Over the past decade, agriculture has gone from sinner to saviour in the context of global warming. For example, the World Bank [1] reported in 2012 that “Some 30 percent of global greenhouse gas (GHG) emissions are attributable to agriculture and deforestation driven by the expansion of crop and livestock production for food, fiber and fuel.” However, an awareness of the potential of, and advocacy for, soils to sequester CO₂ from the atmosphere has been gathering momentum, propelled by the proposal at the 21st Conference of Parties (COP21) of the United Nations Framework Convention on Climate Change (UNFCCC) for the soil carbon content (SOC) of soils globally to be increased by 0.4% per annum (www.4p1000.org, accessed on 20 March 2022). The same World Bank report hoped that, in the global debate about climate change mitigation, “the ‘triple win’ of soil carbon sequestration for increased productivity, improved climate resilience, and enhanced mitigation” would become an integral part of the dialogue.

The World Bank’s call has been taken up by various international consortia such as The Adaptation of African Agriculture [2], Living Soils of the Americas [3] and Advancing Climate Action in the Americas [4]. However, actual mechanisms by which land managers can be rewarded for genuine GHG abatement through soil carbon sequestration (SCS) have been primarily the focus of government or private-sector action in North America and Australia. In these regions, schemes have been developed (see Table 1) whereby land managers are encouraged to implement practices to draw down CO₂ from the atmosphere and store it in soil organic matter (SOM), described as a negative emissions strategy [5]. For a defined area of land, SCS represents the balance between the transfer of atmospheric CO₂ to soil through photosynthetic products, and carbon (C) losses primarily through soil respiration [6]. When the balance favours C accretion (i.e., is positive), net SCS occurs, which is measured by sampling the soil to a specific depth, normally 0.3 m as recommended by the Intergovernmental Panel for Climate Change (IPCC) [7], and measuring the soil C concentration and bulk density of the fine soil mass (particles < 2 mm equivalent diameter, which excludes gravel). Bulk density is necessary because this property can change with time under different soil managements. Because of this, estimates of the soil C stock over time are best based on an equivalent soil mass [7], but are usually scaled up to a C content per unit area (e.g., tonnes (t) per hectare (ha)).

Plant product removal, soil erosion and leaching of dissolved organic C may also deplete the soil C stock, and this will be reflected in its measured value. However, the success of SCS as a negative emissions strategy requires not only that net SCS be positive, but also that net SCS exceeds any increases in emissions of other GHGs that might occur during the operation of the project [8]; that is, net C abatement occurs.

Although the original COP21 proposal referred to an 0.4% increase in soil C in world soils, it is clear from subsequent references [9] that the proposal was focused on agricultural soils and their management. As indicated above, implementation schemes (called protocols) for SCS are now available in several countries ^[6][10][6,10]. One of the longest running schemes is the Emission Reduction Fund (ERF) of the Australian government (www.cleanenergyregulator.gov.au/ERF, accessed on 20 March 2022), under which soil C projects are strongly promoted. If a project achieves genuine net abatement, the landholder is rewarded with Australian Carbon Credit Units (ACCUs). One ACCU is the equivalent of one tonne of carbon dioxide equivalent (CO₂-e) of net C abatement.

Protocol	Additionality Requirement	Permanence Period	Risk-of-Reversal	Leakage	Considers Other GHGs
Climate Action Reserve Enrichment v.1.0	Yes, performance and legal requirement tests	Yes, 100 years or tonne-year accounting for a shorter period	Percentage of credits to a buffer pool	Yes, for displacement of livestock and lower crop yields	Yes, uses modelling or emission factors
Nori Croplands Methodology v.1.1	Yes, project must show increase in SCS over baseline	10 years	Yes, restricted tokens are used for any deliberate reversals	Verify if SOC gains cause losses outside of project boundary	No
Gold Standard Soil Organic C Methodology v 1.0	Yes, performance and legal requirement tests	Permanence within crediting period (5–20 years)	Yes, a percentage of credits go to a buffer pool	Yes, accounts for shifting crop production	Yes, modelling or emission factors if emissions > 5% of baseline
BCarbon	Credits issued for C added after initial testing	10 years, renewable after credits issued	10% of credits to a buffer pool	Potential leakage assessed by life cycle analysis	No

³). For field variabilities ranging from 0.3 to 0.7 standard deviations, the number of samples required ranged from 12 to 62 per ha. The effect of spatial variability, which may be exacerbated by seasonal changes from year to year, can be moderated to an extent by ensuring that samples are taken at the same time each year. However, the sampling intensity required for a 95% level of certainty remains high, so the ERF sets the confidence level for accepting a significant difference between means at 60% ^[20][26]. More intensive soil sampling incurs greater costs. For example, for a 68ha cropping field in central-west New South Wales (NSW), Singh et al. ^[21][27] reported an all-in cost of AUD37/ha (in 2011 dollars) to measure the soil C stock (to 0.3 m) with a standard error ≤2 t/ha. Under its Technology Investment Roadmap ^[16][22] (p. 24), the Australian government proposed the ambitious “stretch goal” of reducing the cost of measurement to AUD3/ha. Hence, much effort has been devoted to developing techniques that are cheaper, with an acceptable degree of precision, such as near- and mid-infrared spectroscopy ^[7][22][7,28]. However, such methods require calibration against soil C concentrations measured by dry-combustion analysis. Other methods, the so-called hybrid methods, seek to reduce the cost of monitoring by coupling direct measurements of soil C with a model of soil C dynamics, as advocated by Powlson and Neal ^[23][29]. A new ERF protocol, released at the end of 2021 ^[20][26], involves using less frequent soil sampling and measurements that are used to check the output of a C model. Other approaches involve the use of remote sensing, in particular, spectral bands ^[7][24][7,30]. Such a method may have some application for bare soil, but not vegetated land, other than for estimating above-ground plant biomass, which may provide an input variable to a soil C model. Prior remote sensing may also be helpful in determining the most effective selection of sites for soil sampling. Oldfield et al. [10] discuss some of the limitations of these “advanced” technologies.

3.2. Examples of Field Measurements of SCS in Australia

Converting cropland to permanent pasture is one of the most promising, eligible changes in land management under the ERF (see Box 1). For example, Badgery et al. ^[25][31] reported on trials on farms in the Cowra Trough, central-west NSW (rainfall 673 mm). Farms were selected on the basis of the soil C increase predicted from a Soil Carbon Calculation Tool ^[26][32] when the farmers changed their management in accordance with ERF requirements. Soil C stock was measured in 2012 according to the ERF protocol (baseline sampling) and again in 2017. Table 23 gives the results for five farms where the management change was from cropping to pasture without organic amendments.

2. The Potential for Increasing Soil C Sequestration

Projections for the Australian Landscape

For an ERF soil C project, operating an approved management practice, the critical issue is by how much can the rate of C inputs be increased relative to the rate of C losses. The main factors governing these input and output processes have been discussed by many authors ^{[6][7][9][11][12][13]}[6,7,9,12,16,19]. In the lead-up to COP26, the Australian government’s Long-Term Emissions Reduction Plan ^[14][20] (p. 55) identified soil C as one of the key low-emission strategies for attaining net zero by 2050. Soil carbon sequestration was envisaged as a mechanism by which emissions from industry that were hard to reduce could be offset by SCS that provided genuine net abatement. Soil C projects were estimated to have the potential to provide at least 17 Mt CO₂-e of accredited offsets annually by 2050, in addition to CO₂ drawn from the atmosphere without accreditation. As modelling for the Plan acknowledges ^[15][21] (p. 79), there is a wide range of estimates for SCS in Australian farmland, depending on assumptions about the effects of biophysical and environmental factors over time, uptake rates by farmers and the costs relative to the benefits (see Costs and Benefits below). For example, with advanced technology (unspecified) and an abatement incentive of AUD80 per t CO₂-e, SCS in Australian farmland was projected to account for 26 Mt CO₂-e annually to 2050. Previously, the first Low Emissions Technology Statement (LETS) ^[16][22] (p. 23) referred to a Commonwealth Scientific and Industrial Research Organization (CSIRO) review ^[17][23] that noted the potential for 35–90 Mt CO₂-e per annum to be drawn down from the atmosphere through improved management of one quarter of Australia’s crop and grazing lands. Estimates of SCS made by some commercial aggregators are considerably higher. For example, Agriprove’s analysis, quoted in the Plan (agriprove.io), indicated that the potential across 36.58 Mha of cropping land and 28.95 Mha of grazing land (not including rangelands receiving <300 mm rainfall) could be at least 103 Mt CO₂-e annually ^[14][20] (p. 56). In an Australian Broadcasting Corporation Science Show of 19 September 2020 ^[18][24], Matthew Warnken of Agriprove stated that some 30 Mha of pasture land would be suitable for “proving” the levels of SOC, delivering approximately 130 Mt of abatement each year.

3. Field Measurements of Soil Carbon Sequestration

3.1. Technical and Financial Considerations

In the field, soil C content varies both spatially and temporally, which creates difficulties for measurement. Once an area of land is delineated (the CEA), soil cores must be sampled to at least 0.3 m depth and the samples analyzed for the organic C concentration. At the same time, soil bulk densities must be measured so that the mean C content per unit volume of equivalent soil mass (the C stock) can be calculated (gravel must be excluded). This is the baseline sampling round. A second round of soil sampling for analysis must be undertaken within five years so that the change in soil C stock can be estimated. Smith et al. [7] suggested that, under some land managements, sampling to more than 0.3 m may be necessary to accurately measure C change in the soil profile. For example, under no-till farming, a decrease in soil C at depth may counterbalance an increase in soil C within the top 0.3 m ^[19][25]. Although the Food and Agriculture Organization has recommended sampling to 1 m [7], this requires specialized equipment and makes the measurement of soil C change prohibitively expensive [10]. The effect of spatial variability on the precision of each soil C mean can be reduced by increasing the number of samples taken in the CEA. For example, for a 50-ha field, Oldfield et al. [10] calculated the number of independent samples needed to estimate with 95% certainty a change of 0.05% in mean soil C concentration over 5 years (corresponding to a sequestration rate of 0.3 t C/ha/year to 0.3 m depth in a soil of bulk density 1 Mg/m