Trade-Off Between System Yield and Area-Scaled Carbon Cost Among Cropping Systems Under Contrasting Water Management on the North China Plain

Abstract

On the North China Plain, the winter wheat season is poorly synchronized with precipitation, making the traditional winter wheat–summer maize system heavily dependent on supplemental irrigation and associated carbon inputs. Based on a split-plot field experiment in Shenzhou, Hebei, from October 2022 to October 2025, this study evaluated the trade-off between annual system yield and area-scaled carbon emission among six cropping systems under conventional irrigation (CK) and rainfed management (R). The winter wheat–summer maize system (WM) maintained the highest grain-oriented annual system yield (22.91 t ha⁻¹ yr⁻¹ under CK), but it also showed the highest area-scaled carbon emission (11.97 t CO₂-eq ha⁻¹ yr⁻¹). The winter wheat–summer maize–spring maize system (WMM) reduced area-scaled carbon cost relative to WM (8.97 vs. 11.97 t CO₂-eq ha⁻¹ yr⁻¹ under CK), whereas its product-scaled carbon footprint remained comparable to or slightly higher than that of WM. Under a unified dry-matter basis, the double silage-maize system (FM) showed the lowest dry-matter-scaled carbon footprint (CF_DM; 193.85 and 175.71 kg CO₂-eq t DM⁻¹ under CK and R, respectively). Soil respiration in 2025 varied mainly with observation date and cropping-system configuration, and soil organic carbon (SOC) stock at the 2025 harvest differed among cropping systems, water-management regimes, and soil depths. Overall, WM remained the highest-yielding option under a grain-supply objective, whereas FM, the ryegrass–early-summer maize system (RM), and the forage winter wheat–early-summer maize system (FWM) were relatively more suitable under multifunctional biomass-supply and low-carbon-transition objectives.

1. Introduction

The North China Plain (NCP) is one of the most important grain-producing regions in China and also one of the agricultural regions most severely affected by groundwater over-extraction. Because precipitation under the monsoon climate is concentrated mainly in summer, the peak water demand of winter wheat is markedly out of phase with natural precipitation. As a result, the traditional winter wheat–summer maize double-cropping system has long relied on supplemental irrigation [1,2,3] to maintain high yields [4,5]. While this system supports regional food security, frequent irrigation, high fertilizer input, and intensive mechanized operations also increase the carbon cost of the field-production stage. Optimizing cropping systems under groundwater constraints therefore requires a better balance among yield, water use, and low-carbon objectives.

Previous studies have shown that irrigation optimization can improve water-use efficiency [5,6,7] and alleviate groundwater pressure to some extent within the winter wheat–summer maize system [1,2,4]. However, when the annual crop configuration remains unchanged, irrigation reduction alone cannot fundamentally alter the co-occurrence of high water consumption and high input intensity. Studies focusing on diversified rotations in the North China Plain have shown that replacing or adjusting the conventional wheat–maize sequence can reduce carbon footprint [8,9,10] while maintaining part of system productivity [11,12,13]. Other studies have further shown that alternative cropping patterns and diversified planting can modify crop output [12,14], soil organic carbon [15], emission characteristics [16,17], and broader sustainability performance [14].

From the perspective of farmland ecosystem processes, changes in cropping systems can influence not only overall system output but also short-term soil-carbon processes, including soil respiration [18,19] and soil organic carbon (SOC) [15], through shifts in crop phenology, root turnover, residue input, and the soil hydrothermal regime. It should be emphasized, however, that input-related carbon emissions and biologically driven soil-carbon processes do not belong to the same accounting boundary: the former reflects management-related burdens during field production, whereas the latter represents ecological process responses within the agroecosystem. If these two dimensions are not explicitly distinguished in the Materials and Methods and Discussion sections, “anthropogenic emission inventories” and “ecosystem process indicators” can easily be conflated.

Despite recent progress, three issues remain prominent in multi-system comparisons. First, time windows and comparison units are often inconsistent, especially when two-year/three-crop systems are directly compared with one-year/one-crop or one-year/two-crop systems. Second, product functional units differ: direct comparison of grain, forage, and silage on a fresh-mass basis can be strongly affected by moisture content and end use. Third, the trade-off between production and emissions often remains conceptual, without an analytical framework that places system yield and area-scaled carbon cost on the same plane. Accordingly, this study focused on a single core question: under groundwater constraints, how do different cropping systems under two water-management regimes trade-off system yield against area-scaled carbon cost, and are these differences accompanied by short-term soil-carbon responses? The specific objectives were to (i) compare annual system yield and area-scaled carbon emission across six cropping systems under conventional irrigation and rainfed management within a unified crop-year/closed-rotation framework, (ii) use product-scaled carbon footprint (CF_prod) and dry-matter-scaled carbon footprint (CF_DM) only as supplementary indicators under comparable production objectives, and (iii) characterize associated short-term soil-respiration and soil organic carbon (SOC) responses. We hypothesized that reducing winter-wheat frequency would lower area-scaled carbon cost relative to WM, that biomass-oriented systems would show lower CF_prod or CF_DM than grain-oriented systems, and that cropping-system reconfiguration would be accompanied by short-term differences in soil-carbon processes.

2. Materials and Methods

2.1. Site Description and Experimental Design

The field experiment was conducted from October 2022 to October 2025 at the Experimental Station of the Institute of Dryland Farming, Hebei Academy of Agriculture and Forestry Sciences, Shenzhou, Hengshui, Hebei Province, China (37.91° N, 115.71° E; 20 m above sea level). A two-factor split-plot design was used, with a nested structure consisting of blocks, main plots (water management), and subplots (cropping system), and each treatment was replicated three times. The study adopted a unified crop-year/closed-rotation accounting framework: October 2022 to October 2023 was defined as crop year 2022–2023, and so on. Figure 1a summarizes the temporal arrangement of the six cropping systems across the three crop years. Because same-source meteorological data were available only from July 2023 onward, Figure 1b shows monthly precipitation and mean air temperature for July 2023 to October 2025.

The main-plot factor consisted of conventional irrigation (CK) and rainfed management (R). The subplot factor consisted of six cropping systems: conventional winter wheat–summer maize double cropping (WM), a winter wheat–summer maize–spring maize system with three crops in two years (WMM), forage winter wheat–early-summer maize (FWM), ryegrass–early-summer maize (RM), double silage maize (FM), and mono-cropped spring maize (M). Rotation calendars, irrigation schedules, major inputs, and functional-unit boundaries are summarized in

Table 1. Rotation calendar, crop-year accounting, field inputs, and functional-unit boundaries of different cropping systems.

System Code	Rotation Calendar and Approximate Sowing/Harvest Window	CK Irrigation Schedule	Annual or Annualized Irrigation Amount (mm)	Fertilizer Rate (Pure Nutrients)	Inputs/Processes Included or Excluded from Emission Accounting	Main Yield Basis (Crop Year)	Unified Functional Units
WM	Winter wheat (early Oct.–early Jun. of the following year) → summer maize (mid-Jun.–early Oct.)	Winter wheat: 3 × 75 mm; summer maize: 1 × 75 mm	300	For each crop season: N 300, P2O5 150, K2O 225 kg ha−1	Included: seed, fertilizer, diesel, machinery operations, electricity for irrigation pumping; not listed separately: plastic film, pesticides	Grain yield	Area-scaled emission; dry matter
WMM	Crop year 1: winter wheat (early Oct.–early Jun. of the following year) → summer maize (mid-Jun.–early Oct.); crop year 2: spring maize (late May–late Sep.)	Winter wheat: 3 × 75 mm; summer maize: 1 × 75 mm; spring maize: 1 × 75 mm	187.5 (annualized from 375 mm over two years)	Same as above	Included: seed, fertilizer, diesel, machinery operations, electricity for irrigation pumping; not listed separately: plastic film, pesticides	Total grain yield (annualized after calculation over the complete two-year/three-crop closed rotation)	Area-scaled emission; dry matter
FWM	Forage winter wheat (early Oct.–early May of the following year) → early-summer maize (early May–late Sep.)	Forage winter wheat: 1 × 75 mm; early-summer maize: 2 × 75 mm	225	Same as above	Included: seed, fertilizer, diesel, machinery operations, electricity for irrigation pumping; not listed separately: plastic film, pesticides	First-season forage fresh/dry matter; second-season maize grain	Area-scaled emission; dry matter
RM	Ryegrass (early Oct.–early May of the following year) → early-summer maize (mid-May–late Sep.)	Ryegrass: 1 × 75 mm; early-summer maize: 2 × 75 mm	225	Same as above	Included: seed, fertilizer, diesel, machinery operations, electricity for irrigation pumping; not listed separately: plastic film, pesticides	First-season forage fresh/dry matter; second-season maize grain	Area-scaled emission; dry matter
FM	Silage maize 1 (late Apr.–mid-Jul.) → silage maize 2 (mid-Jul.–early Oct.)	First season: 2 × 75 mm; second season: 1 × 75 mm	225	Same as above	Included: seed, fertilizer, diesel, machinery operations, electricity for irrigation pumping; not listed separately: plastic film, pesticides	Fresh biomass of double silage maize; converted to dry matter in unified analyses	Area-scaled emission; dry matter
M	Spring maize (late May–late Sep.)	1 × 75 mm	75	Same as above	Included: seed, fertilizer, diesel, machinery operations, electricity for irrigation pumping; not listed separately: plastic film, pesticides	Grain yield	Area-scaled emission; dry matter

.

For WMM, crop year 2022–2023 corresponded to winter wheat–summer maize, crop year 2023–2024 corresponded to spring maize, and crop year 2024–2025 returned to winter wheat–summer maize. In all cross-system comparisons, WMM was first calculated on the basis of its complete two-year/three-crop closed rotation and then converted to an annual mean; incomplete single-year observations were not directly ranked against the other systems.

2.2. Field Management and Measurements

To ensure comparable accounting across systems, this study used the crop year/closed rotation as the basic unit of calculation. Table 1 summarizes the rotation calendar, sowing and harvest windows, irrigation schedules, fertilizer inputs, emission-accounting boundaries, and functional units for each system. Area-scaled carbon emission (CF_area) was used as the primary scale for cross-system comparison; product-scaled carbon footprint (CF_prod) was used as a supplementary indicator within the same production objective; and CF_DM under a unified dry-matter basis was used to further reduce functional-unit differences among grain, forage, and silage outputs.

The winter wheat cultivar was Hengguan 35, the ryegrass cultivar was Jisi 4, the spring maize cultivar was Zhengdan 958, the summer maize cultivar was MY73, and the silage maize cultivar was Fangyu 3201. For each crop season, nutrient input was standardized at 300 kg N ha⁻¹, 150 kg P₂O₅ ha⁻¹, and 225 kg K₂O ha⁻¹. All fertilizer rates reported in this paper therefore refer to a single crop season rather than to the total amount for a crop year. Grain yield and silage-maize yield were determined from a 5 m double-row harvest area, corresponding to approximately 6.0–6.4 m² per plot. The fresh forage output of FWM and RM was converted to dry matter using the measured fresh-to-dry ratio when a unified dry-matter comparison was required.

Soil samples were collected at the major pre-sowing and post-harvest stages corresponding to each system and were divided into the 0–20, 20–40, and 40–60 cm layers. The 2025 cross-sectional SOC comparison was based on post-harvest profile samples, whereas the time comparison was restricted to cropping-system subsets with consistent pre-sowing/post-harvest pairing. SOC concentrations were obtained from archived laboratory analyses at the experimental station and were subsequently used for stock calculation in this study. Straw return or residue incorporation was included in the original input inventory when it actually occurred in a given season, but no extrapolation beyond the field-production boundary was made other than for the explicitly listed inputs.

Soil respiration was measured with an LI-8100A portable infrared gas analyzer (LI-COR Biosciences, Lincoln, NE, USA). PVC collars (inner diameter, 20 cm; height, 10 cm; protruding 5 cm above the soil surface) were installed in each plot after sowing. Measurements were taken at 09:00–11:00 during key crop-growth stages, with additional observations after rainfall or irrigation events. All plots within a given observation campaign were measured within the same fixed morning window; because a formally randomized within-window sequence was not included as an experimental factor, the integrated result in this study is explicitly interpreted as a cumulative soil-CO₂ release estimate based on fixed-time observations rather than as a full-day flux. It was therefore used to compare short-term process responses among systems and was not interpreted as a continuous, year-round net flux measured by automatic monitoring.

The LI-8100A system calculates the instantaneous soil-respiration rate using a closed dynamic chamber method based on the initial rate of change in water-corrected CO₂ mole fraction within the chamber. The calculation is given below.

$$R_s={\displaystyle \frac{V{P}_0\left(\partial C^{\prime }/\partial t\right)}{RS\left(T_0+273.15\right)\left(1-W_0\right)}}$$

(1)

where R_s is the soil respiration rate; V is the total volume of the measurement collar and gas circuit; P₀ is the absolute chamber pressure at the start of measurement; W₀ is the initial water-vapor mole fraction; R is the gas constant; S is the collar-covered soil area; T₀ is the chamber air temperature at the start of measurement; and (∂C′/∂t) is the initial rate of change in the water-corrected CO₂ mole fraction.

Cumulative soil-CO₂ release was estimated using trapezoidal integration between adjacent observation dates.

$$E_{soil}=\sum \left[{\displaystyle \frac{R_i+R_{i+1}}{2}}\times \left(t_{i+1}-t_i\right)\right]\times 0.038016$$

(2)

where E_soil is the cumulative soil-CO₂ release estimated from fixed-time observations; R_i and R_i+1 are soil respiration rates at two adjacent observation dates; t_i+1 − t_i is the interval between the two observations; and 0.038016 is the conversion coefficient used to integrate time-based fluxes into t CO₂ ha⁻¹.

Crop yield was converted to a unified annual system yield. For grain systems, annual system yield was calculated as the sum of grain yield from one or two crop seasons within the crop year. For FWM and RM, annual yield was calculated as the sum of first-season forage output and second-season maize grain yield. For FM, annual yield was the sum of the two silage-maize harvests. In the unified dry-matter analysis, grain crops were converted on the basis of 14% standard moisture content, whereas forage and silage outputs were converted using the measured fresh-to-dry ratio.

2.3. Soil Organic Carbon Stock, Carbon Footprint, and Emergy Evaluation

SOC stock was calculated by layer summation. Carbon footprint was quantified using life-cycle assessment (LCA), with the system boundary defined as the field-production stage from sowing to harvest (cradle to farm gate). Machinery manufacture, infrastructure construction, product processing, transport, storage, and consumption were not included. Carbon-emission accounting included only emissions from the production of agricultural inputs, field operations, and electricity used for irrigation pumping; cumulative soil respiration was not directly incorporated into CF totals, following inventory-based accounting principles commonly used in crop-production assessments [20,21]. Accordingly, CF_area was used to characterize management-related burdens during field production, whereas soil respiration and SOC were used to characterize soil-carbon process responses.

The principal equations used for SOC stock, carbon footprint, and emergy analysis are listed below.

SOC_stock = Σ(C_i × ρ × H_i × 0.1)

(3)

CF_area = Σ(A_i × EF_i)

(4)

CF_prod,i = CF_area,i/Y_sys,i

(5)

CF_DM = CF_area/Y_DM

(6)

where SOC_stock is the SOC stock in the 0–60 cm profile; C_i is the SOC concentration of the i-th layer (g kg⁻¹); ρ is bulk density (set uniformly at 1.47 g cm⁻³ in this study); H_i is the thickness of the corresponding layer (cm); and 0.1 is the unit-conversion coefficient that converts g kg⁻¹, g cm⁻³, and cm into t C ha⁻¹. CF_area, CF_prod, and CF_DM represent area-scaled carbon emission, product-scaled carbon footprint, and dry-matter-scaled carbon footprint, respectively. A_i is the amount of the i-th agricultural input, and EF_i is the corresponding emission factor.

In the emergy analysis, system inputs were divided into natural-resource inputs, purchased-resource inputs, and outputs. Unit emergy value (UEV), emergy yield ratio (EYR), environmental loading ratio (ELR), and emergy sustainability index (ESI) were calculated as supplementary indicators following commonly used agroecosystem emergy approaches [22,23] and broader emergy or energy-efficiency evaluations of cropping systems [24,25]. It should be noted that emergy indicators were used only to provide additional information on resource-use efficiency and environmental loading and were not used to rank the systems within the main “system yield–area-scaled carbon cost” framework of the paper. The carbon-emission coefficients used in the inventory are summarized in

Table 2. Carbon-emission coefficients of agricultural inputs.

Item	Carbon-Emission Coefficient
Wheat seed	0.58 kg CO2-eq·kg−1
Maize seed	1.98 kg CO2-eq·kg−1
Nitrogen fertilizer (N)	7.76 kg CO2-eq·kg−1
Phosphate fertilizer (P2O5)	1.63 kg CO2-eq·kg−1
Potash fertilizer (K2O)	0.65 kg CO2-eq·kg−1
Diesel	0.89 kg CO2-eq·kg−1
Electricity for irrigation	0.6516 kg CO2-eq·kWh−1

, and the main emergy indices and equations are summarized in

Table 3. Main emergy indices and equations.

Emergy Index	Equation	Meaning
Unit emergy value (UEV)	UEV = U/Y	Emergy input required per unit of product
Emergy yield ratio (EYR)	EYR = Y/(EMF + EMT)	Capacity of the system to generate output from purchased inputs
Environmental loading ratio (ELR)	ELR = (EMF + EMT)/EMR	Degree of environmental pressure imposed by the system
Emergy sustainability index (ESI)	ESI = EYR/ELR	Integrated indicator of system sustainability

Note: U is total emergy input, Y is emergy output, EMF denotes non-renewable resource input, EMT denotes purchased-resource input, and EMR denotes renewable-resource input. Definitions, headers, and variable annotations for UEV, EYR, ELR, and ESI are kept consistent throughout the manuscript.

.

2.4. Statistical Analysis

Data processing was conducted in Microsoft Excel 2019(Microsoft Corporation, Redmond, WA, USA), figures were prepared in Origin 2021 (OriginLab Corporation, Northampton, MA, USA), and statistical analyses were performed in Python 3.13 using the statsmodels package (Python Software Foundation, https://www.python.org/). For the split-plot experiment, annual system yield, soil respiration rate, and the stratified post-harvest 2025 soil organic carbon (SOC) dataset were re-analyzed using linear mixed models (LMMs) that reflected the nested structure of blocks, main plots (water management), and subplots (cropping system). Water management, cropping system, and the relevant repeated-measure factor (crop year, observation date, or soil depth) were specified as fixed effects, whereas block, block × water management, and subplot were specified as random effects. Wald χ² tests were used to assess fixed effects, and pairwise comparisons of estimated marginal means were conducted only for measured variables analyzed with the mixed model. The statistical models, hierarchical structure, random-effects specification, and software implementation are summarized in

Table 4. Statistical models, hierarchical structure, random-effects specification, and software implementation details.

Response Variable	Observation Unit	Fixed Effects	Hierarchy/Subject	Working Correlation or Random Structure	Software Implementation
Soil respiration rate (MixedLM)	Subplot × observation date	Water management, cropping system, observation date, and water management × cropping system	Block/block × water management/subplot	— (random-effects structure)	Python statsmodels MixedLM
Annual cumulative soil-CO2 release	Subplot × crop season	Derived by trapezoidal integration of instantaneous fluxes; reported descriptively only	—	—	Excel/Origin summary
System yield (MixedLM)	Subplot × crop year (or annualized closed-rotation mean)	System yield: water management, cropping system, crop year, and their interactions	Block/block × water management/subplot	— (random-effects structure)	Python statsmodels MixedLM
CFarea, CFprod, and CFDM (descriptive summary)	Input inventory × system	Derived CF indices: no inferential testing	—	—	Excel/Origin summary
Stratified SOC data (cross-sectional comparison at the 2025 harvest)	Subplot × soil depth	Water management, cropping system, soil depth, and their interactions	Block/block × water management/subplot	— (random-effects structure)	Python statsmodels MixedLM

Note: MixedLM results are fully consistent with the split-plot hierarchy. Block, block × water management (main plot), and subplot were treated as random effects. Because CF_area, CF_prod, and CF_DM are derived from the input inventory, they are reported descriptively and are not subjected to inferential testing based on field replication. WMM system yield, area-scaled carbon emission, and product-scaled carbon footprint were first calculated on the basis of the complete two-year/three-crop closed rotation and then converted to annual means. Statistical inference for system yield, soil respiration, and the 2025 cross-sectional SOC comparison was based on split-plot linear mixed models, whereas CF_area, CF_prod, and CF_DM were derived from the input inventory and measured yield and were therefore used descriptively.

.

The 2025 cross-sectional SOC comparison was analyzed with an LMM based on post-harvest stratified profile data. The time comparison within the subset of systems with consistent pre-sowing/post-harvest pairing was retained only as a supplementary analysis and was not extrapolated to long-term carbon-sequestration conclusions. CF_area, CF_prod, and CF_DM are derived indicators calculated from the input inventory and measured yield and are therefore reported descriptively rather than subjected to inferential tests based on field replication.

3. Results

3.1. Dynamics of Soil Respiration Rate Under Different Cropping Systems

During the 2025 wheat season, soil respiration rates of common wheat and ryegrass both followed a rise-and-decline pattern over crop development, with clear pulse peaks after irrigation or rainfall events. Soil respiration rates ranged from 3.15 to 6.99 μmol m⁻² s⁻¹ for common wheat and from 2.73 to 6.19 μmol m⁻² s⁻¹ for ryegrass. From regreening to grain filling, respiration gradually increased with rising temperature and enhanced root activity and then declined toward crop maturity.

Across all cropping systems in 2025, soil respiration rates exhibited pronounced seasonal fluctuations and short-term pulses after irrigation or rainfall. In general, systems containing winter crops showed progressively increasing respiration from regreening to grain filling, whereas maize-season peaks occurred mainly during the hot and humid period of July–August. FM reached a relatively high peak in early July, whereas WM remained at a high level in late July, indicating a clear temporal offset in the period of intense soil-CO₂ release among systems. The seasonal trajectories for the wheat and maize seasons are shown in Figure 2 and Figure 3, respectively.

After accounting for block, main-plot, and subplot hierarchies, the split-plot LMM showed significant effects of observation date (Wald χ² = 444.69, p < 0.001) and cropping system (Wald χ² = 18.27, p = 0.003) on soil respiration rate, whereas the main effect of water management (Wald χ² = 0.42, p = 0.517) and its interaction with cropping system (Wald χ² = 2.51, p = 0.775) were not significant. This indicates that differences in soil respiration in 2025 were driven mainly by seasonal progression and cropping-system configuration, whereas the influence of irrigation treatment on instantaneous respiration was expressed more as short-term pulse responses at specific observation dates than as a consistent whole-season effect. The corresponding split-plot LMM results are summarized in

Table 5. Split-plot linear mixed model results for soil respiration rate.

Source of Variation	df	Wald χ2	p-Value
Water management	1	0.42	0.517
Cropping system	5	18.27	0.003
Observation date	12	444.69	<0.001
Water management × cropping system	5	2.51	0.775

Note: Table 5 was based on a linear mixed model consistent with the split-plot design. Fixed effects included water management, cropping system, observation date, and water management × cropping system. Random effects included block, block × water management (main plot), and subplot. p-values are based on Wald χ² tests.

.

Raw soil-respiration observations, cumulative soil-CO₂ release estimates based on fixed-time observations, and the summarized stratified SOC data used in this study are provided in Supplementary Table S3.

3.2. Changes in Soil Organic Carbon Stock Under Different Cropping Systems

At the 2025 harvest, differences in SOC stock in the 0–60 cm profile widened among systems, as shown in Figure 4. WMM showed the highest profile SOC stock (62.88 t C ha⁻¹), followed by M (56.68 t C ha⁻¹). FWM, RM, and WM had 50.88, 50.35, and 47.81 t C ha⁻¹, respectively, whereas FM showed the lowest value (38.45 t C ha⁻¹). Based on an LMM consistent with the split-plot structure, cropping system (Wald χ² = 20.08, p = 0.001), water management (Wald χ² = 4.39, p = 0.036), and soil depth (Wald χ² = 33.78, p < 0.001) all significantly affected SOC stock. The depth × system interaction was significant (Wald χ² = 62.13, p < 0.001), whereas the water × system interaction was not (Wald χ² = 2.59, p = 0.762).

After summing the 0–20, 20–40, and 40–60 cm layers, SOC differences among systems showed a clear profile dependence. Because harvest dates differed among systems, the time comparison was restricted to subsets with consistent pre-sowing/post-harvest pairing: WM, WMM, and FM used the “pre-sowing in 2024–harvest in 2025” pairing, whereas FWM, RM, and M used the “September 2024–September 2025” pairing. Under this restricted condition, the main effect of time was not significant, indicating that short-term SOC changes should be interpreted as directional responses rather than as evidence of long-term carbon-sequestration superiority.

Accordingly, SOC in this study serves mainly as auxiliary evidence of short-term soil-carbon responses, indicating that cropping-system configuration may alter profile carbon distribution and seasonal input–output relationships, but it is not used to rank the six systems in terms of long-term carbon sequestration.

3.3. Carbon-Footprint Characteristics of Different Cropping Systems

Carbon footprint was calculated on a crop-year basis by first summing the input-related emissions of each system within a year and then dividing by the corresponding annual system yield or the unified dry-matter yield. Because WMM is a two-year/three-crop system, cross-system comparison was conducted within a comparable window consistent with its closed rotation, and WMM values were converted to annual means before comparison.

Within the comparable window, FM showed the lowest CF_prod, reflecting the combination of relatively low area-scaled carbon emission and high annual system output. Among the grain-oriented systems, WMM had lower CF_area than WM, but its CF_prod remained comparable to or slightly higher than that of WM under both water-management regimes. These descriptive results indicate that reducing winter wheat frequency can lower area-scaled carbon cost, but this reduction does not necessarily translate into a lower product-scaled carbon footprint when the annual output denominator also declines.

Table 6. Comparable-window annual system yield, area-scaled carbon emission, and product-scaled carbon footprint consistent with the closed WMM rotation.

System	Water Management	Annualized System Yield over the Closed Comparable Window (t ha−1 yr−1)	Area-Scaled Carbon Emission (t CO2-eq ha−1 yr−1, Based on the Input Inventory)	Product-Scaled Carbon Footprint (kg CO2-eq t−1)
WM	CK	22.91 ± 0.11	11.97	522.65 ± 2.59
WM	R	20.43 ± 0.51	10.89	533.20 ± 13.60
WMM	CK	16.67 ± 0.48	8.97	538.28 ± 15.42
WMM	R	14.44 ± 0.41	8.16	565.14 ± 16.08
FWM	CK	33.69 ± 1.79	11.70	347.96 ± 18.12
FWM	R	26.23 ± 1.98	10.89	416.69 ± 32.78
RM	CK	38.93 ± 1.18	11.70	300.72 ± 8.99
RM	R	40.01 ± 2.18	10.89	272.69 ± 15.21
FM	CK	42.10 ± 2.85	6.26	149.11 ± 10.40
FM	R	40.94 ± 2.61	5.44	133.35 ± 8.24
M	CK	11.14 ± 0.17	5.99	537.72 ± 8.02
M	R	9.88 ± 0.23	5.45	551.24 ± 12.67

Note: Table 6 was based on the comparable window of 2022–2024, and WMM values were converted to annual means after calculation over the complete closed two-year/three-crop rotation. Area-scaled carbon emission is a descriptive result derived from the input inventory and unified management records and was therefore not subjected to inferential testing based on field replication. WM = conventional winter wheat–summer maize; WMM = winter wheat–summer maize–spring maize; FWM = forage winter wheat–early-summer maize; RM = ryegrass–early-summer maize; FM = double silage maize; M = mono-cropped spring maize; CK = conventional irrigation; R = rainfed management.

summarizes the comparable-window results, and Figure 5 shows the corresponding CF_prod patterns.

Annual input-related carbon-emission details and crop-season carbon-footprint results are provided in Supplementary Table S1, and the input-inventory coefficients, irrigation-electricity conversion, WMM closed-rotation conversion, and dry-matter conversion procedures are provided in Supplementary Table S2.

3.4. Effects of Water Management on System Yield

Re-analysis with the split-plot LMM showed that system yield was significantly affected by cropping system (Wald χ² = 438.21, p < 0.001) and water management (Wald χ² = 10.11, p = 0.001). Because significant interactions were present, the main effect of crop year was no longer interpreted independently (Wald χ² = 4.39, p = 0.111). In addition, the interactions cropping system × water management (Wald χ² = 13.61, p = 0.018), cropping system × crop year (Wald χ² = 808.39, p < 0.001), and the three-way interaction (Wald χ² = 68.47, p < 0.001) were significant, whereas the water management × crop year interaction was not (Wald χ² = 3.38, p = 0.184). This indicates that responses to irrigation treatment differed among systems and depended on crop year. Overall, conventional irrigation tended to increase system yield in the grain-oriented systems, but the magnitude of the response was not consistent across systems or years. WM remained the most productive grain system, whereas WMM showed reduced annual mean system yield after lowering winter-wheat frequency but still outperformed M.

Table 7. System yield of grain-oriented systems across crop years and the comparable-window annual mean.

System	2022–2023	2023–2024	2024–2025	Comparable-Window Annual Mean
WM-CK	22.66 ± 0.38	23.15 ± 0.27	20.93 ± 1.01	22.91 ± 0.11
WM-R	18.74 ± 0.30	22.12 ± 0.95	18.33 ± 1.71	20.43 ± 0.51
WMM-CK	22.62 ± 0.89	10.73 ± 0.54	21.34 ± 2.80	16.67 ± 0.48
WMM-R	18.81 ± 0.59	10.07 ± 0.31	19.63 ± 2.11	14.44 ± 0.41
M-CK	11.57 ± 0.10	10.70 ± 0.40	10.89 ± 0.59	11.14 ± 0.17
M-R	10.39 ± 0.19	9.38 ± 0.43	9.03 ± 1.75	9.88 ± 0.23

Note: WM = conventional winter wheat–summer maize; WMM = winter wheat–summer maize–spring maize; M = mono-cropped spring maize; CK = conventional irrigation; R = rainfed management.

summarizes the grain-oriented systems, whereas

Table 8. System yield of grain–forage and silage systems across crop years and the comparable-window annual mean.

System	2022–2023	2023–2024	2024–2025	Comparable-Window Annual Mean
FWM-CK	24.21 ± 1.28	43.16 ± 2.61	27.75 ± 0.73	33.69 ± 1.79
FWM-R	20.90 ± 3.34	31.56 ± 1.02	23.23 ± 1.24	26.23 ± 1.98
RM-CK	29.01 ± 1.73	48.86 ± 1.22	29.50 ± 1.61	38.93 ± 1.18
RM-R	27.66 ± 4.80	52.35 ± 2.59	26.36 ± 1.91	40.01 ± 2.18
FM-CK	36.30 ± 2.96	47.89 ± 2.81	32.39 ± 1.35	42.10 ± 2.85
FM-R	37.61 ± 2.61	44.27 ± 2.81	31.15 ± 2.04	40.94 ± 2.61

Note: FWM = forage winter wheat–early-summer maize, in which the first season was whole-plant forage wheat and the second season was maize grain; RM = ryegrass–early-summer maize; FM = double silage maize; CK = conventional irrigation; R = rainfed management. Annual system yield for each treatment was summarized on a crop-year basis.

summarizes the grain–forage and silage systems.

3.5. Comparison Under a Unified Dry-Matter Basis and Supplementary Emergy Results

To further improve functional-unit consistency across systems, the 2024–2025 crop year was converted to a unified dry-matter yield (YDM) and used as an auxiliary functional unit. FM showed both the highest dry-matter yield and the lowest CF_DM under the two water-management regimes (CK: 193.85; R: 175.71 kg CO₂-eq t DM⁻¹). RM and FWM ranked next, whereas WM, WMM, and M had comparatively higher CF_DM under the unified dry-matter basis. These results indicate that FM, RM, and FWM show better dry-matter-scaled mitigation performance when the production objective emphasizes multifunctional biomass supply. These dry-matter-based results are summarized in

Table 9. System dry-matter yield and dry-matter-scaled carbon footprint under a unified functional unit.

System	Water Management	Dry-Matter Yield in Crop Year 2024–2025 (t ha−1)	CFDM (kg CO2-eq t DM−1)
FM	CK	32.39 ± 1.35	193.85 ± 7.88
FM	R	31.15 ± 2.04	175.71 ± 11.08
FWM	CK	26.14 ± 0.61	449.48 ± 10.52
FWM	R	22.10 ± 1.08	495.33 ± 24.16
M	CK	9.36 ± 0.51	642.14 ± 33.71
M	R	7.76 ± 1.51	721.27 ± 140.34
RM	CK	27.77 ± 1.36	423.17 ± 20.28
RM	R	25.04 ± 1.71	437.43 ± 29.14
WM	CK	18.00 ± 0.87	668.15 ± 31.32
WM	R	15.77 ± 1.47	696.93 ± 61.99
WMM	CK	18.35 ± 2.41	661.62 ± 80.63
WMM	R	16.88 ± 1.81	652.41 ± 68.88

Note: CF_DM = dry-matter-scaled carbon footprint; CK = conventional irrigation; R = rainfed smanagement; FM = double silage maize; FWM = forage winter wheat–early-summer maize; RM = ryegrass–early-summer maize; WM = conventional winter wheat–summer maize; WMM = winter wheat–summer maize–spring maize; M = mono-cropped spring maize.

.

The emergy evaluation produced trends generally consistent with the dry-matter-based comparison, but its role in this study was confined to supplementary characterization of resource-use performance and environmental loading. Because emergy indicators do not share the same analytical boundary as the primary line of evidence, these results are presented in the Supplementary Materials and are not used as decisive evidence in the overall ranking of cropping systems. The supplementary emergy indices are presented in Supplementary Table S4.

4. Discussion

4.1. Cropping Systems and Water Inputs Jointly Regulate Soil CO₂ Release Dynamics

Soil respiration in cropland systems is jointly regulated by soil temperature, soil water status, and crop-driven carbon supply. In wheat–maize systems on the North China Plain, previous work has shown that soil respiration responds not only to temperature but also to soil moisture, crop phenology, and plant activity [18,19], whereas precipitation pulses can trigger short-term surges in soil CO₂ release in maize fields depending on soil-water status and tillage background [26,27]. Within this framework, the temporal pattern observed in our 2025 field measurements is ecologically interpretable rather than incidental. Observation date and cropping-system configuration significantly affected soil respiration rate, whereas the whole-season main effect of water management was not significant. This suggests that irrigation treatment influenced soil CO₂ release mainly through short-lived pulse responses at specific dates rather than through a stable season-long shift in baseline respiration. The seasonal offset among systems is also consistent with differences in crop phenology, root turnover, and residue-input timing. Because cumulative soil-CO₂ release was estimated from fixed-time observations between 09:00 and 11:00 and then integrated by the trapezoidal method, the resulting values are best understood as comparative indicators of short-term process intensity under the 2025 observation schedule rather than as continuously monitored annual net carbon losses.

4.2. Pathways Linking Input-Related Emissions, Soil-Carbon Processes, and Yield Allocation

When interpreting carbon-related results, it is essential to distinguish between input-related emissions and soil-carbon processes. In this study, area-scaled carbon emission (CF_area), product-scaled carbon footprint (CF_prod), and dry-matter-scaled carbon footprint (CF_DM) represent management-related burdens during field production and were driven mainly by fertilizer input, machinery operations, and irrigation-electricity use, which is consistent with previous carbon-footprint assessments of cropping systems on the North China Plain [8,9,10]. More recent regional evidence further indicates that diversified rotations can reduce net greenhouse-gas burdens [11,16] while improving broader system sustainability [14,28]. Soil respiration and SOC, by contrast, were used to characterize ecological process responses. The two dimensions may be correlated, but they do not belong to the same accounting boundary. Under this framework, WM remained high in area-scaled emission because of its frequent winter-wheat occurrence, greater irrigation frequency, and higher input intensity. WMM reduced the share of high-water-demand winter wheat and therefore lowered CF_area relative to WM; however, because system yield declined simultaneously, its CF_prod did not decrease relative to WM.

Differences in product-scaled carbon footprint among systems depended not only on emissions themselves but also strongly on the denominator represented by total output. FM, RM, and FWM showed relatively low CF_prod in some comparisons, partly because their area-scaled emissions were lower or not among the highest and partly because multifunctional outputs spread the carbon burden over a larger denominator. This interpretation is also consistent with previous work showing that optimized crop rotation in the North China Plain can reduce carbon footprint without necessarily maximizing the same output dimension across all systems [13,28]. Cross-system interpretation should therefore use CF_area as the primary scale and then use CF_prod and CF_DM as supplementary indicators within the same production objective.

This also explains why silage-based systems with high fresh biomass can appear to have a low product-scaled carbon footprint when functional units are not unified. After conversion to a dry-matter basis, FM still showed the lowest CF_DM, followed by RM and FWM, indicating that their relative advantages were not merely artifacts of a large fresh-mass denominator but reflected genuinely better performance under a multifunctional biomass-supply objective. Accordingly, the quadrant pattern shown in Figure 6 should be regarded as a descriptive tool for system positioning rather than as a one-time comprehensive ranking across all functional objectives.

4.3. Regional Applicability and Management Implications of Grain–Forage Systems

Optimization of cropping systems on the Hebei Plain is better approached through scenario-based judgment rather than by forcing a single overall ranking across all six systems. When grain supply is the core objective, WM still provides the highest grain-oriented system yield, but this is accompanied by stronger groundwater dependence and higher area-scaled carbon emission. Under groundwater constraints, WMM can serve as a transitional alternative pathway, with its main value lying in lowering area-scaled carbon cost rather than comprehensively outperforming WM at every scale. This interpretation is supported by evidence showing that alternating deep- and shallow-rooted crops can mitigate groundwater depletion on the North China Plain [29], while alternative cropping systems in the piedmont region can markedly decrease groundwater consumption and nitrate leaching while keeping yield losses within a relatively limited range [30]. More recent large-scale evidence further indicates that diversified crop rotations can increase food production, reduce net greenhouse-gas emissions, and improve soil health [28]. When the objective shifts toward multifunctional biomass supply and low-carbon transition, FM, RM, and FWM become relatively more favorable.

The relatively high profile SOC stock observed for WMM at the 2025 harvest should be interpreted as a plausible but not yet confirmed response pattern. A reasonable agronomic explanation is that reducing the frequency of winter wheat may have altered seasonal soil-water dynamics and reduced repeated dry-season depletion relative to WM, while the spring-maize phase may have contributed additional belowground carbon inputs. These factors could partly favor carbon retention within the 0–60 cm profile. However, because temporal SOC comparison was restricted to subsets with consistent sampling windows, bulk density was treated uniformly, and equivalent-soil-mass correction was not applied, the present evidence is insufficient to identify a confirmed sequestration mechanism or to infer a long-term carbon-storage advantage of WMM over the other systems. Supplementary emergy results further showed that FM had a higher ESI than FWM and RM, which is consistent with its relative sustainability advantage within the multifunctional biomass-oriented systems.

4.4. Methodological Boundaries and Scope of Interpretation

Without altering the original experimental framework, this study unified the time window, functional units, and carbon-accounting boundary and re-analyzed system yield, soil respiration, and the 2025 cross-sectional SOC comparison using linear mixed models consistent with the split-plot design, thereby incorporating random effects of block, block × water management, and subplot into statistical inference. Three interpretive boundaries are defined here. First, the temporal SOC comparison was restricted to cropping-system subsets with consistent pre-sowing/post-harvest pairing and was not corrected by equivalent soil mass; accordingly, it is used here to characterize the direction of short-term response rather than to infer long-term sequestration advantages. Second, CF_area, CF_prod, and CF_DM are derived from the input inventory and are therefore used to compare management-related carbon costs among systems rather than plot-level inferential differences based on field replication. Third, cumulative soil-respiration release was obtained by trapezoidal integration of fixed-time observations from 09:00 to 11:00 and is therefore treated as an apparent cumulative release index rather than a year-round net flux measured by continuous automatic monitoring.

5. Conclusions

(1)

Under the unified crop-year/closed-rotation framework, WM maintained the highest grain-oriented annual system yield, whereas WMM lowered area-scaled carbon cost relative to WM but did not reduce product-scaled carbon footprint.

(2)

Under a unified dry-matter functional unit, FM showed the lowest CF_DM, and RM and FWM also performed relatively well; these systems were therefore more compatible with multifunctional biomass-supply and low-carbon-transition objectives.

(3)

Soil respiration and SOC provided supplementary evidence of short-term soil-carbon responses to cropping-system configuration. Because the temporal SOC comparison was restricted to subsets with consistent sampling windows and was not corrected by equivalent soil mass, these results should not be used to infer long-term sequestration superiority.