I’ll start with a definition: I think most agriculturalists take ‘cover crop’ to mean a crop grown in the same year as a primary crop such as a harvestable grain or horticultural crop, and that’s what’s assumed in this article. As noted below, some authors also include crops grown alone for the full season, solely for soil quality enhancement.

Current enthusiasm within agriculture in cover-crop usage is well placed. These crops provide clear benefits in protecting soil from water and wind erosion, providing end-of-season livestock feed, enhancing biodiversity, supplying significant quantities of legume N fertility for use by subsequent crops, and removing nitrates from soil at the end of the growing season (though the fate of nitrates thereafter often remains uncertain).

In addition, major attention is being placed on the potential for cover crops to sequester photosynthetically fixed carbon dioxide (CO₂) as soil organic carbon (SOC). This includes discussion on possibilities of paying farmers for doing this.

There have been many research trials completed and several meta-reviews published. This column contains a review of what the meta-reviews say individually and collectively – and what it means for agriculture and net greenhouse gas emissions, especially in a northern climate like Canada.

This article is written primarily for an agricultural audience interested in more depth than generally found in media and agricultural ‘extension’ reports, but less than in found in scientific papers. My column has not been peer reviewed although I have checked analytical details with several of the authors cited below.

I’ll start with my conclusions: The evidence is convincing that cover crops often mean more SOC. This benefit is largely proportionate to how long the cover crop grows, how much cover-crop organic matter is produced, and the average annual temperature of the location. A measurable effect is much less likely when the season of growth is short, e.g., when cover crops are seeded after longer-season harvestable crops and/or when the off-season consists of many months of frozen winter. Most authors assume that more SOC with cover crops means increased carbon sequestration, which is probably true in many cases. But the other possibility is reduced or no loss of soil organic matter when cover crops are present, as compared to soil with no cover crop.

That said, notable weaknesses are common in many of the research summaries, including:

1. Reports based very few years of cover cropping (sometimes only one).
2. The inclusion of data from experiments where the ‘cover crop’ was the only crop grown for an entire season – and, hence, not really a cover crop by the definition stated above.
3. The inclusion of experiments measuring percent SOC or soil organic matter only, with no data provided on soil bulk densities especially for cover-crop treatments in the studies. Procedures have been developed for estimating the missing bulk-density values, but these generally introduce sizable errors.
4. Different sampling depths within or among experiments – range from 2.5 cm to more than 100 cm (though few below 30 cm) – often with simple calculations used to ‘standardize’ to a common depth.

In addition, with one exception which included Chinese literature, none of the reviews included research reports written in languages other than English.

 I’ll refer to these in turn as we look at the results of the various meta-reviews. I should emphasize in advance, that all of the reviews represent a huge amount of work, with authors making major efforts to adjust for weaknesses and inconsistencies in the data sources used. I am most appreciative of that. But at the same time, it’s fair to examine their conclusions in light of how the limitations are addressed.

Now to the specifics.

Eagle et al (2012) of the Nicholas Institute for Environmental Policy Solutions, Duke University, published Greenhouse gas mitigation potential of agricultural land management in the United States: A synthesis of the literature, which includes a chapter called Use winter cover crops. Thirty one field studies were examined with a calculated average annual soil organic carbon gain equivalent to 1.3t/ha/year of CO₂ (range -0.2 to 3.2) or 0.35 t C/ha/year. However most of the data referenced in this review came from California and southeastern states. There were six comparisons from states further north, three from Iowa where the cover crop treatments actually caused a slight reduction in rate of soil organic C accumulation and three from Michigan with gains ranging up to 1.83 t/ha/year of CO₂. The Michigan studies were comparisons of an organic rotation of corn-soybeans-wheat plus cover crops with conventional corn-soybeans-wheat grown without cover crops. When I checked publications/web-sites, cited in these publications as providing more details on methodology, I was unable to find some key details on organic practices used, specifically the source of N fertility. Was the fertility provided by manure or other organic amendments? I don’t know. Hence, I suggest viewing the Michigan results with caution.

Eagle et al also refer to a study from Maryland in which the percent organic matter was higher in a cover crop treatment; they did not include this study in their calculation of mean effects because there weren’t data on bulk density and quantities of soil organic C per unit soil surface area. A summary table published by Eagle et al includes three model calculations and one “expert opinion,” all projecting more SOC with cover crops.

In essence, the Eagle et al review shows consistently more soil organic C with cover-crop usage in southern states, but an uncertain-to-zero benefit in the US Midwest.

Poeplau and Don (2015) published Carbon sequestration in agricultural soils via cultivation of cover crops – a meta-analysis, with the analysis involving 30 studies, 37 sites and 139 plots. The studies were in 11 countries, including Canada and the United States, and across four continents. Number of treatment years ranged from one to 38 and sample depth from 2.5 cm to 120 cm though with only three studies having samples “below the plough layer.” They calculated an average rate of soil C accumulation in cover-crop treatments, compared to plots without cover crops, of 0.32 + 0.08 t/ha/year of organic C. This paper and this quantity have been cited many times in other publications in support of statements that cover crops increase soil organic carbon.

For no other apparent reason other than I’m Canadian, I chose to check details of the four cited studies which were from Canada. They are:

Campbell et al, 1991, Effect of crop rotations and cultural practices on soil organic-matter: microbial biomass and respiration in a thin black chernozem. Can. J. Soil Sci. 71, 363–376. A 30-year study in Saskatchewan

Curtin et al, 2000.Legume green manure as partial fallow replacement in semiarid Saskatchewan: effect on carbon fluxes. Can. J. Soil Sci. 80, 499–505. A nine-year study in Saskatchewan.

Hermawan and Bomke, 1997. Effects of winter cover crops and successive spring tillage on soil aggregation. Soil Tillage Res. 44, 109–120. A one-year study in British Columbia.

N’Dayegamiye and Tran. 2001. Effects of green manures on soil organic matter and wheat yields and N nutrition. Can. J. Soil Sci. 81, 371–382. A five-year study in Quebec.

Of the four studies, three (Campbell et al, Curtin et al and N’Dayegamiye and Tran) all involve ‘green-manure crops’ grown without any other crop for the growing season and not the definition of ‘cover crop’ given at the beginning of this column. I did not check the other 26 cited studies in this review, but it is quite possible that there are additional ‘green-manure’ studies of the same nature included in the analysis by Poeplau and Don.

A second concern for me with this review is that only 30% of the cited studies contained data on soil bulk density. The authors estimated bulk density values for the other 70% using the graph below copied from their paper. It is based on collective data from the 30% of experiments with bulk density data across all countries.

Ignoring the one outlier value (which seriously distorts the R² and p calculations), the line looks like a reasonable fit for the other data. That notwithstanding, the values for bulk density predicted by the graph still appear to have a range of at least +0.05 g/cm3 for a given percent C content. For a soil horizon that is 22 cm deep and about 1.5% SOC, this range in bulk density equates to 3.3 t/ha of organic C, or about 10 times the predicted average accumulation rate of 0.32 t/ha/year advantage for cover-cropped plots.

A third unease for me involves the process used to standardize all organic C measurements to a depth of 22 cm using average measured or predicted bulk density for the assayed horizon – thus mathematically ignoring the normal tendency for SOC concentration to decrease with depth. One might argue a more sophisticated transformation process is not warranted given uncertainties introduced by the bulk density estimates. But it is still another source of imprecision.

In summary, in my view, results of the analysis by Poeplau and Don are sufficient to indicate that cover crops often mean a higher soil C content, but not nearly enough to merit a conclusion of a rate of gain of 0.32 + 0.08 t/ha/year.

As part of an extensive 2019 report called Negative emissions technologies and reliable sequestration, the National Academies of Science, Engineering and Medicine in the United States included a short section recommending cover crops as a means of sequestering soil carbon. However, it only cited two references, Eagle et al (2012) and Poeplau and Don (2015), the two reviews discussed above.

Mcdaniel et al (2014) published a meta-analysis entitled, Does agricultural crop diversity enhance soil microbial biomass and organic matter dynamics? It involved 122 publications with 454 observations. The review featured additional crops inserted into monocrop cultures or simple rotations, and a substantial number of the studies (I don’t think the authors don’t say exactly how many) involved cover crops. About two-thirds of the studies were from North America and the rest distributed globally. Because only 19% of the studies included measurements of soil bulk density to permit the calculation of soil organic carbon content per unit land area, Mcdaniel et al reported only the soil organic carbon concentrations. Because soil samples within individual studies sometimes came from different depths, Mcdaniel et al converted these to an average depth per experiment with no weighting for changes in bulk density with depth. (For their rationale, see Johnson and Curtis, 2001.) There was no adjustment to a common depth across studies.

Mcdaniel et al concluded the inclusion of cover crops increased the soil organic carbon concentration by 8.5% relative to plots with no cover crop. This was a much greater effect than for all other crop rotational comparisons combined in their analysis. The relative SOC boost with cover crops increased with increases in average annual temperature across the various research locations.

The authors defend their use of percent soil organic carbon (i.e., as g/g x 100%) instead of the use of soil organic carbon stocks (t/ha) by the presentation of an appendix graph showing no significant relationship between “% change from monoculture” and soil bulk density. It’s based, I believe, on observations from the 19% of surveyed studies that had bulk density data. Their assumption is in disagreement with the graph presented above from Poeplau and Don showing a trend for bulk density to decline as percent SOC increases.

If bulk density declines as percent SOC increases, this means changes in t/ha of SOM with cover crop addition would be expected to be less than the measured changes in percent SOC.

Kaye and Quemada (2017) in a paper entitled, Using cover crops to mitigate and adapt to climate change. A review, concluded that “Cover crop effects on greenhouse gas fluxes typically mitigate warming by ~100 to 150 g CO_2e/m²/year” with “the most important terms in the budget [being] soil carbon sequestration and reduced fertilizer use after legume cover crops.” Their conclusion on carbon sequestration is based largely on the analyses done by Eagle et al (2012) and Poeplau and Don (2015), which are discussed above.

Norris and Congreves (2018) in a review entitled, Alternative management practices Improve soil health indices in intensive vegetable cropping systems: A review, identified 60 studies where various practices had been used in attempts to improve soil health. They selected 22 of these studies, 11 involving cover crops, for a meta-analysis. The meta-analysis revealed no significant relationship between the presence or absence of cover crops and SOC. (It is assumed that SOC is expressed as a concentration rather than t/ha or equivalent in this review, though I don’t believe the authors actually say so.)

Abdalla et al (2019) published a review entitled, A critical review of the impacts of cover crops on nitrogen leaching, net greenhouse gas balance and crop productivity. Using the Web of Science data base they chose 106 studies involving several key words including, ‘cover crop,’ ‘soil organic carbon’ and ‘SOC.’ The range was global but most studies were from the United States/Canada, Europe and China. The objectives of this meta-analysis included the dynamics of soil nitrogen as well as carbon. For questions about SOC, the authors selected the 47 studies containing data on SOC in t/ha (or equivalent) or the combination of percent SOC and bulk density so that SOC in t/ha could be calculated. The authors adjusted SOC data to a standard depth of 30 cm based on an assumption that SOC decreases with depth at a rate of 2.14% per cm of depth. The rationale for that transformation comes from research by Jobbáy and Johnson (2000). While there is good research showing that the rate of change in organic C content with soil depth is not constant (tillage has a large effect on rate of change; see here for example), this adjustment is likely superior to the assumption of no change with depth to 22 cm in the analysis by Poeplau and Don.

Abdalla et al concluded that across all studies, cover crops increased SOM significantly with a certainty of P<0.001. However, this table copied from their paper, shows no statistically significant relationships at all.

In every grouping in their summary table (legume cover crops, non-legume cover crops, mixed and total), the average change in SOC is less than the standard deviation. It’s essentially the same for total greenhouse gas emissions including nitrous oxide.

In my view, the Abdalla et al paper lacks strong evidence that cover crops enhance soil organic C content.

Bai et al (2019) published Responses of soil carbon sequestration to climate smart agriculture practices: a meta-analysis. They selected 417 peer-reviewed papers studying the effects of biochar, reduced tillage and cover crops on soil carbon concentration. Of those, 64 papers featured cover crops with half of those done in the United States. On average across all studies, cover crops increased SOC concentration by 6% – though less but still statistically significant when the growing season was cooler. Although the authors don’t comment on this finding specifically, one graph in the review paper compares the cover crop effect for the 258 paired comparisons involving irrigated agriculture versus the 25 that were just rain-fed. For the latter, cover crops produced no significant change in SOC concentration. The authors state they did not calculate SOC sequestration for cover crop treatments because of the absence of bulk density data.

Jian et al (2020) published A meta-analysis of global cropland soil carbon changes due to cover crop which involved 131 studies and 1195 comparisons. Sixty percent of the studies were from the United States and Canada with the rest spread globally. The authors included papers written in either English or Chinese. Several sources were used for identifying studies including the references cited in the Poeplau and Don paper. Because of that, and the fact that ‘green manure’ was used as a search word, I assume that the studies used included some with crops grown solely for soil quality enhancement for the entire season but not harvested.

Fifty percent of the studies included measurements of soil bulk density, 21% had bulk density data for the check treatment only – in which case the cover crop treatments was assumed to have the same bulk density – and 29% were studies where an estimate of bulk density was made by Jian et al using the same procedure as used by Poeplau and Don (see above).

Jian et al did not attempt to standardize the data to a common soil sample depth, but just divide the results into two categories, those with measurements up to 30cm in depth and those from deeper depths.

The authors calculated an average soil organic carbon increase of 0.56 t/ha/year with cover crop (including some green-manure) crop usage, compared to the no-cover-crop check treatments. The pattern was generally consistent across crops and crop rotations, but with greater SOC enhancement for clay soils versus those with medium and course texture – and only for soil samples collected at or above 30 cm of depth.  The authors found a statistically significant trend for cover crop SOC enhancement to decrease as the mean annual temperature of the location became cooler and the latitude increased (i.e., for places like Canada).

The cautions I’d have with this review involve the inclusion of some studies with full-season cover crops/green-manure crops, and bulk density estimates made using data on SOC concentrations.

McClelland et al (2020) published results of a meta-analysis entitled Management of cover crops in temperate climates influences soil organic carbon stocks: a meta‐analysis. It involved 40 studies and 181 observation. The studies were mostly from the US and Europe (many from Italy), all from between 23.5 and 66.5 latitude (northern or southern hemisphere). They only used studies where they had reliable information on soil bulk density – although sometimes sourced from external sources – at least for the check treatment. All calculations of soil organic carbon content, expressed as t/ha or equivalent, were standardized to a common depth of 30 cm using the Jobbáy and Johnson (2000) technique described above for Abdalla et al.

The authors calculated an average increase in SOC with cover crops of 1.11 t/ha across all studies, compared to no cover crops. They found the greatest differences when cover crops: provided continuous year-round cover or were early-autumn planted and terminated before winter; with no-till cropping versus conventional tillage; when the above-ground cover-crop biomass accumulation was highest; when soil clay content was higher; and when grown in what they termed subtropical and tropical environments. In total, for studies from locations in temperate environments, there was no statistically significant increase in SOC with cover crop usage. A close relationship was found between the time duration between cover-crop planting and termination dates and the enhancement in SOC.

McClelland et al found that SOC accumulation, compared to no cover crop, was substantially higher for tests/plots accumulating more than 7 t/ha/year of above-ground cover crop biomass compared to those accumulating 7 t/ha/year or less.

One strange result was that authors found no relationship between the duration of cover crop usage (i.e., number of years since cover cropping was initiated) and SOC enhancement. In their discussion, the authors suggest that, because the average cover crop duration among studies was 5.2 years, this means an average annual C accumulation rate of 1.11t/ha divided by 5.2 years = 0.22 t/ha/year.  I suggest that they have a very weak statistical basis for that assumption. (It critically underestimates the measured SOC in tests with only 1-2 years of cover cropping and overestimates in tests from cover crop 8-10 years in existence.)

Crystal-Ornelas et al (2021) published Soil organic carbon is affected by organic amendments, conservation tillage, and cover cropping in organic farming systems: A meta-review. The review included eight comparisons of with and without cover crops. The measurement was only of SOC concentrations, not soil organic carbon stocks, and the overall difference was not statistically significant. However, for three comparisons involving only the upper 15 cm of soil, cover crops meant a significant increase in SOC concentration. There was a reverse trend, though not statistically significant, for five studies with measurements down to 50 cm.

Chahal and Van Eerd (2018), in a paper entitled Evaluation of commercial soil health tests using a medium-term cover crop experiment in a humid, temperate climate, established identical adjacent experiments in two consecutive years at Ridgetown Ontario on sandy loam soil.  Each involves a rotation of several field and horticultural crops with four cover crop treatments and one with no cover crop. While this paper is not a multi-study review like the others, I’ve included it here because of its special relevance to Ontario agriculture where I live and farm. Inclusion also reflects appreciation for the enormous amount of work that this research has entailed.

The authors measured the soil organic carbon content and bulk density for the upper 15 cm of soil in all plots in early September of the sixth year after the initiation of each experiment. There was no effect on cover-crop treatment on soil bulk density. A significant increase in SOC in mg/g of soil with cover crops was measured in both tests, compared to the no-cover-crop controls. See the second column in the table below. (Values followed by the same latter within each test are not significantly different at P = 0.05.)

Using these data and the measured bulk densities of 1.62 (Test A, sampled 2015) and 1.66 (Test B, sampled 2016), I have calculated the equivalent quantities in t/ha of SOC for the upper 15 cm of soil; see values in column 3 of the table. The authors published equivalent numbers (t/ha of SOC), using the same base data in a 2020 paper. These are shown in the fourth column in the table. The values I’ve listed were extracted from a bar graph in the 2020 paper using the software program, Greenshot. I am not clear as to why the values differ between the two papers for Test B but not Test A.

Notwithstanding the above, the data show a consistent cover crop effect on SOC enhancement, relative to no cover crop plots, in both trials and both papers. That’s consistent also with the length of time cover crops grew after short-season horticultural crops were harvested in this study.

Table: Data on soil organic carbon for two cover experiments at Ridgetown Ontario, from Chahal el al papers (2018, 2019 and 2020)

Treatment	SOC, mg/g, 2018 paper	SOC t/ha, 2018 paper	SOC, t/ha, 2020 paper	SOC, mg/g,2019 paper
Test A, planted 2007, measured September 2015
No cover	33.8c	82.1	82.1c	27.7ab
Oats	35.5b	86.3	86.6b	26.8a
OSR	35.8b	87.0	87.0b	29.3ab
OSR + Rye	37.2a	90.4	91.1a	30.9a
Rye	38.0a	92.3	92.8a	29.92b
Test B, planted 2008, measured September 2016
No cover	34.0b	84.7	63.6c	28.2b
Oats	38.1a	94.9	67.3bc	28.2ab
OSR	37.1a	92.4	82.0a	28.0b
OSR + Rye	36ab	89.6	82.0a	28.7ab
Rye	37.5a	93.4	80.5ab	32.2a

However, I have two reasons to be uncertain about the results.

The first is that the quantities shown in columns 2-4 are well above the 3.8% soil organic matter concentration that the authors measured at the beginning of the studies. The quantity 3.8% soil organic matter (SOM) is equivalent to about 22 mg/g SOC assuming SOM is 57% carbon. Twenty-two mg/g SOC equates to about 50-55 t/ha of soil C assuming a bulk density of about 1.6 and soil depth of 15 cm.

The second reason involves a second set of SOM data which the authors obtained from the sample soil samples collected at the same times in September 2015 and 2016. These were sent for analysis at an analytical service lab at Cornell University, with the results published by Chahal et al (2019), and also shown in the fifth column in the table above. These quantities are more similar to the SOC concentration measured when Test A was started. They do show a trend for higher SOC compared to no cover crop with cover crop treatments containing cereal rye. However, the amount of SOC enhancement measured is less than in the 2018 and 2020 papers.

I must emphasize that my comments about divergent data in no way detract from the high regard I have for Dr. Van Eerd and her research team, and their extensive research on soil quality. Even with the uncertainties I’ve raised, the research results indicate that higher SOC did occur some cover crop treatments.

I also want to acknowledge that Dr. Van Eerd was the one who alerted me to the reality that higher SOC values in cover crop treatments do not necessarily mean net carbon sequestration. It can equally means less loss in SOC with use of cover crops compared to no cover crops.

Bottom Line: What’s the overall conclusion?

A very substantial amount of research effort has gone into assessing the effect of cover crop usage on soil organic carbon, with apparent strengths and weaknesses in all of the studies. None of the reviews considered research reports other than those written in English – with the exception of Jian et al which also included reports in Chinese. Given that cover crops are of much interest to applied agriculturalists in non-English speaking countries – and notably Europe – the inclusion of reports in other languages in the meta-reviews would have been useful.

Despite the above, it does seem clear that use of cover crops commonly means more soil organic carbon relative to no cover crops. This is especially so when the growing season is long with average annual temperature relatively high and when the cover crops are grown for a long duration. The latter is more likely when the main cash crop matures very quickly (example, for some horticultural crops) or when the cover crop is in place for most or all of the full growing season (eg., between rows of perennial crops like grapes or as a full-season green manure crop). This also applies for a common practice in temperate winter-wheat growing regions like Ontario where red clover is ‘frost-seeded’ into wheat stands in early spring and allowed to grow until terminated in late autumn.

Conversely, the chance for significant, measurable increases in soil organic carbon seems much less likely when the cover crop is planted after harvest of a full-season, or near-full-season, primary crop and killed later that autumn or before spring planting the following year – and in climates where no crop growth occurs for many months of winter.

Published research also shows that soil organic carbon enhancement is more likely to occur on soils with higher clay contents than those high in sand.

The analysis by McClelland et al showed a distinct difference in relative SOC accumulation for tests/plots accumulating more than 7 t/ha/year of above-ground cover crop biomass versus those accumulating 7 or less. As a starting point, 7t/ha/year might be a useful yardstick for judging whether a detectable cover crop effect on soil organic carbon is to be expected for a given site, or not.

I note that McClelland et al found an even closer statistical fit between length of time from planting to termination and SOC level, but this is difficult to apply where the time from planting to termination may include many months of no growth (eg., winter in most of Canada).

I suggest that none of the studies reviewed above gives a reliable estimate of the rate of annual SOC improvement that might be achieved at any given location with cover crop usage. The limitations in published data include:

1. Reports based very few years of cover cropping (sometimes only one).
2. The inclusion of data from experiments where the ‘cover crop’ was the only crop grown for an entire season – and, hence, not really a cover crop by the definition stated above.
3. The inclusion of experiments measuring SOC or soil organic matter percentages only, with no data provided on soil bulk densities, especially for the cover-crop treatments. Procedures have been developed for estimating the missing bulk-density values, but these introduce sizable errors.
4. Different sampling depths within or among experiments – range from 2.5 cm to more than 100 cm (though few below 30 cm) – often with simple calculations used to ‘standardize’ to a common depth.

In addition, none of the research reviewed included situations where cover crops were used for late-season grazing by ruminant animals.