Another Industry Attempt to Greenwash Beef
A review of the 2020 White Oak Pastures and General Mills study with commentary on the subject of “Regenerative Grazing”
Study: Ecosystem Impacts and Productive Capacity of a Multi-Species Pastured Livestock System | Rowntree et. al (2020)
Review below by: Nicholas Carter | Dr. Tushar Mehta | PlantBasedData.org
In June 2019, we reviewed a study by the environmental consultancy Quantis that was commissioned by White Oak Pastures (WOP). Quantis concluded that White Oak Pastures (WOP) was storing more carbon in its soil than the combined greenhouse gas (GHG) emissions of its pasture-raised cows during their lifetime. Our brief review was critical, partly due to the clear bias in a previous Nestle-funded Quantis study which concluded that allowing plastic bottles in national parks was ecologically benign. However, the main criticism was because of the large body of scientific literature refuting the claims of so-called “regenerative grazing”, an idea popularized by Allan Savory, and which also goes under names such as “holistic grazing” and “rotational grazing”. Subsequently, WOP began to make wide-scale claims about their carbon-negative beef:
Bold claims about the carbon negative beef from White Oak Pastures beef based on a commissioned Quantis study.
In late 2020, a peer-reviewed life cycle assessment was published by Rowntree et. al (2020): Ecosystem Impacts and Productive Capacity of a Multi-Species Pastured Livestock System. In this study, mixed farming of cattle, pigs and poultry birds (mainly chickens) was compared to the conventional farming of each species. Please see below for a summary of the study, followed by our analysis and general comments:
Findings from the study:
WOP grazing methods require 2.5 times more land than conventional meat production of each of the animal species.
WOP claims to sequester 2.29 Mg of C/Ha/yr (carbon per hectare per year), in the form of Soil Organic Carbon (SOC), thus offsetting 80% of the net GHG emissions from their livestock, on a yearly basis.
This made their carbon emissions 66% lower compared to conventional practices for the same species but it was NOT carbon-negative.
The paper also considers the possibility of ascribing the full SOC sequestration to cattle alone, a calculation which would suggest that the cattle were carbon-negative.
Also note, as of this date, WOP continues to advertise the production of “carbon-negative beef”
Off the top, we note that WOP does use well-known methods that show potential improvements over standard North American grazing practices. They apply compost to their land, use legume cover crops to make nitrogen available, and also include local forbs, clovers, and nut-bearing trees. Chemical fertilizers and herbicides are not used. Such methods, amongst others, have long been used in Conservation Agriculture, and are also applied to farming plant-based foods.
WOP also prevents overgrazing by having limits on the number of animals per unit area, i.e. “stocking density”, and rotating animals to prevent overgrazing in any one particular area.
Soil Organic Carbon Measurements
Rowntree et. al (2020) estimated Soil Organic Carbon (SOC) sequestration via a space-for-time substitution rather than directly measuring SOC change over time. This means they took samples from different areas of the WOP land which were estimated to represent the state of their land area in prior years. They did not actually take soil samples each year over the past twenty years. This can make it easy to skew results, depending on the areas that were chosen for sampling and how those areas were used.
Carbon storage on-farm trees was not measured. Note that trees store massive amounts of carbon, but this is not counted when calculating SOC. However, if a significant number of trees are planted, such as reforestation, the grazing area would be reduced.
The paper also makes the following statement: “Importantly, if we were to attribute the soil C sequestration across the chronosequence to only cattle, MSPR beef produced in this system would be a net sink of −4.4 kg CO2-e kg CW−1 annually.”
Evidently, the paper is making a special attempt to promote the idea that grazing cattle may be carbon negative.
Below we will highlight why this suggestion is a skewed and exaggerated interpretation of the available evidence regarding the subject of optimized grazing and human managed grasslands.
Inputs from off-farm sources
Inputs that arose from off-farm sources were not properly accounted for. The study is presented as if the intrinsic qualities of the farm, especially the grazing methods, were responsible for the main SOC sequestration. However, the paper does note that large quantities of hay from off-farmed sources were added as feed and fertilizer during the initial land management and ceased thereafter. This was about 10,000 pounds per acre, for 200 acres according to the data. Resulting cattle manure and hay decomposition act as fertilizer. It is difficult to interpret whether this applied to a single 200 acre area, or if other 200 acres sections received the same treatment in stages.
More importantly, WOP monogastric animals, mainly chickens and pigs, consumed feed crops that were imported from off-farm sources, year after year. According to the supplementary data, the feed quantity was 1.8 million kg (4.034 million pounds) in a sample year. In addition, there was 11793 kg (26000 lbs) of mineral bar per year. See tab 1 of the supplementary data for details.
Considering that the combined carcass weight of chickens and pigs exceeded that of slaughtered cattle on the farm, this manure from off-farm feed is a substantial fertilizer derived from external food sources.
Manure from all animals, including these monogastrics, was used to fertilize the WOP land. This is important since fertilizing soil contributes to vegetation growth, which is the main cause of increased SOC. On its own, this critical missing factor in the paper’s calculations could account for the majority increase in SOC, completely skewing the results, and invalidating their conclusions about “regenerative grazing”. Such lapses are signature characteristics of industry affiliated “research”.
Creative accounting of land use by WOP animals
Further accounting misdemeanours were found through examining the paper and supplementary data. Firstly, note that the land used to grow the monogastric feed discussed in the prior section was not accounted for. WOP consists of 3000 acres, equal to 1214 hectares. However, most or all of the monogastric animal feed was from off farmed sources; this land was not accounted for. Hence, the calculation of feed for the monogastrics should include the land used for their feed production. The promoters of regenerative grazing heavily criticize “row cropping” and monoculture, yet there is no indication that the feed for these animals (roughly 1.8 million kg per year in the year used in their data) was from any form of conservation agriculture. It is possible that chemical fertilizers and pesticides were used, ultimately benefiting the WOP farmland.
There may well be other such liberties taken with data collection, handling and calculations, but our team does not have unlimited time or access to material data from WOP to analyze further. We hope that the largest lapses are thus detected and encourage others to analyze for themselves. The disturbing problem is that few people have ample time to analyze every study, especially those conducted by the hand of industry.
Land characteristics of WOP and SOC Sequestration Potential
WOP is a land area with excellent water and temperature conditions but was severely depleted from prior extractive farming techniques for corn, peanut, soybeans and wheat. As such, initial years of either improved plant or animal farming should sequester SOC, noting that the yearly amounts are known to slow and plateau over time. SOC sequestration is a good strategy in areas such as WOP, but note that it has a limit. Net increased soil carbon sequestration would be less on land that was initially in better condition, closer to its peak carbon level.
This point is important since “regenerative grazing” advocates usually imply SOC sequestration can be used as an unlimited strategy, which is not possible.
Total Land Carbon Storage vs Soil Organic Carbon (SOC)
The paper continues to focus on the misleading paradigm of using soil organic carbon, rather than total land carbon storage, which is the proper measure to use.
Total Carbon stored on a landmass = [Above Ground Vegetation (i.e. plants and entire trees)] + [Above Ground Detritus (logs, leaves, and all other organic matter)] + [Soil Organic Carbon (as measured in this study)] + [Soil Inorganic Carbon]+ [Below Ground Detritus and Live Matter (i.e. carbon in the roots of trees and all other plants)]
Optimized grazing certainly stores more SOC than poor grazing or cropping practices. However, optimized plant agriculture also increases SOC, and produces more protein and calories per unit of land. Both of these practices, however, store far less total land carbon compared to rewilding and reforestation. Note that monogastric animals (i.e. chickens and pigs) also use less land than ruminants (i.e. cattle, sheep and goats) and would allow some saved land for rewilding, but not as much as plant foods for direct human consumption. In such cases, livestock is thought to have a lower impact when diversified and integrated with crop cultivation, although the net benefit is largely due to lower numbers of animals per unit land rather than more intensive grazing practices. Furthermore, grain-fed cattle use less land and feed than grazed cattle, though still more than monogastric animals, to create the same amount of food. Land use change must be quantified when calculating the carbon impact of any product, especially ruminant/cattle meat.
According to Lal (2010), all undisturbed natural ecosystems contain more soil organic carbon than their agricultural counterparts that, on average, sequester 25–75% less. Such natural ecosystems include forests, wetlands, and grasslands. All pasture or cropland was originally a type of natural ecosystem that has been converted to human use.
As such, Rowntree et. al (2020) did not measure carbon opportunity cost of land, which more comprehensive studies do include in LCAs. To understand the magnitude of this opportunity, it was shown that shifts in global food production to plant-based diets by 2050 could lead to sequestration of 332–547 GtCO2, equivalent to 99–163% of the CO2 emissions budget, 19 years of total fossil fuel emissions of upper and middle-income countries. This is very significant during our current climate crisis.
This is accomplished by replacing animal-based foods with optimal plant protein foods, grown on a much smaller fraction of land, but producing the same amount of protein, while then rewilding/reforesting the saved land.
Polycropping with the best area-adapted plant protein foods such as legumes, soy, nuts, seeds, and grains using conservation agriculture presents high potential. Supplying equivalent protein and improved nutrition, while storing more carbon, should be minded as the real benchmark for sustainability.
Rewilding not only stores the most carbon but is also essential for biodiversity and ecosystem services. Biodiversity conservation of grazing versus rewilding was not within the scope of this paper.
The Rowntree et. al (2020) WOP study claims that SOC sequestration is commonly excluded from Life Cycle Assessments (LCAs). This is simply not true, since there are multiple important articles dealing with the topic, as we have referenced.
Bottom line, we should be evaluating any plant or animal agriculture for the amount of land and other resources needed to produce a set amount of protein, calories and other nutrients, using conservation agriculture techniques. The most efficient method will sequester SOC, but also use the least land, thus allowing the remaining land to be rewilded or reforested, and also prevents the conversion of more natural ecosystems into agricultural land, by meeting increased human population demands on a smaller total area of land. As such, there is data showing that an extra 4 billion people could be fed, without causing any further deforestation, if existing land used for animal foods were instead used for more efficient plant-based foods.
Does “Regenerative Grazing” sequester SOC in a linear or unlimited manner?
Rowntree et. al (2020) present the soil organic carbon storage as a linear, yearly event in their graph. However, they admit that the first three years were maximal, then slowed and even peaked by year 13, though they question their own data. Their straight-line could be presented as a curve, such as the red line we superimposed, reflecting nonlinear storage of soil carbon. According to their sampling, there is also variation due to random effects. Yearly fluctuations in carbon storage and release may also be occurring. Small increases in grazing levels may inadvertently release carbon, and the threshold may change as yearly conditions vary, such as rainfall.
It is difficult to base conclusions with only 7 plot points using space-for-time substitution. Numerous spatial samples on a true yearly basis would be better.
Other studies have demonstrated the process of soil carbon saturation at an equilibrium point. The graph below shows a long term example, partly relying on space for time technique, but also including actual samples that take over decades. The amount, rate and duration of SOC storage would apply to the specific geographical area measured where the study was done. Also critical is the stocking density, where smaller numbers of animals allow greater vegetation growth. See the graph below from this study 2014 Do grasslands act as a perpetual sink for carbon?
Smith, P. (2014). Do grasslands act as a perpetual sink for carbon?. Global change biology, 20(9), 2708–2711.
Industry proponents of “regenerative grazing” often implicitly or explicitly exaggerate the SOC benefits of improved grazing techniques, as seen in the recent documentary Kiss The Ground, which makes absurd claims that global grazing could single-handedly solve the entire climate crisis. This would require significant SOC storage year after year, without a reduction in the foreseeable future (i.e. a linear pattern of carbon storage). Unfortunately, the notion is not true.
From Grazed & Confused Report by Table (formerly FCRN)
Once SOC becomes closer to the saturation point, the GHGs (per KG of meat) will be far higher in WOP ruminant meat, more similar to conventional grazing methods.
On page 10 of the Rowntree paper, the authors cite five papers suggesting that components of soil carbon may actually accrue indefinitely. However, when we read these papers, they did not suggest such a claim and in fact, supported the thesis of saturation, with some potential modifiers of the saturation point. As such, this seems to be a misrepresentation of cited sources, and again, an inflated advertising of indefinite SOC sequestration by an industry-funded paper. See Appendix 1 below.
Use of References: “reference dumping” and interpretation of existing literature on optimized grazing
In addition to those mentioned in appendix 1, numerous articles are cited in Rowntree et. al (2020) article, creating the appearance of being comprehensive. Are these references warranted, or dumped into the article for appearance sake? Are their data and conclusions fairly represented? Few may endeavour to review each reference, though we chose a selection from parts of the paper.
As per our review, the findings of these references are often oversimplified and portrayed in a manner implying the importance of grazing for SOC sequestration. However, some of the cited references actually describe the inconsistency around grazing research. They show the inferiority of grazed land compared to natural ecosystems, an overall potential of SOC loss on grazed land depending on various factors, and actually make suggestions for shifting diets away from beef/ruminants. Some also recommend shifting grazing towards intensified cattle farming (i.e. factory farms) as an option to reduce land use and thus restore pasture to natural ecosystems and prevent future deforestation for grazing land. Conant 2017 and Gricom 2017 are two such articles cited in the WOP study, whose findings differ from the way Rowntree 2020 represented them.
Other references in the 2020 WOP study describe the known factors that contribute to improved or optimized grazing. These factors include lower stocking rates (i.e. having fewer animals on an area of land), use of legume cover crops, prevention of overgrazing which may strip vegetation from soil (i.e. having fewer animals and systematically moving their grazing areas), and better manure management to reduce the methane and nitrous oxide emissions that come from manure, and even water flow management or diversion to adjust retention, such as creating ponds. They also show that external fertilizer and water application increase vegetation growth and consequently increase SOC. Ultimately, these are factors that promote dense vegetation growth with least disruption by cattle. Dense vegetation is what stores atmospheric carbon into plant matter, and then to soil. We agree with all of this, noting that the listed references do not demonstrate that large quantities of ruminant meat could be produced while sequestering significant amounts of SOC. Again, the question of “compared to what” arises. The references suggest that improved practices store more SOC compared to conventional grazing, but as previously stated, they do not demonstrate an advantage over grain-fed ruminant animals, monogastric animals, plant foods for direct human consumption, grown by either conventional or conservation agriculture, or rewilding.
We emphasize that the existing literature shows it is not cattle themselves that cause the majority of SOC sequestration during optimized grazing. It is the improved practices, outlined in the paragraph above, as well as exogenous fertilizer or water, that achieve a higher soil carbon steady state despite the grazing of cattle. Literature promoting regenerative grazing often conflates cause with correlation; especially the error of spurious correlation. When the authors make a sample calculation attributing the entire SOC sequestration to cattle, which would peg WOP cattle at −4.4 kg CO2 equivalent per kg of carcass mass, they could also arbitrarily assign the sequestration to the few rabbits they farmed, and then promote the carbon-negative rabbit “supermeat”. Much more likely, the manure from chickens and pigs fed from off-farm sources is a far bigger factor than cows.
Furthermore, just as in medical science, observational data outcomes should include more studies with relevant control and comparison arms, where optimized and standard rewilding practices are compared to optimized grazing and cropping practices, and further bolster existing literature. That is well beyond the WOP study, but we should recognize the need for such studies with lesser involvement of industry affiliation, whether animal or plant-based.
Grazing advocates suggest that cattle hoof action and dung cause increased vegetation growth. Could hoof action benefit degraded land where soil was very compacted? Maybe during the initial year of land rehabilitation, but this should be compared to other methods to achieve the same goal. Also, we have yet to see reliable studies comparing such methods. Regarding fertilization, cattle eating and defecating on the same piece of land results in no gain in net soil nitrogen or nutrient content, unless fed using feed grown from off the land sources, in which case there is transfer of nutrients from one place to another, resulting in loss from one place balanced by gain in another. On the contrary, cattle dung releases both methane and nitrous oxide, which are potent GHGs as mentioned before. The loss of nitrous oxide to the atmosphere is a potential net loss of nitrogen, as are the nutrients in any fecal matter that is washed into streams or ground water. Ecological impacts of manure are a global problem, lessened but not eliminated by optimal management, including the health aspects.
Consider land use from a different angle. WOP is located in an area of Georgia which used to be scrubland and oak savannas. How much carbon would be released from additional deforestation or land degradation if farmers switched from conventional animal farming to “regenerative” cattle grazing, i.e. the main use of WOP land?
Based on surveys, satellite images and categorizations from various U.S. government agencies. Each square on the map represents 250,000 acres of land. https://www.bloomberg.com/graphics/2018-us-land-use/
Observe the amount of land used for pasture in the U.S., and the consequence of switching cattle who are conventionally raised on grain (thus utilizing the landmass denoting “livestock feed”, and switching them to grazing. The chart below shows the current breakdown of land use for animal agriculture in the US:
There’s a single, major occupant on all U.S. land: cows. Between pastures and cropland used to produce feed, 41 percent of U.S. land in the contiguous states revolves around livestock. Land use classifications are based on data published in 2017 by the U.S. Department of Agriculture’s Economic Research Service in a report called the Major Uses of Land in the United States (MLU).
This problem exists on a worldwide scale, since about 45% of the Earth’s habitable surface is used for animal agriculture, but returns about 37% of edible protein. In comparison, plant foods for direct human consumption utilize about 6% of the Earth’s habitable surface and return about 63% of human consumed protein. Of the land used for animals, the largest amount is used for grazing.
Methane (CH4) “All gasses were converted to CO2 equivalents (CO2-e) using current 100-year global warming potentials (CO2 = 1, CH4 = 30.5, N2O = 265)” (Rowntree et. al, 2020).
Emitted methane, the major source of global warming from ruminant animals, lasts for about 10 years in the atmosphere, due to its short half-life. The sun energy it absorbs exerts most of its warming effect during this time.
Methane is considered to have a warming effect that is 28–36 times that of CO2 when considered over a 100-year time scale, but this may undervalue the impact of methane, whether from industry or agriculture. The use of a 100-year timescale is a common way to underestimate the impact of methane, although there is more research needed here.
A 100-year timescale minimizes methane’s impact by at least 2.5 times compared to a 20-year Global Warming Potential (GWP).
The Intergovernmental Panel on Climate Change (IPCC) recognizes the validity of shorter time horizons. The IPCC stated:
“There is no scientific argument for selecting 100 years compared with other choices (20 or 10). . . The choice of time horizon is a value judgement because it depends on the relative weight assigned to effects at different times” (IPCC, 2013, AR5, PP 711–712).
Rowntree et. al (2020) correctly mentions a more fibrous diet (grass-finished) creates much more emissions than grain-finished, but they downplay methane effects in the atmosphere. Here is why methane, even from cattle, matters.
They further described that “Beef cattle account for about 95% of animal and 52% of land emissions.” on WOP. This clearly shows an obvious inefficiency.
This recent study by Matthew N. Hayek shows various methods that significantly underestimate methane when it comes to ruminants.
More on why methane is not just part of the carbon cycle and how others are creating false narratives that are not backed by credible evidence.
Funding & Authors Methods should always be the first area of focus for critique, nonetheless, it is worth considering funding sources in the critical analysis. Industry funding does not discredit a study on its own, but if certain typical patterns emerge, then the effect of an industry motive often has telltale signs.
Rowntree et. al (2020) is definitively an industry-funded study. It was funded entirely by General Mills, which actually purchases meat from WOP, for use in their “Epic Jerky” product. General Mills is trying to incorporate regenerative branding in multiple products. Despite the presence of a local farmer, the connection and interests of massive industry players, including funding, are evident.
As such, WOP is not just a small local farm, but actually part of a powerful and massive multinational company, and an important arm of its branding and marketing.
Individuals from Epic Provisions, a company selling paleo style and regenerative meat products, were also credited in the study. This company is associated with the Savory Institute, which is heavily involved in beef industry promotion and misinformation on the subject.
One of the current authors actually works for General Mills itself, and another was a co-author of the initial Quantis study on WOP, claiming carbon-negative beef, which was again funded by General Mills.
In the supplementary materials of Rowntree et. al (2020), they provide the study “Environmental footprints of beef cattle production in the United States” as a source for quantifying beef emissions in the United States. This study, was funded by the Beef Checkoff Program and one co-author actually works for the National Cattlemen’s Beef Association:
Screenshot directly from the US study on beef emissions used in this study.
This industry study conveniently uses a low range estimate of 17 CO2e/kg for beef, which is much lower than non-industry studies that show 29 CO2e/kg, especially those which factor land use impact that have a mean global average of 60 CO2e/kg. Some papers use carcass weight or body weight (which includes bones and other inedible components). A footprint based on carcass weight will be lower than if you do the conversion to actual meat you get from it. These are ways visible environmental impacts can appear to be minimized.
Essentially there are multiple levels of industry interconnection in the funding and authorship of this study, important lapses in the study, and exaggerated ecological claims. These claims go directly into further marketing of their products, including the notion of “carbon-negative beef”.
Furthermore, just after the publication of this article, as well as recent promotional movies on the subject of “regenerative grazing”, McDonald’s and A&W are promoting their grass-fed and sustainable beef on a massive scale. There is a strong mega corporate branding strategy for “sustainable beef” at work, beyond only General Mills:
Colonialism and Systemic Industry Issues
Our literature review caused notice of another disturbing tendency associated with the meat industry. During a New York Times interview, the owner of WOP, remarked that the “confederate general Robert Lee” was one of his “heroes”, a man who is known to be a racist and involved in slavery. Note carefully that this does not necessarily reflect the personal racial bias of the speaker or on any person at WOP, but that there are systemic issues. The meat industry is often racialized and has a colonial history in America, and currently in turmoil in places like the Amazon, which is being actively deforested for cattle ranching. Cattle take more land per unit product, and a powerful global industry must get that land from somewhere.
Improved grazing and animal farming techniques, as well as improved plant farming techniques, are known to increase soil carbon sequestration, but the results are modest and time-limited. The WOP farming methods likely do have modestly reduced carbon emissions compared to conventional meat production
We can credit the current paper for having the integrity show the increased land use per unit product from WOP.
The paper disproves the “carbon-negative” status of WOP pasture overall.
However, there are serious potential problems with studies funded by large industries, and this study by General Mills shows typical patterns. All studies require critical review, but industry studies even more so.
This study has significant lapses that grossly exaggerate or misrepresent the true SOC sequestration capacity of their farming techniques, as well as the actual land used by their animals, especially cattle. To some degree, it still clings to the exaggerated claim of “carbon-negative beef”.
Grain-fed ruminants, non-ruminant meat, and plant protein foods for direct consumption have progressively lesser GHG and other impacts.
The topic of “regenerative grazing” has long been a marketing tool by industry, who are funding their own studies and using the results for marketing purposes
There are many papers outlining this subject over the years, and the recent report by TABLE (formerly the Food & Climate Research Network), is one of the most comprehensive on the issue. Aptly titled “Grazed and Confused”, here are some of the results:
The benchmark for sustainability should be food systems that use the least land to produce an equal amount of protein and calories, such as plant-based foods grown with conservation techniques. Saved land should be used for rewilding, which is the most important factor for SOC sequestration and biodiversity conservation. This recent textbook, Rethinking Food & Agriculture, is a comprehensive academic review of the subject.
The chart below outlines the potential for the best and worst types of plant and animal agriculture, is a well-referenced summary by Our World in Data.
By Nicholas Carter and Dr. Tushar Mehta
Podcast (Plant Proof) - Can Holistic Grazing Reverse Climate Change? A Review of Kiss the Ground.
Podcast (Plant Proof) - Is grass-fed beef good for our planet?
Here are the five references that are used by Rowntree et al on page 10 of their paper, where they suggest the possibility of indefinite carbon storage on grazed land, rather than saturation. A read of each paper shows that
a) none of the papers are remotely able to make such a strong claim, and
b) most of these papers explicitly support the idea of a soil carbon saturation point, including natural variability, and somewhat modified by cropping or grazing practices.
As such, they are an example of exaggeration and inappropriate use of references to make a claim sound realistic. Please read the section on page 10 of the Rowntree paper on “C storage capacity” which leads them to “question the certainty of soil C saturation in grassland soils”. Then download and read the references below for yourself. All of these papers are in concordance with the references which we have used in our article, suggesting the importance of soil carbon, the potential for improved farming, and concur with the limits that other such papers discuss.
Compares grassland vs. forest carbon balance between soil organic, inorganic, and particulate carbon. It makes no mention of grazing, but considers natural and semi-natural ecosystems, implying low human disturbances. Forests are thought to have higher peak saturation points than grasslands, and forests with certain fungal (mycorrhizal) ecosystems may store indefinite inorganic carbon, which actually confers a greater imperative for forest rewilding rather than grazing land.
The West and Six paper is yet another that suggests no-tillage, improved cropping, legumes, irrigation, improved grazing practices and fertilization increase soil carbon for both cropping and grazing systems. The article definitely says that there are saturation points, but further interventions may raise or lower the saturation point. Grazing may have a longer time to saturation than cropping but there are local variations in all practices. Note that improved grazing likely means both rotation and lower stocking and hence, less meat production per unit land, since the idea is to preserve vegetation.
This paper is specifically a soil chemistry paper regarding measurement techniques with radiolabeled carbon and has no reference to grazing or the peak soil carbon. It does discuss the different components of solid carbon but does not entertain the idea of indefinite carbon storage on grasslands, indicating that it would be a great stretch to use it in such an argument.
This is an article regarding no-till organic agriculture, where corn is grown using either organic or chemical fertilizer. It concludes that organic no-till methods initially store carbon in the surface soil layer, which eventually translocate to deeper layers. Radiolabeled carbon is used. The paper does support a saturation point. This article is mainly about improved plant agriculture and not grazing. Although one could reasonably assume that low-density rotational grazing would have similar results, it does not support an exaggerated claim about the potential of SOC on grazing lands.
The Lavallee paper discusses the complexities regarding soil organic matter and defining as particle, mineral associated, and dissolved components, and their dynamics. It does not make any strong claim about indefinite carbon storage in soil, but rather that optimization requires the understanding of the components mentioned. It does not discuss grazing or cropping, or long term soil carbon storage through human agriculture that would exceed that of rewilding or minimization of human disturbance.