin productive 35 and sustainable farming systems5,6. Increasing SOC stocks, defined as soil carbon sequestration7, 36 can also remove CO₂ from the atmosphere, but global sequestration potential is modest and time37 limited due to steady-states, reversibility, and socio-economic constraints8. Even optimistic 38 scenarios suggest it could offset only a small share of the emission reductions needed for climate 39 targets4. Nonetheless, optimal SOC management is a central component of sustainable soil 40 management and sits on the right side of the climate equation. A significant amount of research 41 has shown that the adoption of sustainable management practices such as diverse crop rotations, 42 organic amendments, or perennial cropping - among others - can slow or reverse declines in SOC 43 stock1,9. 44 However, current understanding of management effects on SOC stocks mostly stems from 45 controlled field experiments9–11, which reveal that both management effects on SOC and the 46 influence of SOC on other ecosystem functions are highly context dependent8,12. Meta-analyses of 47 dozens of controlled experiments allow to isolate the influence of pedoclimatic variability, and 48 have shown that biochar applications, organic fertilizers, perennial crops, and agroforestry have 49 positive effects on SOC in a large majority of cases, while tillage intensity, crop rotation, and 50 mineral fertilizer application have smaller or more ambivalent effects1,13,14. While these provide 51 important scientific insights, meta-analyses of controlled experiments do not capture the complex 52 interactions and variability of farming practices as observed in real farms across different pedo53 climatic conditions12. This because management practices on controlled experiments differ from 54 on-farm reality, as single practices are emphasized for testing, ignoring that these practices often 55 come in bundles15. A recent synthesis of management practices (tillage, crop rotation and 56 fertilization) identified more than 285,000 possible management bundles16, the application of these 57 bundles is then further multiplied by the differing pedo-climatic conditions when considering the 58 outcomes in terms of SOC stock and stock change17. 59 Opportunities for large-scale assessments of SOC stocks have arisen through the increase in 60 national and continental soil monitoring18–20. For example, analysis of changes in SOC stocks 61 between 2009 and 2018 from the European soil monitoring network LUCAS Soil, showed small 62 increases in SOC stocks, especially in grassland soils21. However, in this and other large-scale 63 SOC assessments22–24, agricultural management was not considered, therefore while we can 64 monitor the change in SOC stocks, we do not understand the role of agricultural practices. Clear 65 understanding of how management impacts SOC stocks, and how that varies with pedoclimatic 66 context, is essential for science-informed policy, e.g. defining sustainable management practices25. 67 Thus there is an urgent need for empirical evidence on the impact of real-world management 68 practices on SOC stocks to inform policy and implementation efforts26. 69 Assessments of management impacts on SOC under real farm conditions remain rare, primarily 70 due to two major challenges: limited availability of farm management data and the high complexity 71 of covarying factors (pedo-climatic conditions, social influence, economic lock-ins etc.). The first 72 challenge can be addressed by combining on-farm soil sampling with farmer interviews to 73 document management practices27–29. The second challenge can be mitigated either by limiting 74 geographic scope to constrain variability, sometimes including a paired-site approach where 75 controls are adjacent to the studied practice5,12,30,31, or by focusing on specific cropping systems 76 across broad pedoclimatic gradients17. However, both of these challenges demand labor intensive 77 approaches which restrict sample size and representativeness of earlier studies. 78 To address this knowledge gap, we combined standardized European farm surveys (Farm 79 Accountancy Data Network (FADN), n = 81,688) with large scale soil monitoring data to 80 disentangle the effect of farm management practices on SOC across all 27 European Union 81 member states plus the United Kingdom (Supplementary Figure 1). We first used FADN to predict 82 likely management practices at all agricultural sampling locations of the LUCAS Soil monitoring 83 program (n = 8,834) 32. We then analyzed relationships between management and three SOC 84 metrics: 1) SOC stocks, 2) deviations from benchmarked SOC stocks, and 3) the change in SOC 85 concentration from 2009 to 2018—while taking variability in soil typology and climate into 86 account. We hypothesized that: 87 1. Permanent crops, grasslands, and extensive crops (e.g., temporary grasslands, forage crops) 88 are associated with more positive SOC outcomes than intensive crops (e.g., potatoes, sugar 89 beet or cereals). 90 2. Higher management intensity negatively correlates with SOC outcomes. 91 3. The share of manure in the fertilizer regime, crop rotational diversity, share of leys in the 92 crop rotation, and organic cultivation are positively associated with SOC outcomes, while 93 tillage intensity is negatively associated. 94 4. Pedo-climatic conditions significantly influence the relationship between management and 95 SOC stocks. 96 Results 97 Variability in agricultural management across Europe 98 Analysis of farm management data from 81,688 farms revealed that agricultural management 99 practices exhibited significant variability across Europe, both between and within regions. For 100 fertilizer regimes, total N and K inputs showed a similar spatial pattern, with highest average 101 application rates in NW Europe, while total P inputs were highest in northern Italy, NW and NE 102 Europe (Supplementary Fig 2a-c). The share of N, P, and K from manure followed a similar spatial 103 pattern for all three nutrients, with highest average values in alpine regions and coastal areas of 104 northern Spain (Supplementary Fig 2d-f). However, the proportion of K derived from manure was 105 consistently higher than that of N or P. Organic farming was most prevalent in marginal regions, 106 i.e. with climatic constraints such as in the very north and south or in mountainous regions 107 (Supplementary Fig. 2g). Crop rotational diversity was highest in eastern Germany and Czech 108 Republic (Supplementary Fig. 2h), areas known for large farm sizes. The share of ley and fodder 109 crops in the crop rotation was highest in Sweden, SE France, Portugal, and Italy (Supplementary 110 Fig. 2i). Tillage intensity was generally high, indicating the prevalence of conventional tillage, 111 with lowest values in eastern half of Germany, parts of the UK and France, and southern Portugal 112 (Supplementary Fig. 2j). 113 While spatial patterns in farm management across Europe have been described previously33–36, our 114 study is novel in that the use of individual farm data enabled us to describe cropand altitude115 specific management practices for each region. Andalusia, the southernmost region of Spain, 116 provides an example to illustrate cropand altitude‑specific differences in management practices 117 identified from the farm surveys. While the Andalusian average N input level is 73.5 kg N ha-1 y118 1, our approach uncovered that reported application rates range from an average of 0 for nuts to 119 220 kg N ha-1 for strawberries, with also differences within a crop depending on altitude 120 (Supplementary Fig. 3). Similar intra-regional variability was observed for all other management 121 variables, with organic farming providing another striking example (Supplementary Fig. 3). In 122 Andalusia, only 6% of farms growing wheat below 300 m altitude (n = 191) were organic, whereas 123 67% of those above 600 m (n = 89) practiced organic farming. These crop-region-, and altitude124 specific management practices were used to predict the most likely management at each LUCAS 125 soil sampling location. 126 Of the 257 regions analyzed, our approach identified significant differences in agricultural 127 management in most regions, with 93% of regions showing significant crop and altitude specific 128 differences in N input, 92% in P input, and 95% in K input. Significant differences were further 129 observed in 98%, 97% and 97%, respectively of regions for the share of N, P, and K derived from 130 manure; 88% for organic farming, 97% for crop rotational diversity, 95% for the share of ley, 80% 131 for the share of forage, and 89% for tillage intensity. 132 133 Relationship between land cover and soil organic carbon 134 SOC stocks (χ² = 1608, p < 10-15), benchmarked SOC stocks (χ² = 184, p < 10-15), and the change 135 in SOC concentration between 2009 and 2018 (χ² = 143, p < 10-15) were all significantly dependent 136 on land cover. SOC stocks were highest in grasslands, followed by arable crops with high soil 137 cover such as temporary grasslands and fodder crops (Fig. 1a). SOC stocks were calculated using 138 a mass-corrected fixed depth approach (see method) taking into account variations in concentration 139 of SOC and bulk density. While the concentration of SOC was considerably higher in grasslands 140 than any arable crops, they also presented lower bulk density than arable crops (Supplementary 141 Fig. 4). Generally, soils under arable cultivation, including temporary grasslands, had significantly 142 higher bulk density than grassland or permanent crops (Supplementary Fig. 4). Some permanent 143 crops showed low observed SOC stocks (Fig. 1a and b), as a combination of low bulk density and 144 average to moderate SOC concentration (Supplementary Fig. 4). To account for SOC accumulation 145 potential, observed SOC stocks were benchmarked by dividing by “typical” SOC stock values for 146 that pedo-climatic unit 37. These benchmarked SOC stocks were highest in permanent crops, set 147 aside land and low intensity arable crops such as temporary grasslands and lucerne (Fig. 1b). 148 Lowest benchmarked SOC stocks were found in intensive arable crops such as potato and sugar 149 beet, which are associated with intensive soil disturbance and minimal crop residue return. 150 A majority of sites (55%) showed an increase in SOC concentration from 2009 to 2018, confirming 151 De Rosa et al. 21. Additionally, we found significant differences between individual crops, with 152 most permanent crops and grassland types tending to have lower risk of SOC loss (Fig. 1c). The 153 crops most likely to have SOC losses were potato, rye, oats, and triticale. 154 155 156 Figure 1. Relationship between soil organic C (SOC) metrics and land cover. SOC stocks (a), 157 χ² = 1545, p < 10-15), the benchmarked SOC stocks (b), χ² = 163, p < 10-15), and the change in 158 soil organic C between 2009 and 2018 (c), χ² = 143, p < 10-15) were all significantly dependent 159 on land cover. SOC and land cover data from LUCAS Soil. Only land covers with n > 10 are 160 shown. Total n = 6795, 6795, and 6362 respectively. 161 Management effect on soil organic carbon 162 Mixed linear models with pedoclimatic zone as a random effect showed that management intensity 163 significantly affected all three SOC metrics (Supplementary Table 1). Management intensity was 164 calculated from fertilizer input, share of manure, probability of organic cultivation, tillage intensity 165 (arable only), and share of leys and forage in the crop rotation (arable only). SOC metrics in arable 166 sites showed the strongest response to management intensity (Fig. 2), with a very clear negative 167 effect of intensity on benchmarked SOC. Arable sites with low management intensities showed a 168 benchmarked SOC of 30% above the pedoclimatic typical values, while the most intensive sites 169 were 12% below (Fig. 2b). While management intensity of permanent crops had a slightly negative 170 marginal effect, management intensity of grasslands was positively correlated with both SOC 171 stocks (Fig. 2a) and benchmarked SOC stock (Fig. 2b). The rate of change in SOC showed the 172 same negative correlation with intensity index for all three land cover classes (Fig. 3c). 173 174 Figure 2. Relationship between SOC metrics and management intensity. Marginal effects 175 calculated from linear mixed models with pedoclimatic zones 37 as random effect and intensity and 176 land cover class as fixed effect. The model χ² are 124 (p < 10-15) for SOC stocks (a, n = 6,776), 177 113 (p < 10-15) for benchmarked SOC stocks (b, n = 6,776), and 7.3 (p = 0.007) for the yearly 178 change in SOC concentration (c, n = 6,346). See supplementary table 1 for further model details. 179 Orange = arable, blue = permanent crops, green = grassland, and purple = all sites (no significant 180 interaction effect). Shaded areas show the 95% confidence intervals. The black horizontal line in 181 b) indicates the benchmarked value. 182 Regarding the effect of individual management practices, N input had a negative marginal effect 183 on SOC metrics for arable soils, but a positive effect up to 300 kg N ha-1 for grassland soils (except 184 for the change in SOC concentration which increased linearly with increasing N inputs) (Fig. 3). 185 This observed positive marginal effect of N input on SOC metrics for grasslands explains the 186 positive effect of management intensity overall (Fig. 2). The share of manure in the fertilizer 187 regime had a consistently positive marginal effect on SOC stocks and benchmarked SOC stocks 188 (Fig. 3b, g). The organic farming had consistently positive marginal effect on SOC stocks and 189 benchmarked SOC stocks (Fig. 3c, h), but no significant effect on the change in SOC concentration 190 for arable sites (Fig. 3l). 191 The share of ley and fodder crops in the crop rotation had the strongest positive marginal effect on 192 observed and benchmarked SOC stocks in arable soils (Fig. 3e, j). A slightly negative effect of 193 crop rotational diversity on SOC stocks (Fig. 3d, i) was likely due to rotations with a high share of 194 ley and fodder crops,—which tend to have lower diversity. Tillage intensity was not found to have 195 a significant effect on any SOC metric (Supplementary table 3). 196 197 Figure 3. Relationship between SOC metrics and individual management variables. Marginal 198 effects calculated from mixed linear models with pedoclimatic zones 37 as random effect and 199 management variables as fixed effects. Panels a) to e) show marginal effects on SOC stocks, panels 200 f) to j) on benchmarked SOC stocks, and panels k) and l) on the change in SOC concentration from 201 2009 to 2018. Only management variables that had a significant effect are shown, thus some plots 202 are void. Orange = arable, blue = permanent crops, green = grassland, and purple = all sites (no 203 interaction effect). Shaded areas show the 95% confidence intervals. The black horizontal line in 204 b) indicates the benchmarked value. Crop rotational diversity and the share of ley and fodder crops 205 in the crop rotation only applies to arable crops, thus only orange lines. See supplementary table 206 2 and 3 for model details. 207 Soiland climate-specific management effects 208 To test how management effects compared for different pedoclimatic zones, we calculated the 209 difference in observed SOC stocks between the 10% most optimally managed and 10% least 210 optimally managed fields per soil pedoclimatic zone. Optimal management was defined based on 211 results from the mixed linear models (see Fig. 4 and methods for details), e.g. sites with high values 212 for share of manure in fertilizer mix, organic farming, and share of leys in crop rotation. In arable 213 sites, optimal management consistently had a positive effect on observed SOC stocks compared to 214 suboptimal management, except for Atlantic arable sandy soils, which may be due to the low 215 sample size (n = 118) for this soil type (Fig. 4a). Interestingly, even central arable sandy soils, for 216 which we observed little management variability, had a significant management effect, suggesting 217 that these soils are particularly sensitive to management. The strongest effect was observed on 218 Alpine and boreal soils, where the difference between 10% most optimally managed sites and 10% 219 least optimally managed sites was 41.2 ± 10.2 Mg C ha-1. 220 221 Figure 4. Pedoclimatic zone specific management effect on SOC stock as a function of management 222 variability. A) arable, b) grassland, c) permanent crop. Management effect is defined as the 223 difference between SOC stock observed on the 10% most optimally managed and 10% least 224 optimally managed fields per soil pedoclimatic zone. Optimal management is defined as high 225 probability of organic cultivation, high share of manure, low rotational diversity and high share 226 of leys in the rotation for arable; high probability of organic cultivation, high share of manure, 227 and intermediate levels of N input for grassland and permanent crops (see Fig. 4). Management 228 variability captures how variable management is within the pedoclimatic zone and is calculated 229 as the mean normalized standard deviation per management indicator. Error bars show the 95% 230 confidence interval. See37 for definitions and distributions of pedoclimatic zones. 231 For grassland sites, four out of seven pedoclimatic zones had significant management effects, with 232 strongest effects observed for cold climate semi-natural loamy and clayey soils (29.3 ± 4.8 Mg C 233 ha-1), Mediterranean arable loamy and clayey soils (23.2 ± 8.3 Mg C ha-1), and continental semi234 natural loamy and clayey soils (23.1 ± 3.6 Mg C ha-1) (Fig. 4b). For the other pedoclimatic zones, 235 the combination of practices defined as optimal management for grasslands in this study were not 236 found to correlate with higher observed SOC stocks. For permanent crop sites, a significant 237 management effect of 16.8 ± 7.7 and 15.0 ± 3.3 Mg C ha-1 was observed for Atlantic and Central 238 arable loamy and clayey soils and Mediterranean loamy and clayey soils, respectively (Fig. 5c), 239 with no consistent effects for the other pedoclimatic zone. 240 Discussion 241 Management impacts on SOC 242 While current understanding of how different agricultural practices impact SOC is largely based 243 on analyses of controlled field experiments1,9,10, our study is the first to leverage large-scale, real244 farm data to account for the complexity of farming practices and their interactions with diverse 245 soils and climates. Other studies using on-farm data at smaller spatial scales have also found 246 negative correlations between management intensity and SOC in arable soils, e.g. in the 247 Netherlands5 or Southern Sweden30, which we show at the continental scale. The negative effect 248 of management intensity highlights the combined effect of various management practices that may 249 deteriorate SOC stocks. 250 A key advantage of our approach is that both the soil and farm management datasets used here was 251 designed to be representative18,32, thus capturing the actual variability of soils and management 252 practices at continental scale—an aspect that has been largely absent from previous studies5,17. The 253 outcomes of our research thus provide crucial empirical evidence to guide effective policy design 254 and practical implementation. For example, our analysis shows that the positive correlation of SOC 255 with organic farming14, share of manure in the fertilizer mix1,10, high crop cover such as through 256 temporary grasslands and other perennial crops9,13, and no effect of tillage intensity1,9 previously 257 shown for controlled field experiments also holds for real-farms across Europe. While controlled 258 experiments have shown that crop rotation positively impacts soil carbon storage relative to 259 monoculture1,38,39, our study shows that, in practice, the composition of the crop rotation (e.g. high 260 share of leys or forage crops) is more important than diversity of crops per se. 261 Even less is currently known about management impacts on SOC in grasslands and permanent 262 crops9. Our results show that permanent crops were associated with higher benchmarked SOC 263 stocks compared to arable systems and were more likely to have positive ΔSOC over time (Fig. 264 2), supporting the potential contribution of tree crop systems for increasing SOC stocks1,40. In 265 terms of sustainable management of grasslands, the positive correlation of fertilizer amount with 266 SOC (Fig. 3) adds important nuance to sustainability trade-offs and context-dependency of 267 fertilizer use, confirming similar patterns observed at the regional scale on-farm surveys in 268 England41. 269 While ΔSOC is potentially most interesting from a monitoring perspective, the noise to signal ratio 270 for ΔSOC is particularly high, which is a challenge faced by all SOC monitoring studies. This 271 7. Don, A. et al. Carbon sequestration in soils and climate change mitigation—Definitions and 499 pitfalls. Global Change Biology 30, e16983 (2024). 500 8. Moinet, G. Y. K., Hijbeek, R., van Vuuren, D. P. & Giller, K. E. Carbon for soils, not soils for 501 carbon. Global Change Biology 29, 2384–2398 (2023). 502 9. Wiesmeier, M. et al. Soil organic carbon storage as a key function of soils - A review of drivers 503 and indicators at various scales. Geoderma 333, 149–162 (2019). 504 10. López i Losada, R. et al. Synergistic effects of multiple “good agricultural practices” for 505 promoting organic carbon in soils: A systematic review of long-term experiments. Ambio 506 (2025) doi:10.1007/s13280-025-02188-8. 507 11. Tang, S. et al. Soil carbon sequestration enhanced by long-term nitrogen and phosphorus 508 fertilization. Nat. Geosci. 1–9 (2025) doi:10.1038/s41561-025-01789-y. 509 12. Schepens, J. A. B. et al. From global trends to local realities: SOC accrual of improved 510 management practices in northwestern Europe. Geoderma Regional 42, e00983 (2025). 511 13. Poeplau, C. & Don, A. Carbon sequestration in agricultural soils via cultivation of cover crops 512 – A meta-analysis. Agriculture, Ecosystems & Environment 200, 33–41 (2015). 513 14. Bai, Z. et al. Effects of agricultural management practices on soil quality: A review of long514 term experiments for Europe and China. Agriculture, Ecosystems and Environment 265, 1–7 515 (2018). 516 15. Debernardini, M., Candel, J. & Schulte, R. P. O. From the ground up: exploring European 517 carbon farming through social practice theory. Journal of Rural Studies 120, 103850 (2025). 518 16. Rillig, M. C. & Lehmann, A. Exploring the agricultural parameter space for crop yield and 519 sustainability. The New Phytologist 223, 517–519 (2019). 520 17. Edlinger, A. et al. The impact of agricultural management on soil aggregation and carbon 521 storage is regulated by climatic thresholds across a 3000 km European gradient. Global 522 Change Biology 29, 3177–3192 (2023). 523 18. Orgiazzi, A., Ballabio, C., Panagos, P., Jones, A. & Fernández-Ugalde, O. LUCAS Soil, the 524 largest expandable soil dataset for Europe: a review. European Journal of Soil Science 69, 525 140–153 (2018). 526 19. Pan, G., Xu, X., Smith, P., Pan, W. & Lal, R. An increase in topsoil SOC stock of China’s 527 croplands between 1985 and 2006 revealed by soil monitoring. Agriculture, Ecosystems & 528 Environment 136, 133–138 (2010). 529 20. Román Dobarco, M. et al. Mapping soil organic carbon fractions for Australia, their stocks, 530 and uncertainty. Biogeosciences 20, 1559–1586 (2023). 531 21. De Rosa, D. et al. Soil organic carbon stocks in European croplands and grasslands: How much 532 have we lost in the past decade? Global Change Biology 30, e16992 (2024). 533 22. Levy, P. et al. The effects of land use on soil carbon stocks in the UK. Biogeosciences 21, 534 4301–4315 (2024). 535 23. Wadoux, A. M. J.-C. et al. Baseline high-resolution maps of organic carbon content in 536 Australian soils. Sci Data 10, 181 (2023). 537 24. Xie, Z. et al. Soil organic carbon stocks in China and changes from 1980s to 2000s. Global 538 Change Biology 13, 1989–2007 (2007). 539 25. Alblas, E. et al. Bridging Law and Soil Science to Promote Soil Health. European Journal of 540 Soil Science 76, e70127 (2025). 541 26. Lacoste, M. et al. On-Farm Experimentation to transform global agriculture. Nature Food 3, 542 11–18 (2021). 543 27. Edlinger, A. et al. Compost Application Enhances Soil Health and Maintains Crop Yield: 544 Insights From 56 Farmer-Managed Arable Fields. Journal of Sustainable Agriculture and 545 Environment 4, e70041 (2025). 546 28. Garland, G. et al. Crop cover is more important than rotational diversity for soil 547 multifunctionality and cereal yields in European cropping systems. Nat Food 2, 28–37 (2021). 548 29. Vahter, T. et al. Landscapes, management practices and their interactions shape soil fungal 549 diversity in arable fields – Evidence from a nationwide farmers’ network. Soil Biology and 550 Biochemistry 168, 108652 (2022). 551 30. Williams, H., Colombi, T. & Keller, T. The influence of soil management on soil health: An 552 on-farm study in southern Sweden. Geoderma 360, 114010 (2020). 553 31. Mudge, P. L. et al. Irrigating grazed pasture decreases soil carbon and nitrogen stocks. Global 554 Change Biology 23, 945–954 (2017). 555 32. European Commission. Farm accountancy data network. (2022). 556 33. Ballot, R., Guilpart, N. & Jeuffroy, M.-H. The first map of crop sequence types in Europe over 557 2012–2018. Earth System Science Data 15, 5651–5666 (2023). 558 34. Debonne, N. et al. The geography of megatrends affecting European agriculture. Global 559 Environmental Change 75, (2022). 560 35. Panagos, P. et al. Improving the phosphorus budget of European agricultural soils. Science of 561 The Total Environment 853, 158706 (2022). 562 36. Helfenstein, J. et al. Divergent agricultural development pathways across farm and landscape 563 scales in Europe: Implications for sustainability and farmer satisfaction. Global Environmental 564 Change 86, 102855 (2024). 565 37. Feeney, C. J. et al. Benchmarking soil organic carbon (SOC) concentration provides more 566 robust soil health assessment than the SOC/clay ratio at European scale. Science of The Total 567 Environment 951, 175642 (2024). 568 38. McDaniel, M. D., Tiemann, L. K. & Grandy, A. S. Does agricultural crop diversity enhance 569 soil microbial biomass and organic matter dynamics? A meta-analysis. Ecological 570 Applications 24, 560–570 (2014). 571 39. Zheng, F. et al. Positive effects of crop rotation on soil aggregation and associated organic 572 carbon are mainly controlled by climate and initial soil carbon content: A meta-analysis. 573 Agriculture, Ecosystems & Environment 355, 108600 (2023). 574 40. Kay, S. et al. Agroforestry creates carbon sinks whilst enhancing the environment in 575 agricultural landscapes in Europe. Land Use Policy 83, 581–593 (2019). 576 41. Ward, S. E. et al. Legacy effects of grassland management on soil carbon to depth. Global 577 Change Biology 22, 2929–2938 (2016). 578 42. van Leeuwen, C. C. E., Mulder, V. L., Batjes, N. H. & Heuvelink, G. B. M. Statistical 579 modelling of measurement error in wet chemistry soil data. European Journal of Soil Science 580 73, e13137 (2022). 581 43. Breure, T. S. et al. Revisiting the soil carbon saturation concept to inform a risk index in 582 European agricultural soils. Nat Commun 16, 2538 (2025). 583 44. The State Council of the People’s Republic of China. China’s third national soil survey makes 584 significant progress. (2024). 585 45. Reddy, A. A. The Soil Health Card Scheme in India: Lessons Learned and Challenges for 586 Replication in Other Developing Countries. Journal of Natural Resources Policy Research 9, 587 124–156 (2019). 588 46. Home | Soils 4 Africa - Horizon 2020 programme of the European Union. 589 https://www.soils4africa-h2020.eu/ (2025). 590 47. Australian Government - DAFF. National Soil Monitoring Program. 591 https://www.agriculture.gov.au/agriculture-land/farm-food-drought/natural592 resources/soils/national-soil-monitoring-program (2025). 593 48. Wadoux, A. M. J.-C. et al. On soil districts. Geoderma 452, 117065 (2024). 594 49. Georgiou, K. et al. Soil Carbon Saturation: What Do We Really Know? Global Change 595 Biology 31, e70197 (2025). 596 50. European Commission. Glossary:Farm typology. https://ec.europa.eu/eurostat/statistics597 explained/index.php?title=Glossary:Farm_typology (2025). 598 51. Westbury, D. B., Park, J. R., Mauchline, A. L., Crane, R. T. & Mortimer, S. R. Assessing the 599 environmental performance of English arable and livestock holdings using data from the Farm 600 Accountancy Data Network (FADN). Journal of Environmental Management 92, 902–909 601 (2011). 602 52. Cardillo, C., Di Fonzo, A. & Liberati, C. The Farm’s Orientation towards Sustainability: An 603 Assessment Using FADN Data in Italy. Land 12, 301 (2023). 604 53. Dabkiene, V., Balezentis, T. & Streimikiene, D. Development of agri-environmental footprint 605 indicator using the FADN data: Tracking development of sustainable agricultural development 606 in Eastern Europe. Sustainable Production and Consumption 27, 2121–2133 (2021). 607 54. Statistics Netherlands. Standardised Calculation Methods for Animal Manure and Nutrients: 608 Standard Data 1990-2008. (2012). 609 55. Holland, J. M. The environmental consequences of adopting conservation tillage in Europe: 610 reviewing the evidence. Agriculture, Ecosystems & Environment 103, 1–25 (2004). 611 56. Lüdecke, D. sjstats: Statistical Functions for Regression Models. [object Object] 612 https://doi.org/10.5281/ZENODO.1284472 (2022). 613 57. Hollander, M. & Wolfe, D. A. Nonparametric Statistical Methods. (John Wiley & Sons, Ltd, 614 New York, 1973). 615 58. Fowler, A. F., Basso, B., Millar, N. & Brinton, W. F. A simple soil mass correction for a more 616 accurate determination of soil carbon stock changes. Sci Rep 13, 2242 (2023). 617 59. Wright, M. N. & Ziegler, A. ranger: A Fast Implementation of Random Forests for High 618 Dimensional Data in C++ and R. Journal of Statistical Software 77, 1–17 (2017). 619 60. Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest Package: Tests in Linear 620 Mixed Effects Models. Journal of Statistical Software 82, 1–26 (2017). 621 61. Lüdecke, D. ggeffects: Tidy Data Frames of Marginal Effects from Regression Models. 622 Journal of Open Source Software 3, 772 (2018). 623 62. Helfenstein, J. et al. Farmer surveys in Europe suggest that specialized, intensive farms were 624 more likely to perceive negative impacts from COVID-19. Agronomy for Sustainable 625 Development 42, 84 (2022). 626 627 628 Supplementary info for: 629 Continental-scale evidence of farm management impacts on 630 soil carbon 631 632 Julian Helfenstein1, Nick van Dijk1, Anna Edlinger2, Gabriel Y.K. Moinet3, Sophie van Rijssel1, 633 Alexandre M.J.-C. Wadoux4, Rachel Creamer3, Carmen Vazquez3, Vera L. Mulder1 634 1 Soil Geography and Landscape Group, Wageningen University & Research, Wageningen, the 635 Netherlands 636 2 Wageningen Environmental Research, Wageningen University & Research, Wageningen, the 637 Netherlands 638 3 Soil Biology Group, Wageningen University & Research, Wageningen, the Netherlands 639 4 College of Science and Engineering, James Cook University, Cairns, Australia 640 641 642 Supplementary figures 643 644 Figure 5. Spatial coverage of farm management and soil property datasets. Data from the Farm 645 Accountancy Data Network (FADN, n = 81,688 for the year 2018) was used to predict farm 646 management and (a) shows the number of farm records per NUTS2 administrative region for 2018. 647 The European soil monitoring network (LUCAS Soil) covers around 20,000 sample locations, of 648 which 8,834 were on agricultural land and met our inclusion criteria (b). 649 650 Figure 6. Spatial variability in farm management based on 81,688 representative individual farm 651 observations from 2018. Maps show the area-weighted mean value of a farm management 652 indicator per NUTS2 administrative region. a) – c) total N, P, K input in kg ha-1, including both 653 mineral fertilizer and manure; d) – f) the share of N, P, K respectively derived from manure; g) is 654 the prevalence of organic farming (% of area); h) rotational diversity measured as the Gini655 Simpson index –score close to one means high diversity; i) share of ley and fodder legumes in the 656 crop mix; j) tillage intensity, see methods for details. See Supplementary Fig. 1 for the number of 657 farms (sample size) in each NUTS2 region. 658 659 660