Journal list menu
Grassland management impacts on soil carbon stocks: a new synthesis
Abstract
Grassland ecosystems cover a large portion of Earths’ surface and contain substantial amounts of soil organic carbon. Previous work has established that these soil carbon stocks are sensitive to management and land use changes: grazing, species composition, and mineral nutrient availability can lead to losses or gains of soil carbon. Because of the large annual carbon fluxes into and out of grassland systems, there has been growing interest in how changes in management might shift the net balance of these flows, stemming losses from degrading grasslands or managing systems to increase soil carbon stocks (i.e., carbon sequestration). A synthesis published in 2001 assembled data from hundreds of studies to document soil carbon responses to changes in management. Here we present a new synthesis that has integrated data from the hundreds of studies published after our previous work. These new data largely confirm our earlier conclusions: improved grazing management, fertilization, sowing legumes and improved grass species, irrigation, and conversion from cultivation all tend to lead to increased soil C, at rates ranging from 0.105 to more than 1 Mg C·ha−1·yr−1. The new data include assessment of three new management practices: fire, silvopastoralism, and reclamation, although these studies are limited in number. The main area in which the new data are contrary to our previous synthesis is in conversion from native vegetation to grassland, where we find that across the studies the average rate of soil carbon stock change is low and not significant. The data in this synthesis confirm that improving grassland management practices and conversion from cropland to grassland improve soil carbon stocks.
Introduction
Even accounting for historic losses, grazing lands still contain a substantial amount of the world's soil organic carbon. Integrating data on grassland area (FAO 2015) and grassland soil carbon stocks (Sombroek et al. 1993) results in a global estimate of about 343 Pg C (in the top 1 m), nearly 50% more than is stored in forests worldwide (FAO 2007). These grazing land soil carbon stocks are susceptible to loss upon conversion to other land uses (Paustian et al. 1997) or following activities that lead to degradation (e.g., overgrazing). Over the last decade, grassland area has been shrinking while arable land area has been growing, suggesting continued conversion of grassland to croplands (FAO 2015). Around 20% of the world's native grazing lands have been converted to cultivated crops (Ramankutty et al. 2008), leading to losses of as much as 60% of soil carbon stocks (Paustian et al. 1997, Guo and Gifford 2002). Grassland degradation has also expanded (Bai et al. 2008), likely prompting the loss of grassland ecosystem carbon stocks. Arresting grassland conversion and degradation would preserve grassland soil carbon stocks.
Reversing practices that have led to grazing land degradation can prompt increases in ecosystem carbon stocks, sequestering atmospheric CO2 in grazing land soils. Estimates of the potential for carbon sequestration in grazing lands are substantial, but also highly uncertain (e.g., Smith et al. 2007). These uncertainties stem from a few limitations. First, knowledge about how the world's grazing lands are managed and the propensity of grazing land managers to adopt improved management practices is limited. Data on grazing lands are collected less widely than data on forests and cropland and tend to be of lower resolution and limited extent (Conant and Paustian 2004). Second, information on how grazing land management affects soil carbon stocks has been limited to a subset of regions and management practices.
The purpose of this paper is to synthesize information on grazing land management effects on soil carbon stocks. Substantial work has been conducted since other earlier syntheses (e.g., Conant et al. 2001, Follett et al. 2001). The world's grazing lands are a key element of livestock production systems globally (Bouwman et al. 2005, Erb et al. 2016) providing livelihoods for about 1B of the world's poorest people and one-third of global protein intake (Steinfeld et al. 2006). Here we build on recent regional syntheses (Wang et al. 2011) and studies of particular aspects of grazing land management (e.g., McSherry and Ritchie 2013) to produce an up-to-date, comprehensive source of information on how land management and conversion to grasslands affect this important resource. This new synthesis can undergird global estimates of grassland soil C stocks and changes, which are an essential basis for policy and market decisions on grassland C stock management.
Methods
This work follows the approach taken in our earlier synthesis (Conant et al. 2001). Briefly, the primary literature was surveyed by searching the Web of Science using search terms describing grazing lands (i.e., “grass,” “rangeland,” “pasture,” etc.) and soil carbon (including soil organic matter). The abstracts of these (several hundred) studies were reviewed to determine whether they were likely to contain observations of soil carbon stocks for mineral soil horizons under contrasting management, and if so, the full papers were reviewed and ultimately the relevant data extracted from 126 papers (compared with 115 in our earlier synthesis). We employed the convention used previously to classify the studies (Conant et al. 2001). Almost all studies contrasted a standard/nominal/common management practice with an “improved” management practice that differed from the standard, or occasionally a “degraded” condition compared with a standard/nominal/rehabilitated condition. This convention lumps together a variety of “improved” grazing practices, including lower stocking rates, removal of grazing livestock, several types of rotational or short-duration grazing, seasonal grazing, etc. The “reclamation” study examined the effect of sowing grasses into mining overburden (Brown and Grant 2000).
Data on soil organic matter were converted to soil organic carbon by assuming that 58% of soil organic matter is carbon (Nelson and Sommers 1996). Data on soil carbon stocks (mass per unit area to a specified depth) and soil carbon concentration (mass of carbon per unit mass of soil) are reported separately. If data on soil carbon concentration and bulk density were included, we report carbon stocks. Ancillary information (climate, soils, native vegetation, etc.) were compiled and used for analysis. When these ancillary data for a given study were not included, we used globally available information on climate (Schimel et al. 1996) and vegetation (Matthews 1983) to estimate local conditions. Results from newly synthesized information are reported in conjunction with data from our earlier study (Conant et al. 2001), with noted exceptions. All primary data and ancillary information on location, soil type, climate, etc. are reported in the Supporting Information (Metadata S1 and Data S1).
In cases where measurement uncertainties were reported, meta-analyses were conducted using a random effects model in the meta R package. If the 95% confidence intervals of the effect size for a variable did not overlap with zero, we concluded that the effect of grassland/land use differed significantly between the two treatments. All statistical analyses were conducted in R (R Development Core Team 2010).
Results
Characteristics of studies surveyed
We identified 332 data points across 64 new publications, nearly doubling the number of data points from our previous review, which had 364 (Conant et al. 2001). As in that study, conversion from cultivated land to pasture (36% of newly synthesized studies), fertilization (22%), conversion of native vegetation to pasture (19%), and grazing (19%) studies were the most common type of observation (Table 1). Information on fire, reclamation, and silvopastoralism impacts on soil C were not included previously because we were unable to identify published work. Our search turned up no new studies of earthworms, irrigation, or sowing improved grass species. The largest portion of studies (46%) were based on those in which management practices were compared between fields or farms (which implicitly assume that soil C stocks in the standard/nominal field are unchanging), with many (35%) based on manipulative experiments, and the remainder (19%) based on measured changes in soil carbon over time.
| Treatment | No. studies | No. data points | Portion of data points (%) |
|---|---|---|---|
| Fertilizer | 50 | 208 | 31.7 |
| Conversion: cultivation to grass | 42 | 161 | 24.6 |
| Grazing | 50 | 137 | 21.0 |
| Conversion: native to grass | 32 | 106 | 16.2 |
| Sowing legumes | 7 | 19 | 2.9 |
| Sowing improved grass species | 2 | 6 | 0.9 |
| Grass ley in rotation | 3 | 6 | 0.9 |
| Fire | 1 | 5 | 0.8 |
| Earthworms | 2 | 3 | 0.3 |
| Irrigation | 1 | 2 | 0.3 |
| Reclamation | 1 | 1 | 0.15 |
| Silvopastoralism | 1 | 1 | 0.15 |
Note
- Data include newly synthesized information and data from Conant et al. (2001).
Overall, the characteristics of the studies included in this review differed little from those of our previous assessment (Conant et al. 2001). However, the proportion of studies from warm and tropical climate regimes increased (mean annual temperatures of 17.8°C for this study and 15.4°C for our previous study) and studies tended to sample deeper in the soil profile (44.5 vs. 38 cm previously; Table 2). The proportion of studies representing sub-tropical/tropical grasslands increased as did the number that had taken place in the southern hemisphere (Fig. 1). There were still relatively few studies in sub-Saharan Africa, but the 30 data points from that region was still a substantial increase from the three data points our previous work (Conant et al. 2001).
| Variable | Mean | Median | Minimum | Maximum |
|---|---|---|---|---|
| Mean annual temperature (°C) | 15.4 | 13.2 | −1.5 | 27.9 |
| Mean annual precipitation (mm) | 1004 | 750 | 7.6 | 4900 |
| Latitude (degrees) | 38.4 | 40.5 | 0 | 105.6 |
| Depth (cm) | 38 | 20 | 2 | 800 |
| Duration (yr) | 25.1 | 15 | 0.7 | 200 |
Note
- Data include newly synthesized information and data from Conant et al. (Conant et al. 2001).
The majority of studies (68.2%) examining changes in soil C stocks with management improvements found increased soil C. This was true for most land use/management practices, the two exceptions being conversion from native vegetation (soil C increased for just 46.2% of studies) and grazing management (48.9%). Despite more studies in which soil C declined, soil C increased on average for both conversion from native vegetation and grazing, suggesting that relatively large increases for a small number of studies offset small declines in other studies. There were few studies of irrigation, silvopastoralism, and reclamation, but these practices always increased soil C. Conversion from annual crop cultivation to pasture almost always resulted in increased soil C (91.3% of the time).
Of the studies that reported changes in soil C concentration (i.e., percent C), most (188 out of 270, 69.6%) found increased soil C concentrations with changes in management (Fig. 2). The overall average across all management changes was an increase of 26.1%, representing an average change in soil C concentration of nearly one-quarter of 1% (0.22%). The largest proportional changes in soil C concentration were for reclamation (though based on a single study) and conversion from cultivation to grass (+39.2%; Table 3). Conversion from native vegetation led to declines in soil C concentration in most studies (61%), by an average of 14%. Soil C increased for the majority of studies in other types of comparisons Increase; grazing was the only type of study with a sizable minority (15 of 40 studies) of observed declines in soil C concentration following changes in management. Changes in soil C concentration following changes in management for grazing, fertilization, fire, and grass leys each averaged about a 10% increase over previous practices.
| Treatment | Soil C concentration (%) | Change (%) | |
|---|---|---|---|
| Initial | Final | ||
| Conversion: cultivation to grass | 0.97 | 1.35 | +39.2 |
| Conversion: native to grass | 2.97 | 2.55 | −14.0 |
| Earthworms | 2.24 | 2.89 | +28.8 |
| Fertilization | 3.44 | 3.85 | +11.8 |
| Fire | 1.09 | 1.20 | +10.5 |
| Grazing | 2.62 | 2.89 | +9.99 |
| Grass ley | 2.50 | 2.68 | +10.3 |
| Reclamation | 8.0 | 15.9 | +98.8 |
Note
- Data include newly synthesized information and data from Conant et al. (2001) averaged across depth and experimental duration.
All of the management practices characterized as “improved” tended to lead to increased soil C stocks (Fig. 3), with the average across all studies 0.47 Mg C·ha−1·yr−1. The largest increases driven by conversion from cultivation (0.87 Mg C·ha−1·yr−1) and sowing legumes (0.66 Mg C·ha−1·yr−1) and fertilization (0.57 Mg C·ha−1·yr−1; Fig. 3). Sequestration rates for grazing (0.28 Mg C·ha−1·yr−1) were lower and conversion from native (0.02 Mg C·ha−1·yr−1) led to very low sequestration rates that were not significantly different from zero change. No data subsequent to our previous work (Conant et al. 2001) were collected on sowing improved grass species, irrigation, or introduction of earthworms. About 21% of the fertilization addition studies examined the impact of an organic fertilizer on soil C stocks. Soil C sequestration rates were positive for most fertilization studies whether they examined the use of organic (79% of studies) or inorganic (82%) fertilizers, though the rates were much higher for the organic fertilizers (0.82 Mg C·ha−1·yr−1) than for inorganic fertilizers (0.54 Mg C·ha−1·yr−1).
There was some correspondence between study duration and C sequestration rate, with management changes based on shorter-term studies having higher rates of C sequestration (irrigation [5 yr], sowing legumes [8.3 yr], sowing grasses [9.8 yr], and introduction of earthworms [10 yr]) and vice versa (conversion from native [23 yr], fertilization [26.7 yr], and grazing [38.5 yr]).
Most observations (475 out of 655) occurred within 20 cm of the soil surface as did most of the change in soil C stocks (Fig. 4). The magnitude of the change in soil C stocks generally declined with depth, with an average increase of 23% over initial soil C stocks in the surface 20 cm but just 12% at lower soil profiles.
Discussion
The main conclusions of this updated synthesis of studies examining the effect of grassland management on soil C stocks are consistent with our earlier work (Conant et al. 2001): management practices introduced intending to increase forage production tend to lead to increases in soil C stocks. As the pre-2001 data (Conant et al. 2001) show, responses are varied and practices introduced to increase forage production can lead to soil C losses. Nonetheless, synthesis of existing studies suggests that improving grassland management can lead to soil C sequestration, by an average of 0.47 Mg C·ha−1·yr−1. This is consistent with other recent syntheses (McSherry and Ritchie 2013, Wang et al. 2016). Each of the six types of grassland management improvements and conversion from cultivation to grassland included in this synthesis all led to increases in soil C stocks. Thus these new data serve to expand the number (nearly doubling) range of observations to new places and practices and generally confirm earlier conclusions from other global and regional syntheses (Smith et al. 2007, Wang et al. 2011).
Considerable rates of soil C stock increases (0.28 Mg C·ha−1·yr−1) coupled with the large estimate of global grazing land area (~3.4 billion ha; FAO 2015) and the fact that changes in grazing management can be implemented in any grazed grassland, suggest that grasslands have the potential to sequester large amounts of atmospheric C in the soil, as several authors have suggested (e.g., Smith et al. 2008, Lal 2009). Our synthesis of field data are consistent with this, showing that soil C stocks increase (on average) with improvement in grazing management. However, each of these grazing studies investigated the impacts of a specific grazing management intervention under conditions in which the implemented change in grazing management was warranted and expected to be beneficial. Thus these results do not apply uniformly to all grazing lands and extrapolating the results of this synthesis regionally or globally requires information about where there is scope for improvement of grassland management (Ogle et al. 2004). Also, despite our estimate of an average increases in soil C stocks with grazing improvement, it is not always the case that improved grazing management leads to increased soil C stocks. Even when it does, soil C stock responses vary as a function of climate, soil, and vegetation characteristics (McSherry and Ritchie 2013). Rotational grazing is widely believed to lead to increased productivity and potentially to increased soil C stocks (Briske et al. 2011b). But as is the case for productivity (Briske et al. 2008, 2011a), the few studies available for inclusion in this synthesis are inconclusive.
The grassland management practices evaluated here affect greenhouse gas emissions other than just the net uptake of CO2 (Soussana et al. 2010). Many of the practices evaluated in these studies affect forage productivity, which could have implications for CH4 emissions if changes in forage productivity lead to changes in livestock management and subsequent ruminant and manure emissions (Jones et al. 2005, Soussana et al. 2007, Gleason et al. 2009). Practices that change system nitrogen dynamics, such as fertilizer application, are likely to impact N2O fluxes (Schlesinger 2010). Several aspects of management could influence the amount of C stored in woody biomass (Asner et al. 2004). Among the studies included in this synthesis, measurements of management effects on non-CO2 GHGs were extremely rare, yet it is clear that they are potentially important in developing systems to reduce greenhouse gas emissions from grasslands or to enhance sustainable use of grassland systems (Henderson et al. 2015).
One of the main contrasts with our earlier synthesis is that conversion to pasture from native vegetation led to no significant changes in soil C stocks. This synthesis included 52 studies published since the previous review and soil C declined with conversion from native vegetation to grassland for these studies (by an average of 0.24 Mg C·ha−1·yr−1). All of the new studies examined conversion of native forest to planted pasture land (see Supporting Information for citations). Losses of soil C were observed for both temperate (−0.84 Mg C·ha−1·yr−1) and tropical regions (−0.13 Mg C·ha−1·yr−1). In contrast with the previous synthesis in which more than 70% of conversion from native studies led to increased soil C, just 40% of the new studies did. Losses of soil C stocks in response to forest conversion to grassland compound losses of biomass C.
While our results clearly show a growing body of literature demonstrating that improved grassland management can lead to increases in soil C concentrations and soil C stocks, the results presented here do not address several barriers to developing policies intended to bring about increases in soil C stocks as a greenhouse gas mitigation measure. The impacts of management on changes in woody production are not addressed here nor are changes in forage production, which can have consequence for on-site CH4 emissions and land use/management decisions through leakage/indirect land use change (Gonzalez-Ramirez et al. 2013). Thus while we do not address the several economic and policy challenges to fostering improved grassland management through support for increasing soil C stocks (Conant 2011, 2012, Booker et al. 2013), this synthesis of biophysical work confirms that improvements in grassland management lead to increased soil C stocks in grassland ecosystems under a wide range of conditions.
Acknowledgments
This work was supported through agreements C/10/049 and C/10/050 with the International Livestock Research Institute, grants from the U.S. Environmental Protection Agency EPA-OAR-CCD-09-07, the National Science Foundation award 0842315, a Queensland Smart Futures Fellowship, a grant from the Plant Production and Protection Division of the UN Food and Agricultural Organization, and the Conservation Innovation Grant from the U.S. Department of Agriculture.
