A generic methodological framework for accurately quantifying greenhouse gas footprints of crop cultivation systems

ZHENG Xunhu nd HAN ShenghuiStte Key Lortory of Atmospheric Boundry Lyer Physics nd Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Acdemy of Sciences, Beijing, Chin; College of Erth Science, University of Chinese Acdemy of Sciences, Beijing, Chin

1. Necessity for accurate quantification of crop production GHG footprints

Crop production provides not only staple foods directly for human beings, but also concentrated feeds for livestock, poultry and aquaculture fisheries, and even supplies energy as biofuels. Due to both population growth and increased consumption of animal proteins, the global food demand is likely to double by 2050 against a backdrop of a changing climate and growing competition for land, water, labor, and energy (Chen et al. 2014). In crop cultivation, greenhouse gases (GHGs) are emitted in large quantities to the atmosphere, directly and/or indirectly in some situations, e.g. emissions of methane (CH4) from paddy rice fields and nitrous oxide (N2O) from fertilized croplands. Meanwhile, they are taken up by croplands in other circumstances, e.g. CH4sinks occurring in dry areas and carbon sinks due to the application of organic manure or incorporation of crop residues. Globally, agriculture currently accounts for about 13% of the net anthropogenic GHG emissions (IPCC 2014), which is widely accepted as having contributed significantly to global warming (IPCC 2013). It is clear that humans are facing unprecedented challenges, including the mitigation of, or adaption to,climate change, and agricultural development to meet the increasing worldwide demand for food. The recently agreed goal for mitigating global warming is to keep the temperature rise well below 2 °C on average by the end of this century, as compared to the pre-industrial level (Paris Agreement to United Nations Framework Convention on Climate Change). Among the efforts to meet this goal, an important one is to reduce direct and/or indirect anthropogenic GHG emissions from agricultural food production,which is recognized as a considerable net source with concomitant opportunities for mitigation (IPCC 2014).For this purpose, so-called ‘climate-smart’ agriculture has been proposed as a potentially effective measure to manage these unprecedented challenges (e.g. Grassini and Cassman 2012). This approach seeks to increase or sustain crop productivity while reducing GHG footprints(e.g. Grassini and Cassman 2012; Van Ittersum et al. 2016).‘Climate-smart’ crop cultivation should be especially characterized by low GHG footprints, low synthetic nitrogen consumptions and high yields at the same time (Figure 1).

In the ENVIFOOD Protocol (version 1.0) Food Safety Code released by the European Commission in 2013, GHG footprint is listed first among the 14 indicators for the assessment of environmental and other impacts of food production (Food SCP RT 2013). It is used specifically as an indicator to measure the climate change impact potential(CCIP) exerted by the formation of agricultural product(s)on global warming; namely, the higher a GHG footprint, the larger the CCIP (and vice versa). De Camillis and Goralczyk(2013) even proposed an application of GHG footprints for a fiscal framework to mitigate climate change in food production and consumption. In their framework, they introduce a green value added tax (VAT) rate in terms of GHG mitigation. A green VAT rate of an agriculture-oriented good is a VAT rate corrected by the GHG footprint value due to the formation of the product (Equation (1)).Wide application of this fiscal tool is expected to drastically change the current patterns of food consumption and production towards a product life cycle oriented economy.

Figure 1. A ‘climate-smart’ crop cultivation system is characterized by low greenhouse gas (GHG) footprints, low consumptions of synthetic nitrogen (i.e. new Nr), and high crop yields at the same time.

where VATn(dimensionless) denotes the green VAT of a given food product in thenth year; VAT (dimensionless)is the standard VAT regardless of the GHG footprint for the product; andIn–1andInare the GHG footprints during the formation of the products in thenth?1 andnth years,respectively.

Either the significant contribution of agriculture to anthropogenic GHG emissions or the suggested application of GHG footprints in the determination of green VAT rates suggests a considerable need to accurately quantify or predict GHG footprints for different cropping regimes under various conditions such as climate, soil attributes,and agricultural management practices.

2. Terminology and scope of GHG footprint concept

In some (but not all) circumstances, the meanings of some other terminologies are equivalent to that of GHG footprint, such as carbon footprint (e.g. Gan et al. 2014; Hillier et al. 2009), yield-based GHG emission intensity (namely,yield-scaled life-cycle net GHG emission) for a cropping system (e.g. Chen et al. 2014; Grassini and Cassman 2012)or gross domestic product (GDP)-based GHG emission intensity (namely, GDP-scaled net GHG emission) of an economic entity (e.g. Baumert, Herzog, and Pershing 2005).

As one among a range of ecological/environmental footprints, the GHG footprint is an indicator that specifically measures the impact of agricultural production on climate change (e.g. Biesbroek et al. 2014; Chobtang et al.2017; Food SCP RT 2013). Because the total net GHG emission is quantified by summing up the net emissions of carbon dioxide (CO2), CH4, and N2O in CO2mass equivalents at a given time horizon (CO2eq; hereinafter referred to as the CO2equivalents for the 100-year time horizon, if not specified), a GHG footprint is also referred to as a carbon footprint in some studies (e.g. Gan et al. 2014; Hillier et al. 2009).The so-called GHG emission intensities presented by some researchers (e.g. Chen et al. 2014; Grassini and Cassman 2012) can also be referred to as the GHG footprints, as they are quantified by fully including the emissions and uptakes of all categories within an entire life cycle for the formation of a product or from all sectors of anthropogenic activity of an economic entity. A GHG emission intensity is expressed as the yield-scaled life-cycle net GHG emission (in CO2eq)from a production system or GDP-scaled net GHG emission(in CO2eq) from all anthropogenic activity sectors of an economic entity (e.g. a country, region, or the world). A GDP-scaled GHG emission intensity can be estimated for a regional/national economic entity or the whole world following the IPCC’s GHG inventory methodologies (e.g. IPCC 1997, 2000, 2006). It is usually included by a country in its National Inventories or Communications submitted to the secretariat of the United Nations Framework Convention on Climate Change (http://unfccc.int/national_reports/).Many other researchers, however, have reported signifi-cantly different GHG emission intensities estimated using the net emissions within only a portion of a life cycle for the formation of a product. In such a circumstance, the GHG emission intensity is obviously different from a GHG footprint or carbon footprint.

This paper focuses on addressing the production-based GHG footprints for crop cultivation systems of farms of different scales, cultivation enterprises of different sizes, and crop industries at sub-regional, regional, national, or larger spatial scales, but excludes the GDP-scaled GHG emission intensity or footprint of an economic entity.

The GHG footprint of a crop cultivation system at the farm or enterprise scale is defined as the crop production-scaled net GHG emission during the ‘cradle-to-gate’life cycle (hereinafter referring to the same if not speci-fied as the ‘cradle-to-grave’ life cycle) for a given cropping regime (e.g. Hoefnagels, Smeets, and Faaij 2010), calculated by Equation (2):

whereIdenotes the GHG footprint of a crop cultivation system (Mg or kg CO2eq per unit mass or energy of product(s)) during a particular time period and within the system boundary (Figure 2); and GHGtotaland CP represent the life-cycle net GHG emission (Mg or kg CO2eq) and the crop production mass (Mg or kg matter) or energy (TJ or GJ), respectively, during the same period and within the same boundary.

The products, i.e. the harvested plant organ(s), can be the grains of cereal or oil crops, the stem or root tubers of tuber crops, the melons or fruits of vegetable crops, and so on. The CP of a cultivation system can be expressed in quantities of mass or energy. To quantify the GHG footprint of a cultivation system involving a single crop, the CP is usually measured by the quantity of fresh mass under the crop-specific standard moisture content, in units such as kg or Mg (103and 106g, respectively). For a regime with multiple crops, such as a wheat–onion rotation with both crops being harvested within one year, the products are often not comparable in terms of fresh mass among the plant species. As such, the CP of a cultivation system with multiple crops can be measured by the quantity of energy of all the crops, in units such as MJ, GJ, or TJ (106,109, and 1012J, respectively), while the weight fractions of the energy produced are determined for individual crops.Finally, the mass-based GHG footprints for the products of individual crops in a multi-crop system can then be determined using their energy weight fractions and corresponding energy-to-mass conversion coefficients.

Following the principles of life cycle assessment for environmental management and its framework under the international standard (ISO 14040 2006), the GHGtotalof a crop cultivation system is quantified by fully considering both direct and indirect net emissions of CO2, CH4, and N2O during the whole life cycle. In light of the standard term for a life cycle (ISO 14040 2006), a grazing-free crop cultivation system (the same hereinafter if not specified) is defined as a given farmland area with the plants growing in it, the agricultural inputs to it and the on-farm operations relevant to the formation of the crop product(s). The boundary of such a crop cultivation system starts at the production of the agricultural inputs and ends when the crop products are ready for local consumption as food,feed, industrial materials, and/or other uses on the farm or for export out of the farm for consumption or trade markets elsewhere (Figure 2). The GHGtotalcomprises onand off-farm net emissions (Equation (2)). Both the CP and GHGtotalare quantified at seasonal, annual or decadal time scales depending on the issues to be addressed with the GHG footprints.

The on-farm anthropogenic GHG emissions from a crop cultivation system (GHGon) can be divided into six categories (Figure 2; Table S1a):

(1) The net biogenic CO2emissions of the soil–plant consortium (?ΔOC). This is the opposite of the ΔOC and usually approximates to the multi-year average of the annual reductions in the ecosystem carbon stocks that include the organic carbon in the standing biomass and in the soil. For a cropland free from woody perennials, the opposite of the ΔOC often approximates to the multi-year average of the annual reductions in the soil organic carbon (SOC) stock (?ΔSOC), since the interannual reductions or increases in the carbon stock of the standing biomass are usually too small and thus can be neglected (e.g. IPCC 2006).

(2) The net biogenic CH4emissions (CH4b). This is regarded as the residual of subtracting the CH4uptake (CH4bu) by upland soils from the CH4emissions from managed wetlands (CH4be), such as paddy rice fields.

Figure 2. The boundary of a grazing-free crop production system (confined by the dashed-line frame) in terms of a ‘cradle-to-gate’ life cycle for greenhouse gas (GHG) emissions.

(3) The direct and indirect biogenic N2O emissions(N2Ob). As components of the on-farm N2O emissions (N2Oon), this is composed of (i) the overall direct N2O emissions due to intentional nitrogen inputs to the croplands, not only in the current year (N2Obdc) but also previous years (N2Obdp);and (ii) the indirect N2O emissions occurring over the farmland (N2Obi) due to unintentional inputs of anthropogenic nitrogen via irrigation or atmospheric deposition. The intentional nitrogen inputs are due to application of synthetic nitrogen fertilizers and organic manure and retention/incorporation of crop residues.

(4) The CH4and N2O emissions from on-farm biomass burning (CH4bband N2Obb, respectively).These are abiotic emissions occurring as part of biomass burning in on-farm management of crop residues.

(5) The non-biogenic emissions of CO2, CH4, and N2O in association with fossil-fuel energy consumption by field machinery operations(GHGono). These denote the GHG emissions from the combustion of fossil fuels (mainly diesel)during the driving of on-farm machinery operations, including tillage and soil preparation,irrigation, crop planting and harvesting, fertilizer and manure application, agrochemical spraying/spreading, and so on. The GHGonoonly counts the emissions from direct on-farm combustion of fossil-fuel energy (usually diesel) that occurs during the operation of most machinery.The GHG emissions associated with the generation of electricity consumed occasionally to drive on-farm machinery are excluded. Instead,they are regarded as being part of an off-farm subcategory (Figure 2).

(6) The net abiotic CO2emissions from chemical reactions of applied calcic lime, dolomite, and urea in soils (IC).

Table S1a describes each category and/or sub-category of the on-farm GHG emissions.

The off-farm components (GHGof) associated with crop production on farmland can be divided into two categories (Figure 2; Table S1b):

(1) The off-farm indirect emissions of N2O from the anthropogenic nitrogen applied for the cultivation of crops (N2Oofbi). This originates from the runoff/leaching (N2Oofbil) and gaseous nitrogen loss (N2Oofbig) of applied fertilizer nitrogen and biologically occurs outside the farmland.

(2) The off-farm indirect emissions (in the form of CO2, CH4, and N2O) due to the manufacturing and storage of agricultural input materials and/or the generation of electricity, as well as their transportation to the farm (GHGofi). Agricultural input materials to a crop cultivation system include mineral/synthetic fertilizers and/or organic manure(providing nitrogen, phosphorous, potassium,and/or other nutrients for crop growth), agricultural chemicals/pesticides (herbicides, insecticides, fungicides, nematicides, germicides, and/or bactericides), electricity (providing power for on-farm machinery), and others (seeds, agricultural films, lime, etc.). Thus, the GHGofiis composed of the emissions due to the production, storage and transportation of synthetic fertilizers (GHGofis),organic manure (GHGofim), agricultural chemicals/pesticides (GHGofic), electricity (GHGofie), and other input materials (GHGofio). The GHGofieis composed of the off-farm emissions generated during the mining and transportation of fossil fuels to power plants, as well as the combustion of fossil fuels at power plants to generate electricity, and in building the power grid infrastructure (ISO 14040 2006).

Table S1b describes each category of the off-farm GHG emissions, which exclude those related to the mining, storage, and transportation of fossil fuels to a farm because they are already included in the on-farm emissions related to the energy consumption involved in driving and operating field machinery (Table S2).

3. Generic method to quantify the GHG footprint of a cropping system

3.1. Generic methodological framework

To determine the GHG footprint of a crop cultivation system (Equation (2); Figure 2), quantification of both the CP and GHGtotalis required.

The CP can be quantified by field measurement, survey,or model simulation. Specifically, it is determined by the mass weight of the product’s fresh matter at a standard moisture content (e.g. 15% for maize, 14% for wheat, or 7.5% for rapeseed; Williams, Audsley, and Sandars 2010), or by the energy quantity of the product(s). The mass weight can be converted to an energy quantity using a factor of 17 MJ kg?1oven-dry matter for either grain or non-grain crop products (Grassini and Cassman 2012), together with the moisture content.

The GHGtotalis composed of the net emissions occurring both on- and off-farm (Figure 2; Equation (3)). The on-farm net emissions are determined by Equations (4–17) and the off-farm emissions by Equations (18–27):

Detailed descriptions of the framework equations(Equations (3–27)) are provided in the online supplementary material (Text S1).

As Equations (5–7), (9–11), (14–17), (20–24), and (26–27)formulate, the net emission quantity of the on- or off-farm sub-category, in mass weight of CO2eq or in the carbon or nitrogen weight of a GHG species, is calculated by multiplying the activity level with an emission factor (EF) or the opposite of an uptake factor (UF). For an investigated crop cultivation system, activity levels are surveyed and then converted to quantities with dimensions in correspondence with the EFs’ or UFs’ bases. The base of an EF or UF, i.e.per activity level unit, is per acreage unit of the farmland or per mass, volume, or energy unit of inputs to the farm. An EF denotes the sum of CO2, CH4, and/or N2O, in CO2eq mass weight, or the carbon or nitrogen mass weight of one of the GHGs emitted by per activity level unit, such as those listed in Tables S2 and S3. An UF denotes the carbon mass weight of CO2or CH4taken up by per activity level unit. In principle, an EF or UF may be valuated with different numbers of different dimensions, which can be converted between each other if conversion factors are provided. In a specific case, an EF or UF value may be self-determined via direct measurement, model simulation, or other approaches. It may also be valuated with citation from the literature or with the default recommended by the IPCC guidelines of national GHG inventories (e.g. IPCC 2006). In any case to be reported, it is necessary to describe how an EF or UF value has been determined.

3.2. Activity levels and EFs of the on-farm categories

3.2.1. GHG emissions from on-farm machinery operations

Equation (5) provides the method to quantify GHG emissions due to on-farm fossil-fuel consumption, annually or seasonally, to drive and operate machinery (MO). The MO and their EF, both with a subscripti1, denote the annual or seasonal total quantity of a fossil-fuel species (kg or L)consumed by, or the total land area (ha) being applied with, thei1th machinery operations (tillage, irrigation,fertilizer application, pesticide spraying, harvesting etc.),and the EF as the sum of CO2, CH4, and/or N2O in CO2eq(kg CO2eq kg?1, L?1, or ha?1), respectively. The EF value for a type of machinery operation represents the total emission during the full ‘cradle-to-grave’ life cycle. It integrates both the indirect and direct GHG emissions per activity level unit. The indirect emissions occur off-farm during the mining, storage, and transportation of fossil fuels (mainly diesel, but occasionally coal, liquefied petroleum gas, or natural gas) to the farm. The direct emissions take place during the combustion of the fuels on-farm. Tables S2a–c exemplify the EFs of various on-farm machinery operations, which have been extracted from the literature and may be used in Equation (5) if no better EF value is available or self-determined for a specific operation.

3.2.2. On-farm net emissions of CO2

The changes in the ecosystem organic carbon stock, as well as the chemical reactions of applied limes, dolomites and urea, jointly contribute to the on-farm net emissions of CO2(Equations (4), (6), and (7)).

The net CO2emissions due to changes in ecosystem stock is given as the opposite of the ΔOC (Equation (4)), which counts the multi-year average of the annual changes in the ecosystem carbon stock of a crop cultivation system. As Equation (6) formulates, the ΔOC is calculated by multiplying the cropland area (A), in ha, by the UFC, which in turn denotes the area-based net CO2uptake factor (kg C ha?1yr?1).

Following the IPCC’s methodology (IPCC 2006), the UFCis determined by Equation (S1). For a cropland with woody perennials, such as orchards, vineyards, and agroforestry systems, the UFCis composed of significant changes in carbon stores in both the long-lived biomass and the soil. The amounts of biomass carbon changes depend on land-use changes from other lands to croplands, species types and cultivars, densities, growth rates, and harvesting and pruning practices. For a cropland free from a woody perennial and remaining as cropland for multiple years, UFC≈ UFSOC,as there is no long-term carbon storage in the biomass of annual crops (IPCC 2006). For a cropland converted from other land uses, the changes in ecosystem carbon stocks due to land-use changes also need to be included (IPCC 2006). The UFSOCcan be variable in conjunction with soil properties and management practices, such as crop types,rotation regimes and methods of tillage, drainage, residues management, and organic amendments. Although either the UFCor UFSOCmay be valuated with literature-derived data if their applicability can be proven, they are usually specified by observation, survey, or model simulation,using the gain–loss or stock-difference methods recommended by IPCC (2006). In most cases, it is necessary to specify how either factor is valuated, because both are largely variable with different conditions (e.g. Bayer et al. 2016). As Cui et al. (2014a) illustrate, it takes at least 10 years to stabilize its annual change rates in the ecosystem carbon stocks of a maize–wheat cropland that is consecutively applied with a modified field management practice. The relatively stabilized carbon stock changes between the beginning and end of individual years fluctuate around the annual change rates of SOC, defined as the sum of the carbon stocks in resistant and stable humus, microbial biomass and dissolvable organic matter (Cui et al. 2014a). Assuming a long period is needed for an ecosystem to reach its carbon stock equilibrium after intensive disturbance, such as land-use change or significant changes to field management practices, IPCC(2006) recommend a default time interval of 20 years for counting the average annual change in the carbon stock for the remaining cropland. For quantifying the SOC stock to specify the UFCor UFSOCof a given land area, the recommended default soil depth (30 cm), or other self-specified soil depths, can be adopted (IPCC 2006), with declaration of the exact depth for any case.

3.2.3. On-farm net CH4 emissions

The on-farm net CH4emissions from an investigated crop cultivation system is the balance between the emissions from the managed wetlands and biomass burning and the uptakes by the upland soils (Equation (8)).

Both the wetland emissions and the dryland uptakes are due to biogenic processes and rely on the land areas and the area-based factors of emissions or uptakes (Equations(9) and (11)). Good practice to determine either factor of a specific cultivation system is direct measurement or computation with model(s). In principle, both approaches can successfully relate variable CH4fluxes at different spatial and/or temporal scales with direct driving forces, such as substrate availability, soil redox potential, temperature, and salinity to provide a reasonable EFw(e.g. kg C ha?1yr?1) of a managed wetland (Equation (9)); and the soil permeability, moisture, temperature and ammonium concentration to give a reasonable UFM(e.g. kg C ha?1yr?1)of a dryland area (Equation (11)). The direct driving factors are usually determined by the primary ecological factors as model inputs, including climatic/meteorological conditions, soil properties, vegetation characteristics, land-use legacies, and various agricultural management practices(e.g. Li 2016).

With respect to CH4emissions from paddy rice fields with double cropping (i.e. two rice harvests annually)in a subtropical region, Shao et al. (2017) found signifi-cant but variable increases in annually accumulated CH4emissions (i.e. EFw) in the early years following conversion from a long-term continuous land use for growing upland plants. Meanwhile, in contrast, observations showed relatively stable EFwfor old paddy rice fields (consecutive rice cultivation for 10 years or longer). The experimental data reported in their study provide a hyperbolic equation (Equation (S2); Figure S1). If Equation (S2), as well as its current or further revised parameters, can be proven applicable, the adaptation factor derived from it can be applied to determine the EFwof an investigated managed wetland at a non-equilibrium stage, using Equation (S3).

With regards to CH4emissions in biomass burning as a practice of crop residues management (Equation (10)), the dry weight of burned residues and its carbon content are surveyed, while the recommended default EFmbvalue, e.g.0.9%–1.5% with a mean of 1.2% for CH4–C from the burnt biomass-carbon (IPCC 1997) may be used. The EFmb(kg CH4–C Mg?1dry matter) values for savanna burning correlate negatively with combustion efficiency, i.e. Equation(S4). This supplementary equation is an empirical function adapted from IPCC (2000), and may be used to provide the EFmbvalues if combustion efficiency values can be obtained for an investigated case. An EFmbvalue can also be determined as the product of the CO2EF and the CH4-to-CO2molar ratio of the fume gas from biomass combustion, if they are known (IPCC 2000) via direct measurement or from the literature.

3.2.4. On-farm N2O emissions

Aside from the aforementioned N2O emissions during fossil-fuel combustion to drive and operate machinery,which is included in the GHGonosub-category, the other N2O emissions occurring on-farm are mainly attributable to the biogenic production in microbial processes of nitrogen transformation, such as nitrification and denitrification, and the abiotic process of biomass burning in crop residues management (Equation (12)). As Equation (13) formulates, the biogenic sub-category (N2Ot) is composed of direct emissions due to intentional nitrogen inputs in the current year (N2Obdc), residual nitrogen from the fertilizers applied in previous years (N2Obdp), and indirect emissions due to unintentional nitrogen inputs (N2Obif), which are determined separately by Equations (14–16), respectively.It is noticeable that the N2Otsub-category also includes some abiotic processes producing N2O, such as chemodenitrification, since they are not usually separately observed in practice.

在技術(shù)可行性方面，醫(yī)院只要裝有普通的電腦即可不需要什么復(fù)雜的設(shè)備和機(jī)器。在經(jīng)濟(jì)上，社區(qū)門(mén)診服務(wù)系統(tǒng)肯定比大型的開(kāi)發(fā)成本和維護(hù)費(fèi)用較小，系統(tǒng)性價(jià)比很高。從使用目的來(lái)分析，社區(qū)門(mén)診服務(wù)系統(tǒng)是針對(duì)小醫(yī)院應(yīng)用開(kāi)發(fā)，功能簡(jiǎn)潔，操作方便，非常符合小型醫(yī)院的使用要求。因此社區(qū)門(mén)診服務(wù)系統(tǒng)應(yīng)該是可行的。

To determine the direct emissions due to intentional nitrogen inputs in the current year (Equation (14)), the amounts (Nk) of nitrogen input via applications of synthetic fertilizer, organic manure, and retention/incorporation of crop residues are surveyed or recorded in crop cultivation.A default or literature-derived value can be applied for the EFk(kg N2O–N kg?1N) of a specific case, but only when its applicability can be proven. Since the EFkvalues usually rely strongly on soil properties (such as texture, pH, SOC),meteorological conditions, specifications of fertilization(types and application time and method) and other field management practices (e.g. irrigation, tillage, and liming),the default values recommended for the globe or for a region are not usually appropriate for a specific case of an investigated farmland. Thus, it is good practice to determine the EFkof a specific case by direct measurement or model simulation using well-adapted process-oriented model(s). Simple empirical functions may also be developed and applied to determine the EFkof given cases, such as those reported by Rochette et al. (2008) and Zhang, Gu,and Zheng (2015), if their applicability can be proven for the specific cases of an investigated cultivation system.

Van Groenigen et al. (2010) developed an empirical function (Equation (S5)) to relate the accumulated N2O emissions (N2Ot), annually or in the growing season, with the nitrogen surplus (Nsurplus). If Equation (S5) is proven to be applicable for the investigated crop cultivation system,the N2Otcan be estimated provided both the intentional nitrogen input and the accumulated nitrogen in the aboveground biomass at maturity (Naccum) are given via directly recording/surveying the doses of nitrogen inputs,and direct measurement or model estimation of the nitrogen taken up by the aboveground biomass at physiological maturity (ADM). For instance, theNaccumfor maize may be estimated by Equation (S6) (adapted from Grassini and Cassman 2012) with a given ADM value, if this supplementary equation is proven applicable for the specific case.

If Equation (S5) is proven applicable for directly estimating N2Otfor an investigated crop cultivation system,the steps to separately estimate the N2Obdcand N2Obdp(Equations (14) and (15)) would be skipped. Otherwise, the N2Obdc, N2Obdp, and N2Obifmust be separately estimated.It is noticeable that the concepts for the residual nitrogen of intentional inputs (Nrs) and theNsurplusare different from one another. Quantitatively, theNrsis lower than theNsurplus,since the latter includes the nitrogen remaining in crop roots and that lost via gaseous nitrogen emissions, as well as leaching and/or runoff, whereas the former does not.Irrigation, atmospheric deposition and biological fixation are the main contributors to the quantity (Nuii) of unintentional nitrogen inputs. For a specific cultivation system, theNuiimay be estimated with direct measurements, surveys,or reference to the literature. When the N2Otis quantified either by Equation (S5) or via direct measurement, the N2Obifhas to be estimated separately, since a portion of it should be subtracted from the total indirect N2O emissions in order to avoid double-counting in the estimation of the indirect emissions occurring off-farm (Equation (25)). It is good practice to provide the EFrsand EFuii(kg N2O–N kg?1N for both EFs) values with virtual experiments using welladapted process-oriented model(s). Both EFs may also be provided via direct measurement, for which unfertilized treatments with different time intervals from the last fertilization events can be simultaneously observed. If no data from real experiments or model-based virtual experiments are available for both EFs, the EFrsmay roughly take the average of the EFkvalues, while the EFuiimay roughly take the sum of the IPCC default EF values (e.g. IPCC 2006)for leached and gaseous losses of fertilizer nitrogen (see Section 3.3.2).

When hourly, daily or sub-weekly N2O fluxes from a cropland are measured, they can be integrated to obtain a seasonally or annually accumulated flux. Multiplying this accumulated flux by the investigated cropland area can directly result in the N2Ot. It is good practice to determine the N2Otwith field measurements. In order to obtain reliable N2Otamounts from field observations using the widely applied static chamber and gas chromatograph techniques, the following protocol is required: the N2O flux measurements require (i) high frequencies; (ii) enough random replicates spatially to represent each field treatment well; (iii) an area of at least 0.2 m2with plants covered by a chamber for a spatial replicate if the land is cultivated with crops shorter than 1.5 m; (iv) minimal mechanical disturbance within and near the locations installed with the chamber bases; (v) use of air pressure balance; (vi) generation of air turbulence within the enclosure using a fan if the density of plants is high; and (vii) at least five N2O concentrations observed during a chamber enclosure to enable the calculation of the flux via a nonlinear approach (Wang et al. 2013; Zheng and Wang 2017). In terms of achieving a high observational frequency in manual measurements,good practice is once per day in the morning (0900–1100 local time) during the three to four weeks following each fertilization event and the one week following each rainfall, irrigation or tillage occurrence; and once every three to four days (half-weekly) at other times of the year. This protocol requires at least four spatial replicates for a cropland with non-row cultivation. For row cultivation, the requirement is at least three replicates to represent the crop rows and at least three other replicates to represent the inter-row spaces. The area weights of the row and inter-row spaces are used to scale up the observed fluxes to represent the entire cropland well, while the errors for the weighted mean fluxes are given by propagating the measurement errors for both row and inter-row spaces,using the general algorithm for error propagation in the random theory.

When N2Otis given on the basis of field measurements,it in fact includes the emissions from both natural and anthropogenic nitrogen substrates. As the emissions from natural nitrogen substrates are usually much smaller than those from anthropogenic nitrogen substrates, by about one order of magnitude, these are regarded as neglectable in the methodological framework presented in this paper.

For the N2O emissions from biomass burning as a practice of crop residues management (Equation (17)), the dry weight of burned residues and its nitrogen content are surveyed, while the recommended default EFnbvalue,e.g. 0.5%–0.9% with a mean of 0.7% for N2O–N from the burnt biomass-nitrogen (IPCC 1997) may be used. A casespecific EFnbvalue, instead of the recommended default,may be employed if it is available. Assuming an EF of 1700 kg CO2Mg?1dry matter, the EFnbvalue of burnt residues may be specified if the nitrogen-to-carbon molar ratio of the biomass or the N2O-to-CO2molar ratio of the fume gas from residues combustion are known (IPCC 2000)via direct measurements or by reference to the literature.

3.3. Activity levels and EFs of the off-farm categories

3.3.1. Off-farm GHG emissions due to agricultural inputs

The agricultural inputs to a crop cultivation system are grouped into five categories, which are: (i) synthetic fertilizers for different nutrients of essential nutrients; (ii)organic manure; (iii) agricultural chemicals; (iv) electricity; and (v) the others (Equation (19)). The off-farm GHG emissions subtotal for each sub-category of agricultural inputs integrates the released CO2, CH4, and/or N2O during the processes of manufacturing, storing, and transporting the matter or electric power to the farm, and is expressed in CO2eq mass. It is estimated by multiplying the matter or electricity amount consumed during crop cultivation by its corresponding EF (Equations (20–24)). The consumed amounts of an input materials or consumed electricity are surveyed or recorded in agricultural operations. Their EF values can be self-estimated or taken/adapted from the literature, such as those listed in Table S3. Tables S3a–d exemplify the literature-derived off-farm EF values of (i)fertilizers with single essential nutrients, (ii) compound fertilizes and manure, (iii) agricultural chemicals, and (iv) other inputs, respectively. For a specific case, the EF values listed in Table S3 may be used if no better values are available or can be self-determined. The EF values provided in Table S3 need to be further updated or revised when better or new data become available in the literature. Electricity here is not broken down to different types, since grid electricity is usually used. Thus, only the literature-derived EF (i.e. EFe) of grid electricity is exemplified in Table S3d. The higher EFevalue for the grid electricity of China (Table S3d) is mainly due to the larger fraction of coal-based electric power as compared to western developed countries. For cases using non-grid electric power generated by one energy form alone or by a combination of more than one energy form(such as coal, hydraulic power, nuclear power, natural gas,diesel, solar radiation, wind, tide, geothermal energy, and biofuels), specific values instead of those for grid electricity need to be determined and used for EFe.

3.3.2. Off-farm N2O emissions

Off-farm N2O emissions exclude those occurring during the processes of manufacturing, storing, and transporting the matter or electric power to the farm (Equations(19–24)). They are defined as the indirect emissions from intentionally amended nitrogen but occurring outside the croplands of an investigated crop cultivation system.They are composed of the indirect emissions due to (i)hydrological losses of applied nitrogen or mineralized soil organic nitrogen via leaching and/or runoff, and (ii) gaseous losses of applied nitrogen (Equations (25–27)). The authors assume that approximately 15% of unintentional nitrogen input originates from the nitrogen losses of intentional inputs to the croplands of a crop cultivation system.The indirect emissions occurring on-farm in association with this assumed nitrogen portion is subtracted from the total indirect N2O emissions due to the amended nitrogen and/or soil organic nitrogen losses by leaching/runoff and gaseous releases (Equation (25)) to avoid double-counting.

It is good practice to employ a well-adapted model to simulate, or use field-measured fluxes to estimate, the nitrogen quantities of both hydrological and gaseous removals from the intentionally amended nitrogen pools.IPCC (2006) defines a fraction of all the nitrogen added to,and/or the soil organic nitrogen mineralized in, managed soils for hydrological losses (fleach). It provides a default mean value of 30% for the range of 10% to 80% at the 95%confidence interval (CI) for conditions exceeding the soil water holding capacity as a result of rain or non-drip irrigation. Otherwise,fleachis regarded as zero (IPCC 2006). IPCC(2006) also defines a fraction of gaseous nitrogen losses from all added nitrogen due to nitric oxide emissions and ammonia volatilizations (fgas), and recommends default mean values of 10% (ranging from 3% to 30% at the 95%CI) for the nitrogen in synthetic fertilizers and 20% (ranging from 5% to 50% at the 95% CI) for the nitrogen in organic manure. If the IPCC defaults forfleachorfgasare proven applicable, they can be used to estimate the nitrogen losses as substrates for off-farm indirect N2O emissions.

Usually, it is difficult to directly measure or simulate the EFig(Equation (26)) and EFig(Equation (27)) of a specific case because of the high complexity of the processes involved. Thus, the default values provided most recently by the IPCC as worldwide means are usually used, such as those recommended by IPCC (2006), which are 0.75%(ranging from 0.05% to 2.5% at the 95% CI) for the EFiland 1.0% (ranging from 0.2% to 5.0% at the 95% CI) for the EFig.

4. Advances in the quantification of GHG gas footprints for crop cultivation

4.1. GHG footprints with involvement of carbon stock changes

There are few reports on GHG footprints for croplands with woody perennials. For crop production systems free from woody crops, several studies have instead addressed GHG footprints at site, regional or national scales (e.g. Bayer et al. 2016; Cheng et al. 2015; Grassini and Cassman 2012; Ha et al. 2015; Lai et al. 2016; Li et al. 2016; Wang et al. 2015;Yan et al. 2015; Yang et al. 2014). However, few of the studies on GHG footprints to date have completely taken into account all the categories of net GHG emissions involved in Equations (4–27). Previous studies can be categorized into two groups by involving or not involving organic carbon changes. That is, one involves but the other does not involve the ΔSOC in the quantification of the life-cycle net GHG emission from an investigated cultivation system without woody crop. The first group is reviewed below in this section, and the other in the next (Section 4.2).

The ΔOC of a cropland cultivated with woody perennial(s) or the ΔSOC of a cropland free from woody crops represents the net source or sink fluxes of the atmospheric CO2(Chapin et al. 2006; IPCC 2006). It plays an essential role in determining the quantity of the life-cycle net GHG emission (Zhang et al. 2014) and thus the GHG footprint(Equations (2–4)). So far, however, only a very few studies have involved the ΔSOC in calculating the GHG footprints of crop cultivation systems free from woody perennial.Of these studies, three were implemented in the North China Plain (NCP), which has a typical semi-humid temperate climate (Gao et al. 2015; Yang et al. 2014; Zhang et al. 2016), and one was performed in southern Brazil, which has a typical humid tropical climate (Bayer et al. 2016).In these studies, the researchers investigated site-scale GHG footprints as responses to changes in crop rotation regimes, water and nitrogen management practices, crop residues treatments, and/or tillage methods. In addition to direct measurements of the ΔSOC commonly performed in long-term experimental plots, there were observable differences among these studies.

Bayer et al. (2016), Gao et al. (2015), and Zhang et al.(2016) directly measured both CH4and N2O fluxes fromfield plots for long-term experiments. However, they all neglected the indirect N2O emissions occurring off-farm and thus underestimated the total net GHG emissions and, therefore, the GHG footprints. Yang et al. (2014) did not directly measure any GHG fluxes except for the ΔSOC.Meanwhile, they calculated all the on-farm direct biogenic N2O emissions using the mean EFs of the NCP region for most of the field treatments and the default EFs provided by IPCC (2006) for minor ones. They also used the default EFs of IPCC (2006) to estimate the indirect biogenic N2O emissions. At the same time, They neglected the on-farm CH4fluxes and the N2O emissions induced by the residues of amended nitrogen. Ignoring these GHG items might have influenced the estimated GHG footprints slightly,while directly employing the IPCC default EFs as worldwide averages or regionally averaged EFs, the site-scale values reported by them might have biased the real GHG footprints. In estimating the CO2eq emission quantities for the agricultural inputs and on-farm machinery operations, Bayer et al. (2016), Gao et al. (2015), and Yang et al. (2014) applied EFs referenced in the literature. Zhang et al. (2016), however, directly applied commercial software with default EFs values, for which the sources are not always transparent for readers.

All the differences mentioned above might have influenced the comparability in the GHG footprint values among these studies. Meanwhile, the absence of some GHG emission items, or the use of the EFs of the IPCC defaults or the regional averages, might have biased the reported site-scale footprints. Nevertheless, these limited studies have provided the most accurate site-scale GHG footprints due to involving the ΔSOC directly measured in long-term field experiments. Accordingly, these limited studies have presented some interesting and valuablefindings that are summarized below.

Gao et al. (2015) investigated the footprints of different crop rotation regimes in the NCP via observing the on-farm GHG fluxes and crop yields in four consecutive years. Among the investigated field treatments on cropping regimes and nitrogen and water addition rates, the measured ΔSOC varied from 0.1 to 0.6 Mg C ha?1yr?1and the quantified GHG footprints varied from 0.39 to 0.77 kg CO2eq kg?1. The results showed a 19% increase in crop yields and 40% reduction in GHG footprints due to optimizing nitrogen fertilization (i.e.a 40% reduction in nitrogen addition rates as compared to the farmer’s conventional practice) for the typical winter wheat–summer maize (W/M) rotation regime. In comparison with the W/M regime, the investigated alternatives– namely, the regimes of winter wheat–summer maize–spring maize (W/M-M), winter wheat–soybean–spring maize (W/S-M), and continuous spring maize (M) – significantly reduced the footprints by 30% to 50%. Meanwhile,the W/M-M regime maintained the grain yields, but the W/S-M and M-M regimes reduced the yields by 23% and 30%, respectively. Further taking into account the much less groundwater consumption as compared to the W/M regime (108–159 vs. 264 mm yr?1), Gao and his colleagues recommended the W/M-M regime as a substitution for the most common rotation regime of the NCP.

Instead of following the definition in Equation (2),Yang et al. (2014) defined the GHG footprint of a crop cultivation system as the total net GHG emission per unit cropping land area, accumulated biomass at maturity or economic production expressed in currency. They then compared the footprint values among five cropping regimes that can be adopted in the NCP, which are the rotation systems of sweet potato–cotton–sweet potato–winter wheat–summer maize (SP-C-SP-W/M), typical W/M, ryegrass–cotton–peanut–winter wheat–summer maize (RG/C-PN-W/M), consecutive cotton (C), and peanut–winter wheat–summer maize (PN-W/M). Their results demonstrated a descending order of the GHG footprints among the investigated cropping systems, as follows:W/M ＞ PN-W/M ＞ RG/C-PN-W/M ＞ C ＞ SP-C-SP-W/M.

Zhang et al. (2016) investigated the GHG footprints of the typical W/M rotation regime subject to conventional nitrogen and water management practices in the NCP,which yielded wheat and maize grains at 6.7–7.7 and 9.4–9.6 Mg ha?1yr?1, respectively, and had full crop residues retention for no till and incorporation for other tillage methods. They presented GHG footprint values of 0.286,0.364, 0.360, and 0.334 kg CO2eq kg?1yr?1on average for no till, rotary tillage, sub-soiling, and plow tillage, respectively. These results demonstrate a minor reduction effect from no till (?14% on average) but a marginal increasing effect from both rotary tillage and sub-soiling (about 8%on average), as compared to the conventional plow tillage(adapted from Zhang et al. 2016).

Bayer et al. (2016) compared the GHG footprints amongfive maize-based tropical cropping systems without fertilization but with different cover crops or tillage methods.As their results demonstrated, positive GHG footprints ranging from 0.24 to 0.85 kg CO2eq kg?1appeared in the cases of long-term conventional tillage along with oat or legumes as cover crops or long-term no till along with low crop residues retention and no legume cover crop.On the contrary, negative GHG footprints varying between?0.61 and ?0.22 kg CO2eq kg?1occurred in the cases of long-term no till along with legume cover crops. These negative footprints were due to significantly larger ΔSOC values as compared to the cases with positive footprints.The results also demonstrated much higher maize yields in the regimes with legume cover crops, as compared to the non-legume cover crops (～4.5 vs. 2.1–2.5 Mg ha?1yr?1).These findings imply great importance of a combination of legume cover crops and no till in tropical maize production free from fertilizer application.

4.2. GHG footprint not involving carbon stock changes

Most studies to date have neglected the ΔSOC in the quantification of GHG footprints for crop cultivation systems without woody perennial (e.g. Ali et al. 2017; Chen et al.2011, 2014; Cheng et al. 2015; Cui et al. 2014b; Fumagalli 2015; Grassini and Cassman 2012), despite the ΔSOC being essential for accurately quantifying a GHG footprint as a CCIP indicator.

The literature shows that discarding the ΔSOC always leads to positive GHG footprints for cultivation systems with annual crops. For instance, Ali et al. (2017) reported 0.16–0.44 kg CO2eq kg?1for rain-fed durum wheat under Mediterranean conditions; and Ha et al. (2015), Wu et al. (2014), and Yan et al. (2015) presented 0.17–1.62 kg CO2eq kg?1for maize-based crop cultivation systems.GHG footprints that discarded the ΔSOC of maize-based croplands showed strong positive correlations against nitrogen addition rates in some studies in Canada (Ma et al. 2012) and China (Wang et al. 2015), but significantly negative correlation in other studies in America (Grassini and Cassman 2012), China (e.g. Chen et al. 2011, 2014;Cui et al. 2013) and Italy (Fumagalli 2015). The authors of the literature indicated that the negative correlations provide opportunities to enhance crop yields by optimizing agricultural resources and inputs (e.g. Chen et al. 2011,2014; Grassini and Cassman 2012). Excluding the ΔSOC from the quantification of the life-cycle net GHG emissions, however, would have significantly biased the footprint values to variable extents. For instance, the positive footprints of a typical W/M regime subject to intensive water and nitrogen management practices in the NCP were overestimated by 15%–51% (adapted from Zhang et al. 2016). These most likely biased the footprint values demonstrating no difference at all among different tillage treatments. Such an absence of a difference in the GHG footprints among the tillage methods was contrary to the results derived from the inclusion of the ΔSOC (Zhang et al. 2016). The biases due to excluding the ΔSOC would be even more problematic for unfertilized maize-based cropping systems subject to a wet tropical climate, of which the significantly negative GHG footprints might be overestimated to largely positive ones, whereas largely positive footprints might be underestimated to much smaller positive values (adapted from Bayer et al. 2016). Therefore,the reliability of the above conclusions drawn from GHG footprints excluding the ΔSOC still need confirmation in further studies.

5. Summary and further study needs

The greenhouse gas (GHG) footprint of a crop cultivation system is defined as the life-cycle net GHG emission(in carbon dioxide equivalent over a given time (usually 100 years) horizon (CO2eq)) per unit mass or energy of a crop product. The GHG footprint is an index for indicating the climate change impact potential (CCIP), which is one of the potential environmental impacts exerted by the formation of a crop product. The larger a GHG footprint,the greater the CCIP (and vice versa). The importance of accurately quantifying the GHG footprints of crop cultivation systems is obvious, as GHG footprints are possibly involved in the determination of green value added tax (VAT) for crop-oriented food/drink products. As this review concludes, few studies have fully taken into account the direct and indirect contributors to the life-cycle net GHG emissions from investigated crop cultivation systems, and thus have most likely led to overestimation or underestimation for the GHG footprints. Few studies have addressed the GHG footprints of cultivation systems with woody crops. Most studies addressing the GHG footprints of cultivation systems with annual crops and without grazing have neglected the changes in soil organic carbon stocks, and thus most likely biased the footprints estimates. Meanwhile, there are few studies estimating the on-farm, biogenic GHG emissions using processoriented biogeochemical models. To solve these problems, in this paper the authors propose a generic methodological framework for accurately quantifying the GHG footprint of a crop cultivation system free from grazing. In addition to a brief description on how to measure or simulate some emission factors (EFs), possible values for the other EFs as key parameters of this methodological framework are extracted from the literature and provided for users’ convenience.

With regard to the generic methodological framework(Equations (2–27)) for the quantification of the GHG footprint of a crop cultivation system free from grazing, further study efforts are still needed, as follows:

(1) Produce more complete lists of worldwide on-farm machinery operations and agricultural inputs;

(2) Standardize activity-level metrics for individual components of GHG emissions/sequestrations;

(3) Determine EF defaults of non-biogenic GHG components, i.e. on-farm machinery operations,biomass burning, and agricultural inputs except fossil fuels used to drive and operate machinery,which are stratified by temporal/spatial scales and other variable conditions;

(4) Improve protocols for direct measurement of on-farm biogenic GHG components to determine their EFs;

(5) Develop process-oriented model(s) applicable for simulating EFs of on-farm biogenic GHGs under a wide range of temporal/spatial scales and other conditions; and

(6) Prepare protocols for simulation of the EFs of on-farm biogenic GHGs.

Mixed agricultural systems combining both crop cultivation and livestock production are very widely adopted in practice. Accordingly, future studies are required to expand the methodological framework from the simple one here exclusively for crop cultivation systems free from grazing to a more complex one applicable to mixed production systems consisting of crop cultivation and livestock farming for single or multiple animal type(s). The method to be updated for mixed crop–livestock systems would involve products of crops and animals (meat, milk, eggs, furs etc.)and life-cycle GHG emissions from on-farm and off-farm processes of not only crop cultivation but also livestock production. However, it is currently a challenge to standardize generic method(s) for accurate quantification of the GHG footprints for various widely adopted crop–livestock systems in different regions of the world.

An agricultural production system may exert up to 14 potential impacts on the Earth environment, as described by the 2013 ENVIFOOD Protocol (Food SCP RT 2013), among which the CCIP has been attracting extensive attention and can be quantitatively measured by GHG footprints or net GHG emissions in CO2eq at various scales. Therefore, the generic methodological framework should be further developed for accurately quantifying GHG footprints. A well-developed generic framework is expected to be applicable to the assessment of the CCIP of either simple or complex agricultural systems subject to various conditions of climate, soil, vegetation, and/or management practices at different temporal (e.g. annual or decadal) and/or spatial (e.g. site, regional, national, or global) scales. This is because assessment is the key step to work out strategies to develop climate-smart and thus environmentally friendly agricultural production systems.In addition, a well-established generic methodological framework may also be involved in GHG footprint assessment for the determination of green VATs for agricultural products for food markets, which may drastically change the current patterns of food consumption and production towards a product life cycle oriented economy (De Camillis and Goralczyk 2013).

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported jointly by the National Key R&D Program project of China [grant number 2017YFF0211704] and the National Natural Science Foundation of China [grant number 41761144054].

Ali, S. A., L. Tedone, L. Verdini, and G. D. Mastro. 2017. “Effect of Different Crop Management Systems on Rainfed Durum Wheat Greenhouse Gas Emissions and Carbon Footprint under Mediterranean Conditions.”Journal of Cleaner Production140: 608–621.

Baumert, K. A., T. Herzog, and J. Pershing. 2005. “Navigating The Numbers: Greenhouse Gas Data and International Climate Policy.”World Resources Institute Report.http://pdf.wri.org/navigating_numbers.pdf.

Bayer, C., J. Gomes, J. A. Zanatta, F. C. B. Vieira, and J. Dieckow. 2016.“Mitigating Greenhouse Gas Emissions from a Subtropical Ultisol by using Long-term No-tillage in Combination with Legume Cover Crops.”Soil & Tillage Research161: 86–94.

Biesbroek, S., H. B. Bueno-de-Mesquita, P. H. M. Peeters, W. M. M.Verschuren, Y. T. van der Schouw, G. F. H. Kramer, M. Tyszler,and E. H. M. Temme. 2014. “Reducing our Environmental Footprint and Improving our Health: Greenhouse Gas Emission and Land Use of Usual Diet and Mortality in EPICNL: A Prospective Cohort Study.”Environmental Health13: 27.

Chapin III, F. S., G. M. Woodwell, J. T. Randerson, E. B. Rastetter, G.M. Lovett, D. D. Baldocchi, D. A. Clark, et al. 2006. “Reconciling Carbon-cycle Concepts, Terminology, and Methods.”Ecosystems9: 1041–1050.

Chen, X., Z. Cui, P. M. Vitousek, K. G. Cassman, P. A. Matson, J.Bai, Q. Meng, et al. 2011. “Integrated Soil-crop System Management for Food Security.”P(pán)roceedings of the National Academy of Sciences of the United States of America108: 6399–6404.

Chen, X., Z. Cui, M. Fan, P. Vitousek, M. Zhao, W. Ma, Z. Wang,et al. 2014. “Producing more Grain with Lower Environmental Costs.”Nature514: 486–489.

Cheng, K., M. Yan, D. Nayak, G. X. Pan, P. Smith, J. F. Zheng, and J. W. Zheng. 2015. “Carbon Footprint of Crop Production in China: An Analysis of National Statistics Data.”Journal of Agricultural Science153: 422–431.

Chobtang, J., S. J. Mclaren, S. F. Ledgard, and D. J. Donaghy. 2017.“Environmental Trade-offs Associated with Intensification Methods in a Pasture-based Dairy System using Prospective Attributional Life Cycle Assessment.”Journal of Cleaner Production143: 1302–1312.

Cui, Z., S. Yue, G. Wang, Q. Meng, L. Wu, Z. Yang, Q. Zhang, S. Li,F. Zhang, and X. Chen. 2013. “Closing the Yield Gap could Reduce Projected Greenhouse Gas Emissions: A Case Study of Maize Production in China.”Global Change Biology19:2467–2477.

Cui, F., X. Zheng, C. Liu, K. Wang, Z. Zhou, and J. Deng. 2014a.“Assessing Biogeochemical Effects and Best Management Practice for a wheat–maize Cropping System using the DNDC Model.”Biogeosciences11: 91–107.

Cui, Z. L., L. Wu, Y. L. Ye, W. Q. Ma, X. P. Chen, and F. S. Zhang.2014b. “Trade-offs between High Yields and Greenhouse Gas Emissions in Irrigation Wheat Cropland in China.”Biogeosciences11: 2287–2294.

De Camillis, C., and M. Goralczyk. 2013. “Towards Stronger Measures for Sustainable Consumption and Production Policies: Proposal of a New Fiscal Framework based on a Life Cycle Approach.”The International Journal of Life Cycle Assessment18: 263–272.

Food SCP RT. 2013.ENVIFOOD Protocol, Environmental Assessment of Food and Drink Protocol.Brussels: European Food Sustainable Consumption and Production Round Table(SCP RT), Working Group 1.

Fumagalli, M. 2015. “Alternative Nitrogen Management Practices to Reduce Carbon Footprint of Maize Production.”Italian Journal of Agrometeorology-Rivista Italiana Di Agrometeorologia20: 21–32.

Gan, Y., C. Liang, Q. Chai, R. L. Lemke, C. A. Campbell, and R.P. Zentner. 2014. “Improving Farming Practices Reduces the Carbon Footprint of Spring Wheat Production.”Nature Communications5: 5012.

Gao, B., X. Ju, Q. Meng, Z. Cui, P. Christie, X. Chen, and F. Zhang.2015. “The Impact of Alternative Cropping Systems on Global Warming Potential, Grain Yield and Groundwater Use.”Agriculture, Ecosystems & Environment203: 46–54.

Grassini, P., and K. G. Cassman. 2012. “High-yield Maize with Large Net Energy Yield and Small Global Warming Intensity.”P(pán)roceedings of the National Academy of Sciences of the United States of America109: 1074–1079.

Ha, N., T. Feike, H. Back, H. Xiao, and E. Bahrs. 2015. “The Effect of Simple Nitrogen Fertilizer Recommendation Strategies on Product Carbon Footprint and Gross Margin of Wheat and Maize Production in the North China Plain.”Journal of Environmental Management163: 146–154.

Hillier, J., C. Hawes, G. Squire, A. Hilton, S. Wale, and P. Smith.2009. “The Carbon Footprints of Food Crop Production.”International Journal of Agricultural Sustainability7: 107–118.

Hoefnagels, R., E. Smeets, and A. Faaij. 2010. “Greenhouse Gas Footprints of Different Biofuel Production Systems.”Renewable and Sustainable Energy Reviews14: 1661–1694.

Intergovernmental Panel on Climate Change (IPCC). 1997.Revised 1996 Guidelines for National Greenhouse Gas Inventories: Greenhouse Gas Inventory Reference Manual.Bracknell: IPCC/OECD/IGES.

IPCC. 2000.Good Practice Guidance, Uncertainty Management in National Greenhouse Gas Inventories. Kanagawa: IPCC/IGES.

IPCC. 2006. “Volume 4: Agriculture, Forestry and Other Land Uses.” In2006 IPCC Guidelines for National Greenhouse Gas Inventories, edited by H. S. Eggleston, L. Buendia, L. Miwa,T. Ngara, and K. Tanabe. Kanagawa: IGES.

IPCC. 2013. “Climate Change 2013: The Physical Science Basis.” InContribution of Working Group I to The Fifth Assessment Report of The Intergovernmental Panel on Climate Change, edited by T. F. Stocker, D. Qin, G. -K. Plattner, M. Tignor, S. K. Allen,J. Boschung, A. Nauels, Y. Xia, V. Bex, and P. M. Midgley, 1535.Cambridge: Cambridge University Press.

IPCC. 2014. “Climate Change 2014: Mitigation of Climate Change.” InContribution of Working Group III to The Fifth Assessment Report of The Intergovernmental Panel on Climate Change, edited by O. Edenhofer, R. Pichs-Madruga, Y. Sokona,E. Farahani, S. Kadner, K. Seyboth, O. Edenhofer, et al., 1246.Cambridge/New York: Cambridge University Press.

Lai, L., X. Huang, H. Yang, X. Chuai, M. Zhang, T. Zhong, Z. Chen,Y. Chen, X. Wang, and J. R. Thompson. 2016. “Carbon Emissions from Land-use Change and Management in China between 1990 and 2010.”Science Advances2: e1601063.

Li, C. 2016.Biogeochemistry: Scientific Fundamentals and Modelling Approach.[In Chinese.] Beijing: Tsinghua University Press.

Li, Z., J. Tan, P. Tang, H. Chen, L. Zhang, H. Liu, W. Wu, H. Tang,P. Yang, and Z. Liu. 2016. “Spatial Distribution of Maize in Response to Climate Change in Northeast China during 1980–2010.”Journal of Geographical Sciences26: 3–14.

Ma, B. L., B. C. Liang, D. K. Biswas, M. J. Morrison, and N.B. McLaughlin. 2012. “The Carbon Footprint of Maize Production as affected by Nitrogen Fertilizer and Maizelegume Rotations.”Nutrient Cycling in Agroecosystems94:15–31.

Rochette, P., D. E. Worth, R. L. Lemke., B. G. McConkey, D. J.Pennock, C. Wagner-Riddle, and R. L. Desjardins. 2008.“Estimation of N2O Emissions from Agricultural Soils in Canada. I. Development of A Country-Specific Methodology.”Canadian Journal of Soil Science88: 641–654.

Shao, R., M. Xu, R. Li, X. Dai, L. Liu, Y. Yuan, H. Wang, and F.Yang. 2017. “Land use Legacies and Nitrogen Fertilization Affect Methane Emissions in the Early Years of Rice Field Development.”Nutrient Cycling in Agroecosystems107: 369–380.

Van Groenigen, J. W., G. L. Velthof, O. Oenema, K. J. Van Groenigen, and C. Van Kessel. 2010. “Towards an Agronomic Assessment of N2O Emissions: A Case Study for Arable Crops.”European Journal of Soil Science61: 903–913.

Van Ittersum, M. K., L. G. J. van Bussel, J. Wolf, P. Grassini, J. van Wart, N. Guilpart, L. Claessens, et al. 2016. “Can Sub-Saharan Africa Feed Itself?”P(pán)roceedings of the National Academy of Sciences of the United States of America113: 14964–14969.

Wang, K., X. Zheng, M. Pihlatie, T. Vesala, C. Liu, S. Haapanala,I. Mammarella, ü. Rannik, and H. Liu. 2013. “Comparison between Static Chamber and Tunable Diode Laser-based Eddy Covariance Techniques for Measuring Nitrous Oxide Fluxes from a Cotton Field.”Agricultural and Forest Meteorology171–172: 9–19.

Wang, H., Y. Yang, X. Zhang, and G. Tian. 2015. “Carbon Footprint Analysis for Mechanization of Maize Production Based on Life Cycle Assessment: A Case Study in Jilin Province, China.”Sustainability7: 15772–15784.

Williams, A. G., E. Audsley, and D. L. Sandars. 2010. “Environmental Burdens of Producing Bread Wheat, Oilseed Rape and Potatoes in England and Wales using Simulation and System Modelling.”The International Journal of Life Cycle Assessment15: 855–868.

Wu, L., X. Chen, Z. Cui, W. Zhang, and F. Zhang. 2014.“Establishing a Regional Nitrogen Management Approach to Mitigate Greenhouse Gas Emission Intensity from Intensive Smallholder Maize Production.”P(pán)LoS One9 (5): e98481.

Yan, M., K. Cheng, T. Luo, Y. Yan, G. Pan, and R. M. Rees. 2015.“Carbon Footprint of Grain Crop Production in China – based on Farm Survey Data.”Journal of Cleaner Production104: 130–138.

Yang, X., W. Gao, M. Zhang, Y. Chen, and P. Sui. 2014. “Reducing Agricultural Carbon Footprint through Diversified Crop Rotation Systems in the North China Plain.”Journal of Cleaner Production76: 131–139.

Zhang, W., Y. Yu, T. Li, W. Sun, and Y. Huang. 2014. “Net Greenhouse Gas Balance in China’s Croplands over the Last Three Decades and Its Mitigation Potential.”Environmental Science & Technology48: 2589–2597.

Zhang, W., J. Gu, and X. Zheng. 2015. “Direct Nitrous Oxide Emissions Related to Fertilizer-Nitrogen, Precipitation,and Soil Clay Fraction: Empirical Models.”Atmospheric and Oceanic Science Letters8 (5): 277–282.

Zhang, X., C. Pu, X. Zhao, J. Xue, R. Zhang, Z. Nie, F. Chen, R. Lal,and H. Zhang. 2016. “Tillage Effects on Carbon Footprint and Ecosystem Services of Climate Regulation in a Winter Wheat–Summer Maize Cropping System of the North China Plain.”Ecological Indicators67: 821–829.

Zheng, X., and R. Wang. 2017.Protocols for Chamber-based Manual Measurement of CH4 and N2O Fluxes from Terrestrial Ecosystems.[In Chinese.] Beijing: The Meteorological Press.