Comment https://doi.org/10.1038/s41467-023-37907-w Quantifying the recarbonization of post-agricultural landscapes Stephen M. Bell, Samuel J. Raymond, He Yin, Wenzhe Jiao, Daniel S. Goll, Philippe Ciais, Elsa Olivetti, Victor O. Leshyk & César Terrer Check for updates Despite worldwide prevalence, post-agricultural Challenges in conceiving PALs as carbon sinks The first challenge with PALs is that they are diverse and hetero- landscapes remain one of the least constrained geneous; categorizing them as a single land cover class is not easy. human-induced land carbon sinks. To appraise They include several overlapping terms, some used interchangeably their role in rebuilding the planet’s natural car- (e.g., old fields, idle croplands, set-aside), and others used only in specific cultural contexts (e.g., post-agrogenic lands in the former- bon stocks through ecosystem restoration, we Soviet states). Some PALs also have negative connotations, creating 1234567890():,; 1234567890():,; need to better understand their spatial and difficulties for policy-makers, local stakeholders, and landowners to agree on how to manage them. temporal legacies. Abandoned agricultural lands are probably the most well-known PAL, but “abandonment” can mean different things. In many cases, it is In provisioning human civilization with food, fuel, and fiber for perceived positively as an opportunity for ecosystem regeneration and millennia, agriculture has drastically depleted terrestrial carbon renaturalization, while for others it signals local economic depression, stocks at the expense of natural ecosystems. Our challenge today the loss of traditional rural livelihoods and biodiverse agro-landscapes, is to use more sustainable practices to recapture some of the 116 and an increased risk of wildfires and pests when the land is left Pg of soil organic carbon (SOC) lost since agriculture began, while unmanaged9. Complicating things further, there is still no universal simultaneously ensuring global food security1,2. That being said, definition of abandonment with an agreed minimum time frame. The the cessation of agriculture altogether is still the most efficient Food and Agriculture Organization considers agricultural land to be way to increase carbon stocks and restore ecosystems in tandem abandoned after five consecutive years of disuse, but this rule is not and at large scales. universally applied. Consider the vast expanses of forests that regrew over the 60 Another challenge is that any sequestration and ecological co- Mha of cropland abandoned following the collapse of the Soviet benefits PALs may produce can be rapidly reversed through reculti- Union3. It has been called the world’s largest human-made carbon vation. Of course, many PALs are not permanent by definition and were sink attributed to a single event;4 a title challenged by the climatic never meant to be, such as lands under fallow or shifting agriculture. consequences of the ‘Great Dying in the Americas’ and its 56 Mha But others are simply unprotected: abandoned agricultural lands are at abandoned following the arrival of Europeans5. At more practical constant risk of recultivation in their first few decades and often before scales, intentional efforts to restore agricultural land such as the significant ecological benefits are realized10. Grain-for-Green program in China and the Conservation Reserve These issues notwithstanding, arguably the biggest roadblock in Program in the USA have demonstrated that carbon sequestration is protecting, monitoring, and utilizing PALs as carbon sinks is finding far from being the only advantage6,7. Ecological co-benefits include them. While agricultural lands intentionally converted to pre- reduced soil erosion and water run-off, reduced flooding and disturbance states (i.e., cropland restoration) likely entail an adminis- drought, and improved soil health, water quality, and biodiversity trative record that can be used to validate and update spatial esti- indicators. mates, other kinds of PALs emerge unannounced all over the world These post-agricultural landscapes (PALs) often signify the return every day. Agricultural land abandonment often goes undeclared of ecosystem properties, such as carbon, towards pre-disturbance unless incentives are present and governmental bodies are involved. states or new equilibria through secondary succession. Whether Insufficient data on the different drivers of agricultural cessation— planned or unplanned, they appear in every agricultural region of the which are numerous11—further hinders efforts to map past, present, world and they can drawdown carbon with or without human invol- and predicted PALs. Even if we do map them well, we still need to figure vement. If commitments to halt gross forest area loss by 2030 succeed, out long they have existed to be able to evaluate their ecological recarbonizing PALs will play a key role in reversing global land use impacts accurately. Remote sensing produces our best estimates of change from being a net carbon source to a net sink8. PAL distributions at large geographic scales, but only for the past few Unfortunately, PALs are insufficiently represented in terrestrial decades and with high uncertainties depending on the type of PAL carbon models, both spatially (as a poorly mapped land cover class) investigated and the methods used. and temporally (as uncertain carbon sinks). This hinders our Finally, the rates of carbon sequestration on PALs are also largely ability to monitor, quantify, and leverage them strategically. We unresolved. We know multiple factors modulate SOC and biomass discuss here some of the reasons behind these issues and what can accumulation following agricultural cessation, such as the past and be done to address them so that we can properly evaluate the role present management practices and their legacies (e.g., historical fer- of PALs. tilization rates, crop types cultivated, etc.), soil and vegetation nature communications (2023)14:2139 | 1 Comment properties, climate, and, of course, time. However, we have yet to pastures and natural grasslands with current remote sensing methods. unravel how different combinations of these factors might explain the Agricultural cessation in small fields (i.e., < 0.1 ha) are often beyond the diversity of positive, negative, or negligible SOC changes reported on detection limits of open-access imagery such as NASA/USGS’s Landsat PALs globally. or MODIS archives. Better data, and better use of existing data, is desperately needed Fortunately, our ability to detect and classify PALs is improving. to clarify underlying relationships, model PAL impacts in all agri- Higher spatial, temporal, and spectral resolution remote sensing is cultural regions of the world, and ultimately predict potential carbon now available with the denser time series provided by the ESA’s sinks and sources through time. Unfortunately, the gold standard of Sentinel-1 and Sentinel-2, together with Landsat data. There are also repeated measurements and long-term experimental field sites not increasing options for very high-resolution (e.g., < 5 m) commercial only takes too long, but is logistically and financially prohibitive for satellite imagery that can be used to isolate and process smaller agri- many researchers. We remain severely limited in our ability to promote cultural plots and distinguish between different agricultural manage- PALs as components of sustainable land management strategies. ment practices (e.g., PlanetScope)13. Aside from satellites, platforms like IIASA’s Geo-Wiki (www.geo-wiki.org) demonstrate the potential of Constraining the carbon sink potential of PALs crowd-sourced ground truthing to supplement visual interpretations Since there is still no consensus on the global extent of PALs, the first in training land cover classification models. step to constrain their carbon sink potential is to produce reliable, up- The large volume of data generated by new sensors can also be to-date, spatial estimates of past, present, and predicted instances of processed more rapidly and with more accessibility than ever before agricultural cessation. Even successful mapping approaches struggle using free cloud computing resources like Google Earth Engine14. in some situations because PALs can involve the entire spectrum of Combining these approaches with advances in machine learning agricultural land types, plot sizes, and management practices12. For algorithms being continuously developed with free, open-source example, it is extremely difficult to distinguish between abandoned software libraries has already produced new land use and land cover Fig. 1 | Typical post-agricultural landscape (PAL) trajectory involving land chronosequence method) PALs that share the same environmental and manage- clearing from the primary forest (natural control) for agricultural use (agri- ment characteristics except for the time since agricultural cessation. When sampled cultural control) and ecological succession following agricultural cessation simultaneously and analyzed sequentially, these plots can reveal the temporal (shown here as early-, middle-, and late-stage PALs). The rebuilding of ecosys- dynamics of ecosystem carbon. Active agricultural plots represent the expected tem carbon stocks in PALs is represented by the increasingly darkening top soils baseline of all the plots before agricultural cessation, while natural control plots can and growing plant biomass. However, the ecological and structural legacies of land be used to represent pre-disturbance states and, therefore, idealized PAL end- clearing, tilling, and other destructive practices have lasting impacts on above- and states. Not only are the logistical barriers to paired-plots and chronosequences below-ground properties, preventing a rapid return to pre-disturbance states. much lower than repeated measurements, and so they should be established widely Instead of sampling the same PAL over time (i.e., repeated measurements), a to fill data gaps, but existing published data using these approaches are also a quicker and less expensive alternative is to compare samples from an active agri- largely untapped resource for benchmarking successional carbon modeling. cultural plot and one (i.e., with the paired-plot method) or more (i.e., with the nature communications (2023)14:2139 | 2 Comment critical data gaps (i.e., South America, Africa, and most of Asia based on our global survey of > 3400 published SOC data pairs from PALs). Investing new time, money, and energy may not even be necessary for agricultural areas that are already densely sampled. Previously published paired-plots and chronosequences studies of PALs featuring carbon data are numerous, readily available, and valuable sources of data that can complement well-known, widely used datasets of long- term repeated measurements. They represent past investments in sci- ence that should be repurposed, reused, and archived before being lost19. Valuable secondary soil physicochemical properties (e.g., nitro- gen, phosphorus, particulate and mineral-associated organic matter, etc.) for modeling drivers of SOC dynamics are also often available within the same study. Collecting this information can improve our inferential capabilities and reduce geographic and statistical biases. By extracting and synthesizing disparate time-stamped SOC and auxiliary data from published studies involving paired-plots and chronosequences of PALs, in addition to sampling new sites in underrepresented regions, we can benchmark and validate models of successional carbon dynamics. Widely referenced global syntheses that include categories analogous to recarbonizing PALs are outdated and severely data-poor20,21. More recent regional studies still lack suf- ficient temporal resolution, site-specific parameters, and management information to allow for upscaling. A robust, updated global synthesis exploring recarbonizing PALs is overdue. Fig. 2 | The suitability of an agricultural site will determine if and how it should be managed as a recarbonizing post-agricultural landscape (PAL). New and Leveraging PALs as carbon sinks existing PALs can have a high or low potential for soil and ecosystem carbon (C) Integrating better maps and bigger time-stamped datasets with new sequestration and high or low feasibility when considering various competing approaches in artificial intelligence-driven modeling will allow us to needs (e.g., food production), trade-offs (e.g., wildfire risk), and conflicts (e.g., local investigate the carbon cycle impacts of PALs with more accuracy than land rights). Depending on the carbon data available, the definitions used, and the ever before. Degraded and marginal agricultural lands already repre- land use factors considered, possible land management trajectories include (1) sent some of the best candidates for ecosystem restoration globally. implementing or (2) preventing new PALs and (3) protecting or (4) restoring or These research efforts will provide critical and timely insights for converting existing PALs into more appropriate land uses. Increased temporal data international reforestation and carbon sequestration initiatives. and improved spatial estimates of PALs will support decision-makers in navigating Whenever and wherever PALs do recarbonize, regardless of their scenarios like these. This graphic was created with BioRender.com. carbon sink longevity and permanence, they must be made known and accounted for ref. 22. Of course, not all PALs can be expected to products with unprecedented spatiotemporal resolutions (e.g., recarbonize naturally. Those that are degraded beyond spontaneous Dynamic World)15. Although significant challenges still exist in recovery will need a helping hand through assisted restoration prac- upscaling ecological properties like SOC globally16, we should antici- tices like native tree planting. Afterall, mitigating the negative impacts pate the first reliable maps of historical and existing PALs (and their of climate change by returning carbon to depleted soils will require the durations of existence) and prepare for their integration into terres- use of multiple avenues simultaneously23. trial carbon cycle research. Ultimately, the identification of suitable locations for recar- Therefore, simultaneous with mapping efforts, we also need to bonizing PALs will enable decision-makers to weigh trade-offs and address the lack of temporal SOC data on PALs. Time-stamped data consider different land management trajectories (Fig. 2). Opti- points are crucial for understanding what factors determine if and mized promotion of PALs as carbon sinks might entail incenti- when a given PAL will act as a carbon sink or source following agri- vizing landowners and managers in suitable locations to maintain cultural cessation. Currently, global biogeochemical models are lim- existing PALs or create new PALs through agricultural cessation. ited by relatively poor temporal data (i.e., low quality and quantity) due Payment for predicted and validated carbon sequestration, as one to the logistical and financial challenges of long-term field sites with example, would help deter farmers from seeking alternative repeated measurements, especially in under-resourced regions17. economic pathways for their land that would result in lost Thankfully, space-for-time substitutions like paired-plots and opportunities for climate change mitigation. chronosequences are accessible, inexpensive, and rapid alternatives Future discussions at the international level on promoting PALs as that can help fill the many geographical gaps in temporal ecosystem carbon sinks must carefully account for potential conflicts with food data on PALs (Fig. 1). Although they have notable limitations and production, biodiversity, rural socioeconomic needs, and other sus- potential sources of error (e.g., inappropriate control site selection), tainable development goals. Land tenure can be particularly unclear in they are logistically sound and generally informative for exploring regions with PALs24; ensuring fair and equal land rights considering broad ecological processes in lieu of long-term repeated broad categories of stakeholders is a significant challenge. But most measurements18. Chronosequence and paired-plot sampling in importantly, any agricultural land expansion as a result of PAL pro- underrepresented PAL regions should be especially prioritized to fill motion should be monitored and strictly prevented. It would risk nature communications (2023)14:2139 | 3 Comment negating any benefits realized and become another example of carbon 9. Queiroz, C., Beilin, R., Folke, C. & Lindborg, R. Farmland abandonment: threat or opportunity leakage if payments for ecosystem services are involved. for biodiversity conservation? A global review. Front. Ecol. Environ. https://doi.org/10.1890/ 120348 (2014). Uncontested PALs would be ideal candidates for conservation25. 10. Crawford, C. L., Yin, H., Radeloff, V. C. & Wilcove, D. S. Rural land abandonment is too For contested PALs, it is the responsibility of the research community ephemeral to provide major benefits for biodiversity and climate. Sci. Adv. 8, eabm8999 (2022). to provide accurate and reliable carbon sink data for decision-makers 11. Subedi, Y. R., Kristiansen, P. & Cacho, O. Drivers and consequences of agricultural land to weigh amongst all other competing priorities. Without this infor- abandonment and its reutilisation pathways: a systematic review. Environ. Dev. 42, mation, PALs will continue to be just another uncertain and ultimately 100681 (2022). 12. Yin, H. et al. Monitoring cropland abandonment with Landsat time series. Remote Sens. misused piece of the sustainable land management puzzle. Environ. 246, 111873 (2020). It is unreasonable that in many agricultural regions we still cannot 13. Rufin, P., Bey, A., Picoli, M. & Meyfroidt, P. Large-area mapping of active cropland and short- directly compare the potential costs, benefits, and synergies of sus- term fallows in smallholder landscapes using PlanetScope data. Int. J. Appl. Earth Obs. Geoinf. 112, 102937 (2022). tainable practices (e.g., regenerative agriculture) with the cessation of 14. Gorelick, N. et al. Google Earth Engine: planetary-scale geospatial analysis for everyone. agriculture altogether. Quantifying recarbonizing PALs will help Remote Sens. Environ. 202, 18–27 (2017). advance regional debates on land sharing versus land sparing by filling 15. Brown, C. F. et al. Dynamic World, near real-time global 10 m land use land cover mapping. Sci. Data 9, 251 (2022). critical data gaps on the restoration, and in some cases rewilding, of 16. Meyer, H. & Pebesma, E. Machine learning-based global maps of ecological variables and agricultural lands. the challenge of assessing them. Nat. Commun. 13, 2208 (2022). PALs can diversify the global land sink and provide timely support 17. O’Sullivan, M. et al. Process-oriented analysis of dominant sources of uncertainty in the land carbon sink. Nat. Commun. 13, 4781 (2022). for the UN Decade of Ecosystem Restoration (2021–2030) by fulfilling 18. Walker, L. R., Wardle, D. A., Bardgett, R. D. & Clarkson, B. D. The use of chronosequences in local- to global-level commitments to reverse ecosystem degradation. studies of ecological succession and soil development. J. Ecol. https://doi.org/10.1111/j. But they are in need of an immediate, concerted effort by researchers 1365-2745.2010.01664.x (2010). 19. Vines, T. H. et al. The availability of research data declines rapidly with article age. Curr. Biol. to appraise their role in the terrestrial carbon cycle. PALs are wide- 24, 94–97 (2014). spread, carbon-depleted, and often neglected, despite periodic calls 20. Post, W. M. & Kwon, K. C. Soil carbon sequestration and land‐use change: processes and for conservation and reutilization26,27. Let’s make this the last call. potential. Glob. Chang. Biol. 6, 317–327 (2000). 21. Guo, L. B. & Gifford, R. M. Soil carbon stocks and land use change: a meta analysis. Glob. It’s time to map, model, and manage the world’s best PALs for Chang. Biol. (2002) https://doi.org/10.1046/j.1354-1013.2002.00486.x (2022) carbon sequestration and ecosystem restoration. 22. Matthews, H. D. et al. Temporary nature-based carbon removal can lower peak warming in a well-below 2 °C scenario. Commun. Earth Environ. 3, 65 (2022). 23. Bossio, D. A. et al. The role of soil carbon in natural climate solutions. Nat. Sustain. https:// Stephen M. Bell 1,2,3 , Samuel J. Raymond4, He Yin 5, doi.org/10.1038/s41893-020-0491-z (2020). Wenzhe Jiao1,4, Daniel S. Goll2, Philippe Ciais2, Elsa Olivetti 4,6, 24. Meyfroidt, P. et al. Ten facts about land systems for sustainability. Proc. Natl Acad. Sci. USA Victor O. Leshyk7 & César Terrer 1 119, e2109217118 (2022). 1 25. Xie, Z. et al. Conservation opportunities on uncontested lands. Nat. Sustain. 3, 9–15 (2020). Department of Civil and Environmental Engineering, Massachusetts 26. Poore, J. A. C. Call for conservation: abandoned pasture. Science 351, 132 (2016). Institute of Technology, Cambridge, MA 02139, USA. 2Laboratoire des 27. Schröder, P. et al. Intensify production, transform biomass to energy and novel goods and Sciences du Climat et de l’Environnement, LSCE/IPSL, CEA-CNRS- protect soils in Europe—A vision how to mobilize marginal lands. Sci. Total Environ. 616–617, 1101–1123 (2018). UVSQ, Université Paris-Saclay, 91191 Gif-sur-Yvette, France. 3Institute of Environmental Science and Technology, Universitat Autònoma de Acknowledgments Barcelona, 08193 Bellaterra, Spain. 4MIT Climate and Sustainability S.M.B., D.S.G. and P.C. acknowledge support from the Agence Nationale de la Recherche (ANR) Consortium, Cambridge, MA 02139, USA. 5Department of Geography, grant ANR‐16‐CONV‐0003 (CLAND). S.J.R., W.J. and E.O. acknowledge support from the MIT Climate and Sustainability Consortium (MCSC). Kent State University, 325 S. Lincoln Street, Kent, OH 44242, USA. 6 Department of Materials Science and Engineering, Massachusetts Author contributions Institute of Technology, Cambridge, MA 02139, USA. 7Center for S.M.B., S.J.R., and C.T.: conceptualization; S.M.B.: Writing—original draft; S.M.B., S.J.R., H.Y., W.J., D.S.G., P.C., E.O., and C.T.: Writing—review and editing; S.M.B. and V.O.L.: visualization. Ecosystem Science and Society, Northern Arizona University, Flagstaff, AZ 86011, USA. e-mail: bells590@mit.edu Competing interests The authors declare no competing interests. Received: 28 February 2022; Accepted: 5 April 2023; Additional information Correspondence and requests for materials should be addressed to Stephen M. Bell. Reprints and permissions information is available at http://www.nature.com/reprints References 1. Sanderman, J., Hengl, T. & Fiske, G. J. Soil carbon debt of 12,000 years of human land use. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in pub- Proc. Natl Acad. Sci. USA https://doi.org/10.1073/pnas.1706103114. (2017). lished maps and institutional affiliations. 2. Minasny, B. et al. Soil carbon 4 per mille. Geoderma 292, 59–86 (2017). 3. Schierhorn, F. et al. Large greenhouse gas savings due to changes in the post-Soviet food Open Access This article is licensed under a Creative Commons Attribution 4.0 International systems. Environ. Res. Lett. 14, 065009 (2019). License, which permits use, sharing, adaptation, distribution and reproduction in any medium or 4. Slezak, M. Fall of USSR created world’s largest human-made carbon sink. New Sci. 220, format, as long as you give appropriate credit to the original author(s) and the source, provide a 17 (2013). link to the Creative Commons license, and indicate if changes were made. The images or other 5. Koch, A., Brierley, C., Maslin, M. M. & Lewis, S. L. Earth system impacts of the European third party material in this article are included in the article’s Creative Commons license, unless arrival and Great Dying in the Americas after 1492. Quat. Sci. Rev. 207, 13–36 (2019). indicated otherwise in a credit line to the material. If material is not included in the article’s 6. Hua, F. et al. Opportunities for biodiversity gains under the world’s largest reforestation Creative Commons license and your intended use is not permitted by statutory regulation or programme. Nat. Commun. 7, 12717 (2016). exceeds the permitted use, you will need to obtain permission directly from the copyright 7. Augustine, D., Davidson, A., Dickinson, K. & van Pelt, B. Thinking like a grassland: challenges holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. and opportunities for biodiversity conservation in the Great Plains of North America. Rangel. Ecol. Manag. 78, 281–295 (2021). © The Author(s) 2023 8. Gasser, T., Ciais, P. & Lewis, S. L. How the Glasgow declaration on forests can help keep alive the 1.5 °C target. Proc. Natl Acad. Sci. USA 119, e2200519119 (2022). nature communications (2023)14:2139 | 4