Improved Fertiliser Management To Reduce The Greenhouse-Gas Emissions And Ensure Yields In A Wheat–Peanut Rotation System In China

Over the last century, anthropogenic greenhouse-gas (GHG) emissions have changed the global climate, and agriculture plays an important role in the global ux of GHG. Agricultural management practices, such as split N applications and the use of controlled-release fertilisers have signicantly increased the crop yield and N-use eciency by balancing the N demand of crops and the N availability of soils. However, the impacts of these practices on GHG emissions (in particular in wheat–peanut relay intercropping systems) have not been evaluated in detail. In this study, we used a common compound fertiliser and a controlled-release compound fertiliser (CRF) applied the day prior to sowing, at the jointing stage of wheat, and at the anthesis stage of peanut in various proportions, with a control treatment of 0 kg ha -1 . The ndings demonstrated that treatment JCF70 achieved increases in yields of 9.7% and 14.6% for wheat grain and peanut pod, respectively, compared to treatment JCF100; however, this treatment also signicantly increased soil emissions of CO 2 and N 2 O. In addition, cumulative emissions of CO 2 and N 2 O were higher in the peanut growing season by 74.4 and 31.7%, respectively, than in the wheat growing season owing to the relatively higher soil temperature during the former season. Fertilisation combined with irrigation, was found to be the main cause of GHG emissions. Under the same fertiliser rate and N-management style, JCRF70 further increased the yield of peanut pods and the total combined yield of peanut and wheat by 10.3% and 8.9%, respectively, compared to treatment JCF70. The cumulative CO 2 and N 2 O emissions in treatment JCRF70 were 20.4–45.4% less than those in treatment JCF70. The total global warming potentials of CO 2 and N 2 O were lowest in treatment JCRF70 owing to it providing the highest grain yield. Therefore, N application with three splits, together with the use of a slow-release fertiliser, may be a simple and effective approach to enhance the grain yield while reducing the GHG emissions.


Introduction
Global climate change has been shown to largely result from rising emissions of greenhouse gas (GHG) (IPCC, 2013). The use of mineral nitrogen (N) fertiliser is amongst the largest contributors to emissions of GHG from agricultural production systems, particularly in China (IPCC, 2007;Rochette et al., 2008). Due to China's growing population and limited arable land, chemical fertilisers have been used at an increasing rate to meet agricultural demands (FAO, 2010), however, most research has found that CO 2 emissions increase exponentially with increased N application rate and when N availability exceeds plant demand (Zhong et al., 2016;Dossou-Yovo et al., 2016). In agriculture, the increase in chemical fertilisation not only causes direct economic losses but also increases nitrate (NO-3) leaching and emissions of N 2 O, which is an important greenhouse gas whose global warming potential (GWP) is 298 times that of CO 2 (Guo et al., 2010). In the NCP, wheat-peanut relay intercropping systems are widely used to produce more food for the increasing population. In such systems, local farmers typically apply 100% of the fertiliser in the wheat growing stage since peanut plants are known for their capacity to convert atmospheric nitrogen (N 2 ) into a usable form of N (Zhang et al., 2016). For wheat, the mean N application level is around 300 kg ha − 1 (Ma et al., 2015). However, a level of 240 kg N ha − 1 was suggested to achieve adequate grain yield and N utilisation e ciency (NUE) . Due to the different N demands of plants in different growth stages, excess nitrogen may not only reduce the NUE of crops but may also increase the amount of N that is lost to the environment and may thus contribute to pollution in the atmosphere and in aquatic systems (Guo et al., 2010;Wang et al., 2016). A previous study found that excessive nitrogen could immediately increase the soil NO-3-N contents to a level exceeding the range of crop absorption and utilization at this stage and could produce N 2 O through nitri cation via the oxidation of ammonia (NH 3 ) to nitrite (NO-2-N) Zebarth et al. (2012). Additionally, although peanuts, as legumes, can x nitrogen from N 2 , more than half of the nitrogen needs to be obtained from soil and fertiliser, and a moderate supply of nitrogen can promote the growth of peanut root nodules .
Therefore, nitrogen fertilisation strategies should be improved to obtain high yield in wheat-peanut relay intercropping systems.
Improved fertilisation management can also increase fertiliser NUE and reduce GHG emissions (Liu et al., 2015). Split N fertiliser applications, adjusted ratios of basal-N to top-dressed N, and the application of controlled-release fertiliser (CRF) have been proposed as means to decrease N 2 O emissions (Akiyama et al., 2010;Ye et al., 2013). Speci cally, for a given total N rate, when the rate of N fertiliser application is kept low through split application, lower cumulative CO 2 emissions would be expected relative to a single application. Split application of N fertiliser has been demonstrated to increase the synchronicity between the N demand of crops and soil-N availability, and can therefore improve the NUE (Ribaudo et al., 2011).
Past research has investigated the impact of split N application and the timing of N application on GHG emissions; however, consistent results have not been obtained (Zebarth et al., 2012;Wang et al., 2016;Liang et al., 2017). Yan et al. (2001) and Burton et al. (2008) studied the impact of split N application on N 2 O emission and found that split or delayed application of N could obtain lower or equivalent cumulative N 2 O emissions than a single early N application; however, no signi cant effect was found in the second year. Additionally, in a rain-fed potato experiment, Zebarth et al. (2012) showed that, although split N application led to a signi cant reduction in the nitrate exposure, it did not mitigate the N 2 O emissions from medium-textured soil. Furthermore, a study on maize production using plastic mulching in a semiarid region indicated that N application at a 4:3:3 ratio at the six-leaf, ten-leaf, and grain-lling stages carried a higher risk of increasing N 2 O emissions compared to the application of N at a 4:6 ratio at the six-leaf and ten-leaf stages. This was as a result of the strong and frequent rainfall in the middle and late growing stages of maize . Moreover, Venterea and Coulter (2015) showed that the split application of granular urea to meet crop N demand did not always reduce, and sometimes increased, emissions of N 2 O. These con icting ndings may re ect the potential sensitivity of soil emissions of N 2 O to a wide range of factors that can affect N 2 O production, such as factors related to management, the environment, and soil (Venterea et al., 2012).
New technologies employing CRF are emerging as excellent ways to improve nutrient application and thus to signi cantly reduce environmental risks while achieving high-quality crop yields (Shi et al., 2013). Chi et al. (2020) investigated the impacts of urea, amine, and CRF on soil CO 2 emissions in wheat-maize rotations and found that the use of CRF was associated with lower CO 2 emissions and higher yield.
Additionally, there is increasing evidence that CRF can be used to control the timing of N release from fertiliser. Therefore, the application of CRF in agricultural production can signi cantly decrease N losses from leaching in NO-3-N via the volatilisation of NH 3 and via the emission of N 2 O (Shaviv, 2001; McTaggart and Tsuruta, 2003). However, contradictory results have been obtained regarding the potential of CRF to lower GHG emissions from upland soils in different locations. For example, Ball (2004) reported that CRF reduced the high initial N 2 O ux after heavy rainfall but did not signi cantly mitigate uxes of CO 2 or CH 4 under silage production. Additionally, Yan et al. (2000) found that the ability of CRF to reduce N 2 O emissions from paddy soil was balanced by the increased release of NH + 4-N in the mid-season aeration stage. Meanwhile, Li et al. (2004) found that CRF signi cantly decreased cumulative N 2 O emissions from a rice eld due to its regulation of the timing of N release.
Several studies have investigated the impacts of N fertilisation strategies on GHG emissions from wheat (Liu et al., 2011) and maize  cropping systems in Central China. However, only a few such studies have been conducted on peanut elds. Furthermore, the majority of the previously mentioned studies involved only single-season cropping systems, while a few focused on wheat-peanut relay intercropping systems in the NCP. The goals of this research were to (1) determine the seasonal changes of the NH + 4-N and NO-3-N content of soil and GHG emissions in wheat-peanut relay intercropping systems over the entire year, (2) analyse the effects of split application of N on the crop yields and cumulative emissions of CO 2 and N 2 O in both the wheat and peanut growing seasons, and (3) quantify the combined impacts of split N application and the use of CRF on the global warming potential (GWP) and global warming potential intensity (GHGI). We hypothesised that split N application in combination with CRF would reduce the GHG emissions without sacri cing crop yields in a wheatpeanut relay intercropping system in China.

Experimental area, design, and treatment
Field experiments were performed at the experimental farm of Shandong Agricultural University, Taian, Shandong Province, China, in the NCP (36°09′ N, 117°09′ E) from 2017 to 2019. The study site experiences a temperate continental monsoon climate. The total precipitation during the wheat growth period was 182.6 mm in 2017-2018 and 104.2 mm in 2018-2019, and those during the peanut growth periods were 532.3 mm and 366.8 mm, respectively (Fig. 1). The soil at the experimental farm is sandy loam (Cambisols; FAO/EC/ISRIC, 2003). The soil properties for a depth of 0-20 cm were as follows: pH, 8.13; organic matter, 12.7 g kg − 1 ; total N, 0.96 g kg − 1 ; available phosphorus (Olsen-P), 45.1 mg kg − 1 ; and available potassium (NH 4 OAc-K), 83.1 mg kg − 1 .
The experimental design involved an entirely randomised block with three replicates. Plots were spaced by 3 m × 3 m concrete wall barriers that acted as water barriers. In this study, the N, P 2 O 5 , and K 2 O contents of both the common compound fertiliser (CCF) and the CRF were 20%, 15%, and 10%, respectively, and three splits fertiliser application at a rate of 1500 kg ha − 1 (300 kg ha − 1 pure N) was the main factor that was studied. A N fertiliser was used prior to sowing, at the jointing stage of wheat, and at the anthesis stage of peanuts at ratios of 50-50-0% (JCF100), 35-35-30% (JCF70), and 35-35-30% (JCRF70), with a control (CK) of 0 kg N ha − 1 . Details of the different treatments are provided in Table 1 and Fig. 2. For base N, a N fertiliser was manually spread across the soil surface before sowing and was then ploughed into the soil, while for top-dressed N, a N fertiliser was added in bands in the centre of the rows at a depth of 5 cm for both winter wheat and peanuts. Winter-wheat variety Jimai 22 and peanut variety Shanhua 101 were used. Wheat was sown in the plots in nine rows with a row spacing of 30 cm, and peanuts were sown between the winter wheat rows 15-20 days before wheat harvesting with a spacing of 22 cm between plants (Fig. 2). Diseases, weeds, and pests were adequately controlled in all the treatments. cm) with a water-lled groove was placed under the static chamber to maintain an airtight seal in the top chamber during the air-extraction process. The pedestal were inserted in the soil at a depth of 20 cm after application the basal fertiliser, one row wheat were sown near the edge in the pedestal and the number of wheat seeds was calculated according to design and the inner wheat row in the pedestal is consistent with in the eld. Peanut was sown in the winter wheat rows and one peanut plant were included in the chambers after wheat harvest. The pedestal was removed prior to and reinstalled after top-dressed fertiliser at the same site. Additional chambers (50 × 50 × 80 cm) were used between jointing and harvest to adapt the height of the growing wheat plants at this stage. To minimise internal temperature changes, each chamber was coated with Styrofoam. The chambers contained a sampling outlet in the top portion of the lateral plane, and small fans were placed in the upper part of each chamber to guarantee the full mixing of air within the chamber. One chamber was placed in each plot, with a total of three replicates per treatment.
The rst sampling was performed immediately after the basal fertiliser application of wheat and the following samplings were conducted once a week thereafter until peanut harvest. An additional collection every day for one week was conducted after the second and third fertilisations. Gas samples were  (2) 2.3. Global warming potential (GWP) and global warming potential intensity (GHGI) To calculate the global warming potential (GWP) under different agricultural regimes, the GWP of N 2 O was considered to be 298 times that of CO 2 (IPCC, 2013).
The global warming potential intensity (GHGI) was obtained by dividing the GWP by the grain yield where W is the soil water content (%), and r is the soil bulk density.

Statistical data analysis
All data were analysed by ANOVA using the DPS software (version 7.05; Hangzhou RuiFeng Information Technology Co., Hangzhou, China). The statistically signi cant differences between treatments were determined by the least signi cant difference test (LSD) at the P < 0.05 level. The wheat and peanut yield dates reported in this study were measured over 2017-2019. The data were plotted using the SigmaPlot 10.0 software (Systat Software, San Jose, CA, USA).

Environmental variables and soil condition
The daily air temperature and precipitation, as well as soil temperature and WFPS, are shown in Figure 1. Strong precipitation occurred during the peanut growing season, especially in August. The daily average air temperature during the wheat growing season ranged from -6.9 to 28.8 °C in 2017-2018 and from -4.8 to 29.3 °C in 2018-2019, and those during the peanut growing season ranged from 21.2 to 31.8 °C and from 21.1 to 31.5 °C, respectively (Fig. 1A,B). The seasonal variation trend of soil temperature was comparable to that of air temperature, ranging from 4.8 to 26.7 °C in the wheat growing season and from 21.1 to 31.5 °C in the peanut growing season (Fig. 1C). The mean seasonal soil temperature in the peanut growing season was 13.8 °C greater than that in the wheat growing season. The soil WFPS signi cantly increased after irrigation and heavy rainfall. There was no signi cant difference in soil WFPS between the various fertilisation rates. However, the soil WFPS varied from 15.1 to 76.9% during the wheat growing season, which was generally lower than that in the peanut growing season due to soil evaporation and less precipitation.
The application of N fertiliser was found to signi cantly increase the soil concentration of NH+ 4-N and NO-3-N compared with the control; these concentrations sharply increased after fertilisation in all treatments, especially in the 0-20 and 20-40 cm soil layers (Fig. 3). During the wheat growing season, peak soil NH+ 4-N and NO-3-N concentrations were observed after the jointing fertilisation, the peak values in JCF100 was higher than that in JCF70 and JCRF70. Thereafter, the soil NH+ 4-N content rapidly decreased and remained low until the harvest period of wheat, while the soil NO-3-N content of JCF100 reached another relatively large peak in the 0-20 cm soil layers in the lling stage of wheat. During the peanut growing season, soil concentrations of NH+ 4-N and NO-3-N were signi cantly higher in treatments JCF70 and JCRF70 compared to treatment JCF100 due to the fact that 30% of the total fertiliser was applied in the initial peanut owering stage. Under the same fertiliser rate, the peaks of the soil NH+ 4-N and NO-3-N concentrations were larger in treatment JCF70 than in treatment JCRF70.

CO 2 emission
In the wheat growing season, except for a small peak which occurred after basal fertilisation during 2017-2018, the CO 2 emission uxes remained at a low level in the overwintering period in all treatments ( Fig.   4A,B). However, the CO 2 emission uxes resumed and reached their maximum two and three days after the jointing fertilisation and irrigation events. For the CK, the CO 2 emission uxes was maintained at a low level until the harvest period of wheat, the application N fertiliser can signi cantly increase the ux of CO 2 emission, and this change increased with increasing N application rate, the CO 2 emission uxes were higher by16.3-71.9% compared with CK over the two years ( Fig. 4A-1 The peak CO 2 emission uxes of treatments JCF70 and JCRF70 were higher by 83.6% and 52.9%, respectively, compared with that of JCF100 ( Fig. 4A-2,B-2). Due to a series of precipitation events (totalling approximately 86 mm) which occurred between 27 July and 02 August 2019 (Fig. 1B), an obvious and large peak in CO 2 emissions was observed on 02 August. In addition, the CO 2 emission uxes during the peanut growing season were higher than during the wheat growing season, especially in August and September from 2018-2019. At the same fertiliser rate, the CO 2 uxes were lower in treatment JCRF70 than in treatment JCF70 for both the wheat and peanut growing seasons over the two years.
The application of N fertiliser was found to signi cantly increase the cumulative CO 2 emissions compared with CK; with N fertilisation, the cumulative CO 2 emissions were between 8648 and 16609 kg CO 2 ha -1 in 2018 and 10918 and 16870 kg CO 2 ha -1 in 2019 during the wheat growing season, while those during the peanut growing season were between 9206 and 16940 kg CO 2 ha -1 in 2018 and 16494 and 27700 kg CO 2 ha -1 in 2019 (Table 2). Furthermore, split applications of N were found to signi cantly increase the cumulative CO 2 emissions. The annual cumulative CO 2 emission in treatment JCF70 was 9.0% higher than that in treatment JCF100 over the two years. However, under the same fertiliser rate, the annual cumulative CO 2 emission in treatment JCRF70 was lower by 14.1% compared with that of treatment JCF70. The annual accumulated CO 2 emissions were signi cantly in uenced by the fertilizer treatment and year, but not signi cantly affected by the interaction between the fertilizer treatment and the year (Table 3). Moreover, the results of the correlation analysis (Table 4) showed that the daily CO 2 emission was signi cantly positively correlated with the soil temperature, the WFPS, and the soil NH+ 4-N concentration, with correlation coe cients of 0.43, 0.55, and 0.64, respectively, at the P<0.01 level.  The N 2 O emission uxes exhibited the same trends as the CO 2 emission uxes for both the wheat growing season and the peanut growing season in the two years (Fig. 5A,B). emissions after anthesis fertilisation was lower than that after jointing fertilisation due to the lack of irrigation in this fertilisation event, but emission valus were maintained at a relative high level until the harvest period of peanut. The N 2 O emission uxes reached the peak on the second day for treatments JCF70 and JCRF70 after anthesis fertilisation, and the N 2 O emission ux for JCF70 was higher than for JCRF70 all the time from day two to day six ( Fig. 5A-2,B-2). In all the N fertilisation treatments, the mean N 2 O emission ux was higher in the growing season of peanut than in the growing season of wheat due to the higher temperature and greater amount of rainfall in the former in both years (Fig. 1A,B). The second peak of N 2 O emissions was 70.6% and 49.5% higher in JCR70 and JCRF70, respectively, compared to that in the JCF100 treatment (Fig. 5 a-2 5A,B) and the annual cumulative N 2 O emissions in treatment JCRF70 were lower by 34.3% in 2018 and by 51.8% in 2019 during the wheat growing season, and those lower by 7.9% and 13.0% during the peanut growing season, respectively, compared with treatment JCF70. The annual accumulated N 2 O emissions were signi cantly in uenced by the fertilizer treatment, but not signi cantly affected by the year and the interaction between the fertilizer treatment and the year (Table 3). Additionally, in all treatments, soil variables were found to have no signi cant correlation with daily N 2 O emission uxes, while there was found to be a negative correlation between N 2 O ux and soil temperature (Table 4).

Grain yields and GHGI
Compared to CK, wheat grain yields were found to be signi cantly higher by 36.7-54.3% in 2018 and by 31.8-57.6% in 2019 under all N fertiliser treatments (Fig. 6 A, B). Meanwhile, the wheat grain yield of treatment JCF70 was higher than that of treatment JCF100 despite the fertiliser use being a 30% lower. Compared with treatment JCF100, the grain yields in treatments JCF70 and JCRF70 were higher by 10.9% and 12.8%, respectively, in 2018, and by 16.7% and 19.6%, respectively, in 2019. However, no signi cant difference in grain yield was found between treatments JCF70 and JCRF70. Split N fertiliser application was found to signi cantly increase the pod yield and kernel yield of peanut in both growing seasons compared with CK ( Fig. 6 C, D). Compared with treatment JCF100, the pod yields in treatments JCF70 and JCRF70 were higher by 9.7-21.0% in 2018 and by 14.6-24.8% in 2019, respectively.
Additionally, under the same N fertiliser rate, the pod yields for treatment JCRF70 in 2018 and 2019 were 5.2% and 8.9% higher than for treatment JCF70, respectively.
As shown in Table 2, N application was found to signi cantly increase GWP and GHGI in both the wheat and peanut growing seasons. During the wheat growing season, the GWP and GHGI in JCF100 were higher than that in treatments JCF70 and JCRF70 due to all the fertiliser were applied to wheat. However, there were no statistically signi cant difference of GWP between treatments JCF100 and JCRF70 for the peanut growing seasons and the annual. In addition, the annual lowest GHGI was obtained in treatment JCRF70 due to a higher total grain yield in both years. In comparison with the respective urea treatments, CRF treatments were found to signi cantly decrease GHGI in both the wheat and peanut growing seasons.

Effects of splitting N applications on GHG emissions
For a given total N rate, when the rate of N fertiliser application is kept low through split application, lower cumulative CO 2 emissions would be expected compared with a single application. However, in the present study, throughout the trial period, three splits of N application were found to signi cantly increase the CO 2 emissions, and the high and short-term CO 2 emission peaks were recorded following fertilisation. This nding is consistent with most previous studies (Ward et al., 2017; Wang et al., 2016). The annual cumulative CO 2 emissions in treatment JCF70 were shown to be 5.2% higher than those in treatment JCF100. This is mainly due to the third top-dressed fertiliser at the anthesis stage of peanut, which may indirectly affect the respiration of microorganisms in crops and soil and directly affect soil C content, while CO 2 emissions are comprised of soil respiration and plant respiration, and the CO 2 uxes increase with the soil organic C content (Ward et al., 2017). Except for the third top-dressed fertiliser, the higher temperature and WFPS in the peanut growing season exacerbate this process. Zou et al. (2018) and Rustad et al. (2001) also observed increases in soil CO 2 emissions following warming and precipitation in numerous regions. As we know, that nitrogen availability can be utilized by soil respiration and fertiliser as major factors affecting soil respiration in farmland, directly by in uencing root and microbial activities and indirectly by in uencing physical and chemical soil properties (Ding et al., 2007;Huang et al., 2012;Fan et al., 2015;Gong et al., 2015). Therefore, it can be reasonably assumed that N fertilisation may impact CO 2 emissions by affecting these factors. Additionally, compared to the fertiliser treatments, CK had a smaller CO 2 emission peak at the jointing stage of wheat, the most likely reason for which is that the eld was irrigated after jointing fertilisation. In the NCP, farmers generally irrigate after fertilisation, thereby maintaining the soil moisture in the growing season, while the soil WFPS can signi cantly affect CO 2 emissions (Dossou-Yovo et al., 2016). These results illustrate that the CO 2 emissions peak accompanies irrigation, and fertilisation could improve the peak of CO 2 emissions.
As shown in Table 4, a signi cant positive correlation was observed between soil temperature and cumulative CO 2 emissions. CO 2 emission peaks in the wheat growing season were found to be higher than those in the peanut growing season. Besides a high fertilisation rate, this can be attributed to the fact that the eld was irrigated after jointing fertilisation. However, the cumulative emissions of CO 2 in the peanut growing season were signi cantly larger than those in the growing season of wheat for all treatments. This apparent contradiction has three possible explanations. First, in the NCP, the highest air temperature usually occurs between June and August. The higher soil temperature in the peanut growing season leads to enhanced respiration of peanut plants and soil microbes. Second, the anthesis stage is the key period for the rapid growth of peanut plants. The third fertilisation greatly promotes root growth and the accumulation of aboveground biomass, which leads to the increase of carbon inputs to the soil and the induction of enzyme activity, which may induce the mineralisation of soil organic matter (Du et al., 2018b;Lu et al., 2011). Third, previous studies showed that CO 2 emissions from soil are caused by autotrophic and heterotrophic respiration, which can be affected by WFPS (Buragienė et al., 2019). In our study, a signi cant positive correlation was found between soil temperature and WFPS when several strong precipitation events occurred from late July to early August (Fig. 1A,B), during which the WFPS remained at a high level for several consecutive days (Fig. 1C). The ndings of Wang et al. (2013) were comparable; they reported high CO 2 concentrations in the warm and moist maize growing season and low concentrations in the winter-wheat growing season in a winter-wheat-summer-maize double-crop rotation.
Agricultural production is a primary source of atmospheric N 2 O and contributes around 60% to human N 2 O emissions (Reay et al. 2012), mainly due to the input of excessive N fertiliser. In wheat-peanut relay intercropping systems, the typical farmer's practice is to apply all fertiliser to the wheat, and this kind of centralized application of fertiliser will immediately increase the soil NO-3-N contents to a level that is far beyond the range of wheat absorption and utilization at this stage. Therefore, the excessive soil NO-3-N will produce N 2 O through nitri cation via the oxidation of ammonia (NH 3 ) to nitrite (NO-2-N). N 2 O can also be produced via denitri cation. Many researchers have studied the impact of split N application on N 2 O emission uxes; however, no consensus has been reached on this issue. Cumulative nitrous oxide emissions in the growing season following urea application in three split applications were found to be signi cantly higher compared to a one-time application in the corn stage of a corn-soybean rotation (Venterea and Coulter 2015) and in maize cultivation in a semiarid region in Central China . Additionally, Yan et al. (2001) observed no signi cant effect on N 2 O emissions when using split N application in maize cultivation under low precipitation, however proposed that a signi cant reduction in N 2 O emissions would be expected from split N application under normal rainfall conditions. Conversely, Burton et al. (2008) showed that the split application of N reduced N 2 O emissions in years with heavy rainfall between planting and hilling.
In this study, the N 2 O emissions in fertiliser treatment JCF70 were found to be 38.7% higher than those in treatment JCF100 across all studied years and rotations. Comparable ndings were also reported by fertilisation. This is consistent with the ndings of the present study. A higher N 2 O emission peak was observed on 04 April after fertilisation closely followed by irrigation events compared to the N 2 O emission peak on 27 June which only had irrigation, not rain (Fig. 4A). This can mainly be attributed to the fact that soil concentrations of NH + 4-N and NO-3-N increased following irrigation and fertilisation (Fig. 3), which provided an adequate substrate for nitri cation and denitri cation. This led to rapid rates of hydrolysis and nitri cation during the N fertilisation (Liu et al., 2003) and thus an extremely large quantity of N 2 O was immediately emitted, leading to an emissions peak. In the present study, we observed a positive correlation between emissions of N 2 O and soil temperature in the wheat growing season and a negative correlation in the peanut growing season, which is close to the results of a past study .
However, other research found a signi cant positive correlation between N 2 O emissions and soil temperature (Ding et al., 2007;Allen et al., 2010). The negative correlation that was observed in the present study between N 2 O emissions and soil temperature in the peanut growing season may be due to the high N 2 O emissions immediately after precipitation, when soil temperature is low (Fig. 1C). This demonstrates the speci city of the impact of soil temperature on GHG emissions under diverse environmental conditions and agricultural management styles.

Effects of CRF on GHG emissions
In general, many synthetic N fertilisers are available to agricultural managers. Previous research has concentrated on the impact of various kinds of chemical N fertiliser on CO 2 emissions from soil. These studies found that N fertiliser application results in temporary increases in CO 2 concentrations and that soil respiration differs signi cantly between different forms of N fertiliser. Chi et al. (2020) reported that, during the entire sampling season in wheat-maize rotations, the contents of soil N and C under slowrelease fertilisation was able to effectively control the CO 2 emission ux. In the present study, the cumulative CO 2 emissions of treatment JCF70 were found to be signi cantly higher than those of treatment JCRF70. This may be due to the fact that the rapid release of nutrients in the CCF treatment resulted in the increase of soil N content, thus increasing the CO 2 emissions. Additionally, the cumulative CO 2 emissions of treatment JCRF70 were found to be signi cantly lower than those of treatment JCF70, and as shown in Table 2, there was no signi cant difference in CO 2 emissions between treatments JCF70 and JCF100. These ndings indicate that the use of CRF could reduce cumulative CO 2 emissions from split N application. Previous studies found that nitrogen availability can be utilised by soil respiration and that there exists a process between the application of nitrogen and soil CO 2 emission. During such a process, the available N is slowly released from the CRF during decomposition. This N release is controlled by a polymer membrane (Liang and Liu, 2006). This suggests that CRF could reduce CO 2 emissions by the controlled release of nitrogen.
In our research, the effects of fertiliser management and year on cumulative CO 2 emissions were highly signi cant (P < 0.001), however, no signi cant fertiliser management×year interaction effects were observed on cumulative CO 2 emissions (Table 3). Additionally, as shown in Table 2 The intensity of nitri cation and denitri cation are functions of soil concentrations of NH + 4-N and NO-3-N, respectively (Smith et al., 1998a, b;Dobbie and Smith 2003). Previous studies have shown that there is a positive correlation between nitri cation and denitri cation and the soil mineral N content in cool and humid regions only when this concentration is greater than 5 mg N kg − 1 (Chantigny et al., 1998). Under the same fertiliser rate, the use of CRF has been identi ed as a possible mitigation strategy by the IPCC as it can regulate the timing of N release by diffusion through the fertiliser's coating, which better synchronises with the crop's demand for fertiliser N and thereby reduces soil NO-3-N accumulation after fertiliser application, at least temporarily, and thus decreases the quantity of N that is available for denitri cation (Shaviv et al., 2001;Ji et al., 2013). In our study, both CRF and CCF were found to increase the peak N 2 O emissions from soil, however the former led to a smaller increase than the latter. Compared with treatment JCF70, the cumulative N 2 O emissions in treatment JCRF70 were lower by 34.1% and 13.0% in the wheat and peanut growing seasons, respectively. This was due to the lower concentrations of NH + 4-N and NO-3-N in the CRF treatment compared with the CCF treatment. Our study found no correlation between soil concentrations of NH + 4-N or NO-3-N and N 2 O emissions (Table 4). Similarly, Zhong et al. (2016) and Ji et al. (2013) did not observe a correlation between these parameters, despite the fact that most studies observed that N 2 O emissions are strongly correlated with soil concentrations of NH + 4-N or NO-3-N Shi et al., 2013). These studies stated that N fertilisers provide N resources for N 2 O production, while other environmental characteristics such as soil water content and soil temperature determine N 2 O production (Schuster and Conrad 1992). In the future, it is important to investigate the link between N 2 O emissions and soil concentrations of NH + 4-N or NO-3-N under a variety of soil moisture regimes. In this study, we found that in most cases, the soil NO-3-N concentration was lower in the CRF treatment than in the CCF treatment. Nevertheless, at the end of the peanut harvest stage, higher soil concentrations of NH + 4-N and NO-3-N were observed in the CRF treatment compared to the CCF treatment which supports the results of previous research (Ji et al., 2013;Peng et al., 2011). This proves that CRF released fertiliser N slowly, thus extending nitrogen supply, and released some N in the late peanut growth stage.

Yield, GWP, and GHGI
Our study found that split N application not only signi cantly increased the peanut pod yield but also increased the wheat grain yield, despite a 30% reduction in fertiliser use for wheat (Fig. 6A). This result can be explained by the ndings of our previous study, in which the application of N fertiliser with three splits was shown to increase the plant uptake of N derived from fertiliser and N harvest index (NHI) and reduce N loss, thus increasing the total grain yield relative to a treatment where the total yearly N fertiliser was added to wheat in two splits (Liu et al., 2018). In wheat-peanut relay intercropping rotation systems, local farmers traditionally apply 100% of the fertiliser in the wheat growing season to ensure high wheat yield. The farmers' average N application rate for wheat is signi cantly higher than the recommended rate to satisfy grain yield and NUE (Ma et al., 2015;Li et al., 2011). Such inappropriate application methods not only decrease NUE but also lead to de cient soil fertility following the wheat harvest and result in insu cient soil nutrients for peanut growth in the middle and late growth stages, thus resulting in a low yield (Liu et al., 2019). In the present study, treatment JCF70 better matched soil nutrient availability with crop demand by the application of N with three splits. Furthermore, for a given proportion of N-P 2 O 5 -K 2 O and equal quantities of nutrients, no signi cant difference was observed in the wheat grain yield between treatments JCF70 and JCRF70. However, the peanut pod yield and the total yield of treatment JCRF70 were found to be signi cantly higher by 8.9% and 5.3% respectively, relative to treatment JCF70. Additionally, treatment JCRF70 was found to signi cantly reduce the GWP in both the wheat and peanut growing seasons compared to treatment JCF70. Therefore, the investigated CRF technologies can be expected to have positive effects on environmental quality and crop production in the long-run.
Additionally, to prevent environmental pollution, maintain yield, and achieve economic bene ts, the reduced application of CRF and the combined application of urea and CRF warrants further study.
The goal of agricultural production in a whole agro-ecosystem is to increase the economic output from crops and the sustainable development of agriculture while achieving a win-win situation in terms of the economic and environmental aspects of farmland ecosystems (Song et al., 2013). Therefore, good agronomic practices should lead to higher grain yield and lower GHG emissions. In the current study, the annual GWP of treatment JCF70 was found to be signi cantly higher than that of JCF100, while the GHGI of treatment JCF70 was not found to increase due to a higher grain yield. The same annual GWP was obtained for treatments JCF100 and JCRF70, however the lowest GHGI was observed for treatment JCRF70 due to it having the highest grain yield. Some previous studies also found that CRF could signi cantly reduce the CO 2 and N 2 O emissions in wheat or rice cropping systems compared with urea (Chi et al., 2020;Ji et al., 2013). The ndings of this study show that the slow release of fertiliser reduces GHG emissions and maintains a high yield.
Optimizing N fertiliser inputs in crop rotation systems may be an effective strategy for reducing GHG emissions. Liang et al. (2017) found that, in a double-season rice cropping system, the total grain yield increased by 6.7-13.9% and the GHG emissions signi cantly decreased when the N rate was reduced  2015) reported that a 20% reduction in the optimal rates of N fertilisation and changes to the basal/topdressing ratio could increase yields and reduce GHG emissions. Our previous experimental study (Liu et al., 2018) also showed that treatment JCF70 is environmentally friendly in that it signi cantly decreased the N loss into the environment compared with treatment JCF100. However, we were not able to determine a clear relationship between fertilisation and N 2 O emissions due to the lack of different fertilization levels. The present study clearly showed that the CRF treatment could reduce the GHG emissions during the peanut growing seasons (Table 2). This is consistent with the ndings of Chi et al. (2020) and Shi et al. (2013), who reported a decrease in CO 2 and N 2 O emissions, respectively, in wheat-maize rotations. They found that reducing the N application rate and using controlled-release N fertiliser is a promising management practice to achieve the two objectives of sustaining grain yield and reducing GHG emissions in the NCP. Mainly because there is a rapid rate of hydrolysis and nitri cation after applying N fertiliser in the form of urea, especially for excessive N fertiliser applied in a single application, and almost all peanut and corn growth occurs from June to September, during which period the frequent precipitation and high temperature accelerate the denitri cation process.

Conclusion
In wheat-peanut rotation systems, there is a prominent seasonal variation in soil GHG uxes, and the cumulative GHG emissions are signi cantly affected by the fertilisation treatment, fertiliser type, and soil temperature. In this study, three splits of N application were found to signi cantly increase the total grain yields, however they also increased GHG emissions. Under the same fertiliser rate, in the CRF treatment, the GWP was signi cantly lower and the total grain yield was signi cantly higher compared to the CCF treatment. Thus, the CRF treatment was found to have lower GHG emissions than the CCF treatment. Therefore, to simultaneously maintain crop yield and reduce GHG emissions, three splits of N application and the use of CRF are recommended. Taking economic bene ts into consideration, more study is necessary to elucidate the impacts of reducing the total application rate of CRF or the combined application of urea with CRF on crop yield and GHG emissions due to the high cost of CRF.

Declarations
Author contribution XL and ZL initiated and designed the research. ZL analyzed the data, and wrote the manuscript. CZ, HL and JZ do the experiment, sample plant, XL revised and edited the manuscript. Data availability The data used to support the ndings of this study are available from the corresponding author upon request.
Ethics approval Not applicable.
Consent to participate Not applicable.

Consent for publication Not applicable.
Competing interests The authors declare no competing interests. Daily mean temperature, precipitation, Soil temperature, and WFPS during the wheat and peanut growing seasons in the eld experiment.