Introduction

Globally, the croplands comprise  ~ 12% of the total land area, which has increased by ~ 12% since 1960s (FAO 2011). Croplands in India occupy  ~ 43% of total geographical area (328 million ha, M ha) of the country. The cropland ecosystems have significant potential to store and emit carbon (C) and thus govern the net ecosystem C budget (NECB) and C sequestration potential of soils (Benbi et al. 2019; Liu et al. 2019; Singh et al. 2020). The global C sequestration potential of cropland has been estimated between 0.4 and 1.2 Gt C year−1, while C emissions are rising rapidly by ~ 2.3% year−1 (Meena et al. 2020). Most recent estimates revealed that Indian soils currently had 1370–1770 Tg C, which could be increased to 2460–2650 Tg C by 2050 through a large-scale adoption of best management practices including balanced application of nutrients, manures, agro-forestry and conservation agriculture (Nath et al. 2018). These estimates revealed that a widespread adoption of these practices can enhance C sequestration by 70–130 Tg CO2e year−1 and produce ~ 410–440 million Mg of food grains accounting for  ~ 80–85% of the total requirement by 2050 (Nath et al. 2018). Unlike the forests and grassland ecosystems, croplands are characterized by diverse management practices such as annual cropping systems, regular fertilizer and manure application and irrigation management with significant impact on C cycling (Ceschia et al. 2010; Liu et al. 2019; Singh and Benbi 2020). Therefore, appropriate management of croplands offers opportunities for increasing soil C sequestration (Sainju et al. 2007; Singh and Benbi 2018b), while decreasing emission of greenhouse gases (GHGs) (Linquist et al. 2012; Ma et al. 2013). Application of chemical fertilizers in cropland ecosystems affects crop productivity (Ju et al. 2009), economics, energy input–output balance (Nabavi-Pelesaraei et al. 2014a, 2017; Mandal et al. 2015; Jat et al. 2016) and emission of GHGs (Cai et al. 2007; Yang et al. 2015), which are inter-related and thus impact NECB (Soussana et al. 2010; Ceschia et al. 2010).

Cropland ecosystems release a significant amount of carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) to the atmosphere (Liu et al. 2019). During the past few decades, the global atmospheric concentration of these GHGs has increased at an annual growth rate of 0.5% for CO2, 0.6% for CH4 and 0.25% for N2O (IPCC 2007a). The global technical mitigation potential of croplands estimated at 5.5–6.0 Pg CO2e year−1 by 2030 (Smith et al. 2008; Benbi 2013) is almost equivalent (5.1–6.1 Pg CO2e year−1) to agriculture’s share towards total GHGs emissions (IPCC 2007b). Net emission of CO2e could be decreased by increasing soil C sequestration and by altering the nutrient management practices in crop production (Pathak 2010; Nath et al. 2018), which affects energy input–output ratio (Nabavi-Pelesaraei et al. 2014b, 2019a). Croplands act as both sink and source of C as well as energy (as bio-energy) (Alam et al. 2005; Nabavi-Pelesaraei et al. 2018; Singh et al. 2020). Indeed, agricultural sustainability depends upon the equilibrium between productivity, environment and economics which had close nexus with energy flow (Helander and Delin 2004; Ozkan et al. 2004; Benbi 2018; Nabavi-Pelesaraei et al. 2019b). In the past decade, a new concept of greenhouse gas intensity (GHGI) has been introduced to relate global warming potential (GWP) with economic yield of crops to target a specific GHG mitigation or C sequestration in soil without any change in economic productivity (Shang et al. 2010).

Carbon sequestration potential is predicted from soil respiration (Mosier et al. 2006) or modelling (Li et al. 2006; Zhang et al. 2009; Singh and Benbi 2020), but can also be estimated from NECB (Mosier et al. 2005; Ma et al. 2013; Benbi et al. 2019). The NECB quantifies the exchange of C, energy and mass by crop vegetation within a cropland ecosystem (Running and Coughlan 1988). Although C sequestration potential of croplands is considered modest, the quantification of the net C sequestration potential is difficult because of complex C cycling within an ecosystem and large spatial and temporal variability associated with crop production and soil management regimes (Pan et al. 2004; Smith et al. 2010; Singh and Benbi 2020). There is a considerable uncertainty in NECB of croplands as a function of soil management practices; therefore, the quantification of C budget to elucidate whether a cropland ecosystem is a C source or sink is essential (Chapin et al. 2006; Zheng et al. 2008; Béziat et al. 2009). This approach has an advantage of assessing the effects of individual crops, cropping system, particular climate or a management events on the net biome productivity, while assessing the variation in soil C stocks that level-off the short-term effects associated with soil organic matter fractions with a rapid turnover rate (Ceschia et al. 2010; Liu et al. 2019). Nevertheless, NECB is more appropriate to discern change in a short-term C budget, as it considers the amount of C lost or gained within an ecosystem (Smith et al. 2010; Soussana et al. 2010).

The change in soil C stocks depends upon the balance between C input and output (Benbi et al. 2012; Singh and Benbi 2018a), and the NECB approach considers a precise estimate of balance between C gains and losses (Chapin et al. 2006; Smith et al. 2010). The cropping systems affect the quality and quantities of crop residues returned to the soil and eventually influence C input and GHGs emissions (Kimura et al. 2004; Mosier et al. 2006; Sainju et al. 2008; Khosa et al. 2011). Linquist et al. (2012) reported that the GWP of GHGs emitted from rice (Oryza sativa L.) ecosystem is roughly 4 times greater than wheat (Triticum aestivum L.) and maize (Zea mays L.). Therefore, quantification of croplands’ NECB would enable the estimation of soil C sequestration potential by providing a scientific basis for the assessment of net CO2 exchange within an ecosystem (Chapin et al. 2006; Smith et al. 2010). Most studies conducted earlier focused on the assessment of C budgets of natural undisturbed (forests) and managed ecosystems (Ammann et al. 2007; Du and Liu 2013). However, the studies on cropland ecosystems focused on quantification of C budgets for an individual crop (Verma et al. 2005; Swain et al. 2016; Gao et al. 2017; Kim et al. 2017). The NECB for cropping systems with diversity in management practices, soil types and agro-climatic conditions has not been extensively investigated, which is required for the development of robust GHGs mitigation and C sequestration strategies for a particular cropping system (Zhang and Wang 2013; Benbi et al. 2019).

In a cropland ecosystems, nutrient management-induced change in soil organic C (SOC) pool has been related to C input through root biomass and exogenous C input through organic manures (Benbi and Brar 2009; Singh and Benbi 2018b). In any cropland ecosystem, inorganic fertilizer application has been the major energy input (Nabavi-Pelesaraei et al. 2014b, 2018, 2019b; Singh et al. 2019) significantly influencing GHGs emissions (Nabavi-Pelesaraei et al. 2017; Singh et al. 2020). The crop production and soil management practices adopted by the farmers differ from field to field and from those under well-managed experimental sites (Singh et al. 2019). The farmers’ fields are intensively managed cropland ecosystems and therefore may well be readily available option for assessing the changes in soil C budget (Baker and Griffis 2005). The present study was, therefore, conducted at farmers field in intensively cultivated region in north-western India where three unique cropland ecosystems, viz. rice–wheat, maize–wheat and cotton–wheat, are practiced. Till date, there is a dearth of information on the impact of nutrient management practice on GHGs emissions, C sequestration and NECB in relation to energy flow in these diverse and unique cropland ecosystems. The assessment of GHGs emissions and the amount of C added to SOC pool to quantify the NECB and energy flow nexus will help understand the mechanisms of C cycling in these ecosystems. We hypothesized that differential nutrient management would lead to variable impact on C input in SOC pool and GWP under contrasting moisture regimes to influence NECB in three cropland ecosystems. More specifically, we aimed to investigate the impact of nutrient management practices on GHGs emissions and a change in SOC pool for the estimation of NECB to explicate whether a cropland ecosystem is a net source or sink of C. The present study will help to frame policies for efficient nutrient management practices for increased soil C sequestration, while minimizing GHGs emissions due to optimized energy flow in cropland ecosystems at the farmers’ fields.

Materials and methods

Study area

The study region covered twelve districts of north-western Indian state of Punjab, where the fields were under rice–wheat, maize–wheat and cotton–wheat cropping systems (Fig. 1). Rice–wheat cropping system occupies 2.6 Mha, which constitutes ~ 78% of gross cropped area of the state. Maize and cotton (Gossypium hirsutum L.) occupy ~ 0.14 and ~ 0.51 Mha area constituting 4 and 15%, respectively, of the gross cropped area. The three ecosystems differ in yield potential and rooting system and thus could lead to differential C input to the soil. A detailed field survey with a semi-structured questionnaire (pre-tested on 15 respondents) concerning nutrient management practices for rice–wheat, maize–wheat and cotton–wheat cropping systems was conducted in the selected districts. In total, 101 respondents (37 rice–wheat, 38 maize–wheat and 26 cotton–wheat growers) were interviewed for data collection. The information gathered included crop variety, crop yield, and the rate and source of chemical fertilizers (N, P2O5 and K2O) applied. A brief description on cropping systems, climatic conditions, agroeco-subregion and soil classification for the study region is given in Table 1. Farmers apply irrigation to these crops through underground and canal water depending on visual inspection of the field. In the study region, rice fields had large diversity with respect to irrigation water management, viz. continuous flooding, intermittent flooding with single aeration and intermittent flooding with multiple aerations.

Fig. 1
figure 1

Geographical location of different study sites under rice–wheat, maize–wheat and cotton–wheat cropping systems in north-western India

Full size image
Table 1 Details about climatic conditions, agro-eco-subregion and taxonomy of soils under different districts in Punjab (north-western India)
Full size table

Energy auditing and estimation of energy indices

Energy input under different cropping systems was estimated from the quantity of fertilizer N, P2O5 and K2O used and their respective energy equivalents (Table 2). Energy coefficients of 60.6 MJ (kg N)−1, 11.1 MJ (kg P2O5)−1 and 6.7 MJ (kg K2O)−1 were used for calculating the energy input (MJ ha−1) through different fertilizers. Energy input through fertilizer N, P2O5 and K2O was summed to estimate total energy input (EI). The output energy was estimated from quantity of above-ground biomass produced by multiplying with their respective energy equivalents (Table 2). The energy output for each crop was obtained by multiplying economic yield by their respective energy equivalents. The amount of energy produced from the above-ground biomass (grain plus straw in case of unhusked rice and wheat, grain plus stalk in case of maize, and seed cotton plus sticks in case of cotton) yield of different crops was summed to estimate total energy output (EO).

Table 2 Energy equivalents of different fertilizer nutrient inputs and outputs (economic product and residues) in rice–wheat, maize–wheat and cotton–wheat cropping system
Full size table

Different energy indices, viz. energy productivity (EP), specific energy (ES), energy ratio or energy use efficiency (ER) and net energy gain (NEG) for an individual crop, were calculated using the following equations (Eq. 1–6).

$$E_{{\text{I}}} \left( {{\text{GJ}}\;{\text{ ha}}^{ - 1} } \right) = \sum \left( {E_{{\text{N}}} + E_{{{\text{P}}_{2} {\text{O}}_{5} }} + E_{{{\text{K}}_{2 } {\text{O}}}} } \right)$$
(1)
$$E_{{\text{O}}} \left( {{\text{GJ}} \;{\text{ha}}^{ - 1} } \right) = \sum \left( {E_{{{\text{Economic}}\;{\text{yield}}}} + E_{{{\text{Residues}} \;{\text{yield}}}} } \right)$$
(2)
$$E_{{\text{P}}} \left( {{\text{kg}} \;{\text{MJ}}^{ - 1} } \right) = \frac{{Y_{{{\text{Economic}} + {\text{residues}}\;{\text{ yield}}}} }}{{E_{{\text{I}}} }}$$
(3)
$$E_{{\text{S}}} \left( {{\text{MJ}} \;{\text{kg}}^{ - 1} } \right) = \frac{{E_{{\text{I}}} }}{{Y_{{{\text{Economic}} + {\text{residues}} \;{\text{yield}}}} }}$$
(4)
$$E_{{\text{R}}} = \frac{{E_{{\text{O}}} }}{{E_{{\text{I}}} }}$$
(5)
$${\text{NEG}} \left( {{\text{GJ}}\;{\text{ ha}}^{ - 1} } \right) = E_{{\text{O}}} - E_{{\text{I}}}$$
(6)

where ‘EN’, ‘\(E_{{{\text{P}}_{2} {\text{O}}_{5} }}\)’ and ‘\(E_{{{\text{K}}_{2 } {\text{O}}}}\)’ represent energy input (each in GJ ha−1) through fertilizer N, P2O5, and K2O, respectively, for rice–wheat, maize–wheat and cotton–wheat cropping system. The ‘\(E_{{{\text{Economic}}\;{\text{yield}}}}\)’ and ‘\(E_{{{\text{Residues}} \;{\text{yield}} }}\)’ represent the energy output (each in GJ ha−1) as economic yield (grain/seed cotton) and residues yield (straw/stalk/sticks) (each in kg ha−1) under rice–wheat, maize–wheat and cotton–wheat ecosystems.

Estimation of GHG emission, GWP and GHGI

The environmental impact of nutrient management in rice–wheat, maize–wheat and cotton–wheat ecosystems was assessed by estimating GHGs emissions, GWP and GHGI. The fertilizer-induced GHGs (CO2 and N2O) emissions were estimated for each site under different cropping systems using CO2 equivalents for N, P2O5 and K2O from the literature. The emission factors for GHGs emission, viz. 1.3 kg C (kg N)−1, 0.2 kg C (kg P2O5)−1 and 0.15 kg C (kg K2O)−1 (Lal 2004), were used. The estimated CO2 and N2O emissions were converted to CO2 equivalent (CO2e) using 100-years GWP of 1, 265 and 25 for CO2, N2O and CH4, respectively (IPCC 2013) using Eq. 7. The N2O emissions were calculated by multiplying the fertilizer-N input with a respective conversion factor (Eq. 8) (IPCC 2007a, b). The study region has been divided among three broad categories, viz. rice area fell into continuously flooded (0–10%), intermittently flooded with single aeration (30%) and intermittently flooded with multiple aerations (60–70%) based on soil type, water availability, rice varieties cultivated and local climatic conditions (Benbi 2018). Therefore, the CH4 emission from rice fields for estimating GWP of a rice–wheat cropping system was based on region specific emissions of 55.7, 34.5 and 23.3 kg CH4 ha−1 for continuously flooded (scenario 1), intermittently flooded with single aeration (scenario 2) and intermittently flooded with multiple aerations (scenario 3), respectively (Khosa et al. 2011). Since there is no CH4 emission under upland cropping systems, viz. maize–wheat and cotton–wheat, the GWP was calculated based on CO2 and N2O.

$${\text{GWP}} \left( {{\text{kg}}\;{\text{ CO}}_{{2{\text{e}}}} \;{\text{ha}}^{ - 1} } \right) = \left( {{\text{CO}}_{2} } \right) + \left( {{\text{CH}}_{4} \times 25} \right) + \left( {{\text{N}}_{2} {\text{O}} \times 265} \right)$$
(7)

The N2O emissions from N applied through chemical fertilizer were estimated using Eq. 8.

$${\text{N}}_{2} {\text{O}} \;{\text{emission}} = {\text{N}} \times {\text{ EF}}_{1} \times \frac{44}{{28}}$$
(8)

where N2O emissions (kg N2O ha−1) from fertilizer N application and ‘EF1’ = emission factor (0.01 for N2O emissions from N inputs, kg N2O-N kg−1 N input).

The GHGI was estimated from the ratio of GWP to that of the economic yield (grain in case of rice, wheat and maize and seed cotton in case of cotton) (Eq. 9) and was expressed as kg CO2e kg−1 economic yield (Shang et al. 2010). Greenhouse gas intensity for maize and cotton was estimated based on their relative equivalent yield.

$${\text{GHGI}} = \frac{{\text{ GWP}}}{{{\text{Economic}}\;{\text{yield}}}}$$
(9)

Carbon equivalent emissions (CEE) and carbon efficiency ratio (CER) were calculated by using Eqs. 10 and 11 (Bhatia et al. 2005).

$${\text{CEE}} \left( {{\text{Mg}} \;{\text{C}} \;{\text{ha}}^{ - 1} } \right) = {\text{GWP}} \times \frac{12}{{44}}$$
(10)
$${\text{CER}} = {\text{grain}}\;{\text{yield}}\;{\text{or}}\;{\text{seed}}\;{\text{cotton}}\;{\text{yield}}\;{\text{in}}\;{\text{terms}}\;{\text{of}}\;{\text{ C /CEE}}$$
(11)

The C concentrations of 38, 39, 39 and 40% in the rice, wheat, maize grain and seed cotton, respectively, were used for estimating CER (Huang et al. 2007).

Calculation of net ecosystem carbon budget

Based on the published studies (Smith et al. 2010; Ma et al. 2013), we summarized the different components for NECB (Eq. 12).

$${\text{NECB}} = {\text{GPP}} - R_{{\text{E}}} - {\text{Harvest}} - {\text{CH}}_{4} + C_{{{\text{Below } - \text{ ground}}\;{\text{biomass}}}} + C_{{{\text{Litter}}}} + C_{{{\text{Rhizodeposit}}}}$$
(12)

where ‘GPP’ is the gross primary production and was inferred from net primary production (NPP) via ratio of NPP/GPP (Luyssaert et al. 2007). The NPP–GPP ratio of 0.58 deduced from MODIS model was used for estimating GPP (Zhang et al. 2009). Total NPP for different crops and cropping systems was estimated as a sum of NPP for different eco-systems’ components, viz. actual economic yield, above-ground biomass yield, below-ground biomass yield, litter and rhizodeposits (Eq. 13).

$${\text{NPP}} = {\text{NPP}}_{{{\text{Economic}}\;{\text{yield}}}} + {\text{NPP}}_{{{\text{Above } - \text{ ground}} \;{\text{yield}}}} + {\text{NPP}}_{{{\text{Below } - \text{ ground }}\;{\text{yield}}}} + {\text{NPP}}_{{{\text{Litter}}}} + {\text{NPP}}_{{{\text{Rhizodeposits}}}}$$
(13)

The \({\mathrm{NPP}}_{\mathrm{Economic yield}}\), \({\mathrm{NPP}}_{\mathrm{Above}-\mathrm{ground yield}}\) and \({\mathrm{NPP}}_{\mathrm{Below}-\mathrm{ground yield}}\) were estimated using equations (Eqs. 14–16) (Huang et al. 2007).

$${\text{NPP}}_{{{\text{Economic}}\;{\text{yield}}}} = {\text{Economic}} \;{\text{yield }} \times {\text{ dry}}\;{\text{ matter}}\;{\text{fraction}} \times C_{{{\text{fraction}}\;{\text{of}}\;{\text{economic }}\;{\text{yield}}}}$$
(14)
$$\begin{aligned} & {\text{NPP}}_{{{\text{Above } - \text{ ground}} \;{\text{yield}}}} = {\text{Economic}} \;{\text{yield }} \times {\text{ dry}}\;{\text{ matter}}\;{\text{fraction }} \\ & \quad \times {\text{ ratio}}\;{\text{of}}\;{\text{residue}}\;{\text{ to}}\;{\text{ economic}}\;{\text{yield}} \times C_{{{\text{fraction}}\;{\text{of}}\;{\text{residue}}}} \\ \end{aligned}$$
(15)
$$\begin{aligned} & {\text{NPP}}_{{{\text{Below } - \text{ ground}}\;{\text{yield}}}} = {\text{Economic}}\;{\text{ yield }} \times {\text{ dry}}\;{\text{matter}}\,{\text{fraction}} \\ & \quad \times \left( {1 + {\text{ratio}}\,{\text{of}}\,{\text{residue}}\,{\text{to}}\,{\text{economic}} \,{\text{yield}}} \right) \\ & \quad \times {\text{ ratio}}\;{\text{of}}\;{\text{roots}}\;{\text{to}}\;{\text{shoot}} \times C_{{{\text{fraction}}\;{\text{ of}}\;{\text{ roots}}}} \\ \end{aligned}$$
(16)

The \({\text{NPP}}_{{{\text{Litter}}}}\) was estimated as 5% of the total above-ground + root dry biomass as a leaf litter (Kimura et al. 2004). The \({\text{NPP}}_{{{\text{Rhizodeposit}}}}\) was estimated as 15% of total biomass for rice and 18% wheat, maize and cotton (Bronson et al. 1998; Gregory 2006). The dry matter fraction (in Eq. 15) of 0.85, 0.85, 0.78 and 0.90 for rice, wheat, maize and cotton, respectively, was considered (Huang et al. 2007). The ratio of residue to economic yield of 1.32, 0.72, 1.27 and 1.61, respectively, was used in the calculations. The ratio of roots to shoot for rice, wheat, maize and cotton used in the calculations was 0.10, 0.11, 0.09 and 0.06, respectively (Huang et al. 2007).

The ecosystem respiration (\(R_{{\text{E}}}\)) of the cropland ecosystem represents the sum of autotrophic respiration (\(R_{{\text{A}}}\)) and soil heterotrophic respiration (\(R_{{\text{H}}}\)). The \(R_{{\text{E}}}\) values of 9210 kg C ha−1 for rice, 8410 kg C ha−1 for maize, 5580 kg C ha−1 for wheat were used in the calculations (Yang et al. 2015; Gao et al. 2017). Since the information on RE for cotton ecosystem was not available, we assumed the same value as for maize ecosystem. The net ecosystem exchange (NEE) of CO2 between the agro-ecosystem and atmosphere referred as how much of C is lost or gained by an ecosystem was estimated using the following relationships (Eqs. 17–20) (Smith et al. 2010; Gao et al. 2017).

$${\text{NEE}} \left( {{\text{kg}}\;{\text{C}}\;{\text{ ha}}^{ - 1} } \right) = {\text{GPP}} - R_{{\text{H}}} - R_{{\text{A}}}$$
(17)
$$R_{{\text{E}}} = R_{{\text{H}}} + R_{{\text{A}}}$$
(18)
$$R_{{\text{A}}} = {\text{GPP}} - {\text{NPP}}$$
(19)
$$R_{{\text{H}}} = R_{{\text{H}}} - R_{{\text{A}}}$$
(20)

Harvest includes biological and residue yield removed from the field and converted to C by crop specific C content of 42, 49, 47 and 39% for rice, wheat, maize and cotton, respectively (Huang et al. 2007). The residue yield was estimated using residues: economic product ratios of 1.32, 1.72, 1.27 and 1.61, respectively.

We estimated the change in soil organic carbon (∆SOC) considering apparent average conversion rate of organic C to SOC of 213 g kg−1 (Xie et al. 2010). Therefore, the ∆SOC was estimated from the NECB using a coefficient of 0.213 using Eq. 21.

$$\Delta {\text{SOC}} = 0.213 \; \times \;{\text{NECB}}$$
(21)

The amount of C added into SOC pool was estimated (using Eq. 22) based on the assumption that 18, 15, 15 and 10% of above-ground crop residues and all the root biomass of rice, wheat, maize and cotton, respectively, were left in the soil. A survey of farmers’ field in the study region conducted by Beri et al. (2003) showed that between 80.3 and 83.9% of rice straw is burnt in situ; therefore, only 18% of rice residue gets incorporated into soil. However, fraction of above-ground biomass of wheat, maize and cotton burnt or removed from the field in the study region were taken as 0.85, 0.85 and 0.90, respectively.

$${\text{Amount}}\;{\text{of}}\;{\text{C}}\;{\text{added}}\;{\text{to}}\;{\text{SOC}}\;{\text{pool}} \left( {{\text{kg}}\;{\text{ C}}\;{\text{ ha}}^{ - 1} } \right) = \left( {F \times {\text{C}}_{{{\text{residues}}}} + {\text{C}}_{{{\text{root}}}} } \right) \times F_{{\text{R}}}$$
(22)

where ‘F’ is the fraction of above-ground biomass leaf in the soil and ‘\({\text{C}}_{{{\text{residues}}}}\)’ and ‘\({\text{C}}_{{{\text{root}}}}\)’ are the residue C and root C, respectively. The C content in the plant roots was considered 40%. The ‘FR’ is a fraction of residue and root C converted to SOC.

Since farmyard manure (FYM) application has not been a regular practice and is applied at 10–20 Mg ha−1 (before sowing of summer crop) once in 2–3 years, its contribution towards GHGs emission, C budget and energy input was not taken into consideration.

Rice equivalent yield of non-rice crops

To compare the total system productivity, GHGs emissions, GHGI, CER and input–output energy relationships of different cropping systems, the yield of non-rice crops (maize and cotton) were converted into rice equivalent yield (REY) (Eq. 23).

$$\begin{aligned} & {\text{REY}}\left( {{\text{Mg}} \;{\text{ha}}^{ - 1} } \right) \\ & \quad = \frac{{{\text{Yield}}\;{\text{of}}\;{\text{non } - \text{ rice}}\;{\text{crop}} \left( {{\text{Mg}}\;{\text{ha}}^{ - 1} } \right) \times {\text{MSP}}\;{\text{of}}\;{\text{non } - \text{ rice}}\;{\text{crop }}\left( {{\text{US}}\$ {\text{Mg}}^{ - 1} } \right)}}{{{\text{MSP}} \;{\text{of}}\;{\text{ rice}}\;{\text{ crop}} \left( {{\text{US}}\$ {\text{ Mg}}^{ - 1} } \right)}} \\ \end{aligned}$$
(23)

where ‘MSP’ is the minimum support prices of rice, wheat, maize and seed cotton fixed by Government of India (GOI) in 2016–17 (1US$ = 60 INR). Minimum support price of 245.0 US$ Mg−1 rice, 270.0 US$ Mg−1 wheat, 227.5 US$ Mg−1 maize grain and 693.3 US$ Mg−1 seed cotton was used in calculations. However, the comparison of different cropping systems for their NPP, NECB, NEP, amount of C added into SOC pool and the ∆SOC pool was based on actual yield of non-rice crops.

Calculations for gross returns above fertilizer cost

Gross returns above the fertilizer cost (GRAFC) for different crops were estimated as a difference in product of grain yield (Mg ha−1) and MSP (US$ Mg−1) and the amount (A) of fertilizer nutrients (N + P2O5 + K2O) applied (Mg ha−1) and price of N, P2O5 and K2O (US$ Mg−1) using Eq. 24. For the estimation of GRAFC for maize and cotton, rice equivalent yield was considered in the calculations to compare the three cropping systems. The MSP (245.0, 270.8, 227.5 and 393.3 US$ Mg−1 rice, wheat, maize and seed cotton, respectively) decided by the GOI for 2016–17 was used in the calculations.

$${\text{GRAFC}}\left( {{\text{US}}\$ {\text{ha}}^{ - 1} } \right) = \left[ {{\text{Grain}} \;{\text{yield}} \times {\text{MSP}}} \right] - \left[ {A_{{\left( {{\text{N}} + {\text{P}}_{2} {\text{O}}_{5} + {\text{K}}_{2} {\text{O}}} \right)}} \times {\text{price}}_{{\left( {{\text{N}} + {\text{P}}_{2} {\text{O}}_{5} + {\text{K}}_{2} {\text{O}}} \right)}} } \right]$$
(24)

where ‘\(A_{{\left( {{\text{N}} + {\text{P}}_{2} {\text{O}}_{5} + {\text{K}}_{2} {\text{O}}} \right)}}\)’ represent the amount of N + P2O5 + K2O applied and \({\text{price}}_{{\left( {{\text{N}} + {\text{P}}_{2} {\text{O}}_{5} + {\text{K}}_{2} {\text{O}}} \right)}}\) represent the market selling price of N + P2O5 + K2O. The market selling price (in India) was US$ 94.7, 366.7 and 183.3 Mg−1 for urea, diammonium phosphate (DAP) and muriate of potash (MOP), respectively,

Gross returns above the fertilizer cost for rice–wheat (GRAFCRW), maize–wheat (GRAFCMW) and cotton–wheat (GRAFCCW) cropping systems were estimated as a sum total of GRAFC for two crops grown in a sequence using Eqs. 25–27.

$${\text{GRAFC}}_{{{\text{RW}}}} \left( {{\text{US}}\$ {\text{ha}}^{{ - 1}} } \right) = {\text{GRAFC}}_{{\text{R}}} + {\text{GRAFC}}_{{\text{W}}}$$
(25)
$${\text{GRAFC}}_{{{\text{MW}}}} \left( {{\text{US}}\$ {\text{ha}}^{{ - 1}} } \right) = {\text{GRAFC}}_{{\text{M}}} + {\text{GRAFC}}_{{\text{W}}}$$
(26)
$${\text{GRAFC}}_{{{\text{CW}}}} \left( {{\text{US}}\$ {\text{ha}}^{{ - 1}} } \right) = {\text{GRAFC}}_{{\text{C}}} + {\text{GRAFC}}_{{\text{W}}}$$
(27)

Statistical analyses

Statistical analysis of data was performed by one-way analysis of variance (ANOVA) with SPSS 16.0 for windows (SPSS Inc., Chicago, U.S.A.). The study sites were treated as replicates (random effects) and the crops (rice, maize and cotton) or the cropping systems (rice–wheat, maize–wheat and cotton–wheat) as treatments (fixed effect). Mean separation for crops or cropping systems was evaluated at 95% confidence interval using Tukey’s post hoc test. Mean treatment differences for crops significant at p < 0.05 were marked using small letters, while the significant difference for cropping systems was marked using capital letters. Different environmental indices, viz. GWP, GHGI, CEE and CER, estimated for three contrasting scenarios (scenario 1: continuously flooded, scenario 2: intermittently flooded with single aeration, and scenario 3: intermittently flooded with multiple aerations) were distinguished with subscript notations as GWP1, GWP2, GWP3, and so on.

Results

Fertilizer use in crops and cropping systems

Farmers fertilize rice crop with 170.2 kg N, 28 kg P2O5 and 18.2 kg K2O ha−1, maize crop 156.8 kg N, 37.6 kg P2O5, and 19.7 kg K2O ha−1, and cotton with 146.1 kg N, 34.3 kg P2O5 and 21.9 kg K2O ha−1 (Table 3). Fertilizer-N use was significantly (p < 0.05) lower in cotton by 14.2% than in rice ecosystem. Compared to maize, fertilizer-N use in cotton was lower by 7.3%. Fertilizer-N use in rice and maize did not differ significantly. On the other hand, fertilizer-P2O5 use in rice was significantly lower by 9.6 kg (34%) and 3.3 kg P2O5 ha−1 (10%), compared with maize and cotton, respectively. Conversely, fertilizer-K2O use in rice, maize and cotton did not differ significantly. The three cropping systems did not differ significantly for fertilizer-N, fertilizer-P2O5, fertilizer-K2O and fertilizer-N + P2O5 + K2O use (Table 3).

Table 3 Fertilizer nitrogen (N), phosphorus (P2O5) and potassium (K2O) use (kg ha−1) in rice–wheat, maize–wheat and cotton–wheat cropping systems in north-western India
Full size table

Crop yield and gross returns above the fertilizer cost

Economic and residues yield of rice was significantly higher than maize and cotton (Table 4). For rice ecosystem, economic yield was higher by 2.38 Mg ha−1 (52%) and crop residue yield by 3.37 Mg ha−1 (58%), compared with maize. The REY of cotton was significantly higher by 1.88 Mg ha−1 (44%) compared with maize. The comparison of cropping systems based on REY of non-rice crops revealed that economical yield of rice–wheat was significantly higher than others. The GRAFC were significantly lower for maize compared with rice and cotton (Fig. 2). For cotton, the GRAFC were significantly higher than rice by 2534 US$ ha−1 and maize by 3275 US$ ha−1. The GRAFC were lowest for maize–wheat and highest for cotton–wheat cropping system. The GRAFC for rice–wheat were significantly higher by 732 US$ ha−1 compared with maize–wheat cropping system.

Table 4 Economic and residues yield of rice–wheat, maize–wheat and cotton–wheat cropping systems in north-western India
Full size table
Fig. 2
figure 2

Gross returns above fertilizer cost (GRAFC) for rice–wheat, maize–wheat and cotton–wheat cropping systems. Line bars indicate standard error from mean. Mean values for a crop or cropping system followed by different letters are significantly different (p < 0.05) by Tukey’s post hoc test. Small letters differentiate crops, and capital letters differentiate cropping systems

Full size image

Effect of fertilizer use on GHGs emission and environmental indicators

The CO2 emission was significantly lower in cotton by 108 kg CO2e ha−1 (13%) and 52 kg CO2e ha−1 (6.6%) compared with rice and maize, respectively (Table 5). The CO2 emission due to fertilizer application in rice was significantly higher by 56 kg CO2e ha−1 (7%), compared with the maize ecosystem. The three cropping systems, however, did not differ significantly for CO2 emission. The fertilizer-N application-related N2O emission was significantly higher by 17% in rice, compared with cotton ecosystem. However, the N2O emission in three cropping systems did not differ significantly. The GWP1 was significantly higher by 1.6 Mg CO2e ha−1 year−1 (123%) for rice, compared with cotton ecosystem. The GWP for maize and cotton ecosystems did not differ significantly. The GWP1 of rice–wheat was significantly higher by 1.5 Mg CO2e ha−1 year−1, compared with other two cropping systems. Maize–wheat and cotton–wheat cropping systems did not differ significantly for GWP. The GWP2 and GWP3 for rice estimated for scenario 2 and scenario 3 considering CH4 emissions from rice fields that were intermittently flooded with single and multiple aerations, respectively, was ~ 18.1 and 27.6% lower compared with GWP1 considering CH4 emission when rice fields remained continuously flooded throughout the growing season. The GWP2 and GWP3 for annual rice–wheat cropping system were lower ~ 12.1 and 18.3%, respectively, compared with GWP1. The GHGII was significantly higher by 0.09 kg CO2e kg−1 in rice, compared with maize ecosystem. It was significantly lower for cotton, compared with rice and maize ecosystem. The comparison of three cropping systems revealed that GHGI was significantly higher for rice–wheat, compared with other two cropping systems. The GHGI of maize–wheat was significantly higher than the cotton–wheat cropping system (Table 5). The CEE1 was significantly higher for rice and lowest for cotton, with maize in between. It was significantly higher for rice–wheat, compared with other two cropping systems, which themselves did not differ significantly. Conversely, the CER was significantly lower for cotton–wheat (7.03), compared with maize–wheat (9.59) and rice–wheat (8.07) cropping system. These results showed that GWP, GHGI and CER for rice and rice–wheat cropping system decreased considerably following judicious irrigation management of intermittent flooding and aeration.

Table 5 Greenhouse gas (GHG) emission, global warming potential (GWP), greenhouse gas intensity (GHGI), carbon equivalent emission (CEE) and carbon equivalent ratio (CER) under rice–wheat, maize–wheat and cotton–wheat cropping systems in north-western India
Full size table

Soil respiration and ecosystem carbon budget

The autotrophic respiration (RA) was significantly higher for rice by 2.0 Mg C ha−1 and 3.4 Mg C ha−1 as compared to maize and cotton ecosystems, respectively (Table 6). Rice–wheat had 46% higher RA than the cotton–wheat cropping system. For a maize–wheat cropping system, RA was significantly higher than cotton–wheat, but significantly lower than rice–wheat. In contrast to RA, RH exhibited a reverse trend with the significantly higher value for cotton, and the lowest for rice. Similarly, rice–wheat system had significantly lower RH, compared with maize–wheat and cotton–wheat cropping system. The RH for maize–wheat was 24% higher than rice–wheat, but 19% lower than the cotton–wheat cropping system. The NPP of 7.3 Mg C ha−1 for rice was higher by 2.7 Mg C ha−1 (59%) and 4.7 Mg C ha−1 (181%), compared with maize and cotton ecosystems, respectively. Over the whole annual cycle, atmospheric CO2 assimilated into NPP together totalled 14.1 Mg C ha−1, 11.5 Mg C ha−1 and 9.7 Mg C ha−1, respectively, for rice–wheat, maize–wheat and cotton–wheat cropping systems (Table 6). Rice–wheat had 23% and 45% higher NPP compared with maize–wheat and cotton–wheat cropping system, respectively. The NPP through the above-ground residue biomass yield comprised the largest proportion (48–50%) of the total NPP in three cropping systems, the highest being for rice and the lowest for cotton-based ecosystem (Fig. 3). Similarly, the NPP through different components was significantly higher for rice–wheat, followed by maize–wheat and the lowest for cotton–wheat cropping system.

Table 6 Estimated autotrophic respiration (RA), heterotrophic respiration (RH), net primary production (NPP), gross primary production (GPP), harvest and net ecosystem exchange (NEE) for rice–wheat, maize–wheat and cotton–wheat cropping system in north-western India
Full size table
Fig. 3
figure 3

Net primary production (NPP) through economic yield, above-ground biomass, below-ground biomass, litter and rhizodeposition under rice–wheat, maize–wheat and cotton–wheat cropping systems. Mean values for a crop or cropping systems followed by different letters are significantly different (p < 0.05) by Tukey’s post hoc test. Small letters differentiate crops, and capital letters differentiate cropping systems

Full size image

The GPP was higher by 4.6 Mg C ha−1 (58%) and 8.0 Mg C ha−1 (174%) in rice, compared with maize and cotton ecosystems, respectively (Table 6). Within a cropping system, the GPP of rice–wheat was higher by 4.6 Mg C ha−1 (23%) and 7.6 Mg C ha−1 (46%) than maize–wheat and cotton–wheat cropping systems, respectively. The harvest component was also significantly higher for rice and rice-based cropping, compared with maize and cotton grown in a sequence with wheat. The NEE of CO2 was significantly affected by cropland ecosystems. These results showed a negative value for NEE of − 0.5 Mg C ha−1 for maize and − 3.8 Mg C ha−1 for cotton, in contrast to positive value of NEE for rice ecosystem (Table 6). Over an annual cropping system, the NEE of CO2 was estimated to be 9.5 Mg C ha−1, 5.8 Mg C ha−1 and 2.7 Mg C ha−1, respectively, for rice–wheat, maize–wheat and cotton–wheat cropping system. As compared with cotton–wheat, rice–wheat and maize–wheat cropping system had significantly higher NEE by 3.5-and 2.1-times, respectively. The NECB was significantly lower for cotton, followed by maize and the highest for rice ecosystem (Fig. 4). It was negative for rice (scenario 1), maize and cotton, but positive for wheat. It was 2856, 2895 and 1801 kg C ha−1, respectively, for wheat sown in rotation with rice, maize and cotton. The NECB for a rice–wheat cropping system was 2427, 2448 and 2459 kg C ha−1, respectively, under scenario 1, scenario 2 and scenario 3. The comparison of cropping systems revealed a significantly higher NECB for rice–wheat, compared with maize–wheat and cotton–wheat cropping system.

Fig. 4
figure 4

Net ecosystem carbon budget (NECB) for rice–wheat, maize–wheat and cotton–wheat cropping systems. Line bars indicate standard error from mean. Mean values for a cropping systems followed by different letters are significantly different (p < 0.05) by Tukey’s post hoc test. Small letters differentiate crops, and capital letters differentiate cropping systems

Full size image

Amount of organic C added into SOC pool and a change in SOC pool

The amount of C added to SOC pool in rice ecosystem was significantly higher by 92.8 and 161.3 kg C ha−1 compared with maize and cotton ecosystems, respectively (Fig. 5). The amount of C added to SOC pool through wheat grown in three cropping systems did not differ significantly. The rice–wheat system had significantly higher C addition to SOC pool, compared with maize–wheat and cotton–wheat cropping systems. Maize–wheat had significantly higher C addition of 64.7 kg C ha−1 than cotton–wheat cropping system. The change in SOC pool exhibited a trend similar to the amount of C added to SOC pool. The rice–wheat system with highest amount of C added to SOC pool had significantly higher build-up of SOC as compared to the other two investigated cropping systems (Fig. 6). Similarly, maize–wheat system also had a positive value of SOC pool, although it was significantly lower by 511 kg C ha−1 than the rice–wheat cropping system. Conversely, cotton–wheat had a loss of SOC to the extent of − 817 kg C ha−1.

Fig. 5
figure 5

Amount of carbon added into soil organic carbon (SOC) pool in soils under rice–wheat, maize–wheat and cotton–wheat cropping systems. Line bars indicate standard error from mean. Mean values for a crop or cropping systems followed by different letters are significantly different (p < 0.05) by Tukey’s post hoc test. Small letters differentiate crops, and capital letters differentiate cropping systems

Full size image
Fig. 6
figure 6

Change in soil organic carbon (∆SOC) pool for rice–wheat, maize–wheat and cotton–wheat cropping systems. Line bars indicate standard error from mean. Mean values for a cropping systems followed by different letters are significantly different (p < 0.05) by Tukey’s post hoc test

Full size image

Energy use in crops and cropping system

The EI associated with chemical fertilizer application in cotton was lower by 12% and 6.9%, respectively, compared with rice and maize ecosystems (Table 7). The difference in EI among three cropping systems was statistically non-significant. The EO based on rice equivalent yield of non-rice crops was significantly higher for rice, followed by cotton and the lowest for maize ecosystem. Cotton–wheat had significantly higher EO, compared with rice–wheat and maize–wheat cropping systems. The comparison of rice–wheat and maize–wheat systems revealed that rice–wheat had significantly higher (27%) EO compared with the maize–wheat cropping system. The EP was significantly higher for rice, compared with maize and cotton ecosystems. Within an annual cropping system, rice–wheat had significantly higher EP by 0.52 kg MJ−1 than maize–wheat cropping system. Unlike EP, the ES for cotton was significantly lower, compared with rice and maize ecosystems. The ES for maize ecosystem was higher by ~ 0.79 MJ kg−1 (28%) than the ES for rice. In a cropping system, ES for maize–wheat was higher by ~ 23% compared with rice–wheat and cotton–wheat systems. The ratio of energy output and input (ER) was significantly higher for cotton than for rice and maize ecosystems. The comparison of cropping systems revealed that cotton–wheat had significantly higher ER, compared with the other two investigated cropping systems. Our results showed that cotton–wheat had significantly higher energy efficiency by 33 and 61%, respectively, compared with rice–wheat and maize–wheat cropping systems. The rice and cotton ecosystems had a significantly higher NEG of 195 and 199 GJ ha−1, compared to maize (111 GJ ha−1) ecosystem. Over an annual cropping cycle, cotton–wheat had higher NEG by 221 and 304 GJ ha−1, respectively, compared to rice–wheat and maize–wheat cropping systems (Table 7).

Table 7 Input energy (EI) related to fertilizer nutrient application, output (economic + residues yield) energy (EO) and input–output energy relationships, viz. energy productivity (EP), specific energy (ES), energy ratio (ER) and net energy gain (NEG) for rice–wheat, maize–wheat and cotton–wheat cropping system in north-western India
Full size table

Discussion

Fertilizer-induced GHG emissions

The concept of GHGI that relates agricultural practices to GWP and economic yield of a crop was used to address global challenges of increasing food production with a concomitant decrease in GHGs emission (Mosier et al. 2006; Shang et al. 2010). The CO2 and N2O emissions in rice were significantly higher, compared with cotton ecosystem because of higher fertilizer-N use in rice. However, the GHGs (CO2 and N2O) emission in three cropping systems was at par, because of non-significant difference in chemical fertilizer (N + P2O5 + K2O) use among the three cropland ecosystems. For a rice–wheat cropping system, we estimated GWP by multiplying the region specific CH4 emissions based on irrigation water management, soil type and rice varieties cultivated (Khosa et al. 2011) with a factor of 25 (IPCC 2007a), as there were contradictory reports on the effect of fertilizer-N application on CH4 emission under rice ecosystems (Zou et al. 2009; Banger et al. 2012; Linquist et al. 2012). CH4 is a very small source of GHGs emissions in a upland cropping systems (maize–wheat and cotton–wheat in this study) and CH4 emission makes only a little contribution when calculating net GWP (Huang et al. 2013). Therefore, our estimates for GWP were based on CO2 and N2O emissions for upland ecosystems. Earlier, Liu et al. (2012) reported < 1% contribution of CH4 emission in GWP calculations from an experiment on N and straw application in a maize–wheat cropping system. In the published literature, a contrasting information exists on the effect of fertilizer-N application on increased (Banger et al. 2012), decreased (Zou et al. 2009) and no-change (Cai et al. 2007) in CH4 emission. Using the IPCC default coefficients, the estimated GWP of rice–wheat cropping system varied between 3.62 and 4.43 Mg C ha−1 year−1 for three scenarios based on water management regimes. These estimates of GWP in the present study were within the range (3.5–7.1 Mg C ha−1 year−1) reported by Grace et al. (2003) for irrigated rice–wheat cropping system in Indo-Gangetic Plains, using IPCC default GHG emission coefficients. The GWP of rice–wheat cropping system in our study was much lower than in double rice cropping system (13.4–26.0 Mg C ha−1 year−1) (Shang et al. 2010). It could be ascribed to CH4 emission in large amounts under double rice cropping system, contributing towards GWP. These results showed that CH4 emission comprised between 74 and 99% of total GWP of an annual rice–wheat cropping system. Our results highlight the possibility of 12–18% reduction in GWP of a rice–wheat cropping system with irrigation water management following intermittent flooding and aeration of rice fields. The GHGI varied between 0.62 ± 0.01 and 0.74 ± 0.01 kg CO2e kg−1 grain in rice–wheat cropping system for three scenarios in this study which was nearly the same to the estimates of 0.51–0.72 kg CO2e kg−1 grain among different fertilizer treatments in a rice–wheat cropping systems in China (Ma et al. 2013). Earlier, Qin et al. (2010) reported almost similar range (0.24–0.74 kg CO2e kg−1 grain) of GHGI for a rice–wheat cropping system. However, the higher value of GHGI (3.22 kg CO2e kg−1) simulated by denitrification-decomposition (DNDC) model in Chinese waterlogged paddies has been reported (Li et al. 2006).

The GWP of maize–wheat cropping system observed in the present study was similar to that reported for Chinese soils (Huang et al. 2013). In a maize–wheat cropping system, Huang et al. (2013) reported that GHGI across different fertilizer-N and straw management treatments ranged between 0.18 and 0.56 kg CO2e kg−1, which was nearly the same to that observed in the present study (0.64 ± 0.02 kg CO2e kg−1). The lower GWP for maize–wheat and cotton–wheat cropping systems compared with rice–wheat cropping system was ascribed to the fact that CH4 is only a very small sink under upland cropping systems, viz. maize–wheat and cotton–wheat cropping. Carbon equivalent emission under rice–wheat cropping system of 1.21 ± 0.02 Mg CO2e ha−1 year−1 observed in the present investigation was lower compared to previous estimates of 3.82 Mg CO2e ha−1 year−1 (Bhatia et al. 2005). It could be ascribed to high GWP (13.9 Mg CO2e ha−1) in the study by Bhatia et al. (2005), compared with 2.9 Mg CO2e ha−1 in the present study. The cotton–wheat cropping system had lower CER, compared with maize–wheat and rice–wheat. It indicates that maize–wheat and rice–wheat cropping system had higher amount of C fixed (as crop yield) per unit of the amount of C loss (as GHGs emission) compared with cotton–wheat cropping systems. This suggests that rice–wheat and maize–wheat cropping systems had higher technical C sequestration and GHGs mitigation potential.

Net ecosystem carbon budget and its components

The NPP, an important component of NECB accounts for exchange of C, energy and mass by plant vegetation (Running and Coughlan 1988), reflects the difference between accumulative photosynthesis and RA by plants per unit of time and space (Lieth and Whittaker 1975). The differences between RH and NPP referred as NEE describe how much of C is lost or gained for an ecosystem. The Chinese rice ecosystem is reported to have a NECB of 6340 ± 1490 kg C ha−1 year−1, much higher than this study (− 429 kg C ha−1 year−1) (Gao et al. 2016). It was ascribed to the fact that rice residue after harvesting was removed from the fields in our study, in contrast to in situ incorporation reported in study by Gao et al. (2016). In addition, the higher NPP of 12.1 ± 1.29 Mg C ha−1 year−1 in Chinese rice ecosystem compared with 7.3 ± 0.06 Mg C ha−1 in the present investigation leads to comparable differences in NECB. Rice residue retention offers a significant reduction in C footprints by ~ 12.9%, compared with conventional system of open field burning (Singh et al. 2020).

Under temperate conditions, a negative NECB (− 1679 to − 1969 kg C ha−1) during the dried fallow season and a positive NECB (375–661 kg C ha−1) have been reported in a mono-rice ecosystem (Kim et al. 2017). Swain et al. (2016) reported that although GWP-based CO2–C equivalent of CH4 emission was the sources of C loss from lowland rice ecosystem, but considering different components of NECB, rice ecosystem had a considerable potential to store 1.04 Mg C ha−1 year−1 and therefore, could act as net C sink. A study by Swain et al. (2016) contradicts our results because the harvest was a positive component of NECB in Swain et al. (2016), while a negative component in our study. Nonetheless, compost application of 5 Mg ha−1 year−1 (245 g C kg−1) was the additional C input (e.g. in Swain et al. 2016), compared with our study where manure-C input was not included as a NECB component. The NECB of rice, wheat and a rice–wheat cropping system was 2790, 2710 and 5500 kg C ha−1 year−1, respectively, in fertilizer-N applied plots and 4190, 5140 and 9330 kg C ha−1 year−1, respectively, in manure + fertilizer applied plots (Yang et al. 2015). The lower NECB of rice–wheat cropping system in the present study compared to Yang et al. (2015) inculcates the contribution of C input as harvest. Hollinger et al. (2005) showed that considering biomass export and fuel combustion may transform a soybean crop from a sink to a C source. Zheng et al. (2008) reported that slight C gains occurred during rice growing season and C loss during the non-rice growing season and these C losses were related to low rates of crop residue incorporation and lack of organic manure application. Kim et al. (2017) reported that the annual and seasonal NECB changed in a quadratic fashion by N fertilizer rate, maximized at 113–127 kg N ha−1 and decreased thereafter. The annual NECB was negative (− 1192 to − 1434 kg C ha−1 year−1), irrespective of the N application rate (Kim et al. 2017).

In maize ecosystem, GPP of 8.0 ± 0.11 Mg C ha−1 observed in the present investigation was lower, compared with GPP of 13.1−18.0 Mg C ha−1 year−1 for spring maize in NE, USA (Verma et al. 2005). As high as 17.96 ± 0.92 Mg C ha−1 year−1 of GPP for irrigated maize ecosystem has also been reported (Suyker and Verma 2012). In north-east Italy, Alberti et al. (2010) reported a nearly similar GPP range of 8.05–9.98 Mg C ha−1 year−1 for maize ecosystem that has been observed in the present study. Our results also corroborate the findings of Gao et al. (2017), who reported a similar range of GPP (7.8–8.9 Mg C ha−1 year−1) for spring maize in Shouyang, Shanxi, China. In the present study, the NEE of − 0.5 Mg C ha−1 year−1 for maize ecosystem was in the range between − 0.03 and − 0.62 Mg C ha−1 year−1 in China (Du and Liu 2013). A negative NEE value of − 10.2 g C m−1 d−1 for maize has also been reported earlier (Béziat et al. 2009). The NEE represents the C budget and accounts for C release and absorbed by the ecosystem through assimilation and respiratory processes, respectively, and a negative NEE value indicates a C sink (Prescher et al. 2010). A positive NEE value, however, indicates a C source, where C budget of an ecosystem comprised NEE, non-respiratory C losses (i.e. harvest) and non-photosynthetic C gain (i.e. through the application of organic manures) (Ammann et al. 2007; Schulze et al. 2005). The comparison of three land-use systems, viz. managed forest (spruce), a grassland and a cropland, revealed that for a forest site, NEE ranges from − 698 to − 444 g C m−2, in contrast grassland (− 177 to − 62 g C m−2) while cropland (− 115 to − 32 g C m−2) apparently had distinct lower C sink (Prescher et al. 2010). Our results indicate that maize ecosystem was a source of 2879 kg C ha−1 year−1, while wheat was a net sink of 2884 kg C ha−1 year−1. Therefore, the maize–wheat cropping system behaved as a C sink of 5.3 kg C ha−1 year−1. Wang et al. (2015) reported that annual maize–wheat cropping behaved as C source of 770 kg C ha−1 year−1, winter wheat being C sink, while summer maize as C source. In northern China, Jun et al. (2006) reported that maize–wheat cropping was a C source with NPP of 1080–3410 kg C ha−1 year−1. In Nebraska (USA), Grant et al. (2007) reported that maize–soybean cropping system was a net C source because of failure of positive maize net biome production to offset negative soybean net biome production in the next season.

In the present study, rice ecosystem appeared as C source of 397–429 kg C ha−1 year−1, while wheat was a C sink of 2856 kg C ha−1 year−1, and a rice–wheat cropping system was a net sink of 2427–2459 kg C ha−1 year−1. Almost a similar range of NECB (560–1640 kg C ha−1 year−1, mean = 1020 kg C ha−1 year−1) in a rice–wheat cropping system has been reported earlier (Zhang et al. 2014). Ceschia et al. (2010) reported a mean NECB value of 1120 kg C ha−1 year−1 for winter wheat. Soils under rice–wheat cultivation had a high capacity for soil C sequestration (Benbi and Brar 2009; Lu et al. 2009; Shang et al. 2010; Singh and Benbi 2020), which plays important role in both crop productivity and the global C cycle (Huang and Sun 2006). A lower positive value of NECB (~ 900 kg C ha−1 year−1) of different crop rotations in Europe has been reported (Janssens et al. 2003; Kutsch et al. 2010). In a cereal-based cropping systems, Osborne et al. (2010) reported that in a situation where only economic yield (grains) is being exported from the fields, while most of residue biomass remains there, C input from decomposing residues enriches the SOC pool. In a rice–wheat cropping system, rice residue retention on soil surface or incorporation into soil had 4.6–4.8-times higher C added through rice residue + roots to SOC pool, compared with the conventional tillage coupled with open field in situ residue burning (Singh et al. 2020). In contrast, the situation with residue biomass exported from the fields after crop harvest, photo-synthetically assimilated C is lost, and has been a negative component of NECB. The situation of C export as harvest is more interesting in the present study, where the farmers prefer to burn the rice residue, use wheat and maize residues as animal feed, and cotton sticks as fuel. Among the three cropping systems, cotton–wheat is the largest C source, followed by maize–wheat and the lowest by rice–wheat cropping system. It could be attributed to (1) the lower economic yield of wheat, maize and cotton, compared with rice, and (2) non-availability of farmyard manure in requisite amounts. Therefore, it is likely that NECB in such situations will be negative, and an ecosystem will act as a C source. As such, if C loss as harvest is not been compensated by the application of manures, NECB will correspond to C losses from the soil. Ceschia et al. (2010) reported that 15 representative crop species common to Europe covering a large climatic gradient had a negative NEP (− 2.84 ± 2.28 Mg C ha−1 year−1), and most cropping systems behaved as atmospheric C sinks. Average NECB of 1.38 ± 2.39 Mg C ha−1 year−1 corresponds to an annual loss of about 2.6 ± 4.5% of the SOC content (Ceschia et al. 2010). Therefore, a cropland ecosystem may be a source or C sink, and that an entire cropping system is considered to estimate a C balance.

Energy input–output relationships

Input energy related to the application of chemical fertilizers (N + P2O5 + K2O) in three cropping systems did not differed significantly. It could be ascribed to non-significant difference in the rate of chemical fertilizers applied in three cropland ecosystems. Although the rate of fertilizer-N application was significantly higher in rice, an expected increase in EI due to fertilizer-N application was compensated by decreased rate of fertilizer-P2O5 in rice, compared with maize and cotton. Further, fertilizer-K2O application rate in three crops (rice, maize and cotton) was statistically non-significant, resulting in a non-significant difference in EI through chemical fertilizers in three ecosystems. These results corroborate earlier research findings, showing that fertilizer-N are major energy input in rice cultivation (Nabavi-Pelesaraei et al. 2014a, b, 2019a; Singh et al. 2019). Energy output calculated based on rice equivalent yield was significantly lower for maize, compared with rice and cotton ecosystem. It was also observed that EO was highest for cotton–wheat, followed by rice–wheat and the lowest for maize–wheat. The ER and NEG were significantly higher for cotton–wheat, compared with rice–wheat and maize–wheat ecosystem. Higher EP for cotton–wheat than in maize–wheat cropping system could be attributed to significantly higher rice equivalent yield of cotton, with a similar fertilizer-N rate in both the crops. A positive energy balance in crop production has been ascribed to maximized energy output with the lowest energy input (Shahan et al. 2008).

Conclusions

Rice–wheat cropping system had higher NPP and NECB, followed by maize–wheat and the lowest for cotton–wheat cropping system because of high C fixation ability per unit of C loss. Rice–wheat system behaved as the largest, while maize–wheat as a moderate net C sink, as opposed to cotton–wheat that behaved as a net C source. Although rice–wheat ecosystem had higher technical C sequestration and GHGs mitigation potential, these results highlight overwhelming significance of crop residue retention and irrigation water management for long-term sustainability of rice–wheat system. Irrigation water management with intermittent flooding and multiple aerations will not only help reducing the GWP, but will also check the groundwater depletion. There exists an immense potential to further increase the NECB of rice–wheat cropping system through in situ management of crop residue with co-benefits of C sequestration in soil.