The surface soils were collected from 0 to 15 cm soil depth from three predominant soil types of India according to the USDA soil taxonomy 1) Alfisol (mixed hyperthermic VerticHaplustalfs) was collected from the farmer’s field (21.96°N latitude and 77.74°E longitude) in Betul district of Madhya Pradesh under maize-fallow cropping sequence, 2) Inceptisol (mixed hyperthermic Typic Haplustept) was collected from the farmer’s field (26.41°N latitude and 80.23°E longitude) in Kanpur district of Uttar Pradesh under rice-wheat cropping system, and 3) Vertisol (Isohyperthermic Typic Haplustert) was taken from the experimental farm (23.31°N latitude and 77.41°E longitude) of ICAR-Indian Institute of Soil Science under soybean-wheat cropping system. In the laboratory, air-dried soils, after carefully removing the recognizable gravels and debris (≥2 mm), were passed through a 2 mm sieve by gently breaking the clods along planes of weakness by hand, thus preserving soil aggregation. The wheat crop residue was collected after the harvest of wheat from an experimental plot under a long-term soybean-wheat cropping system at a recommended dose of fertilizer in the research farm of the Indian Institute of Soil Science. For the residue treatment, the wheat residues were pulverized after air drying and sieved to a size of 2 mm. An elemental analyzer (NC analyzer, Thermofisher, Flash 2,000 model) was used to determine total C and N concentrations in soil and residues. We used sulphuric acid-perchloric acid digestion and molybdenum antimony colorimetric estimation (Page et al., 1982) to determine total P concentration in soil and residue. The basic properties of soil and wheat residues are presented in Table 1.

Soil microcosms were prepared by taking 40 g of soil in a 100 mL beaker placed inside 461 mL glass jars with septum-fitted lids for gas sampling. After a 10-day pre-incubation at 70% moisture and room temperature, the experiment was conducted with varying residue and nutrient inputs at 80% field capacity. Wheat residues were applied at four levels: 0, 90, 180, and 270 mg (dry weight), which was equivalent to 0 (WR0), 5 (WR5), 10 (WR10), and 15 (WR15) Mg ha⁻¹, respectively. The crop residues (<2 mm) were completely mixed with soil (<2 mm) for incubation based on a 10 cm incorporation depth in the field. After residue addition, the nutrient solution was added to all the WR treatments at three levels: 1) NP0: no nutrient, 2) NP1, nutrients (N and P) were added to balance the residue C/nutrient stoichiometry of C/N/P= 100: 8.3: 2.0 (Fang et al., 2020) to achieve 30% stabilization of added residue C input at 5 Mg/ha (R5), and 3) NP2 = 3 × NP1. Therefore, the nutrient addition altered nutrient stoichiometry in WR treatments, and the achieved stoichiometry in each treatment combination is shown in Supplementary Table S1. Nutrient solutions corresponding to nutrient levels were prepared using AR grade urea and potassium dihydrogen phosphate, and 10 M sodium hydroxide solution was used to modify the pH of the nutrient solution to 7. All treatments were replicated three times and incubated at 30°C incubation temperature at 80% FC moisture content. The incubation temperature was selected because the aboveground mean annual surface temperature is ca. 30°C in sub-tropical and semi-arid regions during different crop-growing seasons (Lenka et al., 2022). Triplicate empty glass jars were also incubated to account for the headspace GHGs. Field capacity (FC) was measured at matric potentials of −33 kPa using sieved (<2 mm) soil samples in pressure plate extractors from Soil Moisture Equipment Corp., Santa Barbara, CA, with FC moisture content at 0.14 m³m⁻³. Soil moisture was maintained through regular weighing and water addition to compensate for evaporation losses during gas sampling intervals.

The GHG fluxes from different treatments were measured in gas chromatography (Agilent Technologies model 7890A). Headspace gases were drawn from the incubation jars using a syringe and immediately transferred to a 10 mL evacuated glass at frequent intervals for 96 days of incubation. The CH₄/CO₂/N₂O flux rate was calculated as the change in headspace N₂O/CH₄/CO₂ concentration using the ideal gas law and molecular weight. Cumulative N₂O, CH_4, and CO₂ emissions were determined by linear integration of daily fluxes. The global warming potential of CO₂-equivalent was calculated by multiplying the cumulative N₂O and CH₄ emissions by their respective radiative forcing potentials using the following equation (Singh et al., 2024):

GWP (mg CO₂ eq. kg⁻¹ soil) = CH₄ (mg kg⁻¹ soil) × 27.2 + N₂O (mg kg⁻¹ soil) × 273 + CO₂ (mg kg⁻¹ soil) × 1

After the incubation period of 96 days, soil samples were analyzed for relevant soil properties such as soil minerals nitrogen (NO₃-N and NH₄-N), alkaline phosphatase, soil microbial biomass carbon (SMBC), and labile SOC following standard analytical procedures. The moisture content was determined gravimetrically using the oven-dry method in a part of the moist composite soil samples. Soil mineral nitrogen was estimated after extraction in 2M KCl, and subsequent analysis employed standard methods (Kempers, 1974). Labile SOC calculations utilized the potassium permanganate oxidation method (Blair et al., 1995; Islam et al., 2003) and measured alkaline phosphatase (alk-P) as an indicator of P demand (Alef and Nannipieri, 1995). Soil MBC was determined by a fumigation-extraction method, and a conversion factor of 0.45 was applied to determine MBC (Joergensen and Brookes, 1990).

All data underwent tests for normality and homogeneity of variance, with transformations applied as needed. Statistical analyses were performed using SPSS software (version 21.0, SPSS Inc., Chicago, IL, United States), setting a significance threshold of p = 0.05. The general linear model univariate ANOVA was employed, followed by Tukey’s HSD test for multiple comparisons of means. The relative effect of management variables (residue and nutrient input, soil moisture), post-harvest soil parameters (NO₃-N, NH₄-N, MBC, labile C, and alk-P) and inherited soil parameters (pH, CEC, clay, sand, silt, TC, TN, legacy P and K) on soil GHG emission (CO₂/CH₄/N₂O) was evaluated using Pearson correlation (two-tailed significance) and partial least squares (PLS) regression models. The variable influence on projection (VIP) score of value >0.8 (Gómez-Gener et al., 2018) was used to identify the potential drivers of soil GHG emission. The PLS model fitting and cross-validation details are presented in the supplementary file. The model performance was analyzed using the root mean square error (RMSE), adjusted (R ²) coefficient of determination, and p-value between observed and predicted values due to their robustness, simplicity, and widespread use. Finally, the effect size (VIP scores) were grouped into two broad categories: anthropogenic (WR input, the stoichiometry of C: N, C: P, N input, MBC, NO₃-N, NH₄-N, SMBC, labile C, alk- P) and natural drivers (pH, CEC, clay, sand, silt, TC, TN, legacy P, and K). All the graphs were constructed using Origin Pro, version 2024b (Origin Lab Corporation, Northampton, MA, United States).

Total C mineralization was significantly influenced by the main effects of soil type, wheat residue (WR) input, and the interactive effect of soil type and WR only (Supplementary Table S2; Figure 1). The total cumulative C mineralization ranged from 416.02 mg C kg⁻¹ soil (WR15 + NP2 in Alfisol) to 27.33 mg C kg⁻¹ soil in treatment WR0 + NP0 in Inceptisol over 96 days of incubation (Table 2). Averaged across nutrient and residue input, C mineralization was significantly the highest in Alfisol, 1.3 times Vertisol, and 1.4 times Inceptisol; however, the effect of Vertisol was comparable to Inceptisol. Wheat residue input significantly enhanced total cumulative C mineralization; the order was WR15 (317.46 mg C kg⁻¹) > WR10 (204.85 mg C kg⁻¹) > WR5 (118.67 mg C kg⁻¹) > WR0 (41.80 mg C kg⁻¹) averaged across nutrient input and soil types. High nutrient input increased the total cumulative C mineralization, but the effect was comparable to NP0 across soil type and WR input.

The cumulative N₂O flux was significantly influenced by the main factors of wheat residue input, nutrient levels, and soil type (Table 2; Supplementary Table S2). Unlike total C mineralization, there was a significant three-factor interaction of wheat residue, nutrients, and soil type on the cumulative N₂O flux over 96 days of incubation. For example, nutrient input (NP2, C/N/P = 100:8.3:2.0) had the highest N₂O flux of 128.41 μg N kg⁻¹ soil in the high-residue (WR15) input treatment Vertisol compared to the lowest value (−20.82 μg N kg⁻¹ soil) in WR10 + NP0 (C/N/P > 100:8.3:2.0) under Inceptisol (Table 2; Figure 3). The negative flux indicates N₂O consumption. The cumulative N₂O flux over the 96-day incubation period was similar in the Alfisol (17.60 μg N kg⁻¹ soil) and Vertisol (15.53 μg N kg⁻¹ soil) across residue and nutrient input levels. Nutrient (NP) input significantly enhanced the cumulative N₂O flux; the trend was NP2 > NP1 > NP0. However, the effect of wheat residue was inconsistent with WR15 ≈ WR0 > WR5 ≈ WR10 across nutrient input and soil type, indicating a nonlinear response of cumulative N₂O emission with WR rate (Figure 2; Supplementary Figure S2). The negative N₂O flux of cumulative N₂O emission in Inceptisol at all WR levels (cf. WR0) indicates the sink capacity of the soil (Figure 3; Supplementary Figure S2).

High input of wheat residue (WR15 and WR10) significantly decreased cumulative CH₄ flux compared with WR0 and WR5, the order being WR5 (6.86 μg C kg⁻¹ soil) > WR10 (2.99 μg C kg⁻¹ soil) ≈ WR15 (0.13 μg C kg⁻¹ soil) ≈ WR0 (0.05 μg C kg⁻¹ soil) across nutrient management and soil type. The main factors of soil type and residue input significantly influenced the cumulative CH₄ flux (Supplementary Table S2). In addition, a significant interaction of WR × soil type occurred; that is, high residue input (WR10 and WR15) only decreased cumulative CH₄ flux in Vertisol and Inceptisol (Figure 4A). However, in Alfisol, cumulative CH₄ flux increased with WR input to WR10 (@10 Mg/ha) and then declined at WR15. Significant interaction effect of soil type x NP (p = 0.053) showed different responses of soil type to increment in nutrient input (Supplementary Table S2); the response was linear in Alfisol, curvilinear in Vertisol, and nonlinear in Inceptisol (Figure 4B), across residue input. The cumulative CH₄ flux ranged from 23.42 μg C kg⁻¹ soil WR5 + NP0 (C/N/P > 100: 8.3: 2.0) in Inceptisol to −6.64 μg C kg⁻¹ in treatment WR15 + NP0 (C/N/P > 100: 8.3: 2.0) under Vertisol, negative values indicate CH₄ consumption from ambient (Table 2).

The cumulative GWP was significantly the highest in the Alfisol, 1.2 times Vertisol, and 1.4 times Inceptisol across residues and nutrient inputs (Figure 5; Supplementary Figure S3). GWP increased significantly (p < 0.001) with the residue input in all soil types (Figure 5), and the trend followed the order WR15 (1,142.9 mg CO₂-C eq. kg⁻¹ soil) > WR10 (769.1 mg CO₂-C eq. kg⁻¹ soil) > WR5 (474.5 mg CO₂-C eq. kg⁻¹ soil) > WR0 (154.8 mg CO₂-C eq. kg⁻¹ soil) (Supplementary Figure S3). The interactive effect of WR × soil type x NP was significant (p = 0.047), and WR x soil type (p = 0.007) on cumulative GWP over 96 days of incubation (Supplementary Table S2). Therefore, the highest interactive effect was in the treatment WR15 (@15 Mg/ha) + NP2 (C/N/P = 100: 8.3: 2.0) in Alfisol (1,489.76 mg CO₂-C eq. kg⁻¹ soil) and the lowest in WR0 (no residue) + NP1 (C/N/P < 100: 8.3: 2.0) in Inceptisol (96.70 mg CO₂-C eq. kg⁻¹ soil) (Table 2).

The results of this study demonstrated that crop residue input was one of the major drivers of total carbon (C) mineralization (CO₂ flux) in all three soils (Fang et al., 2018; Li et al., 2018), with the highest residue application (WR15) resulting in the most significant increase in carbon release. The order of C mineralization (WR15 > WR10 > WR5 > WR0) across all soil types highlights the direct relationship between organic matter availability and microbial decomposition processes (Li et al., 2021a). The additional carbon from increasing levels of wheat residue input likely provided a readily accessible energy source for soil microbes, accelerating the mineralization process. This observation is consistent with the substrate-induced respiration theory, where adding organic residues stimulates microbial activity, leading to increased carbon turnover (Fontaine et al., 2004; Lenka et al., 2019; Liu et al., 2023a). The accelerated C mineralization could be further attributed to crop residue-induced soil carbon (C) priming by stimulating microbial activity, which alters the decomposition of native soil organic carbon (SOC) (Lenka et al., 2019; Lenka et al., 2021; Salehin et al., 2024). This priming could be higher in high residue input (cf. low residue) (Fang et al., 2018). While nutrient input generally increased C mineralization, the effect was relatively minor compared to the influence of soil type and residue input. In soils with adequate nutrient levels (Ponce-Mendoza et al., 2010), additional nutrient inputs may not significantly enhance microbial activity beyond a certain threshold (Conde et al., 2005). The comparable effect of high nutrient input (NP2) to the NP0 treatment across soil types and residue levels suggests that nutrient availability was not the primary limiting factor for C mineralization in this study (Ma et al., 2020).

Instead, the physical and chemical properties of the soil and the quantity of organic residue appeared to play more critical roles. These findings align with previous research that highlights the importance of soil types and residue C inputs in driving microbial activity and subsequent carbon mineralization rates (Fang et al., 2019; Lenka et al., 2019; Liu et al., 2023b; Singh et al., 2024). The addition of wheat residue (15 Mg/ha) with nutrient input (NP2) led to higher cumulative C mineralization in Alfisol compared to Inceptiosl in treatment WR0 plus NP0 over the incubation period, likely because (i) nutrient input improved the C: N imbalance of soil microorganisms due to the high C: N ratio of wheat residue input (79.54: 1.00), thus eliminating the N limitation of microorganisms (Fang et al., 2020; Song et al., 2022) as a result, microorganisms in NP2 increased C mineralization compared to NP0, which was used to satisfy energy requirements when accessing nutrients. (ii) The lower pH (5.70) and high sand content (53%) (Table 1) in Alfisol accelerated the use of total C (residue plus soil) by microorganisms by enhancing their activity, driven by better soil structure, aeration, and nutrient availability (Song et al., 2022). The result was also supported by higher labile C (Table 3) in Alfisol receiving high residue and nutrient input. The strong positive correlation between C mineralization (CO₂ fluxes) and labile C (Supplementary Figure S7) provided the required energy source for higher microbial activity (Fang et al., 2020; Lenka et al., 2022) compared to Vertisol and Inceptisol. Our results demonstrated that Alfisols having higher sand content (Table 1) have greater macropores and oxygen supply (Pathak et al., 2013; Chandrakala et al., 2021), promoting more active microbial communities that can efficiently decompose organic residues. In contrast, the lower mineralization rates observed in Vertisols and Inceptisols may be due to their higher clay content, which can limit microbial access to organic substrates by physically protecting organic matter within soil aggregates (Chowdhury et al., 2021; Lenka et al., 2021; Ren et al., 2024). This significant interaction of WR and soil types indicated the inherent properties of the soil modulate the effectiveness of residue inputs in promoting C mineralization. In soils with favorable conditions (e.g., Alfisols), residue additions can lead to a marked increase in microbial activity and carbon mineralization. However, in soils with less favorable conditions (e.g., Inceptisols), the same residue inputs may not be as effective due to limitations in microbial activity or substrate accessibility (Lenka et al., 2019). The partial least square regression analysis demonstrated that anthropogenic management practices (residue and nutrient inputs) and acquired soil properties substantially impact carbon mineralization (CO₂ fluxes) compared to natural soil factors. The PLS analysis indicated that wheat residue input (WR) and nutrient input were the most influential variables, followed by labile carbon, NO₃-N, and microbial biomass carbon (MBC). The model shows that anthropogenic factors account for 72% of the variance in C mineralization. In contrast, natural factors, such as soil type, contribute only 28% (Figure 7). This finding underscores the critical role of targeted management practices in regulating soil carbon dynamics (Fontaine et al., 2004), highlighting the potential to optimize carbon mineralization through appropriate residue and nutrient management strategies considering the basic soil characteristics.

Our experiment showed that Vertisol (cf. Incetisol) exhibited a significant increase in denitrification activity when higher levels of residue (WR15) and nutrients (NP2) were applied (Table 2; Figure 2). This could be attributed to i) the enhanced availability of substrates for microbial nitrification and denitrification processes (Abalos et al., 2022; Lenka et al., 2022) ii) the high clay content (Table 1) in Vertisol (Gebremichael et al., 2022) iii) enhanced microbial biomass carbon (MBC) and nitrate (NO₃-N) concentrations (Table 3) (Li et al., 2021a), iv) high soil pH (Khalifah and Foltz, 2024), and less available soil P (Gebremichael et al., 2022) (Table 1) and v) high alkaline phosphatase activity (Yang et al., 2021). The presence of smectite and vermiculite clay types in Vertisols (Rakhsh et al., 2017; Fang et al., 2018; Fang et al., 2019) is related to higher mineral-associated carbon (Fang et al., 2019), creating an ideal environment for the processes of ammonification and nitrification, resulting in high concentrations of NH₄-N, NO₃-N, and N₂O losses compared to Inceptisol. Further, Inceptisol demonstrated negative N₂O flux, indicating consumption, possibly due to a significant amount of legacy phosphorus (19.7 mg kg-1), a high soil pH (8.49), and less clay content (13%) compared to Vertisol and Alfisol soils (Table 1). These characteristics of Inceptisol contributed to reduced microbial activity, as evidenced by lower levels of MBC and alkaline phosphatase. The results supported the previous finding of the association of soil legacy P limitation on enhanced N₂O emission (Li et al., 2021a; Gebremichael et al., 2022) in Vertisol (cf. Inceptisol). The addition of nutrients (N and P) changed the stoichiometry of nutrients (carbon/nitrogen/phosphorus), eliminating the P limitation, with increased P levels leading to lower emissions of N₂O, emphasizing the importance of nutrient management in controlling greenhouse gas release. However, the dominant influence of nitrogen over phosphorus on N₂O production plays a crucial role in determining the overall impact of phosphorus when residues and NP nutrients are applied together. Because the relative effect of soil legacy P estimated through PLS model analysis was lesser than soil NO₃-N and N input (Figure 7). Excessive residue without sufficient nutrient input resulted in nitrogen immobilization (Table 3). This immobilization led to a decrease in N₂O emissions. Conversely, when nutrient inputs were balanced, it stimulated microbial activity, resulting in increased emissions. The results of our study support the conclusions of previous researchers that maintaining a high ratio of carbon, nitrogen, and phosphorus (C: N: P) in the soil promoted the growth and activity of microbial biomass, resulting in the conversion of nitrogen from its mineral form to its organic form, which reduced the risk of nitrogen losses through denitrification and leaching (Lenka et al., 2019; Lenka et al., 2022; Li et al., 2021a; Li et al., 2021b). Consequently, a residue input will preserve the nitrogen (N) in its organic state, making it gradually accessible for microbial and plant development. The nonlinear relationship between N₂O fluxes and residue input indicated the interaction of carbon and nitrogen availability, soil microbial activity, and the physical characteristics of the soil (Fang et al., 2020; Singh et al., 2024; Xu et al., 2024). The observed interaction between residue and nutrient input with soil type suggests that the impact of residue carbon input on reducing soil N₂O emissions was influenced by both the C/N/P stoichiometry and the fundamental soil properties, such as clay percentage, pH, and phosphorus content (Supplementary Figure S7). The strong link between N₂O emissions and acquired soil parameters, such as MBC, NO₃-N, and alkaline phosphatase (Alk-P), further highlighted the crucial role of soil management in affecting N₂O fluxes.

The PLS regression analysis reaffirmed our findings and offered valuable insights into the variables that affect soil nitrous oxide (N₂O) emissions (Figure 7; Supplementary Figure S8) (Shah et al., 2024; Xu et al., 2024). The PLS model, which accounted for a significant amount of the variation in N₂O emissions, emphasized the influential impact of management factors and acquired soil qualities compared to inherent soil characteristics. The elevated VIP scores for parameters such as alkaline phosphatase (alk-P), nitrate-nitrogen (NO₃-N), nitrogen input, and MBC suggested that these factors drive N₂O emissions. The high ranking of alk-P indicates that nutritional stoichiometry, specifically the availability of phosphorus (Li et al., 2021a; Gebremichael et al., 2022), substantially influences the microbial activities that produce N₂O fluxes. This is reinforced by the significance of NO₃-N, which plays a direct role in nitrification and denitrification, the primary biological processes responsible for N₂O production (Weitz et al., 2001; Khalifah and Foltz, 2024). Consequently, the presence of limited phosphorus in Vertisols (3.7 mg kg⁻¹) and Alfisol (8.5 mg kg-1) (compared to Incetisol: 19.7 mg kg⁻¹) and the high concentration of nitrate nitrogen (NO₃-N) greatly increased the emission of nitrous oxide (N₂O) in both soils (Figure 2). The importance of nitrogen input has been identified as another crucial component, emphasizing how the management of nutrients directly impacts the release of N₂O by influencing nitrogen availability in the soil (Wang C. et al., 2021; Lenka et al., 2022; Singh et al., 2024).

The model also recognized the importance of nutrient stoichiometry (C: N and C: P), essential for maintaining a balance between carbon and nitrogen cycling. This stoichiometry directly affects the efficiency of N₂O production or reduction processes (Liu et al., 2021; Lenka et al., 2022; Song et al., 2022). The VIP scores indicated that the influence of inherited soil properties, such as clay content and soil pH, on N₂O emissions was lesser than the impact of management practices and the consequently acquired soil attributes. The relative contribution of the anthropogenic drivers is more significant than natural, with a share of 62% of the total N₂O variance in the PLS model. Our study indicated that the effects of soil type on N₂O emissions could be counteracted by effective management strategies, including those that promote microbial activity and increase nutrient availability. Optimizing residue and nutrient management strategies will be crucial in efficiently reducing N₂O emissions in various soil types.

Higher organic C and N availability from high residue input (WR10 and WR15) likely suppressed methanogenesis and enhanced oxidation of CH₄, resulting in CH₄ consumption (Choudhary et al., 2024). This suggestion is consistent with a substantial reduction in the cumulative CH₄ flux in high wheat residue (WR10 and WR15) compared to low residue inputs (WR0 and WR5). The effect was more pronounced in Vertisol, followed by Alfisol and Inceptisol. The negative correlation (though insignificant) between CO₂ and CH₄ emissions (Supplementary Figure S7) aligned with the previous finding of a positive relationship between CH₄ uptake and high CO₂ emission in the upland ecosystems (Li et al., 2022; Wu et al., 2024). Across soils, the nonlinear response of CH₄ emissions to residue input suggested that while low and moderate residue input might stimulate CH₄ production, the excessive residue could lead to a saturation point or shift in the microbial activity that reduces CH₄ emissions (Wang X. D. et al., 2021; Singh et al., 2024). Additionally, the Pearsons correlation analysis also found that CH₄ fluxes were negatively correlated (Supplementary Figure S7) with several inherited (legacy) soil properties, such as soil organic carbon, total nitrogen, available potassium, and clay content, indicating a greater edaphic (natural) control on CH₄ emission/uptake. The Vertisol was rich in clay content compared with Alfisol and Inceptisol (Table 1) and low in legacy P, showing significantly higher cumulative CH₄ uptake over the incubation period (Figure 4) (Cui et al., 2023; Shumba et al., 2023). Among the acquired properties induced by varied residue and nutrient input, MBC, NH₄-N, and alkaline phosphatase negatively correlated with CH₄ fluxes. Our study observed high residue input (WR10 and WR15) resulted in higher MBC, NH⁴-N, and alkaline phosphatase activity related to less CH₄ fluxes (high CH₄ oxidation) in the treatments (Lai et al., 2017; Fang et al., 2018; Yang et al., 2021; Shah et al., 2024). The correlations analysis suggested that inherent soil characteristics and management-induced changes can significantly influence CH₄ emissions. Notably, there was no significant correlation between CH₄ fluxes and direct residue or nutrient inputs, suggesting that the impact of these management practices on CH₄ emissions is mediated through changes in soil properties rather than direct effects. Moreover, the interaction between soil type and nutrient input (p = 0.053) revealed that soil types respond differently to increments in nutrient input across varying residue levels (Nguyen et al., 2013; Liu et al., 2021; Singh et al., 2024). Alfisol exhibited a linear response, Vertisol a curvilinear response, and Inceptisol a nonlinear response to nutrient additions, highlighting the complexity of nutrient-soil interactions and their influence on CH₄ fluxes. Overall, our results suggest that Vertsiols were CH₄ sinks (−2.33 μg C kg⁻¹ soil) compared to Alfisol (2.70 μg C kg⁻¹ soil) and Inceptiosl (7.16 μg C kg⁻¹ soil) across varied levels of residue and nutrient input. As estimated through the PLS model analysis (Joshi et al., 2024), the relative effect size of the studied variables on CH₄ emission further supported our results. The best-fitting PLS model identified that soil CH₄ flux responded strongly to inherited (legacy) soil properties with a predominant effect of the soil phosphorus, clay, silt + clay content, potassium, and total nitrogen (Figure 7) (Yu et al., 2017). The relative contribution of anthropogenic and natural factors are 44% and 56%, respectively, to the total variation in CH₄ flux estimation. This contrasts with CO₂ and N₂O emissions, where anthropogenic management factors (residue and nutrient input) and acquired soil properties played a more significant role. Our results confirmed previous results that edaphic variables had greater control than anthropogenic factors on CH₄ fluxes at the global scale (Gatica et al., 2020). The lower predictive power of the PLS model for CH₄ fluxes (r ² = 0.35) compared to CO₂ (r ² = 0.88) and N₂O (r ² = 0.59) suggests that CH₄ emissions are governed by more complex or less direct factors that the model does not fully capture (Xu et al., 2024). This finding underscores the need for a more nuanced understanding of the interactions between soil management practices and CH₄ fluxes, particularly across different soil types.

Normalizing the non-CO₂ (CH₄ and N₂O) GHG emission to CO₂ emissions gives a single metric that helps identify the best management practices and soil type relevant to climate change mitigation and feedback. Among the three soil types examined, Alfisol exhibited the highest cumulative GWP, 1.2 times greater than Vertisol and 1.4 times greater than Inceptisol across all residue and nutrient input levels. That suggests that Alfisol may be more prone to greenhouse gas (GHG) emissions due to its inherent properties when subjected to residue and nutrient management practices. Though high residue input (WR10 and WR15) generally decreased CH₄ and N₂O emission, the increasing trend of GWP with higher residue input levels (WR15 > WR10 > WR5 > WR0) is likely due to the increased availability of organic carbon from the residue, which accelerates microbial activity microbial biomass carbon and leads to higher CO₂ emissions, the primary contributor to GWP, that aligned with previous studies (Li et al., 2022; Singh et al., 2024). The response of GWP to residue and nutrient input was modulated by soil type (p = 0.047), further emphasizing the complexity of the factors that influence GWP in soils (Rakhsh et al., 2017; Li et al., 2021a). Across all treatments, the integrated application of WR15 plus NP2 had the highest cumulative GWP of 1,489.76 mg CO₂-C eq. kg⁻¹ soil in Alfisol. In contrast, the lowest GWP (96.70 mg CO₂-C eq. kg⁻¹ soil) was recorded in the treatment with no residue input (WR0) and moderate nutrient input (NP1) in Inceptisol. These results underscore the need for tailored residue and nutrient management strategies to mitigate GWP in agricultural soils. While residue inputs are essential for maintaining soil health and fertility, their impact on GHG emissions and GWP must be carefully managed, especially in soils like Alfisol, which are more susceptible to high GWP. The study suggests that reducing residue input or optimizing nutrient management could effectively reduce GWP in such soils, thereby contributing to more sustainable agricultural practices. Additionally, wheat residue return to soil compared with residue burning could emit fewer GHGs (Deshpande et al., 2023) considering the potential benefits of wheat residue return on improving soil health, soil organic carbon, and crop yield, plus the associated co-benefits in improving essential ecosystem services (Singh et al., 2020; Li et al., 2022; Zhang et al., 2023; Ma et al., 2024).

The study highlights the importance of understanding the interactions between residue management, nutrient inputs, and soil type in influencing GWP. By adopting management practices that consider these interactions, it may be possible to reduce the climate impact of agricultural soils while maintaining their productivity and ecological function (Lenka et al., 2017; Wang X. D et al., 2021; Li et al., 2022).