Our previous blog, Regenerative Agriculture: Good for Soil Health, but Limited Potential to Mitigate Climate Change, generated a spirited discussion. Here we provide further elaboration on our conclusions. For information on ways to reduce agriculture’s GHG impact please see our blog on 6 Ways the US Can Curb Climate Change and Grow More Food.

The term “regenerative agriculture” is a broad and not-yet-clearly-defined term. It can refer to a range of practices with both climate and non-climate benefits. Our previous blog post focused specifically on the limited potential to mitigate climate change by removing carbon from the atmosphere and storing it in soils.  Our title could have been clearer about this soil carbon focus. One goal of our post was to shift attention away from soil carbon to other ways regenerative practices can achieve climate benefits. For example, in Africa, practices such as agroforestry offer much promise to boost productivity, increase above-ground vegetative carbon and meet rising food demands without clearing natural forests. Similarly, our report, Creating a Sustainable Food Future, identified improved grazing pastures and silvopastoral systems as critical to meeting climate mitigation goals, especially across the tropics.

Implemented in the right way, preserving the huge, existing reservoirs of vegetative and soil carbon in the world’s remaining forests and woody savannas by boosting productivity on existing agricultural land (a land sparing strategy) is the largest, potential climate mitigation prize of regenerative and other agricultural practices. Realizing these benefits requires implementing practices in ways that boost productivity and then linking those gains to governance and finance to protect natural ecosystems. In short, “produce, protect and prosper” are the most important opportunities for agriculture.

By contrast, most claims for the climate mitigation role of regenerative practices focus on their potential to build soil carbon. As our blog post noted, we do not consider this potential to be large.  Among the comments we received was a critical letter from one group of scientists (Paustian et al. 2020) and a supportive letter from another group (Powlson et al. 2020)1. Here we elaborate our thinking:

First, we address the realistic potential to sequester carbon in soils by examining a range of identified practices. Put simply, many practices that sequester soil carbon at the field level involve taking crops out of production. Given growing demand for crops, such practices are unlikely to be scalable because they create a need to expand cropland elsewhere to make up for the foregone production. And this in turn drives conversion of natural ecosystems and release of carbon stored in vegetation and soils.

Second, we address the best-known, large global estimates of potential soil carbon gains. We find that they tend to have limited documentation and analysis of the feasibility of the massive expansion of practices they are relying on to sequester carbon. They also rely largely on practices that decrease crop production or on practices on working lands that more recent science has shown to be ineffective or less effective at sequestering soil carbon, such as no-till farming.

Third, we elaborate on issues related to the use of manure, implications of yield changes and nitrogen limitations.

Finally, we explain our concerns why overestimating potential soil carbon gains could undermine efforts to advance effective climate mitigation in the agriculture sector.  

We welcome further discussion and hope to convene a meeting of interested parties with various perspectives soon.

I. Examining the Practices

In the simplest language, practices that sequester significant carbon at the field level mainly involve taking crops out of production and therefore cannot be easily scaled, given growing needs for food. All else held equal, efforts that reduce or take land out of agricultural production will require plowing up other land elsewhere to replace the forgone production, which releases carbon, offsetting gains at the field level. And the main practices long thought to sequester carbon that maintain crop and grazing production — namely no-till farming and grazing management — are now known to be far less effective or even ineffective, and often face major adoption challenges. Other practices have promise and warrant continued work, such as cover crops, but their total achievable potential to sequester carbon in soil is currently uncertain.

In the following, we evaluate key practices in more detail:

Practices That Increase Soil Carbon at the Field Level, But Do Not Generate Overall Climate Benefits

Increasing soil carbon at the field level does not necessarily generate a global climate mitigation benefit. Climate benefits depend on a full accounting of all greenhouse gas (GHG) effects, including those related to agricultural expansion to make up for any reduced production. Large claims of climate mitigation benefits through soil carbon typically fail to account for all GHG effects. We call this limitation a “GHG accounting error.” Four categories of soil carbon practices are subject to this error:

  1. Taking agricultural land out of production: The practices that sequester the most carbon on individual fields are those that take those fields out of agricultural production. These practices, including peatland restoration, play an important role in broadly cited papers such as Paustian et al. (2016) and Smith et al. (2008). These papers are also cited in Paustian et al. (2020), the critical comment sent to WRI. This comment also cites the soil carbon sequestration potential of turning marginal cropland into perennial grasses and trees, and into grassed waterways and buffer strips, both of which fall into this category because the “regenerated” land no longer produces food. Other soil carbon papers cited in Paustian et al. (2020) assign a large role to avoided deforestation (Soussana et al. 2019; Bossio 2020), which does not sequester carbon, but avoids its loss.

    We agree on the importance of these measures. But, one, they are not about building soil carbon on working agricultural lands, and therefore are not the focus of our blog post. Two, their climate benefits rely on additional efforts to boost yields or reduce demand. Much of Creating a Sustainable Food Future is about how to stop deforestation and rewet peatlands to avoid roughly 5 billion tons of annual GHG emissions as well as the actions needed to restore forests. But we do not think such practices are usefully grouped with regenerative agriculture because they are about ceasing agriculture rather than improving agriculture itself. In addition, given the projected global need for at least 50% more crops per year by 2050 relative to 2010, removing cropland from production in one field without additional efforts to make up the food deficit requires plowing up land elsewhere to replace the food production. That releases carbon stored in both soils and plants, offsetting soil gains at the original field scale.   

    To reforest lands not just at the field level, but also on a net global basis requires a distinct set of efforts to reduce the need for agricultural land, such as moderating meat demand, reducing food loss and waste, avoiding competition from bioenergy for food crops and land, and boosting crop and pasture yields. Doing so also requires finance and good land use governance. With such efforts, restoration can generate net carbon gains. Co-mingling such comprehensive land preservation and restoration efforts with soil carbon gains on working lands under a common umbrella term of “regenerative agriculture” sets up high expectations for carbon savings — which predominantly come from landscape protection and restoration rather than agricultural practices. But because it is called regenerative agriculture, it focuses action on the smallest contributors to carbon sequestration — namely, practices on working agricultural lands.

  2. Turning cropland into grazing land or perennials: Another practice that sequesters soil carbon at the field level is to convert cropland into grazing land. “Integrating livestock” into crop production by rotating crops and grazing in different years is often presented as a core principle of regenerative agriculture.2 Cropland can also be converted to grazing permanently.3 Two of three papers cited in one response to our blog post (Paustian et al. 2020) focus on converting cropland to grazing land permanently (Ogle et al. 2005; Guo & Gifford 2002), and the other identifies doing so as the most beneficial grazing practice (Conant et al. 2017).

    The opportunity to convert annual cropland into grazing is also limited by the growing need for annual crops. There are places where incorporating a grazing rotation holds promise, such as parts of the United States and Canadian Great Plains, where doing so can break pest cycles and where livestock grazing is still common. But given demands for crops, turning one hectare of crops into grazing will usually require plowing up more grazing or forest land at another place to replace the crops, releasing carbon and likely resulting in no net climate mitigation overall. And even if overall diet changes or yield gains could free up cropland globally, that land could alternatively sequester carbon through reforestation, so its use for grazing would still have a high opportunity cost.

  3. Adding manure and organic amendments: Adding manure or other organic amendments can build soil carbon in a field but that does not mean it increases net soil carbon sequestration overall. Because the world generates only so much manure — and even if temporarily stored, nearly all manure is eventually added to some field somewhere — adding more manure to one field mainly involves adding less to another. Suggestions that composting municipal solid waste could be an alternative source of carbon exaggerate the scale of that resource, which could, at most, add around 11 million tons of carbon to soils in the United States each year, of which only some part would likely remain long-term.4 They also overlook that there are other potential uses for municipal solid waste, such as aviation biofuels, and that the globally agreed Sustainable Development Goals have set a target for cutting in half per capita  food waste by 2030.

  4. Increased fertilization:  In some locations, increasing fertilizer use has led to increased soil carbon (Poulton et al. 2017), which helps to explain soil carbon gains in China. But the pollution and GHG emissions associated with the manufacture and use of that nitrogen can overwhelm any climate mitigation benefits from soil carbon gains (Gao et al. 2018).

    Likewise, better nutrient management or more legumes do not generally help build soil carbon (Soussana et al. 2019). Better nutrient management generally means applying less nitrogen to avoid water pollution and emissions. And the appropriate goal of increasing legumes in most places is to replace fertilizer, not to add more. Practices that limit erosion can be beneficial for soil carbon, but they do so by maintaining existing soil carbon, not building more.

No-till and Grazing Land Management

For practices that do not change the uses of cropland or grazing land, the principal large estimates of soil carbon sequestration have relied on no-till farming and changes in grazing management. Whatever the original justification for these large estimates, recent science has greatly reduced estimates of their potential.

No-till: As reflected in Ogle et al. (2005), cited by the Paustian et al. (2020) critique of our blog post, science once supported substantial soil carbon sequestration claims for no-till farming, but scientific understanding of no-till has changed. Since that time, as summarized in Creating a Sustainable Food Future, scientists realized that what causes soils to retain carbon is extremely complex and little understood. Most significantly, new studies and reanalysis of old data show that once soil carbon is measured properly and below the top-soil layer — even to a medium depth of around 30-35 centimeters — gains in soil carbon at the top are compensated in general by losses below, resulting in no overall carbon gains or only small carbon gains.5

Three authors of the Paustian et al. (2020) critique agreed in Soussana et al. (2019) last year that this new science undercuts claims that no-till would build soil carbon: “Meta-analyses conducted in recent years and covering the entire soil column suggested no significant positive difference in change in SOC [soil organic carbon] on average in no-till soils, although some increase in organic matter (and hence C) concentrations in the 15-20 cm layer of top-soil is usually observed” (Soussana et al. [2019]). Griscom et al. (2017), also cited, makes the same point.

Soussana et. al. (2019) also endorses the findings that using no-till for five years or less is likely to increase emissions of nitrous oxide, another potent GHG. As we discuss in our report, in the United States, no-till farmers typically plow up their land more frequently than that, 6 and there is broad agreement that doing so eliminates most or all of any soil carbon benefits. This science therefore implies that no-till is not sequestering carbon in the United States, and that it could be increasing GHG emissions through nitrous oxide.

Grazing management: Although scientific reviews in the early 2000s were hopeful about soil carbon gains from grazing management, as we summarize in Creating a Sustainable Food Future, “subsequent analyses have shown that the impact of improved rangeland management practices on soil carbon is highly complex, site-specific, and hard to predict.” Consistent with our finding, several authors, including an author of Paustian et al. (2020), published an article this year called “Soil carbon sequestration in grazing systems: managing expectations” (Godde et al. 2020),7 which articulates heavy limitations.

Much of the work prior to 2007 suggesting large soil gain potential on grazing land was led by Richard Conant, a co-author of Paustian et al. (2020). But Conant has also made valuable contributions showing that the potential to do so on working grazing lands is more limited than previously thought. For example, Conant co-authored a report in 2015 (Henderson et al. (2015),8 which lowered estimates of global technical potential for carbon sequestration in grazing lands to 300 million tons of CO2 — one-fifth of previous estimates — despite the vast 3 billion hectares global grazing lands occupy. This lowered estimate did not address economic potential, which would further reduce the ultimate potential sequestration. While half this potential could come from adding more legumes into some pastures, the paper warned that there is a “high risk of the practices, particularly legumes, increasing soil-based GHGs if applied outside of [a] relatively small effective area.” The paper also warned that scientists lack the tools to clearly identify which lands would gain soil carbon from legumes and which would not.

Similarly, as Conant et. al (2017) summarized persuasively in a review paper, the big soil carbon gains per hectare are from turning cropland into grazing land (which, as we noted previously, is limited by the world’s growing need for crops) or by adding legumes, limited for reasons set forth in the Henderson et al. (2015) paper. The remaining potential gains just from changes in grazing management are much smaller. And as Godde et al. (2020) noted, they are “derived from a limited number of observations and practices occurring in particular contexts and regions and cannot be extrapolated to global grazed area since sequestration rates are highly context-dependent.”

Based on these papers, we do not sense much disagreement about the science, but rather the significance we assign to the potential soil carbon gains from transforming cropland into grazing land given the need for more crops.

Cover Crops and Agroforestry

This category of practices focuses on those valuable measures that increase the growth of vegetation on existing cropland or pasture, typically through cover crops or by adding trees and shrubs. Despite our enthusiasm for these practices for other potential climate benefits, we do not believe the evidence to date justifies expansive estimates of their soil carbon potential, either in relation to the potential tons sequestered per hectare or to the total number of hectares that could be managed with these practices.

For agroforestry, soil carbon claims are based on limited and in some cases implausible data on soil carbon gains per hectare.9 There also are feasibility challenges of practicing agroforestry in mechanized cropping systems. We also consider the focus on unwarranted soil carbon gains potentially distracting given that the predominant and clearer carbon gains occur in the trees and shrubs themselves.

Cover crops are today used on only a small percentage of cropland. Although cover crops are likely to sequester some soil carbon, for a variety of reasons we address in an endnote, we do not consider estimates of the quantity of soil carbon gains per hectare from cover crops reliable at this time.10 Cover crops may also increase emissions of nitrous oxide, which could offset some or all of carbon sequestration.11 More information is needed about these issues and on cover crop adoption potential and ways to overcome barriers to adoption.12

Despite our reluctance to assign a soil carbon “climate wedge” to these practices based on the evidence today, we consider them of exceptional value for water quality and soil erosion control, while agroforestry clearly can sequester carbon in vegetation. In some cases, both practices can boost yields, with intensive silvopasture particularly promising. Although researchers and leading farmers have not yet fine-tuned cover crop use to the point where a sizeable share of farmers is ready to adopt them, they have been finding ways to make use of cover crops easier, and use has been growing. As these efforts continue, we hope evidence will emerge to warrant claims of sizable and feasible soil carbon gain potential. Regardless, because of the vital other role of these practices, we strongly urge creative programs to expand cover crops and agroforestry while innovating around the challenges.13

II. Large Global Estimates of Soil Carbon Sequestration Potential

Some responses to our blog post cited large estimates of global soil carbon potential. We have now more closely reexamined these estimates. To evaluate an estimate, it is necessary to know the different levels of soil carbon gains that estimate is assuming for different practices in different countries, types of land and soils, the adoption rates estimated for different types of farming and locations, and the evidentiary justification for those estimates. Unfortunately, when we further searched for this information, we found shortcomings in the documentation behind the estimates, as well as estimates that either rely on taking cropland out of production or on estimates of no-till or grazing benefits that are no longer justified.

IPCC 2007 estimates

By far the most cited estimates of soil carbon sequestration on working agricultural lands come from estimates done for the 2007 IPCC report, which are relied on for example in Paustian et al. (2016). They amount to 3 billion tons of CO2 per year in technical potential, divided almost equally between cropland and grazing land management. Estimates of economic potential are roughly half. The technical potential was more than half of estimated agricultural mitigation emissions at the time, which may explain why soil carbon sequestration has dominated much thinking for agricultural climate mitigation since then.

However, it is not possible to evaluate the bases for these large estimates because there is no ultimate documentation. Neither the original IPCC report nor subsequent papers that rely on it, such as Paustian et al. (2016), explain these estimates independently, and they instead cite to Smith et al. (2008).  (Technically, the IPCC report cites a version of Smith et al. available in 2007). Smith et al. (2008), in turn, explains that the technical potential estimates for cropland management result from modeling of some kind using a dataset of carbon sequestration gains from Ogle et al. (2005). Yet there is no documentation available for this modeling to understand the quantities of mitigation assumed, adoption rates, or other bases for these estimates. For the technical potential to sequester carbon on existing grazing land, there is no identified source at all, as Ogle et al. (2005) did not address grazing land improvements on existing grazing land.

For the global economic potential (1.6 billion tons of CO2 per year), Smith et al. (2008) identifies a 2006 report of the U.S. Environmental Protection Agency (EPA) as the source of the estimates.14 But that EPA report only addresses non-CO2 mitigation, such as methane and nitrous oxide, not the carbon sequestered in soils. Through email correspondence, we confirmed that economic estimates were ultimately derived from a rarely used, global version of the U.S. land use and emissions model FASOM-GHG. Further personal correspondence with the lead author of Smith et al. (2008) and with the two lead FASOM modelers indicate that no document exists that explains the bases for these global model results. As a result, it is not possible to learn the mix of practices assumed or estimated that would sequester this carbon, the assumed carbon gains per hectare, the estimated adoption rates or costs in different parts of the world, or the evidentiary basis for any of these estimates. This lack of documentation must also have precluded other scientists from reviewing the basis for these large soil carbon potential estimates during the IPCC review process.

Despite these limitations, the sources identified as influencing these global estimates make clear that the vast majority of soil carbon sequestration from cropland management was supposed to derive from the adoption of no-till farming, which, according to these estimates, is low-cost.15 As discussed above, new science undercuts these earlier claims.

The assumptions for the grazing management estimates are even harder to trace because not only does Ogle et al. (2005) not address them, but the U.S. version of the FASOM-GHG model declined to incorporate carbon sequestration on pasture into its estimates on the grounds that “limited data are available on the cost of adopting practices and corresponding carbon and other GHG effects.”16 From other evidence,17 we infer that the authors must have made simple assumptions about carbon gains and costs per hectare of grazing land globally despite this lack of data. Whatever the basis, as discussed above, these estimates are no longer consistent with the science of the limited, uncertain gains from changes in grazing management on existing grazing lands.

Other Large Global Estimates

Other large global estimates are no more convincing. For example, the major global initiative to build soil carbon today is the “4 per mille” initiative, which claims a potential to sequester a stunning 9 gigatons or more of carbon dioxide per year. The principal paper justifying this estimated potential (Minasny et al. 2017) lists case studies in which practices have sequestered carbon. But almost two-thirds of these case studies rely on adding manure or other soil amendments or on reducing crop area, practices that increase carbon in a field but, as noted above, but do not lead to net climate mitigation benefits. Moreover, the paper only assumes large global gains if these practices are broadly applied at some unspecified levels and in unspecified ways. That soils can store more carbon and that practices exist to sequester carbon in individual fields do not tell us how feasible it is to expand those practices and by how much, or whether they even result in net climate mitigation benefits after all land and GHG effects are taken into account.

Other papers cited in response to our blog post are also unconvincing. Fuss et al. (2018) only cites aggregate estimates by other papers without explanation. Another paper just assumed for purposes of analysis that farmers could implement unspecified practices to add carbon to soils, but even so found that the maximum potential storage in soils is limited (Sommer & Bossio 2014).

We have also discovered that old large global estimates continue to be cited even by those disagreeing with them. For example, even though Soussana et al. (2019) appears to endorse the new understanding of no-till, it also continues to cite large 2007 IPCC estimates of sequestration from cropland management that were based on no-till.18 Godde et al. (2020) cites large estimates of grazing management from the IPCC even while explaining that the science on which they are based is not valid.


Simple estimates of technical potential are valuable as preliminary bounding exercises, but they cannot be properly used as estimates of feasible potential. Many typical estimates, such as those used in Paustian et al. (2016) and Griscom et al. (2017) for cover crops, multiply an estimate of average carbon gains per hectare for a practice by large numbers of hectares to simulate a scenario of broad adoption. These estimates are seductive. If there is a large technical potential, say 1 billion tons of CO2, the intuition may be that the world can achieve at least some meaningful fraction of that, say 300 million tons. But the concept of technical potential is too malleable to justify that latter assumption. For example, the technical potential exists to eliminate all emissions from cars if people replaced all driving with walking, but that is not a meaningful estimate because doing so is not feasible.

To be meaningful, the analyzed technical potential must also be feasible. For example, in focusing attention on restoring peatlands, our Creating a Sustainable Food Future report reasons that drained peatlands emit around one billion tons of CO2, annually but only occupy roughly 0.5% of global agricultural land. The carbon savings per hectare are therefore very large, and the areas are small enough that governments can restore them through their own direct efforts. We also find that vast areas of drained peatlands have relatively limited agricultural uses.

Large estimates of soil carbon potential have also not addressed the challenge of persistence. Unlike carbon in trees, which will tend to persist on its own, soil carbon is not something farmers can add to their soils once and leave alone. Soil carbon is constantly at war with bacteria, fungi and other microbes that eat organic matter and return carbon to the atmosphere. To maintain soil carbon, farmers must generally maintain the practices that built it. Soil carbon in agricultural lands is therefore significantly more unstable than soil carbon in the forest because farmers often need to change what and how they farm with changing economics. A proper feasibility analysis needs to factor the risk of those changes into its projections, but we cannot find such analyses. No-till is an example. The great majority of U.S. farmers — either because of weeds, soil compaction or changing crops — feel that they need to regularly plow up their no-till, losing most or all of any previous carbon gains even if they occur. We have not seen any study that attempts to estimate adoption of indefinite, continuous no-till in a way that reflects these constraints.

Obviously, feasibility estimates are judgment calls and reasonable people can differ, but “technical” potential estimates alone reflect no judgment call. And judgment calls about what to push are important because the world has little time to reduce emissions and only so much capacity and resources to focus.

III. Other Issues Related to Manure, Yields and Nitrogen


In our blog post, we use claims of soil carbon gains by adding manure (a source of carbon) to a field as an example of a GHG accounting error because doing so comes at the expense of not adding that manure to another field. Paustian et al. (2020) agrees with this point but dismisses it on the grounds that such manure-based practices are an unimportant part of soil carbon sequestration estimates. We disagree. They were roughly 30% of the studies cited in Minasny et al. (2017) to justify the 4 per mille initiative. Paustian et al. (2016), Ogle et al. (2005) and Lal (2004) also cite manure or organic amendments. Regardless, the reliance on manure is just one example of the accounting errors, whose most important example is reliance on practices that reduce crop production.


Some commentators pointed out that regenerative agriculture practices do not necessarily decrease yields and may even increase yields. We agree, which is why we characterized yield effects as “mixed.” But that does not refute our point that negative yield effects can indeed occur — and need to be accounted for both in estimates of feasibility and in whether practices actually generate climate mitigation benefits when factoring in the need to replace crops. We do not see evidence that they have been factored into calculations of mitigation benefits from soil carbon sequestration.


Our blog post points out that building soil carbon is limited by nitrogen. In response, despite agreement that nitrogen is necessary to build soil carbon, others claim that abundant surplus nitrogen available in many areas is adequate. While we agree that some surplus nitrogen is potentially available, nitrogen supply still remains a major limitation.

One implication of the need for nitrogen is that nitrogen limitations help explain why soil carbon gains have been challenging in much of Africa. Africa is relevant because many claims have been made about the potential to regenerate African soil through payments to farmers for soil carbon gains. Yet in Africa, mining of nitrogen from soils is occurring, and even with enhanced use of legumes, fertilizer additions will still be necessary to build soil carbon. (The most rapidly growing legume in Africa, soybeans, fixes most of its nitrogen, but is still a net nitrogen remover from soils [Salvagiotti 2008]. Therefore, it does not contribute nitrogen to building soil carbon.) Unfortunately, as some researchers have acknowledged, adding fertilizer is not a way to mitigate climate change through soil carbon because of the cost and associated emissions from nitrogen manufacture and use (Soussana et al. 2017). Yet in addition to strong food security reasons to boost soil carbon by adding nitrogen, such efforts can contribute greatly to climate change mitigation by helping to boost yields (a “land sparing” strategy) and thereby reduce the need to clear forests and savannas to produce more food.

A second implication is that in areas with surplus nitrogen, soil carbon gains need to focus on those practices, particularly cover crops, that maintain roots after the main crop stops growing and therefore intercept nitrogen leaching from soils. The reason farmers apply surplus nitrogen is that much of it escapes to the air and water. Other practices that claim to sequester soil carbon need to demonstrate that they can, in fact, trap nitrogen, too. That is not clear for practices like no-till.

A third implication should be great skepticism of large soil carbon gains that exceed any plausible available nitrogen. For example, despite the potential of cover crops to trap some nitrogen, many reported soil carbon gains greatly exceed any plausible quantity of nitrogen available, and in fact rival those that can be sequestered not just by soils, but trees themselves in re-growing forests.19

It is possible that we may discover that some practices increase carbon-to-nitrogen ratios over sustained periods and thus can sequester carbon in excess of the maximum implied by the available nitrogen. But until that is proven, many large estimates of soil carbon gains should be treated with skepticism.

IV. Why Does This Matter?

Properly estimating potential soil carbon gains matter, we believe, for three reasons:

First, we are concerned that overly optimistic expectations around the potential to sequester carbon in working agricultural lands risks diverting effort from more promising opportunities for reducing GHG emissions from agriculture. These alternatives may include the use of regenerative practices such as improving grazing, which can protect forests and their carbon, but often only if linked to governance that protects forest — the “produce, protect and prosper” strategy discussed above. We need to support such practices for the right reasons so that we design policies to realize their potential benefits. Other promising mitigation opportunities include practices beyond common definitions of regenerative agriculture, such as restoring peatlands, improving manure management, reducing “enteric” emissions from belching cows through feed improvements, and increasing efficiency in the uses of nitrogen and other chemicals.

Second, practices that do not account for all GHG effects when calculating mitigation benefits could take the world in the wrong direction. For example, some propose that converting cropland to grazing land can make beef “carbon neutral,” which if true would make eating beef better for the climate than eating lentils. These claims are based on the theory that beef’s higher emissions are offset by soil carbon gains.20 This claim overlooks the fact that converting cropland to grazing land requires replacing the crop production elsewhere. Others have argued for “lighter grazing” to make beef production carbon neutral (Soussana et al. 2014), but such efforts also require replacing the reduced meat or milk production somewhere else. These approaches also undermine the critical need to increase output of milk and meat per hectare to meet rising food needs without deforestation and, ideally, free up land for reforestation (see Griscom et al. 2017 and our own report, Creating a Sustainable Food Future).

Third, overestimating the potential to sequester soil carbon based on what the world currently knows could undercut the need for research and development of new perennial systems or other innovative measures that might actually sequester soil carbon in more feasible ways. For example, agronomists and companies are working on oilseeds from trees that could replace soybeans (Brasch 2014) and ways of pressing protein feeds out of nitrogen-rich grass/legume combinations. By replacing annual crops with perennials, either practice could build soil carbon. But why bother to invest in their research if we already claim to know how to increase soil carbon at scale?

Given the 11-gigaton mitigation gap between expected agricultural emissions in 2050 and those needed to hold global warming below 2 degrees C (3.6 degrees F), it is critical to focus on actions with the greatest mitigation promise. Given the challenges discussed above, the realistic ability to sequester additional carbon in working agricultural soils is limited. Because what causes carbon to remain in soils is not well understood, further research is needed, and our views may change as new science emerges. Soil carbon research should continue. But today’s policy and actions must be based on what we know now. Based on that, some regenerative practices can help stabilize existing soil carbon and build resilience to climate change. They may even contribute to climate mitigation where they boost yields or increase vegetative carbon (as in trees). But these practices should not be relied upon for large-scale mitigation through soil carbon gains. Instead we should pursue a multi-pronged approach, such as the one laid out by WRI and others in Creating a Sustainable Food Future.


Abdalla, M., Hastings, A., Cheng, K., Yue, Q., Chadwick, D., Espenberg, M., Truu, J., Rees, R.M. and Smith, P. (2019). A critical review of the impacts of cover crops on nitrogen leaching, net greenhouse gas balance and crop productivity. Glob. Ch. Biol. 25(8):2530-2543.

Brasch, S., Is this Asian tree a new hope for Florida’s citrus growers, Modern Farmer (March 28, 2014) https://modernfarmer.com/2014/03/invasive-tree-new-hope-floridas-citrus-growers/

Brown V. Can responsible grazing make beef carbon-neutral? Civil Eats (April 10, 2018); https://civileats.com/2018/04/10/can-responsible-grazing-make-beef-climate-neutral/

Campbell, J., J. Sessions, D. Smith, K. Trippe, Potential carbon storage in biochar made from logging residue: Basic principles and Southern Oregon case studies, Plos One 203475 (2018).

Cardinael, R., T. Chevallier, A. Cambou, C. Béral, B.G. Barthès, C. Dupraz, C. Durand, E. Kouakoua, and C. Chenu. 2017. “Increased Soil Organic Carbon Stocks under Agroforestry: A Survey of Six Different Sites in France.” Agriculture, Ecosystems & Environment 236 (January): 243–55. doi:10.1016/j.agee.2016.12.011.

Conant, R., C. Cerri, B. Osborne, K. Paustian, Grassland management impacts on soil carbon stocks: a new synthesis, Ecological Applications 27:662-668 (2017).

Fuss S. et all., Negative emissions—Part 2: Costs, potentials and side effects, Environ. Res. Lett. 13:063002 (2018).

Gao, B., T. Huang, X. Ju, B. Gu, W. Huang, L. Xu, R.M. Rees, D.S. Powlson, P. Smith, and S. Cui. 2018. “Chinese Cropping Systems Are a Net Source of Greenhouse Gases despite Soil Carbon Sequestration.” Global Change Biology 24 (12): 5590–5606. doi:10.1111/gcb.14425

Gerber et al., Tackling climate change through livestock: A global assessment of emissions and mitigation opportunities (FAO, Rome 2013).

Giller K. et al., Beyond conservation agriculture, Frontiers in Plant Science 6:870 (2015).

Godde C.M. et al., Soil carbon sequestration in grazing systems: managing expectations, Climatic Change doi:10.1007/s10584-029-02673-x (2020).

Griscom W. et al., Natural climate solutions. PNAS 114(44):11645-11650 (2017).

Henderson et al. Greenhouse gas mitigation potential of the world’s grazing lands: Modeling soil carbon and nitrogen fluxes of mitigation practices, Agr. Eco. & Env. 207:91-100 (2015).

IPCC 2007; Smith, P., D. Martino, Z. Cai, D. Gwary, H. Janzen, P. Kumar, B. McCarl, S. Ogle, F. O’Mara, C. Rice, B. Scholes, O. Sirotenko, 2007: Agriculture. In Climate Change 2007: Mitigation. Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [B. Metz, O.R. Davidson, P.R. Bosch, R. Dave, L.A. Meyer (eds)], Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Kravchenko, A. N. and Robertson, G. P. (2011). Whole-profile soil carbon stocks: the danger of assuming too much from analyses of too little. Soil Sci. Soc. Am. J. 75:235–240.

Lorenz, K., and R. Lal. 2014. “Soil Organic Carbon Sequestration in Agroforestry Systems. A Review.” Agronomy for Sustainable Development 34 (2): 443–54.

Lubowski et al. “Land Use” in Agricultural Resources and Environmental Indicators, 2006 Edition (EIB-16 Economic Research Service/USDA 2006).

Minasny, B., B.P. Malone, A.B. McBratney, D.A. Angers, D. Arrouays, A. Chambers, V. Chaplot, et al. “Soil Carbon 4 per Mille.” Geoderma 292:59–86. (2017).

Pittelkow C. et al., Productivity limits and potentials of the principles of conservation agriculture, Nature 517:365-368 (2015).

Poorter, L_. et al._ Biomass resilience of Neotropical secondary forests_. Nature_ 530, 211–214 (2016).

Poulton P. et al., Major limitations to achieve “4 per 1000” increases in soil organic carbon stock in temperate regions: Evidence from long-term experiments at Rothamsted Research, United Kingdom,” Global Ch. Biolo. 24:2563-2584 (2017).

Powlson, D.S., A.P. Whitmore, W.T. Goulding, Soil Carbon Sequestration to Mitigate Climate Change: A Critical Reexamination to Identify the True and the False.” European Journal of Soil Science 62: 42–55 (2011).

Powlson, D.S., M.J. Glendining, K. Coleman, and A.P. Whitmore. “Implications for Soil Properties of Removing Cereal Straw: Results from Long-Term Studies.” American Society of Agronomy 103 (1): 279–287 (2011).

Powlson, D.S., A. Bhogal, B.J. Chambers, K. Coleman, A.J. Macdonald, K.W.T. Goulding, and A.P. Whitmore. “The Potential to Increase Soil Carbon Stocks through Reduced Tillage or Organic Material Additions in England and Wales: A Case Study.” Agriculture, Ecosystems, and Environment 146: 23–33 (2012).

Powlson, D.S., C. Stirling, M. Jat, B. Gerard, C. Palm, P. Sanchez, and K. Cassman.  “Limited Potential of No-Till Agriculture for Climate Change Mitigation.” Nature Climate Change 4 (8): 678–683 (2014).

Powlson, D.S., C.M. Stirling, C. Thierfelder, R.P. White, and M.L. Jat., “Does Conservation Agriculture Deliver Climate Change Mitigation through Soil Carbon Sequestration in Tropical Agro-ecosystems?” Agriculture, Ecosystems & Environment 220: 164–174 (2016).

Salvagiotti F, Cassman KG, Specht JE, Walters DT, Weiss A, Dobermann A, Nitrogen uptake, fixation and response to fertilizer N in soybeans: a review. Field Crop Res 108:1–13 (2008)

Sant-Anna S. et al., Changes in soil organic carbon during 22 years of pastures, cropping or integrated crop/livestock systems in the Brazilian Cerrado, Nutr Cycl Agroecosyst 108:101–120 (2017).

Schlesinger W., R. Amundson, Managing for soil carbon sequestration: Let’s get realistic, Glob. Ch. Biology 25:386-389 (2018).

Soussana, J.F., S. Lutfall, P. Smith, R. Lal, C. Chenu, and P. Ciais, “Letter to the Editor: Answer to the Viewpoint ‘Sequestering Soil Organic Carbon: A Nitrogen Dilemma.’” Environmental Science and Technology 51: 11502 (2017).

Soussana, J.-F., S. Lutfalla, F. Ehrhardt, T. Rosenstock, C. Lamanna, P. Havlík, M. Richards, et al. “Matching Policy and Science: Rationale for the ‘4 per 1000 - Soils for Food Security and Climate’ Initiative.” Soil and Tillage Research 188: 3–15 (2019).

Shackley S., J. Hammond, J. Gaunt, R. Ibarrola, The feasibility and costs of biochar deployment in the UK, Carbon Management 2:235-356 (2011).

Thorbecke, M., and J. Dettling. 2019. “Carbon Footprint Evaluation of Regenerative Grazing at White Oak Pastures.” February 25. https://blog.whiteoakpastures.com/hubfs/WOP-LCA-Quantis-2019.pdf

US-EPA 2006 Global anthropogenic non-CO2 greenhouse gas emissions: 1990–2020, United States Environmental Protection Agency, EPA 430-R-06-003, June 2006. Washington, DC: US-EPA.

van Groenigen, J.W., C. van Kessel, B.A. Hungate, O. Oenema, D.S. Powlson, and K.J. van Groenigen. 2017. “Sequestering Soil Organic Carbon: A Nitrogen Dilemma.” Environmental Science & Technology 51: 4738–4739.

Woolf, D., Amonette, J. E., Street-Perrott, F. A., Lehmann, J. & Joseph, S., Sustainable biochar to mitigate global climate change. Nature Comm. 1:56 (2010).


  1. This letter is from nine scientists, whose peer-reviewed papers on the subject include Powlson et al. (2011); Powlson et al. (2014); van Groenigen et al. (2017); Giller et al. (2015); Pittelkow et al. (2015); Sant-Anna et al. (2016). ↩︎

  2. Thorbecke & Dettling (2019); https://blog.whiteoakpastures.com/hubfs/WOP-LCA-Quantis-2019.pdf ↩︎

  3. Thorbecke & Dettling (2019); https://blog.whiteoakpastures.com/hubfs/WOP-LCA-Quantis-2019.pdfsee also papers cited in Creating a Sustainable Food Future p. 399, n. 303. ↩︎

  4. Municipal solid waste sent to landfills in the United States in 2017 was 139.6 million tons, of which almost exactly half was food, wood, paper or yard waste (U.S. EPA, National overview, facts, and figures on materials, wastes and recycling  https://www.epa.gov/facts-and-figures-about-materials-waste-and-recycling/national-overview-facts-and-figures-materials (accessed July 8, 2020)). Assuming 60% dry matter, that yields 42 million tons of dry matter, with roughly 21 million tons of carbon, and typically half of carbon is lost during composting, yielding roughly 10.5 million tons of carbon. This is the maximum amount that could be added to soils if all of this material could be separated. In reality, much of the material sent to landfills would be difficult to separate because it represents material already left over after two-thirds of paper and yard trimmings are already recycled. ↩︎

  5. As discussed in Paustian et al. (2018) and in Creating a Sustainable Food Future, a landmark paper by Baker in 2008 found that when soil carbon changes were measured to a full meter depth, claimed gains from no-till evaporated. In response, Kravchenko & Robertson 2011 pointed out how hard it is to discern soil carbon changes at that depth statistically. However, that paper affirmed the potential statistical reliability of measuring carbon at medium depth, e.g., 30 cm. In Creating a Sustainable Food Future, we rely on later meta-analyses by Powlsen et al. (2014) and Powlsen et al. (2016) which relied on measurements to that depth and found that in roughly half of studies, no-till generates a small carbon gain and in roughly half, no gain..  ↩︎

  6. There is limited data on persistence of no-till (and some U.S. Department of Agriculture (USDA) studies incorporate “ridge till” along with “no-till,” which is not the same).  However, in Creating a Sustainable Food Future (Searchinger et al. 2019), we summarized the evidence as follows: “For example, in one complicated analysis of Iowa using data from the 1990s, the authors estimated that the probability of no-till persisting for even two consecutive years was only 8%, with the vast majority of farmers practicing no-till for a single year. A study by the USDA using more recent data estimated that only 13% of cropland in the Upper Mississippi River basin was in no-till for three consecutive years, the maximum period for which data could be assessed?”  ↩︎

  7. Goode et. al. (2020). The main argument of the article is that for carbon sequestration in grazing land, “its global mitigation potential is lower than often implied.” Mitigation efforts “are associated with large uncertainties and will ultimately depend on the economic cost-benefit relation and feasibility of implementing the different strategies.” ↩︎

  8. Henderson et al (2015). This modeling was originally done for the FAO and published in Gerber et al. (2013), which estimated roughly double the technical potential, up to 600 million tons. But by the time the work was published in peer-reviewed form, the authors had cut their estimate in half. ↩︎

  9. Typical estimates such as those in Fuss et al. (2015) rely on Lorenz & Lal (2015) for the proposition that agroforestry can sequester gigatons of carbon dioxide per year. But Lorenz & Lal (2015) states that “data on the SOC [^soil organic carbon] sequestration rates are scanty.” Those presented in the paper come only from seven sites in three studies (Table 1), and we believe some are too implausible to be relied upon. For example, one study cited claimed 4.16 tC/ha/y in soil organic carbon sequestration, which exceeds typical estimates of the carbon that regrowing tropical forests take up in vegetation or in both vegetation and soils (see Porter et al. (2016); Searchinger et al. (2018).). At a 12:1 carbon ratio, this figure also implies soils would absorb 346 kg of nitrogen on each hectare each year, which is many times higher than all or nearly all crop nitrogen surpluses (Zhang et al. 2015). ↩︎

  10. One reason is that measurements of soil carbon gains from cover crops are typically taken after only two or three years, and that is not long enough either for reliable measurements or for the carbon in the dead cover crop plants to fully transform into more long-lasting soil carbon. Reported results also include implausible estimates that far exceed any nitrogen likely available to build soil carbon and that, if accurate, more plausibly are counting temporary vegetative carbon gains in soils rather than longer lasting soil organic matter. For example, Abdalla et al. (2019) (table 2), estimates average carbon gains from non-leguminous cover crops of 1.4 tC/ha/, with gains within one standard deviation of 2.9 tC/ha/y.  To put these estimates in perspective, the second claims in effect that cover crops sequestered as much carbon in soil as most forests are likely to sequester in a year in trees and soils combined. In addition, if we assume a long-term carbon nitrogen ratio of 12:1 and that cover crops can reduce losses of nitrogen from leaching by 50% (and overall reductions in losses by 40% after factoring in losses to the air), the mean estimate implies a surplus nitrogen level for the field of 290 kg/ha/y and the larger estimate a surplus of 480 kg/ha/y. Except in rare locations, these numbers exceed the quantity of nitrogen applied or fixed to almost all crop fields, let alone that remaining after much of the nitrogen is removed in the crops (Zhang et al. (2015).  The other available meta-analysis for cover crop soil carbon (Poplau & Don 2015) has smaller estimates for soil carbon gains and does have studies that persist for 15 years or more, but its mean estimate also appears to be significantly influenced by implausibly large soil carbon gains in sites analyzed only for two or three years. ↩︎

  11. As summarized in Abdalla et al. (2019), many papers have found nitrous oxide increases through cover crops while others have not. Abdalla et al. (2019) used a statistical test of 12 studies and 28 treatments to find no consistent, statistical pattern (a paired t-test). We think that particular statistical test inappropriate for this kind of distribution, and the results reported would justify an alternative conclusion. Nitrous oxide emissions tend to happen only in large quantities on specific days when conditions are optimum. The results reported found that several treatments with cover crops had high-spike emissions, and the average of all cover crop results was a meaningful increase in nitrous oxide compared to non-cover crops. Those are the types of inconsistent but occasionally high emissions results one would expect if cover crops increased nitrous emissions. ↩︎

  12. Those can include potential adverse yields if not terminated in time in the spring and limited time to establish cover crops in the fall. In addition, a third of U.S. cropland could not have cover crops because it is already used for hay, cropland-pasture, winter wheat and planted fallow. For example, as of 2012, out of 353 million acres planted, there were 56 million acres of hay, 13 million hectares of cropland used for pasture, 13 million acres of planted fallow (Ribaudo 2012), and 41 million acres of winter wheat, which equal 35%. USDA, National Agricultural Statistics Survey Data  https://www.nass.usda.gov/Charts_and_Maps/Field_Crops/index.php ↩︎

  13. Paustian et al. (2016) includes large technical estimates of potential soil carbon gains from biochar, but these are technical potential estimates only and therefore do not inform feasible potential.  We believe the biochar estimate, based on Woolf et al. (2010), relies heavily on many sources of biomass that this paper assumed are carbon-free, but come with high costs because they sacrifice carbon storage potential elsewhere. For example, any land capable of producing biomass to turn into biochar is also capable of providing other benefits, and thus has an opportunity cost. Forest harvest residues, in addition to being a prime target of biofuel estimates for airplanes, would, if left in the forest, continue to store carbon in vegetation for some time and then add carbon to soils. The only paper we have found estimating net gains from turning these residues into biochar found that doing so would increase carbon in the atmosphere for decades (Campbell et al. 2018). We think the most carbon-beneficial potential would come from using crop residues for biochar, but the most careful estimate of their use placed the cost at $210-$300 per ton of CO2e  (after backing out an assumed subsidy) even in England, where high residue production and mechanization should keep the cost relatively lower (Shackley 2011). Despite a high cost, we think the most realistic opportunities are to make and use biochar where it can regenerate degraded soils enough to significantly boost yields. ↩︎

  14. Smith et al. (2018) also cited to three other papers, but they were all domestic, U.S. papers and cannot explain global estimates. They also were cited as inputs to the analysis by the IPCC, not the actual analysis. ↩︎

  15. Ogle (2005) largely focuses on estimated soil carbon gains from no-till farming. An email from Pete Smith confirms that assumed cheap changes in tillage were the primary source of soil carbon gains in the economic estimates. That is also evident from the 2005 document (EPA 2005) describing the U.S. version of the FASOM-GHG model, which the principal FASOM modeler (Bruce McCarl) pointed us to for information while confirming that there was no written documentation of the international results. The carbon sequestration possibilities in the U.S. FASOM-GHG included crop tillage change, crop mix change, crop fertilization change, and grassland conversion. EPA 2005 p. 3-13, Table 3-5. However, the report later stated that the agricultural soil carbon sequestration predicted in the model for the United States “reflects the relatively low opportunity cost associated with adopting reduced tillage.” EPA 2005, p. 4-13. ↩︎

  16. U.S. EPA, Greenhouse Gas Mitigation in U.S. Forestry and Agriculture (2005), p. 3-14. ↩︎

  17. Without citing a source, the document does express the view that 0.7 tons of carbon could be sequestered per hectare per year for all pasture and range in the United States, which is high given that most of the grazing land in the country is quite dry. U.S. EPA (2005), p. 3-14. (The EPA paper suggests it consider this estimate the technical potential only on private pasture of 590 million acres, but 590 million acres is actually the estimate of total U.S. pasture and range, much of which is publicly owned [^Lubowski et la. 2006]). ↩︎

  18. “For croplands, the economic potential could reach 62% of the technical potential with a price of $100 USD per tCO2.” Soussanna et al. (2019), p. 10, citing Smith et al. (2008). ↩︎

  19. As one illustration, one estimate of the surplus nitrogen in corn/soybean rotations in the United States is 50 kg of N per hectare per year (email from Professor Xin Zhang based on work published as Zhang et al. 2015). If 40% of that could be trapped by cover crops (assuming 20% is lost to the air and then half of the remainder is still leached), and that N could be turned into soil organic matter at a carbon/nitrogen ratio of 12:1, the maximum gain would be 0.24 kg of carbon per hectare per year. That is far less than estimates of close to 3 tC/ha/y presented in Abdalla et al. (2019), and similar numbers are reported for some of the data in Poplau & Don (2015), which influence their mean estimates. ↩︎

  20. Thobecke & Dettling (2019). V. Brown (2018); Soussana et al. (2014). ↩︎