2.3.1. Field measurements

Gas samples were collected to evaluate CH₄ and N₂O for daily and seasonal variations using a closed chamber technique. Following the recommendations provided by Minamikawa et al. (2015), rectangular chambers were employed so that the area it covers is equal to that occupied by ten rice plants at the Sri Lankan paddy site. To maintain roughly the same head space between the plant and the top of the chamber and prevent the suppression of rice development, chambers 35 cm wide and 75 cm length were constructed using acrylic and transparent polythene in three different height stages (250 cm, 500 cm and 950 cm). A small, battery-powered fan was fixed inside the chamber’s top to mix the inside gases and maintain a constant target gas concentration (Prinn et al., 1994). Additionally, the chamber had a separate gas sampling port, vent hole, and rubber stopper to avoid any drastic inside pressure changes that the chamber deployment could cause. Throughout the entire paddy season, chamber bases were firmly pressed to the soil at each experimental plot at a depth of around 5 to 10 cm, and gas samples were collected by placing the chamber on the base and ensuring a gas-tight closure using a water seal. In temperate regions of Asia, Minamikawa et al. (2012) suggested that daily mean CH₄ flow measurements be carried out once per day during mid-morning, roughly 10:00 (09:00–11:00) since these measurements resulted in acceptable estimates (i.e., 10%). Emissions during this period are assumed to be typical of the daily average emission rate, which is then used to calculate the total seasonal emissions (Cai et al., 1997; Zou et al., 2005). In Sri Lanka, gas samples were collected during 9–11 am local time at 0, 15, and 30 min intervals with a 60 ml syringe fitted with a stopcock valve and stored in 6 ml evacuated glass vials.

Chambers of similar dimensions were used in the Indian experimental sites. Gas samples were drawn from the chambers in the same manner as described above, using a 50 ml syringe. Gas samples were transferred to vacuum-sealed 135 ml vials with rubber stoppers, and analyses were completed within three days of sampling.

In both Sri Lankan and Indian experimental sites, weekly gas sampling was done, along with additional measurements regarding the N fertiliser event and the transition of the AWD season. Water depth inside the base and base height above the soil surface were measured on each gas sampling day. Piezometers installed at each plot were used to guarantee the water depth using the AWD approach. Soil redox potential was recorded in Sri Lanka and India using a portable battery-operated Eh meter by inserting the electrode into the soil under investigation to a root-zone depth of 5 cm. Grain yield was measured at the harvest stage at each plot under both treatments. The concentration measurements of CH₄ and N₂O were completed using a gas chromatograph equipped with a thermal conductivity detector (TCD) using helium as the carrier gas. The Indian team calculated fluxes as per the standardised protocols (Goldenfum, 2009; Venterea, 2010). Further, Ly et al. (2013) noted that cumulative emissions were computed by integrating the area under the curve of adjacent measurement points.

In the Japanese experimental site, seasonal measurements were conducted weekly for seasonal variation and a campaign was conducted for diurnal variation only for CH₄. Measurements were taken four times each during the panicle heading and formation stages and seven times during the ripening stage. Measurements of CH₄ emissions were performed during the hours of 7:00 am to 11:00 am local time. Chambers (30 cm × 60 cm × 60 cm) were placed on each plot and an additional acrylic frame with 60 cm height was placed below the chamber to extend the chamber height after the middle of July to accommodate the growth of rice plants.

The CH₄ concentration inside the chamber was continuously measured by using a high-precision portable gas analyser (G4301, Picarro Inc, Santa Clara, CA, USA) at intervals of 0.9 s. When ebullition did not occur to get a steady linear increase in CH₄ concentration in the chamber, the monitoring period was around 6 min, resulting in a continuous linear increase in the chamber’s CH₄ content. The monitoring period was approximately 10–15 min when frequent ebullitions were noticed. Sensors (TEROS-12, METER) were inserted at 5 cm and 10 cm soil depth to measure soil temperature and volumetric water content where ±3 cm of the inserted depths were contained in the measured region. All data were recorded using a logger (ZL6 Basic, METER) at 1 h intervals. After monitoring, intact soil core samples were extracted from each plot at 5 and 10 cm depths. Sensor calibration for volumetric water content was done using the volumetric water content determined by oven-drying the intact soil cores. Ebullition and CH₄ fluxes through the rice plants were separately computed in accordance with Kajiura and Tokida (2021). In a kernel density distribution of the flux, the lowest flux showing local maxima was considered as the diffusive CH₄ flux (i.e., plant-mediated flux and molecular diffusion of dissolved CH₄ across the water–atmosphere interface). The difference between the total and plant-mediated CH₄ fluxes was then used to compute the ebullition CH₄ flux. It should be emphasised that due to the poor solubility of the CH₄, the diffusive flow of dissolved CH₄ was neglected.

2.3.2. Statistical analysis

All statistical analyses were performed using Minitab®; 17.1.0. One-way analysis of variance (ANOVA) and paired t-test were used to analyse statistical significance in seasonal emissions and grain yield, respectively. To test differences among treatments, Tukey’s HSD (Honest Significant Difference) test was performed with a significance level of 0.05.

3.1.1. Seasonal emissions in Sri Lanka

The seasonal CH₄ and N₂O emissions from the Bathelegoda, Sri Lanka experimental site under both CF and AWD, with and without plants, are shown in Figure 4. Note that they are shown for both the wet and dry seasons.

Figures 4a and 5a indicate the seasonal dynamics of CH₄, demonstrating the complex interplay between methane generation, oxidation and transportation. Due to the constant anaerobic conditions present in CF/P as opposed to the intermittent aeration in AWD/P, emissions rates of CF/P were higher than those of AWD/P, as was expected. The large error bars show the spatial variability in the fluxes. Throughout the whole season, as discussed by Pandey et al. (2014) and Oo et al. (2018), three peaks were noticed under with-plant (P) treatments (Figure 4a). Generally, high CH₄ flux is caused by the favourable oxygen status and C availability in the paddy soil due to the release of more plant-borne carbon sources, such as root exudation continuing at high levels or decaying roots. Soil-redox (Eh) ranged from −100 to −200 mV during the CF in WS, as also evidenced by Yagi and Minami (1990) and Wang et al. (2017), who mentioned that the production of CH₄ starts at a redox potential of −150 to −160 mV. Lower CH₄ emissions under drainage have been consistently reported in many past studies, indicating a negative correlation between CH₄ emission and soil Eh (Corton et al., 2000; Wassmann & Aulakh, 2000). In this study, a reduction in CH₄ emission was observed under enhanced redox conditions in AWD, which ranged between (−125) mV – (+510) mV. Decrement of CH₄ emissions from AWD/P was caused by less developed aerenchyma of rice plants that have previously undergone AWD treatment, as compared to those under anaerobic conditions permanently.

With the transition from aerobic to anaerobic conditions due to the AWD episode ceasing, a rapid increment of CH₄ emission was detected under the AWD/P combination. Generally, in paddy systems, transportation and atmospheric emission of methane occurs mainly via three channels: diffusion of dissolved gas through water-air and soil-water interfaces, ebullition as gas bubbles, and plant transport (i.e., by diffusion and conversion to gaseous CH₄ in the aerenchyma and cortex) and subsequent release to the atmosphere (Davamani et al., 2020; Le Mer & Roger, 2001). Literature studies have concluded that more than 90% of CH₄ is emitted through rice plants under normal CF conditions (Xie & Li, 2002). In comparison, 70.3% of emissions occurred through the plants during the wet season, while 29.6% occurred through the soil surface (Figure 1a), suggesting that soil-dominant diffusive emissions are instrumental in alternative wetting and drying periods. Both CF/NP and AWD/NP treatments showed very low emissions, around 15%, in comparison to CF/P and AWD/P (see Table 2), thus corroborating the literature observations. This is primarily due to the fact that methane tends to oxidise in the suboxic zone (i.e., at the interface between oxic and anoxic regimes) where the concentration gradients of methane and oxygen essentially overlap, for instance at the surface of flooded rice soils and in the rhizosphere of rice plants. Therefore, as reported in many studies (Conrad & Rothfuss, 1991; Inubushi et al., 1992), 65–80% of generated methane becomes oxidised en route in NP plots.

In contrast to the wet season, CH₄ flux shows high emissions at the beginning of the dry season, as indicated by Figure 5a, and gradually declined to approximately zero towards the end due to the limited presence of soil organic matter with the reduction of carbon supply because of the ageing of plants. Low Eh values in the field, ranging from −150 mV to −200 mV, have enhanced CH₄ formation by methanogenic bacteria under flooded conditions while at −75 mV to +550 mV under drained conditions. Contrary to Li et al. (2009a,b), CH₄ emissions under AWD treatments were not observed to be zero due to the emission of trapped methane in soil pores (Bubier & Moore, 1994).

Recognisable N₂O emissions were not detected during the wet season (Figure 4b) and the rice growing period fluctuated near zero. High emissions were detected for all four treatments at the start of the wet season as a result of the unexpected drying situation at the field (water level = −15 cm below the surface).

Seasonal N₂O emission increased in the dry season compared with the wet season in numerals (Figure 5b) and is strongly influenced by water regime and fertilisation (Aung et al., 2018). However, no significant difference (p > 0.05) in cumulative emission for the dry season was observed among the treatments. Besides, N₂O emission during the dry season (Figure 5b) varies throughout the season in such a way that shows a gradual increment, then peaking at the reproductive phase and followed by a gradual decrease to zero emission. However, a considerable increment of N₂O emission could be observed after first flood irrigation and puddling with the application of basal fertiliser to increase soil biological fertility and concentration of mineral elements and lack nutrients prior to planting.

After the first peak, emissions fluctuated near zero and the second fertilisation was done at 17 DAT under flooded conditions. A temporal peak of N₂O emission was observed at 20 DAT, although it lasted only a few days. It is worth noting that although nitrification and denitrification processes are influenced by soil moisture (Davidson et al., 1986), N₂O production is suppressed under very high moisture content in soil (Tirol-Padre et al., 2017). In this study, continuously irrigated rice paddies have shown quite small N₂O emissions than that of AWD treatment under proper application of N fertiliser, as mentioned by Furukawa et al. (2007) and Tirol-Padre et al. (2017). Brentrup et al. (2000) have documented that N₂O emission increases under moisture alternations such as successive moist and dry episodes. In agreement with the literature, during the alternative wetting and drying period, N₂O fluxes peaked when the water table dropped below the soil surface under high soil redox conditions and during the transition from aerobic to anaerobic just after fertilisation. Hence, throughout the dry season, peaks from AWD/P treatment have increased, as reported in previous studies (Hou et al., 2012; Tirol-Padre et al., 2017).

Statistical details for both wet and dry seasons for the Sri Lankan site are mentioned in Table 2.

3.1.2. Emission dynamics in Indian paddy plots

CH₄ and N₂O emissions measurements in Indian paddy sites for the wet season are shown in Figures 6 and 7, respectively.

Regardless of treatments, the methane emissions observed at 30 DAT peak can be attributed to the interaction between microbial biomass dynamics and organic matter decomposition under shifting soil redox conditions (ranging between −153 to −159 mV). Mid-season drainage substantially reduced methane emissions, while CF without plant plots emitted high methane emissions due to the organic matter accumulation and absence of plant-mediated processes such as oxygen transportation, allowing methanogenic microorganisms to dominate and increase methane production (Conrad, 2007; Khalil & Rasmussen, 1992).

Although measurements of N₂O emissions yielded negative results under CF plots, with the onset of the AWD period, emissions peaked ten days after drainage. Notably, plots without plants exhibited higher N₂O emissions than plots with plants. Furthermore, in terms of supplementation, plots enriched with vermicompost and the recommended dose of fertiliser (RDF) experienced elevated N₂O emissions, followed by plots supplemented solely with RDF, and the control plots exhibited the lowest N₂O emissions. A large number of denitrifying bacteria in the earthworm gut and the surrounding vermicompost substrates is generally attributed to the increased N₂O emissions in vermicompost-amended soils (Wang et al., 2014).

Figures 8 and 9 show the CH₄ and N₂O emission variation in the dry season for Indian paddy plots, respectively.

Figures 8a and b, which exhibit CH₄ emissions data in the dry season, showed the trend around 30 DAT; emissions gradually approached zero. Differences between the plots with and without plants were negligible, demonstrating that the introduction of vermicompost did not exhibit any statistically significant impact on methane emissions (Sharma et al., 2016).

The results demonstrated a noticeable increase in N₂O emissions following fertilisers applied in divided dosages. All treatments show a similar pattern in N₂O fluctuations, while the highest N₂O dynamics belongs to the vermicompost amendment plot when compared to synthetic fertiliser and control plots. As mentioned by Sharma et al. (2016), applying vermicompost increases the soil’s physical properties, like soil aggregates and structure, thus enhancing the aeration in the soil profile. Intriguingly, the N₂O emissions rates of the plots without plants were higher than those with plant plots, while control plots without any supplementation showed the least emissions, almost nil (Figure 5). This observation underscores the interplay between nutrient availability, fertilisation strategies, and the subsequent manifestation of N₂O emissions.

Comparison between results from Sri Lankan and Indian paddy sites regarding seasonal emissions under CF and AWD and grain yield are shown in Table 3.

3.1.3. Methane emission dynamics of Japanese rice paddies

The Japanese experimental sites were used in this study to closely monitor methane ebullition under fully flooded site conditions, with and without the effect of straw incorporation. The seasonal and diurnal CH₄ flux for the plots containing plants is depicted in Figure 10.

The plots with straw treatment showed higher fluxes via both plant-mediated and ebullition during the heading stage than without straw (Figure 10a); the decomposition of applied rice straw enhanced CH₄ production during the earlier part of the growth period. During the ripening stage, there were high CH₄ emissions via ebullition (Figures 10a and 10b) and the proportion of CH₄ flux as ebullition to the overall flux was greater than 60%. There are two possible reasons for this behaviour. First, there was an increase in the supply of organic matter because of the plant roots. Second, there was a dysfunction of aerenchyma in the paddy rice, causing a reduction of plant-mediated emissions. Total flux and ebullition tended to rise in the early morning and fall in the late afternoon throughout the diurnal variation (Figure 10c). Within the day, the CH₄ flux through the plant was essentially constant (data not shown). The magnitude of the ebullition CH₄ flux at all growth stages was not explained by any of the factors considered in this study: soil temperature, atmospheric pressure and average volumetric air content, according to the results of the multiple regression analysis. Figure 11 shows the correlation between the plant-mediated flux for the plots with plants and volumetric air content averaged in 0–10 cm depth at heading and panicle formation stages. The average volumetric air content was positively correlated with both with-straw and without-straw plots. This implies that the increase in the contact area between rice roots and CH₄ bubbles in the soil enhanced CH₄ transport through the plant.