Effects of different management patterns on greenhouse gas emissions from single-season rice fields**

. Different cultivation practices affect both rice yield and GHG emissions. In this study, GHG emissions, environmental factors, and soil factors on an annual scale for four common single-season rice-cultivation methods were monitored and the direct and indirect drivers of GHG emissions were analysed. The results showed that there were significant differences in the aver - age soil temperature, water content and dissolved oxygen content for the different methods used, but not in terms of the NH 4+ -N and NO 3– -N content. The highest average methane flux was obtained using the waterlogging in the non-rice season with straw return mode method (4.20±0.16 mg m −2 h −1 ), which was a significantly higher result than that produced by the other methods. The main factors influencing CH 4 emissions was atmospheric temperature for waterlogging in the non-rice season with straw return and crayfish farming, NO 3 − -N for waterlogging in the non-rice season without straw return, NO 3 − -N, and NH 4+ -N for waterlogging in the non-rice season with straw return, and NO 3 − -N, NH 4+ -N, and soil water for non-rice season with straw return. The average fluxes of both CO 2 and N 2 O were highest with the drainage in the non-rice season with straw return treatment, but the differences between the treatments were not significant. None of the factors that were determined had a significant direct effect on CO 2 emissions under either cropping system. The main factors affecting N 2 O emission were NH 4+ -N and NO 3 − -N in the WSC treatment, NH 4+ -N in the waterlogging in the non-rice season without straw return treat-ment, and NH 4+ -N, NO 3 − -N, and dissolved oxygen content in the drainage in the non-rice season with straw return treatment. The results obtained have the potential to form an important basis for the establishment of agronomic measures to reduce and control GHG emissions from rice fields.


INTRODUCTION
The greenhouse effect is considered to be the main cause of global warming.Nitrous oxide (N 2 O) and methane (CH 4 ) are the main greenhouse gases, which are 265 times and 28 times more potent than CO 2 in terms of their warming potential in a 100-year period, respectively (IPCC, 2014).The process of N 2 O production in the soil is determined by temperature, moisture, pH, nitrogen availability and other factors (Brown et al., 2011), and that of CH 4 production is mainly determined by oxygen content but also by soil temperature, pH, moisture and salinity (Serrano-Silva et al., 2014).N 2 O and CH 4 formation is also clearly affected by denitrification (or nitrification) and a methane-producing substrate, such as nitrate, ammonium, and acetate (Gao et al., 2012).As multiple environmental factors affect both N 2 O and CH 4 emissions, changes in either environmental factor can affect the production and emission rates of these gases (Mer and Roger, 2001).It may be observed that soil environmental factors have an impact on greenhouse gas emissions alone or through interaction, and that the impact of a single factor is limited by other factors.But no consistent conclusion may be arrived at concerning the direct or indirect impact of a certain factor on GHG emissions.
Rice fields are an important source of CH 4 and N 2 O (Ma et al., 2007), in which methane emissions account for 5~19% of the total global emissions (IPCC, 2007).The amount of greenhouse gas emissions from paddy fields is largely dependent on agricultural management measures.In order to improve the rice yield and maintain soil fertility, the application of chemical fertilizer and crop straw during rice cultivation is a conventional management measure, but these additions significantly increase the CH 4 and N 2 O emissions from paddy fields, and the extent of the increase mainly depends on the water condition of the paddy fields (Lee et al., 2010;Li et al., 2014).
Integrated paddy farming is an efficient form of paddy ecosystem cultivation (Ruddle, 1982).At present, the global area of rice-fish co-cultivation is approximately 2.54 million ha, among which the rice-fish model has the largest area and the rice-crayfish model produces the greatest economic benefits.To date, research concerning greenhouse gas emissions has focused mainly on the rice-fish pattern, and few studies have addressed the rice-crayfish system.It is interesting to note that both positive and negative effects on paddy field CH 4 emissions have both been reported in integrated paddy farming.Thus, Datta et al. (2009), Frei et al. (2007a, b), and Bhattacharyya et al. (2013) believe that rice-fish farming significantly increases CH 4 emissions.But Yuan et al. (2009), Liu et al. (2006), andZhan et al. (2008) found that it reduces CH 4 emissions.With reference to the rice-crayfish system, Xu et al. (2017) and Sun et al. (2019) found that it is capable of reducing CH 4 emissions.It may be observed that the GHG emission patterns are varied in terms of the different integrated paddy farming patterns (Ling et al., 2021).
The area under rice cultivation in China is > 3×10 7 ha all year round (CAYEC, 2017), and the planting area of a single-season of rice in the southern rice-producing areas has risen to include approximately two-thirds of the total rice cultivation area (Liu et al., 2013).Therefore, in this study, the different soil and environmental factors and the characteristics of greenhouse gas emissions in singlecropping rice fields under four management modes were simultaneously studied.In addition, the direct, indirect, and comprehensive effects of these factors on greenhouse gas emissions were assessed.The results provide basic data for the accurate evaluation of GHG emissions from rice fields in China.At the same time, they provide a guiding significance for clarifying the main determining factors of GHG emissions in various circumstances of rice paddy cultivation and for formulating emission reduction measures for different rice-cropping patterns.

MATERIALS AND METHODS
The experiment was carried out in Houhu Town, Qianjiang City, Hubei Province, China, which has a north subtropical monsoon-influenced humid climate, an annual average temperature of 16.1℃, a frost-free period of 246 days, and an annual average rainfall of 1 100 mm.The experimental site was in the low lake area of Jianghan Plain, with a static groundwater level of 40-60 cm in winter.The soil is classified as a tide rice soil which developed from lake sediment.At the time of our experiment the physical and chemical properties of the top soil (0-20 cm) were 26.43 g kg −1 organic matter, 0.21% total nitrogen, 129.50 mg kg −1 alkaline nitrogen, 9.13 mg kg −1 fast-acting phosphorus, 178.67 mg kg −1 fast-acting potassium, and the soil had a pH of 7.12.
The experiment was conducted using four common management practices in one-season rice fields in the Hubei Province: (i) waterlogging in the non-rice season with straw return and crayfish farming (WSC), (ii) drainage in the nonrice season with straw return (DS), (iii) waterlogging in the non-rice season without straw return (W), and (iv) waterlogging in the non-rice season with straw return (WS).The experiment commenced in 2014, and GHG emission monitoring was conducted from 2015-2018.The rice variety planted was the single-season variety "Jianzhen2".
Flooding treatments: WSC, W, WS were conducted in the first year after the rice harvest.The field was flooded, and the water level gradually rose from approximately 20 cm to 50-55 cm.The fields were drained one week before transplanting (end of May) and dried for 5-7 days.Each treatment was repeated on three plots, and each plot had an area of 100 m².A 60 × 40 cm (width × height) ridge was created around the plots and wrapped with mulch to prevent any straining of water and crayfish.A crayfish trench (3.0-4.0 × 0.8-1.0m (width × depth)) was dug on one side of the WSC plot, and an anti-escape net was set around the plots.Seed crayfish were released in October 2014 at a rate of 200 kg h m −1 (about 40 individuals kg −1 ), and no additional crayfish were added in subsequent experiments, the experiment relied on the natural reproductive capacity of the initial crayfish.Crayfish feed was only provided from March to May each year, with 46.6, 11.0, and 10.5 g kg −1 of nitrogen, phosphorus, and potassium, respectively, as the main components.The crayfish were harvested twice a year.The fertilization and field management measures for the different cultivation models are shown in Table 1.
Soil temperature data were recorded with a data logger (HOBO UTBI-001, Onset Computer Corporation, Bourne, MA, USA).In order to avoid any effects arising from the frequent collection of soil samples, the soil was collected from five sampling sites in each paddy field at a frequency of one soil sample for every two gas samples.The soil samples were collected at a depth of 0-20 cm and immediately preserved at 4℃.The soil water (SW) content was determined using the drying and weighing method (Bao, 2000).The concentrations of NH 4 + -N and NO 3 − -N were analysed using an AA3 Auto Analyser (Bran+Luebbe GmbH, Norderstedt, Germany) according to classical colorimetric methods.The concentration of DOC was analysed using a TOC Analyser (vario TOC, Elementar Analysen systeme GmbH, Langenselbold, Germany) according to standard procedures.
The gas samples were analysed using stainless steel static-box sampling and indoor gas chromatography (Wu et al., 2018).The static chamber was divided into three parts: base, middle chamber, and top chamber, with a small fan placed at the top, a small sampling hole was situated in the top panel of the top chamber, and a thermal insulation layer was wrapped around the chamber.Samples were taken every 7-10 days during the non-rice season, every 5-7 days during the rice-cultivation period, and twice a week after agricultural operations such as fertilization or drainage, the sampling time was 8:00-10:00 a.m.daily.
In the idle period and the early stages of rice growth (plant height > 50 cm), only the top chamber was used for collecting gas samples, and the middle chamber was added when the rice-plant height exceeded 50 cm.One gas sample was collected with a 30 mL medical syringe at 0, 5, 10, 15, and 20 min after the cover was removed.The samples were taken to the laboratory for analysis within 24 h.In order to avoid the frequent disturbance of the soil, the sampling base was fixed for the entire sampling period (except during field preparation).A probe-type electronic thermometer was used for detection and manual recording.
The gas flux was calculated as: where: F is the gas flux, i.e., F CH 4 (for CH 4 , mg m −2 h −1 ), F CO2 (for CO 2 , mg m −2 h −1 ), and F N2O (for N 2 O, mg m −2 h −1 ), ρ is the density of the gas under standard conditions (kg m −3 ), V is the effective volume of the confinement box (m 3 ), S is the base area (m 2 ), dC/dt represents the change in gas concentration in the confinement box per unit time, and T is the average temperature in the confinement box.
All data were processed using Excel software, and statistical analyses were performed using SPSS software (SPSS 20.0, SPSS Inc., Chicago, IL, USA) and R. The significance level for the statistical analysis was set at p < 0.05.The plots were created using Sigma plot, R, and AI software.
The average SW content (p = 0.072), the average temperature (p = 0.027), and the DOC content (p = 0.063) showed significant differences among the four modes, but the NO 3 --N concentration (p = 0.3) and the NH 4 + -N concentration (p = 0.15) did not (Table 2).
The CH 4 fluxes showed similar trends with different treatments (Fig. 1a).The CH 4 emissions increased with increasing temperature before transplanting, they peaked at the end of July during rice cultivation, and then decreased, with a second peak emission period in September and then a decrease until the rice harvest and idle period.Emissions were lower during the flooding period from November of the first year until March of the second year after that they continued to increase with increasing temperatures.The average gaseous CH 4 fluxes at different stages were highest with treatment WS, which were significantly higher than those for the other treatments during the flooding stage (stage A) and the rice-planting stage (stage B).In the idle stage (stage C), CH 4 flux was highest with treatment WSC, which was significantly higher than that under treatments W and DS.The mean flux throughout the monitoring cycle was highest with treatment WS, which was significantly higher than that with WSC (p < 0.0005), W (p < 0.0005), or DS (p < 0.0005) (Table 3, Fig. 2b).The flux dynamics of the CO 2 emissions were consistent among the various treatments (Fig. 1c).Emissions were very low in the flooding stage and increased gradually after transplanting, reaching a peak at the end of July, and then decreased until the rice harvest and idle period.The average CO 2 flux did not differ among treatments at the flooding stage (stage A), but was highest in treatment W and lowest in WS.At the rice-planting stage (stage B), the average CO 2 flux was lowest in treatment W, which was significantly lower than that in WS and DS, but not significantly different from that in the WSC treatment.At the idle stage (stage C), emissions were highest with treatment WSC, but did not differ significantly between treatments.Throughout the monitoring period, the average CO 2 fluxes were higher for the WSC and DS treatments and lowest in the WS treatment, but the differences were not significant (Table 3, Fig. 2a).
The dynamics of the N 2 O emission fluxes were complex (Fig. 2b).N 2 O flux was low in the non-rice flooding phase which lasted from February to the end of May in 2016, but varied more from November to the end of May in both 2016-2017 and 2017-2018.While the degree of variation was high in 2015 during rice cultivation, it was low  -N had a significant direct positive effect on CH 4 flux, with a path coefficient of 0.899 (p < 0.001), and also SW had a significant direct negative effect on CH 4 flux, with a path coefficient of -0.458 (p < 0.001).
In treatment WS (straw return and flooding), the whole model explained 49.2% of the variance in CH 4 flux, and also NO 3 − -N and AT had significant direct positive effects, with path coefficients of 0.649 (p < 0.001) and 0.261 (p < 0.05).The whole model explained 19.0% of the variation in CH 4 flux in treatment WSC, and also AT had a significant direct positive effect, with a path coefficient of 0.364 (p < 0.05).Overall, SW had a positive effect on DS, which became negative when the field was flooded in the non-rice season.The effect of AT on the CH 4 emissions flux gradually increased after flooding as elements such as straw return and crayfish farming influenced the system, while the effects of SW and NO 3 − -N on CH 4 gradually decreased.These results showed that the drivers of CH 4 fluxes differed among the cultivation systems tested.The multiple stepwise regression and quadratic regression results (Table 4) showed that the R 2 of all of the treatments decreased after the optimized regression, which indicates that the explanatory power became lower as the number of factors decreased, while the significance increased.For the optimized model, the dominant factors were AT for WSC,  The model explained 24.1 and 14.7% of the variation in CO 2 emission fluxes for DS and W (Fig. 4).No factors had a significant direct effect under either cropping system.The whole model explained 17.8 and 22.7% of the variance for WS and WSC, respectively, and NO 3 − -N had a significant direct negative effect on CO 2 emission fluxes, with path coefficients of −0.327 (p < 0.05) and −0.402 (p < 0.05), respectively.There was no significant direct effect due to other factors.Overall, temperature had a nega-tive effect in treatment DS, but when flooding occurred in the non-rice season (W, WS, WSC), temperature had a positive effect on the CO 2 flux.As elements such as straw and cultured crayfish gradually entered the system, the effects of NO 3 − -N and AT on the CO 2 emission fluxes gradually increased, while the effects of NH 4 + -N on CO 2 production gradually decreased.
After all of the factors were taken into consideration, the model only had a limited explanatory power.After excluding the less influential factors using a multiple stepwise regression method, the results showed that CO 2 emission fluxes were mainly influenced by AT, DOC, and NO  5).The stepwise regression models for W and DS were not significant, thereby indicating that important factors were omitted in these two systems.
The  After excluding the less influential factors using multiple stepwise regression, the results (Table 6) showed that the R 2 value decreased for all treatments, and the abil-

DISCUSSION
Flooding and straw return, alone or in combination, can significantly increase the average CH 4 emissions flux in rice or non-rice seasons by creating an anaerobic environment and providing a substrate (Conrad, 2007;Jiang et al., 2019;Li et al., 2007;Xu et al., 2017;Zhang et al., 2009).The results showed that the average flux of CH 4 was highest in the WS treatment with both straw return and flooding measures, but applying the WS treatment together with crayfish farming (WSC) can reduce CH 4 emissions, which is consistent with previous studies (Li et al., 2007).Although the large amount of rice straw entering the system as feed increases carbon input, the digging behaviour of crayfish greatly increases the water-soil contact area, and the diurnal foraging behaviour of crayfish and nighttime fishing behaviour of farmers may increase the level of dissolved oxygen in the water column.These factors could lead to a change in the carbon-to-nitrogen ratio, and also to an increase in the overall level of dissolved oxygen due to an expansion of the soil and water interface, and to an increase in the redox potential leading to the oxidation of CH 4 , resulting in lower CH 4 emissions (Abutoama and Abdulhalim, 2017;Bodelier, 2011).
The effect of water management on the CO 2 emissions from farm systems is complex.The respiration of paddy systems depends mainly on soil microorganisms (Raich and Potter, 1995), and moisture severely affects microbial respiration (Keith et al., 1997;Orchard and Cook, 1983;Turcu et al., 2005).For example, soil CO 2 emissions from flooded farmland are much lower than those from alternating wet and dry conditions (Kuzyakov and Siniakina, 2001), and paddy drying can promote soil CO 2 emissions (Nishimura et al., 2008;Palmer et al., 2014).In addition, water-layer thickness also affects CO 2 transport and emissions (Wu et al., 2009).Returning rice straw to paddy fields is common, and studies have shown that returning straw during non-rice-season drainage enhances soil microbial populations and stimulates soil CO 2 emissions in rice fields (Yin et al., 2008).Returning straw to rice fields under continuous flooding throughout the rice season increases CO 2 emissions (Shen et al., 2014).
In this study, we found that straw return did not increase CO 2 emissions under flooding conditions in the non-rice season while the rice season was managed normally, this suggests that the synergistic scientific management of paddy moisture and straw can effectively control CO 2 emissions.The direct positive effects of NH 4 + -N and NO 3 − -N on CO 2 fluxes were significantly reduced or changed to negative effects, and the direct positive effect of AT was strengthened by flooding in the non-rice season under straw return conditions.The relationship between the effects of SW and AT on NO 3 − -N changed (from positive to negative or from negative to positive), and these shifts in mutual effects may influence CO 2 emissions.
Crayfish activity in rice-crayfish systems disturb the soil much more than in rice-duck or rice-fish systems (Brown et al., 2004;Fanjul et al., 2011;Musgrove and Geddes, 1995;Sarr et al., 2001;Stieglitz et al., 2000;Wang et al., 2010), this results in a substantial increase in the aerobic and anaerobic soil interface (Fanjul et al., 2007), which could theoretically lead to a decrease in CH 4 emissions and an increase in CO 2 emissions.Our study confirmed that crayfish farming with flooding and straw return increases CO 2 emissions, but not significantly.The direct effects of AT, DOC, and NO 3 − -N on CO 2 fluxes were significantly enhanced by crayfish farming, most likely due to the increase in DOC with increases in temperature and crayfish activity, as well as to the interrelationship among these three factors.In this study, the DS fields were drained in the non-rice season, i.e., the soil surface became more oxidized, whereas other treatments were flooded in the non-rice season (stage C took place in the non-rice season but remained for a short time and could be classified as the rice season).In the rice season, all treatments had the same water management regime.A comparison of DS and WS showed that the overall N 2 O emissions were slightly reduced by flooding after straw return, mainly in stage A. WS fields were always anaerobic, and denitrification caused a further reduction of N 2 O to N 2 as an electron acceptor (Chapuis-Lardy et al., 2007;Ryden, 1983), resulting in lower N 2 O emissions.The DS fields were aerobic for the most part with occasional anaerobic conditions and alternated between nitrification and denitrification, resulting in higher N 2 O emissions.The NO 3 − -N content was the main factor influencing N 2 O emissions from DS fields.Many studies have shown that N 2 O emissions in paddy farming systems are regulated by the Eh value which is related to the amount of dissolved oxygen in a system (Bhattacharyya et al., 2013;Li et al., 2007;Liu et al., 2016;Yuan et al., 2009;Zhan et al., 2008), mainly because of the influence of redox conditions on nitrification and denitrification.In this study, N 2 O emissions were lower in the WSC treatment, mainly because the NO 3 − -N content had a significant negative effect on N 2 O emissions, and the NO 3 − -N content was significantly higher in the WSC treatment than that in the WS treatment (Table 4).In crayfish farming, aquatic grass planting and crayfish hole-digging behaviour may lead to a significant expansion of the watersoil interface and the total aerobic surface area (Kristensen et al., 2012).In this case, NO 3 − -N is not completely reduced but rather, NO 2 − may be accumulated which is toxic to microorganisms and may reduce the production efficiency of microbial N 2 O.The application of quicklime for disinfection at this stage can increase the pH value, which is significantly and negatively correlated with N 2 O emissions (Datta et al., 2009).This result explains, to some extent, the reduction in N 2 O emissions in the WSC treatment.CONCLUSIONS 1.Both straw return and flooding significantly increased CH 4 emissions, but crayfish farming suppressed the increase in CH 4 emissions to a significant extent due to straw return.
2. Inorganic nitrogen (mainly NO 3 − -N) in the three treatments of (waterlogging in the non-rice season without straw return, waterlogging in the non-rice season with straw return mode, drainage in the non-rice season with straw return) without crayfish significantly affected CH 4 emissions, while temperature was the main factor affecting CH 4 emissions in the crayfish culture system.
3. The highest total CO 2 emissions were found in cultured crayfish (waterlogging in the non-rice season with straw return and crayfish farming) and non-rice-season drainage (drainage in the non-rice season with straw return) with straw return.The lowest total CO 2 emissions were found in non-rice-season flooding (waterlogging in the non-rice season without straw return) without straw return.Both temperature and NO 3 − -N had a negative effect on CO 2 emission fluxes in all non-rice-season flooding systems (waterlogging in the non-rice season without straw return, waterlogging in the non-rice season with straw return mode, drainage in the non-rice season with straw return) and a positive effect in drainage in the non-rice season with straw return.
4. The highest N 2 O emissions were found in drainage in the non-rice season with straw return and the lowest in waterlogging in the non-rice season without straw return.Straw return under flooded conditions promoted N 2 O emissions, and crayfish farming (waterlogging in the non-rice season with straw return and crayfish farming) significantly reduced them.Soil NO 3 − -N content had a significant positive effect on N 2 O emissions in drainage in the non-rice season with straw return and a significant negative effect in waterlogging in the non-rice season with straw return mode and waterlogging in the non-rice season with straw return and crayfish farming, the increase in NO 3 − -N content may have been the main factor regulating N 2 O emissions in waterlogging in the non-rice season with straw return and crayfish farming.
Many factors and agronomic measures affect CH 4 and N 2 O emissions in paddy fields, but there are still certain patterns among similar types of agronomic parameters.Although these findings cannot be generalized to include all soils (and zones) as variations in soil type and temperature will produce some different effects, this study provides a basis for the selection of agronomic measures for the targeted reduction and control of GHG emissions for rice cultivation.

Fig. 1 .
Fig. 1.Greenhouse gas emissions flux during the observation period: a -annual dynamics of the CH 4 flux, b -annual dynamics of the N 2 O flux, c -annual dynamics of the CO 2 flux.

Fig. 3 .
Fig. 3. SEMs of soil factors and environmental factors as predictors of CH 4 flux.
and SW for DS, which is consistent with the structural equation model results.

Fig. 4 .
Fig. 4. SEMs of soil factors and environmental factors as predictors of CO 2 flux.

−
effects of environmental factors on N 2 O emission fluxes are shown in Fig. 5.The results showed that the model explained 47.8% of the variance for the DS treatment, where NO 3 − -N and AT had significant direct effects on N 2 O emission fluxes, with path coefficients of 0.612 (p < 0.01) and -0.452 (p < 0.05), respectively, while the other factors had no significant effects.The model explained 45.9% of the variation for treatment W, with significant effects occurring on N 2 O flux due to NO 3 with path coefficients of -0.480 (p < 0.01) and −0.360 (p < 0.05), respectively, and no significant effects observed for other factors.The model explained 17.5% of the variation in the N 2 O flux after straw was returned to the field (WS).NH 4 + -N had a significant direct negative effect on the N 2 O flux, with a path coefficient of -0.355 (p < 0.05), while other factors had no significant effect.The model explained Ta b l e 5. Relationships between soil CO 2 fluxes and environmental factors during the observation period determined by multiple linear regression analysis and optimization of regression

Fig. 5 .
Fig. 5. SEMs of soil factors and environment factors as predictors of N 2 O flux.
ity of the model to explain the variance in the N 2 O flux decreased, but the overall significance of the retained factors increased.Stepwise regressions also showed that the main influences on the N 2 O flux were NH 4 and DOC in the DS treatment.
Ta b l e 1. Rice cultivation and fertilization management practices (test period: June 2015-May 2018) Ta b l e 3. Mean values of the greenhouse gas emissions flux A, B, and C represent the different stages of flooding, rice planting, and post-harvest to pre-flood.Other explanations as in Table2.
Ta b l e 6. Relationships between soil N 2 O fluxes and environmental factors during the observation period were determined by multiple linear regression analyses and the optimization of regression