1 Introduction

Global warming is mainly attributed to the elevated atmospheric concentrations of greenhouse gases (GHGs) including carbon dioxide (CO2), nitrous oxide (N2O), and methane (CH4) by anthropogenic activities (O’Neill et al. 2021; Wu et al. 2022). If the concentrations of GHG continue to increase at the current rate, the average global temperature would increase by 3.8 °C by the end of the century (IPCC 2021). Agriculture accounts for approximately 12% of GHG emissions, mainly from soil cultivation and crop production (Raihan and Tuspekova 2022). Therefore, reducing GHG emissions in agroecosystems is critical to alleviate greenhouse effects and achieve carbon neutrality goals.

Tillage practices, such as conventional tillage (CT) and no-tillage (NT) can alter the intensity of GHG emissions through perturbations of carbon, nitrogen, and water dynamics in the agroecosystems (Osborne et al. 2010). NT is considered an eco-friendly cropland management practice with lower soil disturbance (Krauss et al. 2017) and higher soil organic matter stocks as compared to CT (Stewart et al. 2016), leading to reduced GHG emissions. However, inconsistent results of the effects of NT on GHG emissions were observed among individual studies, which showed that NT significantly increased, decreased, or did not affect (Bayer et al. 2015) GHG emissions from agricultural soils. Specifically, Dencső et al. (2021) reported that NT had higher CO2 emissions than CT in all laboratory experiments. On the contrary, in another study, cumulative CO2 emissions were reduced by almost half during the maize season in NT than those in CT (Salamanca-Fresno et al. 2022). CH4 as another important GHG, acted as a sink or a source (Maucieri et al. 2021) in agroecosystems. Pu et al. (2022) found that NT increased CH4 uptake in the summer maize season. Conversely, NT showed a slight but non-significant tendency to increase CH4 emissions in maize-cultivated fields (Maucieri et al. 2021). For N2O, NT generated 32–59% more emissions than CT, and decreased the emissions by 22.6% in winter wheat seasons compared to CT (Pu et al. 2022), or had no significant effects on cumulative emissions from a maize field (Borzouei et al. 2021). The high degrees of inconsistency in the results of previous studies make it a significant challenge to reveal the actual effects of tillage practices on GHG emissions and seek effective GHG emission mitigation strategies. Therefore, a further investigation on the effects of soil tillage on GHG emissions is needed.

GHG emissions are mediated by soil processes that are affected by soil physical and chemical properties including moisture content, temperature, bulk density, pH, and soil organic matter, etc. (Alskaf et al. 2021; Hu et al. 2022). In prior studies, increased soil CO2 emissions were observed with increased soil organic matter and moisture content by NT (Bilandžija et al. 2016) and negative relationships of CO2 emissions were found with soil bulk density due to reduced gas diffusion and microbial activities (Beare et al. 2009). In a meta-analysis, CH4 emissions were found to significantly increase with elevated soil moisture that benefited anaerobic conditions under NT (Feng et al. 2018). Similarly, elevated soil moisture promoted N2O emissions under NT as compared to CT (Tellez-Rio et al. 2015). According to a previous meta-analysis and other observations, soil temperature and pH also played an important role in controlling GHG emissions (Mei et al. 2018). For instance, soil CO2 emissions were significantly enhanced in alkaline soils and inhibited by acidic soils under NT, whereas N2O emissions were positively related with acidic and neutral soils (Shakoor et al. 2021). However, inconsistent observations, i.e., enhanced and inhibited emissions were observed for CH4 (Huang et al. 2018).

At present, most studies only focused on the influence of tillage practices on GHG emissions in agroecosystem during the growing season (GS) and the effects of tillage practices on GHG emissions in the entire year were unclear (Pu et al. 2022). Only a few studies differentiated GHG emissions between GS and non-growing season (NGS) based on an annual observation mostly due to difficulties in conducting field work in long and severe winters during NGS, especially in cold regions (Voigt et al. 2017). Yet, the greater impacts of tillage practices are likely to occur in the NGS when plant uptake is negligible and the activity of microorganisms and subsequent soil GHG fluxes are sensitive to soil properties induced by tillage (Zona et al. 2016). NGS is thus defined as the period of the stagnation of crop growth stagnation or when crops grow slowly, which occurs at low temperature or in winter season as suggested by prior studies (Wang et al. 2021). Over- or under-estimations of interannual GHG emissions usually occur owing to the shortage of NGS data. Effective GHG emissions should be monitored locally for the entire year, covering GS and NGS, to reflect the variation of management impacts on the environment over time (Wang et al. 2019a). In prior studies, the agroecosystem CO2 emissions in NGS accounted for more than 50% (Natali et al. 2011). However, soil CH4 uptake rates were not consistent seasonally, with CH4 uptake in NGS more sensitive to the temperature than that of GS (Lin et al. 2015). N2O emissions to the soil temperature showed significant seasonal differences in GS, with no significant differences observed in NGS (Cantarel et al. 2012).

The cultivating land of North China Plain accounts for approximately 21.3% of the total area in China, which is the main grain-producing territory (Kan et al. 2020). In this area, the main planting method is wheat–maize rotation, and the unreasonably intensive tillage practices are important GHG emissions sources to the atmosphere. Most of prior research on the responses of GHG emissions to tillage practices (i.e., CT and NT) was conducted under short-term, discontinuous conditions, focusing on GS alone, which resulted in large differences from studies based on long-term, continuous NGS-GS situations. This work employed in situ real-time monitoring technologies to accurately monitor GHG emissions by different tillage practices in continuous NGS and GS over 3 years seasons.

The objectives of this study were: 1) to determine the responses of soil CO2, CH4, and N2O emissions to CT and NT in NGS and GS seasons and 2) to identify the main pathways of reducing soil GHG emission under NT in soil processes. It was hypothesized that 1) NT decreased soil CO2, CH4, and N2O emissions as well as global warming potential (GWP) in typical rotation systems and 2) GHG emissions in NGS, although were small, were important for accurate prediction of annual GHG emissions.

2 Materials and methods

2.1 Study site

This study was carried out on a wheat–maize rotation system from 2019 to 2021 at the test field of Shandong Yucheng Agro-ecosystem National Observation Research Station in Shandong Province of China. The start date of the tillage study was in July 2018. The daily precipitation and air temperature in 2019, 2020, and 2021 are illustrated in Fig. 1. The total precipitation and days of rainfall events in 2019, 2020, and 2021 were 340.5, 634.0, 1036 mm and 42, 49, 71 days, respectively. The climatic conditions and soil basic properties of this study area were described in detail in our previous study (Li et al. 2022b). Yearly cropping at this site was a rotation of winter–wheat (Triticum aestivum L.) and summer–maize (Zea mays L.).

Fig. 1
figure 1

Air temperature and precipitation daily during study year at Yucheng Station

Full size image

2.2 Experimental design

According to a previous study (Du et al. 2021), the experiment was conducted with a randomized complete block design. The whole-plot treatments were two tillage practices (CT vs. NT) with three replicates. The experiment started in July 2018 and the selected period was from Jan 1, 2019 to Dec 31, 2021 to study the effect of GS and NGS of the whole year on soil GHG emissions. The GS was defined to last from Apr 1 to Oct 31, while NGS included two time periods, i.e., Jan 1 to Mar 31 and Nov 1 to Dec 31. The planting and harvesting dates were Nov 1 of the previous year and Jun 10 of the current year for wheat, and Jun 20 and Oct 15 for maize. Thus, a portion of the wheat period was in GS and all of the maize period was in GS.

For CT treatments, a small rotary tiller was carried out to loosen the soil and remove fallow vegetation with a soil depth of 15–20 cm each year before the sowing of wheat only. For NT treatments, glyphosate was used to kill the surface weeds, which was a standard practice in the testing location. The soil surface of wheat and maize was mulched with manually collected maize and wheat straw at a rate of 6 Mg ha− 1 for 15 days after sowing.

2.3 Crop management

Before the experiment, winter–wheat and summer–maize rotations were extensively excised in the examined region. The selected wheat and maize varieties were widely used “Jimai 22” and “Denghai 605” in the local region, respectively. During the wheat sowing period, basal fertilizers [Nitrogen (urea, 46.4% nitrogen content, 105 kg N ha− 1), phosphorus (750 kg P2O5 ha− 1), and potassium (160 kg K2SO4 ha− 1)] were applied to maintain the necessary nutritional needs. The other part of the nitrogen fertilizer (105 kg N ha− 1) was applied to the soil surface at the time of wheat greening period by top-dressing. For maize, 210 kg N ha− 1 of nitrogen fertilizer was spread onto the soil surface by top-dressing after rain on July 28, 2019, July 29, 2020, and July 28, 2021. The wheat seedlings were irrigated during the wheat rejuvenation period at a rate of 5 m3 per plot for each treatment. Irrigation during the maize period depended on the precipitation of the year and the soil was irrigated only in 2019 (Jul 10) due to the low rainfall. Further details of the crop management were described in our previous study (Li et al. 2022a).

2.4 Measurement and analysis of GHG fluxes

In May 2018, transparent chambers were installed in each experimental plot. The transparent chambers were fixed in the center of each experimental plot and were not moved, but the chambers were opened and closed automatically within a specified time for air sampling. Gas samples were obtained every hour, during which the chamber was closed for 200 s. After sample collection, the chamber was kept open to eliminate the influence of heat on GHG emissions. All chambers were continually monitored for 23 h d− 1 (except for daily calibration from 23:00–00:00) using an automatic control system. Further details of the measurement process and the entire monitoring system were described in our prior study (Du et al. 2021).

2.5 Calculation of GHG fluxes, cumulative soil GHG fluxes, and GWP

According to our prior study, the soil daily GHG fluxes and cumulative soil surface GHG emissions were calculated by the general equation described before (Li et al. 2022b). The GWP was used to quantify the total effects of all three GHGs measured at the same measure in terms of CO2 equivalent emissions (Tellez-Rio et al. 2017). The IPCC factors (IPCC 2021) were used to calculate the GWP (Mg CO2-eq ha− 1 yr− 1) over a 100-year time horizon according to following equation:

$$GWP=\left({\textrm{CO}}_2\times 1\right)+\left({\textrm{CH}}_4\times 34\right)+\left({\textrm{N}}_2\textrm{O}\times 298\right)$$
$${\textrm{CO}}_2=-44/12\times SOCSR$$
$$SOCSR=\left({C}_{\textrm{t}}-{C}_0\right)/T\times \overline{\textrm{BD}}\times D\times {10}^{-1}$$

where SOCSR represents the soil organic carbon sequestration rate (Mg C ha−1 yr−1); C0 represents the SOC content (g C kg−1) before the trial that began in 2018; Ct represents the SOC content (g C kg−1) in 2019, 2020, and 2021, respectively; \(\overline{\textrm{BD}}\) represents the average soil bulk density during 2019–2020 (g cm− 3); and D represents the soil depth (cm). The unit transformation factor was 10−1.

2.6 Auxiliary measurements

The fumigation extraction method was used to measure soil microbial carbon (SMC) and nitrogen (SMN) (Vance et al. 1987). The detailed assay was described in our previous study (Li et al. 2022a). Soil bulk density was determined by the ring knife method with three replicates. Soil pH was determined by the potentiometric method with a water to soil ratio of 5:1. Soil organic carbon and total nitrogen were determined by an element analyzer (UNICUBE, ELEMENTAR, Inc., Germany). A straightforward linear calibration was carried out using the drying technique before inserting the vertical soil temperature and moisture sensor (CS655, Campbell Scientific, Inc., Logan, UT, USA) into the soil at a depth of 30 cm. After the sensor was placed and calibrated, the test chamber was put together. A different data logger (CR1000, Campbell Scientific, Inc., Logan, UT, USA) captured data of soil temperature and soil moisture.

2.7 Statistical analysis

Data was processed using Excel 2016 (Microsoft Corp., USA). Means and standard deviation were computed with three repetitions. Data in the figures were presented as mean values ± standard deviations. The mean values, standard deviations, significance, and correlation coefficients were estimated using SPSS 23.0 (SPSS Inc.). The figures were constructed using Origin 2021. A structural equation model was designed to test the impacts of NT on soil GHG emissions and GWP, which was performed by AMOS 24.0 (AMOS China, Beijing, China). The effects of CT and NT on soil CO2, CH4, N2O fluxes as well as GWP were analyzed using one-way analysis of variance (ANOVA) followed by a test for least significant differences (LSD) at P < 0.05. A linear fit and principal component analysis model was used to assess the effects of soil temperature and moisture on soil GHG emissions and GWP.

3 Results

3.1 Soil physicochemical properties and crop yields

Averaged from 2019 to 2021, soil moisture increased 0.78% in NT as compared to CT (Table 1). Annually, soil moisture increased 0.64%, 0.93%, and 0.66% in 2019, 2020, and 2021 under NT compared with those under CT, respectively. No significant differences were observed for soil temperature between CT and NT on either average scale or annual scale. Relative to CT, NT increased soil bulk density significantly by 10.44% during the three cycles. Mean soil pH within the three tested years increased 0.49% in NT compared to that in CT, which was consistent with the outcomes in 2019. The year 2021 had the highest precipitation but exhibited the lowest soil moisture, which was mainly attributed to elevated evaporation induced by high atmospheric and soil temperatures in that year and the timely drainage after heavy rainfalls. The soil bulk density in the year of 2021 was higher than that in the other years for both CT and NT due to soil slumping caused by yearly soil settlement in the test plot and possible soil particle adhesion due to the heavy rainfalls.

Table 1 Effects of CT and NT on soil moisture (SM), soil temperature (ST), soil bulk density, soil pH, SOC, TN, SMC, and SMN from 2019 to 2021
Full size table

Averaged from 2019 to 2021, SMC, and SMN had significant increases of 6.60% and 4.13%, respectively, in NT compared to those in CT, respectively (Table 1). Soil organic C (SOC) increased 10.11% and total nitrogen (TN) decreased 2.00% in NT compared to those in CT during the three observed years. The higher soil SOC values in 2021 than the other years were attributed to the higher precipitation, which inhibited the mineralization process. Concentrations of individual phylogenetic groups within the soil microbial community vary at the global scales and significant variation in soil and microbial element ratios between vegetation types (i.e., forest versus grassland) are also commonly observed (Cleveland and Liptzin 2007). Therefore, we believed that the higher SMC/SMN ratio of this study as compared to results from other studies was attributed to the difference of geographical locations, climatic factors, crop, and terrestrial ecosystem type, etc.

Compared to CT, NT improved the aboveground biomass and grain yields of wheat by 21.3% and 13.3%, respectively, from the averaged results and no significant differences were found for maize yields (Table 2). Meanwhile, the effect of NT on wheat and maize yields showed temporal stability to some extent with increased crop yields in each year.

Table 2 Effects of tillage practices on aboveground biomass and grain yields of wheat and maize from 2019 to 2021
Full size table

3.2 Soil GHG emissions and GWP

3.2.1 CO2

Daily CO2 emission dynamics showed similar trends under the two different tillage practices, with a variation rage of 0.25–98.34 g CO2 m− 2 d− 1 for CT and 0.34–67.37 g CO2 m− 2 d− 1 for NT (Fig. 2a). In this tested area, soil was a source of CO2 emissions, which was proved by the positive values of the daily dynamic variations. The fluctuation range of CO2 emissions was obviously higher in GS compared to NGS.

Fig. 2
figure 2

Effect of CT and NT treatments on soil (a) CO2, (b) CH4, (c) N2O fluxes, and (d) GWP during three annual cycles from Jan 1, 2019 to Dec 31, 2021

Full size image

Averaged from 2019 to 2021, the cumulative CO2 emissions dramatically decreased 35.43% for NT compared to those in CT (Fig. 3a). On the annual scale, the cumulative CO2 emissions decreased 63.73% and 1.41% in 2019 and 2020 in NT as compared to those in CT. In 2021, NT increased the cumulative CO2 emissions insignificantly compared to CT. On the seasonal scale, NT decreased the cumulative CO2 emissions by 35.76% and 34.33% in GS (P < 0.05) and NGS (P < 0.05) across three tested years. In 2019, NT reduced the cumulative CO2 emissions for GS and NGS compared to CT, which was consistent with that of GS in 2020. The gradual decrease in cumulative CO2 emissions from 2019 to 2021 was mainly caused by the gradual increase in precipitation. Previous studies also found that increased precipitation suppressed soil CO2 emissions due to weak soil aeration and enhanced CO2 dissolution (Knorr et al. 2008). There are two peaks occurred in 2019 during the maize growing season, where soil CO2 emissions fluctuated widely and had large values relative to the wheat period because maize was a C4 plant and had a high soil temperature during its growth.

Fig. 3
figure 3

Effect of two treatments (CT and NT) on annual cumulative soil (a) CO2, (b) CH4, (c) N2O fluxes, and (d) GWP in NGS, GS throughout the year during three annual cycles from Jan 1, 2019 to Dec 31, 2021. * Represents a significant difference at P < 0.05 in the same season or whole year according to the least significant difference (LSD) multiple-range test

Full size image

3.2.2 CH4

The CH4 fluxes presented no strong seasonal pattern, which switched between sink and source of CH4 and varied widely among CT and NT treatments with fluctuating ranges of − 10.14–11.26 mg CH4 m− 2 d− 1 and − 5.01–8.46 mg CH4 m− 2 d− 1 (Fig. 2b). Averaged from 2019 to 2021, soil was a net source for CH4 on an annual basis as shown in Fig. 3b.

Relative to CT, the cumulative CH4 emissions decreased 67.33% in NT across 3 years (Fig. 3b). The cumulative CH4 fluxes of GS was a net sink (− 43.24 ± 3.86 mg m− 2, mean ± SD) in NT and a net source (84.94 ± 19.08 mg m− 2) in CT. For NGS, the cumulative CH4 fluxes was a net source in both treatments. Relative to CT, NT inhibited CH4 emissions and switched from a net source to a net sink of NGS and GS in 2019. A similar seasonal pattern was observed in 2020 and 2021, with higher CH4 emission rates in NGS and lower uptake values in GS for both tillage practices.

3.2.3 N2O

Daily N2O flux dynamic variations showed a clear seasonal pattern, which had higher emissions or uptake rates in GS and lower rates in NGS (Fig. 2c). On an average annual basis, the cumulative N2O emissions decreased in NT compared with those in CT and even switched from a net source to a net sink, which was consistent with every tested year (Fig. 3c). Meanwhile, soil was a net sink for N2O in NGS and GS under NT, and a net sink in NGS and a net source in GS under CT.

In 2019, NT inhibited N2O uptake in NGS and N2O emissions in GS compared to those in CT (Fig. 3c). Compared to CT, NT promoted N2O emissions in NGS and N2O uptake in GS in 2020. The tested soil was a net source of N2O fluxes in NGS and GS under CT, but a net sink for N2O under NT for both seasons in 2021.

3.2.4 GWP

The daily GWP dynamics also showed a clear seasonal pattern with higher GWP in GS and lower GWP in NGS, which was similar to CO2 fluxes (Fig. 2d). The trends indicated that CO2 fluxes dominated the annual GWP of all three GHG fluxes. The agroecosystem was found to be a net source of GHGs in both CT and NT (Fig. 3d). The data in Fig. 3 have error bars. Because some error bars are very small, they do not show clearly in the figure, such as the CO2 and GWP data.

In comparison with CT, NT decreased the cumulative GWP by 37.25%, 34.85%, and 37.84% on the average annual, NGS and GS basis, respectively (Fig. 3d). The cumulative GWP decreased 63.73% and 4.93% in 2019 and 2020 in NT compared to those in CT. No significant differences were observed in 2021. On the seasonal scale, NT decreased the cumulative GWP by 64.43% and 63.72% in NGS and GS compared to CT in 2019. In 2020, NT decreased the cumulative GWP by 9.38% compared to CT only in GS.

3.3 NGS contributions for soil GHG fluxes and GWP

The cumulative CO2 emissions in NGS accounted for 16.9% and 17.3% of its annual total accumulated fluxes in CT and NT, respectively (Fig. 4). Meanwhile, the cumulative CH4 fluxes in NGS contributed to 15.6% and 16.4% of its annual budget in CT and NT, respectively. The cumulative N2O emissions in NGS as a percentage of the whole year were 21.6% in CT and 13.8% in NT. For GWP, the percentage of NGS in the whole year were 16.6% and 17.3% under CT and NT, respectively.

Fig. 4
figure 4

Pie diagram for assessing the percentage of NGS in a whole year for soil cumulative CO2, CH4, N2O fluxes, and GWP under CT and NT

Full size image

3.4 Factors regulating GHG fluxes and GWP

Principal component analysis found that soil temperature and soil moisture explained 83.9%, 70.9%, 73.0%, and 83.8% of the variations in soil CO2, CH4 and N2O fluxes, and GWP, respectively (Fig. 5a–d). The soil temperature, moisture and GHGs flux data in Fig. 5 are from observed results of the 3 years 2019, 2020, and 2021. Relative to soil moisture, soil temperature showed a higher correlation with soil CO2 emissions and GWP. CH4 fluxes showed a negative correlation with soil temperature and a positive correlation with soil moisture. N2O fluxes had opposite observations of CH4 fluxes.

Fig. 5
figure 5

Principal component analysis for assessing the effect of soil temperature (ST) and moisture (SM) soil (a) CO2, (b) CH4, (c) N2O fluxes, and (d) GWP under CT and NT

Full size image

Structural equation model analysis showed that soil GHG emissions were controlled by a variety of soil physical and chemical parameters, such as temperature, moisture, pH, bulk density, and carbon and nitrogen content, etc. (Fig. 6). The overall concept of the structural equation model is to show that the physical properties of the soil ultimately affect soil CO2 emissions by influencing the chemical properties of the soil. The indicators in the pink boxes on the left are the soil physical properties and the yellow boxes are the soil chemical properties. Soil compactness is reflected by soil bulk density to a certain degree, which further affects the soil heat dissipation and flow. In this study, soil bulk density was closely linked with soil temperature. Among different factors, soil temperature showed a significant positive correlation with bulk density, and a significant negative correlation with pH and SOC content. Soil bulk density displayed a significant negative correlation with soil moisture and total nitrogen content. Moreover, SOC exhibited a significant positive relationship with both soil moisture and pH. A significant negative correlation between soil moisture and total nitrogen was also observed based on the structural equation model. SOC and total nitrogen contents showed negative (P < 0.05) and positive (P > 0.05) correlations with GHGs emissions, respectively. Compared to soil temperature (r = − 0.135; P < 0.05), soil moisture (r = 0.316; P < 0.05) was a major factor in controlling the intensity of GHG emissions. Meanwhile, SOC (r = − 0.325; P < 0.05) had a greater effect on the variations of GHG emissions as compared to soil total nitrogen content (r = 0.171; P > 0.05).

Fig. 6
figure 6

Effects of soil physicochemical properties on soil greenhouse gases emissions based on structural equation model. Solid and dashed lines represented positive and negative effects in the fitted structural equation model. Line thickness was proportional to the strength of the effects of the variables. Green solid or dashed lines represent significant relationships (* P < 0.05; ** P < 0.01). SM: soil moisture; ST: soil temperature; BD: soil bulk density; SOC: soil organic carbon; TN: soil total nitrogen

Full size image

4 Discussion

Based on in situ observation, this study provided evidence that NT induced reduced CO2 and CH4 emissions, and enhanced N2O uptake as compared to CT in a Chinese wheat–maize rotation system. NT significantly reduced GWP and showed a similar trend to that of CO2 dynamics, indicating that CO2 fluxes dominated the total GHG balance. Notably, NT reduced CO2 emissions and GWP in 2019 and 2020, but increased them in 2021. Therefore, the stability of NT as a conservation tillage practice needs to be further verified on long-term scales. Soil temperature had a stronger correlation with GHG emissions and GWP compared to soil moisture, but structural equation modeling analysis showed that soil moisture was the dominant factor controlling the intensity of GHG emissions. We concluded that a stronger correlation did not mean a dominant factor and higher contribution. We also showed the percentage contributions of NGS in a whole year for soil CO2, CH4, N2O fluxes and GWP in both tillage practices, which were in a range of 13.8%–21.6%. This is a response that has not been characterized in most previous studies.

4.1 Tillage practices effects on GHG fluxes and GWP

On the annual basis, NT decreased CO2 emissions by 35.43% compared to CT (Fig. 3a), which was consistent with our first hypothesis. The findings were also consistent with a previous study in the Spain from 2011 to 2017, which demonstrated that soil CO2 fluxes were low and stable in a monoculture maize system of NT (Dachraoui and Sombrero 2021). This study also reported that NT reduced cumulative CO2 emissions by almost half during the maize GS. However, soil CO2 emissions were also increased by NT as compared to moldboard plow in Andosols with higher soil organic matter and water content, which contradicted the findings of our study. Similar results have been found in the studies of Alskaf et al. (2021). The current work in a wheat–maize crop system supported findings of other studies such as Mangalassery et al.'s (2015) showing that NT had higher soil moisture content, which contributed to decreased CO2 emissions. Zapata et al. (2021) observed that soil temperature and moisture were the most important factors in determining the magnitudes of soil CO2 emissions among the environmental variables. In the prior study, soil moisture was found to be a controlling factor for CO2 emissions compared to soil temperature (Poll et al. 2013), which was consistent with our findings. NT showed different effects on soil CO2 emissions, which were mainly attributed to different soil compositions, properties and crop types in different geographical locations.

The CH4 emissions were significantly inhibited by 67.33% in NT compared to that in CT (Fig. 3b), which was in line with the findings of Pu et al. (2022) and the first hypothesis. Positive and negative CH4 fluxes occurred irregularly during the study period in this study (Fig. 2b). Such high variability of CH4 fluxes within tested years was also observed in other studies (Jin et al. 2017). In our study, soil moisture showed a weak positive correlation with CH4 fluxes (Fig. 5b), which supported the potential CH4 oxidation under low soil moisture (Prajapati and Jacinthe 2014). Tu and Li (2017) discovered that soil temperature and CH4 fluxes showed a negative correlation with enhanced CH4 oxidation by methanotrophs, which was similar to our results (Fig. 5b). However, higher or no significant differences (Bertora et al. 2020) in soil CH4 emissions in NT relative to CT have been reported. Some of the inconsistencies with NT effects on soil CH4 emissions were mainly due to the interactions with other management practices or the soil microclimate. The negative correlation between soil CH4 emissions and bulk density was observed in our study as a consequence of greater compaction, which affected soil aeration and gas diffusivity, leading to a lower CH4 oxidation intensity (Tullberg et al. 2018). Given the lack of consensus on the effect of NT on soil CH4 emissions, and potential complex interactions with environmental and management factors, further investigation is required.

The cumulative N2O emissions were decreased by 339.07% in NT compared to that in CT as a N2O sink (Fig. 3c), as observed by the findings of Wang et al. (2019b). Meanwhile, the reduction of N2O emissions under NT was also observed in the study of Yuan et al. (2018). The lower N2O emissions in NT were attributed to the decreased nitrification resulted from the lower topsoil pH (Li et al. 2015). On the contrary, NT increased N2O emissions compared with CT, which was mainly due to increased soil bulk density, higher water content, and increased denitrification (Niu et al. 2019). In our study, NT increased soil bulk density and soil moisture, but reduced N2O emissions, indicating that nitrification was the dominating nitrogen cycling process and the higher soil moisture accelerated the dissolution of N2O in this tested area. Soil N2O fluxes were negatively correlated with soil carbon content and positively correlated with soil nitrogen content in this study (Fig. 6), consistent with those reported by Pareja-Sánchez et al. (2020). Tierling and Kuhlmann (2018) found that alkaline soils were conducive to high N2O emissions due to more suitable growth conditions for both nitrifiers and denitrifies, which supported our findings (Table 1). Specifically, soil physicochemical properties, climate conditions, cropping system, and the number years were the dominant factors of the effects of NT on N2O emissions.

Compared to CT, NT significantly decreased GWP by 37.25% (Fig. 3d), which was in line with those observed by Shang et al. (2021) and the first hypothesis. In central China, NT was observed to significantly decrease GWP by 27.7% (Guo et al. 2021). The effects of tillage practices on GWP and GHG emissions were influenced by multiple in situ factors (e.g., climatic condition, soil texture, and soil components). Sapkota et al. (2017) showed that NT decreased more GWP than CT with the increase of soil organic carbon. GWP represented the extent of GHG emissions and was influenced by the intensity of CO2, CH4, and N2O fluxes. In our study, CO2 fluxes dominated the balance of GWP variations. A meta-analysis of 90 observations showed that NT reduced GWP through reduction in CH4 emissions, which conflicted with our study. Chatskikh et al. (2008) proved that GHG emissions were decreased by NT through inhibiting net CO2 and CH4 emissions and stimulating N2O uptake, as observed by this study. Similarity, Cooper et al. (2021) also found that net GWP was 30% lower in NT systems, due to enhanced soil carbon stocks and the proportion of sequestered recalcitrant carbon increase and thus lowered CO2 fluxes. In this study, a significant negative correlation between GWP and soil carbon content was also found using structural equation model (Fig. 6).

4.2 The necessity of NGS in accurately predicting soil GHG variations

Most previous studies only explored the effects of tillage practices on soil GHG emissions during GS (Horák et al. 2020), which could result in over- or underestimation of actual GHG emissions in comparing with in-situ observations. In this study, we explored the percentage contributions of NGS in the whole year for soil GHG emissions and GWP. A range of 16.9%–17.3% for CO2, 15.6%–16.4% for CH4, 13.8%–21.6% for N2O, and 16.6%–17.3% for GWP (Fig. 4) were observed, which was in line with our second hypothesis. In a prior study, CH4 emissions during NGS were found to contribute approximately one-quarter of the annual total CH4 emissions (Bao et al. 2021).

The emissions and fluctuation range of CO2 in NGS were obviously lower than those of GS (Fig. 2a and Fig. 3a). For CH4 and N2O, a different seasonal pattern was found in NGS and GS. The net source of CH4 emissions in NGS switched into net sink during GS under NT in 2020 and 2021 (Fig. 3b). Meanwhile, the cumulative N2O fluxes also switched from a net sink in NGS to a net source in GS under both tillage practices in 2019, as observed in Pu et al. (2022). Gong et al. (2021) also found the difference and concluded that the switch of soil CH4 and N2O fluxes as sink or source between NGS and GS was closely associated with soil temperature. A study by Lin et al. (2015) in alpine meadows of the Qinghai–Tibet Plateau observed that soil CH4 uptake increased with the increase of temperature. However, the increase was not consistent seasonally, which was more pronounced in NGS than that of GS. The switch between sink and source for CH4 and N2O emissions in NGS was attributed to climatic factors, such as temperature and precipitation. Therefore, when studying the variations of soil GHG emissions over time and the interannual cumulative effects, NGS should be treated as an important period. Otherwise, the seasonal variation pattern and GHG fluxes would be misjudged.

4.3 Relationships between soil physicochemical properties and GHG fluxes

Relationships between GHG fluxes and soil properties have not yet been consistently concluded and remain elusive in the literature. To address these “hot spots”, we performed structural equation model to assess these relationships and seek main pathway of the effects of soil physicochemical properties on GHG emissions. We found that NT increased soil moisture and pH, thus enhancing SOC stocks and decreasing nitrogen stocks, eventually inhibiting GHG emissions based on structural equation model (Fig. 6). Bai et al. (2019) reported that NT was widely adopted to reduce GHG emissions due to increased soil carbon content, which was consistent with our results.

The microbial decomposition of soil organic matter was promoted by CT due to enhanced soil disruption, disturbance of soil aggregates, and improved aeration, consequently causing carbon loss in the form of CO2 (Plaza-Bonilla et al. 2014). Lu et al. (2016) also found that NT decreased CO2 emissions with carbon protection associated with improved soil aggregation and reduced soil temperature, which was in line with our result that soil temperature had a negative correlation with soil carbon stocks (Fig. 6). Soil acted as a source of CH4 emissions in this studied site, depending on the soil carbon content, soil temperature and moisture, pH and bulk density (Shakoor et al. 2020). NT increased bulk density with soil compaction and adversely affected CH4 emissions. Meanwhile, soil mineralization processes were inhibited by increased soil bulk density and decreased soil temperature, thus reducing the methanogenic bacteria activity and CH4 emissions (Zhang et al. 2015). Similarity, NT led to anaerobic microsites to limit availability of decomposable soil organic carbon as well as mineral nitrogen to microbial communities, thus decreasing N2O emissions (Ruan and Philip Robertson 2013). In this study, N2O manifested itself as a soil sink, which was probably attributed to the increased soil moisture and subsequent accelerated dissolution.

5 Conclusions

Most previous studies investigated the effects of GS on CO2 emissions, but ignored NGS, which resulted in over- or underestimation of soil CO2 emissions. This study conducted structural equation modeling to reveal the main processes by which tillage practices affected soil CO2 emissions. Our results elucidated that the differences of GHG emissions were induced by tillage practices and farming season and NGS needed to be considered to accurately quantify soil GHG emissions. Moreover, in response to the two hypotheses, the following conclusions were also obtained: (1) NT practice, as compared to CT, significantly decreased the emissions of CO2, N2O and CH4 and GWP by 35.43%, 67.33%, 339.07% and 37.25% during three annual cycles, respectively; and (2) GHG emissions in NGS accounted for a high percentage of the entire year, ranging from 13.8%–21.6% under CT and NT. Meanwhile, we concluded that a higher correlation did not mean greater importance or a greater contribution, and thus needs more attention in future studies. In exploring the influence of tillage practices on soil GHG emissions, only soil physicochemical properties were measured in this study. Therefore, attention should be paid to soil microbial scales in future studies to better reveal the intrinsic mechanisms. Taken together, NT can be extensively adopted as an eco-friendly tillage practice, but the stability needs to be verified on long-term scales. It is expected that more researchers will continue the exploration in the future to identify the uniform law and perform reliability verification.