Farmers’ Participatory Evaluation of Alternate Wetting and Drying Irrigation Method on the Greenhouse Gas Emission, Water Productivity and Paddy Yield in Bangladesh

In dry season paddy farming, the alternate wetting and drying (AWD) irrigation improves water productivity, paddy production, and has the potential to decrease greenhouse gas (GHG) such as methane (CH4) and nitrous oxide (N2O) emissions when compared to continuous flooding (CF). However, there is a lack of research in Bangladesh on the effects of water management on CH4 and N2O emissions. During November 2017–April 2018, participatory on-farm trials were conducted at Feni and Chattogram districts of Bangladesh. Total 105 farmers comprising 20-hectare of land (62 farmers at Feni and 43 farmers at Chattogram district, each location having 10 hectare of land). We compared irrigation water and cost reductions, paddy yield, and CH4 and N2O emissions from paddy fields irrigated using AWD and CF irrigation methods. The CH4 and N2O emissions were determined using the Cool Farm Beta-3 methodology, and the global warming potential (GWP) was estimated using the Intergovernmental Panel on Climate Change-2014 standard approach. The mean results of randomly selected 30 farmers from two locations (15 of each) showed that AWD remarkably decreased irrigation water consumption by about 24% and increased water productivity by 224%. We estimated 23% savings for irrigation costs in AWD. By this time, AWD improved paddy production by 3% over CF. The AWD irrigation resulted in a 47% reduction in cumulative CH4 emissions having a lower CH4 emission factor (0.74 kg ha day) than CF (1.39 kg ha day ). There was no obvious difference in N2O emission between AWD and CF. When compared to CF, AWD decreased the overall GWP by 27% and lowered the GHG intensity by 42%. The CH4 and N2O emissions did not differ substantially between Feni and Chattogram.

Typically, pump owners offer irrigation water on a contractual basis for the duration of each season, so ensuring the continuity of farmers. He determines the complete irrigation program independently and at a set cost. The owner begins irrigating farmers' fields in his command area on one side and proceeds serially from one plot to the next until he reaches the last plot. Generally, farmers want to save as much water as possible throughout their irrigation shift. Additionally, most farmers feel that maintaining standing water in rice fields at all stages/phases is necessary to guarantee a larger harvest. They are spending around 2500−5000 liters of water to produce one kilogram of rice Bouman, 2009). But scientifically, this quantity of water does not require from a physiological standpoint as continuous standing water is needed only at transplanting, blooming and grain filling stage/phase to avoid water stress (Kürschner et al., 2010).
This usual approach of continuous flooding (CF) leads to substantial surface runoff, flow and infiltration, accounting for around 80% of total water consumption (USDA, 2019). By 2025, Asia's available water supplies per capita are anticipated to decrease by 15-54% from 1990 levels (Subedia and Poudel, 2021). Like other ricegrowing regions on the Asian continent, Bangladesh is already experiencing water constraint, which means farmers need water-saving technology to produce rice with less water . Due to Bangladesh's frequent water shortages, particularly during the dry Boro season, sufficient water to irrigate rice fields is increasingly scarce. Additionally, due to global warming, severe changes in the pattern of precipitation and drought have become more prevalent in recent decades, posing a substantial danger to manage water for rice farming soon (IPCC, 2019). The increased quantity of atmospheric greenhouse gases (GHGs) viz., CH4 and N2O is a significant contributor to global warming and climate change. Rice farming is accounting for about 11% and 6% of global CH4 and N2O emissions, respectively (Ciais et al., 2013;IPCC, 2019) while in Bangladesh it contributes 33% of agricultural GHG emissions (FAO, 2018). According to Wassmann et al. (2019), irrigated rice is the most potential source for emitting 70-80% of global CH4 followed by monsoon rice (15%).
The International Rice Research Institute (IRRI) developed an alternating wetting and drying (AWD) irrigation strategy (Price et al., 2013) to save water and mitigate the emission of CH4 and N2O in rice fields instead of continual flooding (CF). Compared to CF, AWD reduced CH4 and N2O emissions by 45-90% (Linquist et al. 2014), and irrigation water usage by 15-35% without decreasing rice productivity (Siopongco et al., 2013). AWD was also found to lower arsenic deposition in grain by 64% (Linquist et al., 2014). Leaching losses of soil N may also be decreased by reducing the percolation loss of irrigation water in AWD (Peng et al., 2011). AWD may increase the soil P status by increasing the number of aerobic microorganisms  with a greater organic matter content by earthworm activities (Carrijo et al., 2017), hence, resulted in strong root anchoring, nutrient absorption, increased number of productive tillers, and eventually increased grain production (Yang et al., 2007). Apart from the saving of irrigation water by 70% and CH4 emissions by 97%, AWD was rebuked for 33% yields loss while N2O emissions were more than quadrupled (Lagomarsino et al. 2016).
As discussed previously, the disparate effects of AWD on irrigation water use, GHGs emissions, and grain yields underscore the importance of additional research to increase our understanding of the relationships between cultivation practices, local environments, rice growth, and GHGs emissions. This information will be essential in assisting agricultural extension agencies and smallholder farmers for the implementation of AWD. The purpose of this on-farm study was to determine the prospective for AWD to reduce CH4 and N2O emissions and its effect on rice production and irrigation water saving in the farmers' rice fields at Feni and Chattogram districts of Bangladesh.

Experimental site and season
On-farm participatory research trial was conducted at total 20 hectare land, 10 hectares under 62 farmers' paddy fields at Feni (N: 22˚53"38'; E: 91˚32"5') and 43 fields at Chattogram district (N: 23˚32"14'; E: 90˚24"18') of Bangladesh during November 2017-April 2018. The locations have an average climate characteristic with annual mean rainfall of 498 mm, maximum and minimum air temperature of 40 and 24°C, respectively. The paddy soil is classified as clay-loam and loam, respectively. The soil properties are listed in Table 1.

Land preparation and transplanting
A two-wheel tractor was used which included four rotary tillage passes and cross plowing, followed by two days of sun drying, and finally inundation and leveling. The fields were plowed and puddled thoroughly to about 10-cm depth before transplanting. 35 days aged seedlings of BRRI dhan28 were planted at 20 × 20 cm spacing of rice hills to each plot.

Installation of AWD pipes and water flow meter
In each experimental field 15 farmers plots were selected randomly at different distance from water pump.
PVC made AWD pipes were installed 10 days after transplanting @ 10 pipes in each bigha (1335 m 2 ) of land. A farmer was treated as a replication in every location with two treatments such as of AWD and CF (Continuous Flooding). We installed a water flow meter at the front of outlet pipe of irrigation pump ( Fig. 1) to measure the amount of water and time to irrigate the field AWD plots.

Irrigation management
Field plots under AWD were irrigated following the principle of 'safe AWD' (Lampayan et al., 2015), where floodwater depth inside the AWD pipes was monitored every day and plots were re-flooded up to 5 cm when water depth dropped to 15 cm below the soil surface (Fig. 2). From one week before to one week after flowering (52-77 days), the field was kept flooded, topping up to a depth of 5 cm. After flowering, during grain filling and ripening (79-100 days), the water level was dropped again to 15 cm below the soil surface before re-irrigation.
AWD was suspended for 14 days after installation of pipes to assist suppression of the weeds by the ponded water and improve the efficacy of herbicides (pretilachlor). In CF irrigation method, fields were continuously flooded until two weeks before harvesting and field were irrigated regularly as and when needed.

Crop management
Two to three rice seedlings hill -1 of 35 days old of BRRI dhan28 were transplanted at 20 × 20 cm distance.
Fertilizer management were adopted as per government recommendation. Phosphorus (triple super phosphate) and potassium (muriate of potash) were applied during final land preparation at 85 and 25 kg ha -1 , respectively.
Sulfur (gypsum) and zinc (zinc sulphate) were applied to all plots as basal at the rate of 11 and 3 kg ha -1 , respectively. For nitrogen, prilled urea was applied as broadcast in three equal splits at 7-10 DAT, at maximum tillering and panicle initiation stages.

Measurements
Measurements such as crop growth duration (days), number of productive tillers m -2 , number of grains panicle -1 , 1000-grains weight (g), grain and straw yield (t ha -1 ) were collected. Crop was harvested a maturity from the central 2 × 1.5 m area from three spots of each plot. The yield was calculated at 14% moisture content.
The irrigation water savings was determined based on the numbers of required irrigation, and amount of water needed based on the readings of water flow meter. All these measurements were done for both AWD, and CF irrigation method.

Data analysis
Analysis of variance of the water productivity, paddy yield attributes and yield, seasonal cumulative emission of CH4 and N2O gases, GWP, and GHGI was performed with the Statistical Tool for Agricultural Research: STAR 2.0.1 (IRRI, 2014). All pair-wise mean comparison of treatments was done with The Duncans' Multiple Range Test at p ≤ 0.05 level of significance.

Water productivity and irrigation cost
Irrigation method exerted a significant effect (p ≤ 0.05) on the water productivity both at Feni and Chattogram location of the study (Table 2). Data demonstrated that, the frequency of irrigation per hectare of land was approximately 20 times lower in AWD (65 and 56 at Feni and Chattogram, respectively) relative to CF (85 and 73 at two locations, respectively). One hectare of land under AWD method required total 3873 and 3382 m 3 irrigation water at the Feni and Chattogram, respectively. These amounts were 24% less than that of CF method at both the locations (5152 and 4454 m 3 ha -1 , respectively). At Feni, the water productivity (WP) of AWD and CF method were 1.71 and 0.47 kg m -3 , respectively; while at Chattogram the values were 1.50 and 0.52, respectively. The mean values of WP for two locations revealed that, AWD required about 624 L irrigation water (excluding rainfall) to produce 1 kg paddy. On the contrary, CF required 2021 L irrigation water. On average of two locations, 224 % higher WP was estimated in AWD over CF.

Yield attributes and yield of paddy
Rice yield was influenced significantly (p ≤ 0.05) by the irrigation methods at both the locations (Table 3) of the present on-farm study which might have attributed to the significant variation of the number of productive tillers m -2 area. The AWD produced about 24% higher number (1233 at Feni and 1311 at Chattogram) of productive tillers relative to CF (979 and 1045 at two locations, respectively). In the present study, on average of two locations, we found about 3% higher paddy yield in AWD (5.96 and 6.24 t ha -1 ) over the CF (5.78 and 6.06 t ha -1 ). The number of paddy grains panicle -1 and the weight of 1000-paddy grains did not vary significantly by the AWD and CF method. The paddy under AWD matured about one week earlier than CF across the locations.

CH4 emission
There was a significant effect (p ≤ 0.01) of AWD and CF irrigation method on the emission of CH4 gas from the paddy field at both on-farm study locations (Fig. 3). A substantially higher amount of total emission was usually found in CF, followed by AWD. We estimated 93.2 and 83.7 kg less CH4 ha -1 in AWD at Feni (94.2 kg ha -1 ) and Chattogram (106.5 kg ha -1 ) which was about 49 and 44% smaller than that of CF (187.4 and 190.1 kg ha -1 at two locations, respectively). Data indicated the CH4 emission factor for AWD was lower (0.74 kg ha -1 day -1 ) compared to CF (1.39 kg ha -1 day -1 ).

N2O emission
The emission of N2O was not varied significantly (p ≥ 0.05) by the irrigation methods at both Feni and Chattogram (Fig. 4). But numerically about 7% higher amount of N2O was found in AWD both at Feni (10.7 kg ha -1 ) and Chattogram (9.98 kg ha -1 ) than that of CF (9.96 and 9.31 kg ha -1 , respectively).

The Global Warming Potential (GWP)
The GWP was affected significantly (p ≤ 0.05) by the AWD and CF irrigation method at both the locations (Fig. 5). The higher share of GWP was found in CF than in AWD. The CF irrigation method produced 2231.6 kg higher CO2 eq. ha -1 GWP at Feni (5435 kg CO2 eq. ha -1 ) and 2096.12 kg higher CO2 eq. ha -1 GWP at Chattogram (5516.1 kg CO2 eq. ha -1 ) over AWD (3203.4 and 3419.98 kg CO2 eq. ha -1 , respectively) which were about 70 and 61% higher than that of AWD at Feni and Chattogram, respectively. Overall, the total GWP attributed to CH4 emissions was 95.04% in AWD and 97.19% in CF.

The intensity of GHG emission (GHGI)
The impact of irrigation method was significantly different (p ≤ 0.05) on the GHGI at both Feni and Chattogram ( Figure 6). We found that, the GHGI of AWD was 43% and 40% smaller at Feni (537.41 kg CO2 eq. ton -1 ) and Chattogram (546.32 kg CO2 eq. ton -1 ) than that of CF (940.31 and 910.23 kg CO2 eq. ton -1 ). Data revealed that, production of each ton of paddy under AWD method attributed 537.41kg CO2 at Feni and 546.32 kg CO2 at Chattogram, while CF was responsible to emit 940.31 and 910.23 kg CO2, respectively.

Impact of irrigation method on water productivity
AWD substantially (p ≤ 0.05) increased the quantity and amount of irrigation water used (Table 2).
Consequently, AWD (mean across locations, 1.60 kg m -3 ) had a 224% greater water productivity than CF (0.50 kg m -3 ). Water conservation is therefore a significant advantage of AWD at this study locations, since it needed 20 less number irrigation than CF. A similar conclusion was reached by (Rahman and Bulbul, 2014) who found that a season-long standing depth of water is not required for good rice yields and noted the highest 85.55 kg cm -3 water productivity in AWD. Again, a past study of Anbumozhi et al. (1998) shown an increase in water productivity of 1.26 kg m -3 in AWD plot when compared to CF (0.96 kg m -3 ). Feng et al. (2007) concluded that AWD for rice should be more widely used due to its potential to increase water productivity 3.27 kg mm -1 in AWD that is 19% higher than CF. They found irrigation water savings were 40-70% without any yield loss by applying AWD. Water conservation in AWD systems may be ascribed in part to decreased percolation and seepage. Without flood water, percolation and seepage are substantially decreased; nevertheless, these losses are largely dependent on the hydrological characteristics of the soil (Carrijo et al., 2017). For example, Sharma et al. (2002) determined that 51% of total water input in a rice field is lost via percolation, whereas Linquist et al. (2015) found that about 15% of applied water is lost through percolation and seepage in clayey Californian rice soils. In this study, AWD used 25.7 % less water on average than CF. AWD exposes fields to intermittent flooding allowed to recede until the soil reaches a specific moisture level, at which point the field is flooded. When compared to CF systems, AWD has been shown to minimize water inputs by 23% (Bouman and Tuong, 2001).
AWD substantially decreased irrigation water consumption by 34% (Carrijo et al., 2017) and 23.4% (Chidthaisong et al., 2018) when compared to CF. Around 23.4-42.6% of water was found to be saved in AWD without compromising paddy yields . Additionally, a group of researchers observed irrigation water savings of 35% ) and 30% (Devkota et al., 2013) when AWD is used instead of CF.
Saving about 24% of irrigation water resulted in a 23.41% reduction in irrigation expenses in our research.
This result is consistent with earlier findings that AWD may help decrease irrigation expenses by lowering pumping costs and fuel usage (Lampayan et al., 2015). Reduced irrigation was linked with a decrease in irrigation costs between 12 and 15%, indicating a significant benefit of AWD irrigation for resource-scarce farmers (Shahe Irrigation System should be implemented.

Impact of irrigation method on paddy yield
We observed a 3% increase in rice production in AWD compared to CF (Table 3), which may be attributable to a 24% in productive tillers. Although the number of panicles and the weight of 1000 grains were both constant numerically in AWD and CF. Increased paddy yields under AWD are primarily due to decreased redundant vegetative growth and improved canopy structure and root growth (Chu et al., 2015;Zhou et al., 2017); elevated hormonal levels, most notably increases in abscisic acid levels during soil drying and cytokinin levels during rewatering; and enhanced carbon remobilization from vegetative tissues to grain (Zhang et al., 2010;Zhang et al., 2012). Yang et al. (2009) observed an increase in rice yields under AWD due to a rise in the percentage of productive tillers, a decrease in the angle of the uppermost leaves, which allows more sunlight to penetrate the canopy, and a shift in shoot and root activity. In Nepal, a group of researchers found no significant difference in rice yields between AWD and CF, with AWD saving 57% of irrigation water (Howell et al., 2015). Rice fields with a 120-200 times greater soil oxygen content and more carbon release from the rice roots under AWD than under CF result in increased microbial populations and biomass in the rice rhizosphere, as well as increased rice production (Tian et al., 2013;Subedi and Poudel, 2021). The strong root development under AWD vs. CF more effectively absorbs water and nutrients, resulting in a greater rice grain production (Wang et al., 2016). Drying the rhizosphere modifies plant hormone signaling and increases grain filling rate (Zhang et al., 2010). The considerably greater number of productive tillers in AWD than in CF contributes to AWD's higher yield (Howell et al., 2015).
It is still disputed if the AWD irrigation system has the potential to reduce or sustain grain yields. AWD may result in higher nitrogen losses through nitrification and denitrification, resulting in decreased plant nitrogen absorption (Pandey et al., 2014). Increased numbers of tillers and effective tillers under AWD may have resulted in increased competition for plant resources between tillers and panicles, resulting in substantially reduced grain weight, quantity, and filling (Peng and Bouman, 2007). In comparison, a reduced tiller count under AWD was offset by increased grain weight and a higher percentage of grain filling per panicle, resulting in improved yield (Bouman and Tuong, 2001). A meta-analysis of 56 research including 528 side-by-side comparisons between AWD, and CF showed that AWD reduced rice grain production by 5.4 percent owing to water stress (Carrijo et al., 2017). However, Rahman and Bulbul (2014) asserts that a little amount of water stress on the plant does not result in a reduction in grain production. He found that water levels 15 to 25cm below ground level in AWD had no effect on the total number of filled grains, while 5 cm standing water in CF did, and that such standing water throughout the season is not necessary for good rice yields. Additionally, certain research from southeast China indicates that using an AWD irrigation technique may result in an improvement in grain production (Chu et al., 2015;Zhou et al., 2017). The factors outlined above may have resulted in an increase in paddy yield in AWD over CF in this research. The variations across research are due to differences in soil hydrological conditions and irrigation techniques used at different times . This demonstrates the need for further research on the impact of AWD on rice production in Bangladesh.

Impact of irrigation method on the emission of GHGs and the intensity of GHG
Irrigation method under AWD and CF influenced the amount of CH4 emitted by rice production. In this research, AWD irrigation substantially (p ≤ 0.05) decreased CH4 emissions on average about 47% (49% at Feni and 44% at Chattogram) when compared to CF irrigation (Fig. 3). These findings corroborate earlier findings (Ku et al., 2017;Islam et al., 2020). When AWD irrigation is managed correctly, significant reductions in CH4 emissions are anticipated. AWD's efficacy in lowering CH4 emissions is dependent on the efficiency of water management, the kind of soil, and other cultivation techniques (Xu et al., 2015). Intermittent aeration oxygenates the soil, resulting in the oxidation of CH4 by methanotrophs, resulting in a decrease in CH4 emission.
According to some estimates, up to 80% of the CH4 generated during the rice growing season is oxidized by methanotrophs (Singh et al., 2010). In comparison, CF rice cultivation anaerobicifies the soil environment, lowering the redox potential, which promotes the anaerobic breakdown of complex organic substrates by methanogens, which ultimately results in CH4 generation over AWD (Minamikawa et al., 2006).
The technique of irrigation had no significant effect (p ≥ 0.05) on the fluctuation of N2O in this research (Fig.   4). Although N2O emissions from rice fields grown under AWD were about 7% higher than those from paddy fields cultivated under CF conditions in Feni and Chattogram. Changing water regimes from CF to AWD influences the intensity of nitrification and denitrification, depending on the availability of oxygen. The topsoil layer becomes aerobic throughout a drying cycle, while the bottom soil layer stays anaerobic even when the water level reaches 15 cm below the soil surface. Thus, large amounts of N2O are generated because of microbial nitrification of NH4 + and denitrification of NO3 - (Islam et al, 2020). While N2O generation declines at very high moisture levels, it rises in fields with repeated wet and dry spells (Brentrup et al., 2000). By contrast, some prior research indicates that CF enhances N2O emission, and the higher the soil moisture, the larger the N2O emission (Baggs et al., 2000;Yano et al., 2014). By contrast, the reduced N2O emission peaks under CF conditions are most likely the result of additional denitrification to N under severe anaerobic conditions (Zou et al., 2005).
AWD irrigation reduced GWP by 43% as compared to CF irrigation. These results demonstrate that CH4 emissions are completely responsible for the global warming potential of rice fields. Although N2O has a much higher radiative force than CH4, its emissions are very insignificant. Thus, CH4 is the main source of greenhouse gas emissions in rice cultivation, accounting for more than 90% of total GWP emissions (Sander et al., 2014;Janz et al., 2019). In this study, the total GWP related with CH4 emissions was 95% AWD and 97% in CF, while N2O contributed only by 1% to GWP. These results are consistent with previous studies (Sander et al., 2017;Islam et al., 2018;Oo et al., 2018). As was previously observed for GWP, AWD irrigation showed the potential to reduce GHGI by 41% when compared to CF irrigation (Islam et al., 2018;. Therefore, the most successful strategies for lowering GWP and GHGI in rice production should focus on reducing CH4 emissions.

Conclusion
We studied the efficacy of AWD in terms of water savings, paddy production, and GHG emissions in farmers' paddy fields at Feni and Chattogram districts in Bangladesh. The irrigation water consumption was significantly decreased by 24% in AWD with 224% greater water productivity. Hence, the irrigation costs were saved by 23.41% than the CF. By this time the paddy yield was improved significantly by 3% in AWD compared to CF.
The AWD decreased seasonal CH4 emissions by 47% than CF but no effect on seasonal N2O emissions. In AWD, the intensity of GHG was 42% lower relative to CF. Simultaneous accomplishment of increased grain production, water conservation, and acceptable GHG emission reduction is a prerequisite for AWD adoption by existing local farmers since water and environmental conservation are not reflected in the farmers' profit in the country. Field experiments demonstrating AWD's capability should be conducted in Bangladesh under a variety of agroecological zones, soil types, and farmer management circumstances. To make AWD profitable to farmers, rather than pump owners, a community-based, prepaid card-metering subsurface irrigation system should be established.

Declaration of competing interest
The authors state that they have no known conflicting financial or personal interests that may seem to have influenced the work described in this article.