The experimental trials were conducted in two representative regions of rice cultivation in Colombia: Saldaña at the Lagunas Experimental Center (3° 55’ 59” North, 75° 1’ 1” West) in the Tolima region and in Aguazul at the La Primavera Experimental Station (5° 28’ 54” North, 72° 38’ 8” West) in the Casanare region. Tolima represents the central region, which is primarily irrigated rice production. For our field experiment, irrigation water was sourced from the Saldaña River Irrigation District during the first semester of 2022. Casanare represents the plains region, which consists of both rainfed and irrigated rice systems. In our field experiment, rice was grown under rainfed conditions in the first semester of 2022 and under irrigation in the second semester between 2022-2023, with irrigation water sourced from the Charte River.

Field experiments at both sites followed a 2 x 2 factorial arrangement with two irrigation systems (continuous flooding and intermittent irrigation two rice varieties per region. In Tolima, the varieties were Fedearroz 67 (F-67) and Fedearroz 2000 (F-2000), while in the Casanare region, we used FL Fedearroz Itagua (F-Itagua) and Fedearroz 70. The factorial designed was implemented within a randomized complete block design, with four replication per variety in each treatment. This resulted in four plots per variety under flooded irrigation and four plots per variety under intermittent irrigation in both regions. The total experimental area covered 1200 m², with individual plot sizes of 50 m². In both regions, the implementation of the intermittent irrigation treatment, fertilizer applications, and water replenishment occurred with soil moisture levels maintained near or above field capacity. During fertilizer application, soil moisture ranged from 29% to 37% under intermittent irrigation treatment in Tolima, and from 35% and 38% in Casanare. Irrigation predominates in the Tolima production system. Conversely, in the Casanare region, both irrigated and rainfed rice are cultivated under diverse climatic conditions and water resource availability. Notably, in the rainfed system, the intermittent irrigation treatment was tailored to depend solely on rainfall, eliminating the need for supplemental irrigation. Consequently, achieving the intermittent irrigation regimen entailed draining surplus water only during fertilizer application to sustain soil moisture levels at around field capacity.

For soil preparation during the rice growing season in Tolima and Casanare, a 24-inch disc plow (harrow) passed over the soil two times, followed by two passes of a Micro-grader. Subsequently, a 14-point furrower created rows spaced 0.17 m apart in the trial plots. Weather stations located at the experimental centers in the Tolima and Casanare regions provided climatic data for the trials, including temperature, precipitation, relative humidity and atmospheric pressure. Table 1 provides further details of the agricultural practices, including planting and harvesting dates, fertilizer type and application.

In Tolima, from June to September 2022, 363 mm of rainfall occurred over 47 events. During the vegetative phase (60 days), 275 mm fell in 33 events; the reproductive phase (35 days) saw 63 mm from five events, and the maturation phase (30 days) received 25 mm over eight events (Figure 1). The daily average temperature was 27 °C (max 30 °C/min 24 °C). Relative humidity averaged 81%, peaking at 84% during the vegetative phase and dropping to 76% later. Solar energy varied: 408 cal cm⁻² d⁻¹ (vegetative), 439 cal cm⁻² d⁻¹ (reproductive), and 445 cal cm⁻² d⁻¹ (maturation).

In Casanare’s rainfed system (first semester of 2022), 1283 mm of rain fell over 72 events. The vegetative phase saw 776 mm (61 events), the reproductive phase 425 mm (35 events), and maturation 82 mm (13 events). The daily average temperature was 25 °C (max 29 °C/min 22 °C), with an average humidity of 91% during the vegetative and reproductive phases, dropping to 88% in the maturation phase. Solar energy was 308, 362, and 405 cal cm⁻² d⁻¹ for the vegetative, reproductive, and maturation phases.

In the second semester for irrigated rice, 61 mm of rainfall occurred over nine events: 49 mm in the vegetative phase (5 events) and 12 mm during maturation (4 events). Average temperatures were 26 °C (max 29 °C/min 24 °C), with 79% humidity (82% in vegetative, 75% in reproductive and maturation phases). Solar energy was 403, 464, and 370 cal cm⁻² d⁻¹ for the vegetative, reproductive, and maturation phases.

The closed static chamber technique was used to determine CH₄ and N₂O emissions, following Chirinda et al. (2017). Leak control, calibration, and validation procedures were implemented as described in Loaiza et al. (2024a), including chamber vent systems to prevent pressure differences, battery- powered fans to ensure gas homogenization. Each chamber consisted of two parts made of polyethylene: a base with a height of 40 cm and a lid with a volume of 114 liters and a height of 80 cm. Before sowing rice seeds, one base was inserted into the soil in each plot, covering nine or twelve rice seedlings. The lids were equipped with fans for air mixing, a steel thermometer for temperature recording, a gas sampling port, and a vent installed on the lid to maintain equilibrium with external pressure variations. During early morning hours between 8 and 10 am throughout the rice growing season, four gas samples were collected from each chamber using a polyethylene syringe at 15-minute intervals. Immediately after collection, the gas samples were transferred to pre-evacuated 5.9 ml glass Exetainer vials (Labco Ltd.). The sampling period covered approximately 80% of the growing rice season in both regions to obtain accurate and reliable data on CH₄ and N₂O emissions.

The concentrations of each gas were determined using gas chromatography (Shimadzu GC-2014) with a Flame Ionization Detector (FID) for CH₄ and a ⁶³Ni Electron Capture Detector (ECD) for N₂O. The detection limit was 0.060 ppm for CH₄ and 0.100 ppm for N₂O. The gas fluxes were calculated based on the linear increase in the gas concentration observed throughout the sampling period using the following Equation 1:

where F (mg m^-2 h^-1) is the CH₄ or N₂O flux, ΔC/Δt (ppm h^-1) is the linear change in CH₄ or N₂O concentration observed over time, M is the molecular mass of the gas (16 for CH₄ and 44 for N₂O), V (m³) and A (m²) are the chamber volume and area covered by the chamber. V_m is molar volume of gas (L mol^-1) determined through the ideal gas law. The seasonal cumulative fluxes for CH₄ and N₂O emissions (kg ha^-1) were calculated by linear interpolation between sampling dates. The global warming potential was calculated in terms of carbon dioxide equivalent (kg CO₂ equiv. ha^-1) over a 100-year time frame using the IPCC guidelines and radiative forcing potentials of 27.2 for CH₄ and 273 for N₂O (IPCC, 2021).

Aboveground rice biomass in both regions was assessed at various phenological stages, including primordium, tiller, and flowering. Sampling involved randomly placing 0.25 m² quadrants within the treatment plots and collecting all aboveground biomass components, such as stems, leaves, and panicles. These biomass samples were dried at 70 °C for 24 hours until a constant weight was achieved, following the methodology outlined by Yepes et al. (2011). At physiological maturity, a 20 m² area was harvested from each plot to determine rice grain yields. The harvested grains were dried in an oven at 70 °C for 72 hours, and the reported grain yield reflects a moisture content of 14%.

The statistical analyses were conducted using R Studio software, with the ADE4 and Agricolae libraries employed for data processing. The normality of datasets was assessed based on sample size, employing the Shapiro-Wilk test for datasets with less than 50 observations and the Kolmogorov-Smirnov test for datasets with more than 50 observations, both at a 5% significance level. Data that met the normal distribution criteria underwent one-way and two-way ANOVA, followed by Tukey’s HSD post-hoc tests for group comparisons. Other data were analyzed with the non-parametric Kruskal-Wallis test, followed by Dunn’s test for post-hoc analysis. In the ANOVA, irrigation treatment (flooded vs. intermittent) and rice variety (two per region) were considered as independent variables, while CH₄ and N₂O fluxes, cumulative seasonal emissions, grain yield, and aboveground biomass were the dependent variables.

The relationships between daily CH₄ and N₂Oemissions and climatic conditions were analyzed through multivariate principal component analysis (PCA), co-inertia analysis, and permutation Monte Carlo tests to compare production systems and assess the climate’s effect on emissions. In the PCA, daily CH₄ and N₂O emissions were treated as dependent variables, while climatic conditions (air temperature, rainfall, relative humidity, and solar radiation) were included as explanatory variables (Supplementary Tables S1 and S2). These analyses were conducted using the R Studio environment and the ADE4 library (Posit team, 2023; Thioulouse et al., 2018).

In Tolima, CH₄ emissions peaked under continuous flooding, reaching 37 mg m⁻² d⁻¹ 59 days after germination, spanning both the reproductive and maturation phases of rice varieties. In contrast, intermittent irrigation consistently resulted in emissions below 10 mg m⁻² d⁻¹, with no significant differences observed between varieties for either treatment (p > 0.05). N₂O emissions notably increased during the fourth fertilization, 70 days after germination, near the maturation phase under continuous flooding. Intermittent irrigation effectively mitigated N₂O emissions for both varieties on the same date, with significant differences observed between varieties for both treatments (p < 0.05). Specifically, variety F-67 exhibited the highest emissions under flooded conditions, whereas variety F-2000 demonstrated higher emissions under intermittent irrigation (Figure 2).

In Casanare, CH₄ emissions peaked 76 days after germination under continuous flooding during the rainfed season. The pattern was similar for both varieties under intermittent irrigation, with significant differences between varieties (p < 0.05). N₂O emissions were highest during the second and fourth fertilizations under continuous flooding, remaining below 4 mg m⁻² d⁻¹ with intermittent irrigation (Figure 3). During the irrigated season, CH₄ emissions peaked 43 days after germination under continuous flooding, while intermittent irrigation showed consistently negative emissions throughout the cycle. For N₂O, peaks occurred during the second and third fertilizations, with an increase under continuous flooding at each fertilization compared to consistently negative emissions with intermittent irrigation (Figure 4). Significant differences (p = 0.001) were also observed between the two growing seasons in the Casanare region.

Both N₂O and CH₄ emissions showed significant differences between treatments (p < 0.05) in both regions. Principal component analysis revealed a direct relationship between CH₄ emissions, air temperature, and solar energy, and an indirect relationship with relative humidity, during the planting season from May to September 2022 in both regions (Tables S1 and S2). The average explained variance was 70%.

Flooded treatment consistently resulted in significantly higher cumulative CH₄ and N₂O fluxes than intermittent irrigation across all regions, varieties, and seasons. The exception was Casanare Season I with the F-70 variety, where higher emissions were observed under flooding. Under intermittent irrigation, CH₄ emissions were reduced by up to ~100%, effectively suppressing methane release compared to continuous flooding across both regions. Cumulative N₂O flux reductions ranged from 54% to 78% in Tolima and 6% to 46% in Casanare during the first season, reaching 100% in the second season. The most substantial N₂O reductions under intermittent irrigation were seen with F-67 in Tolima (78%) and F-Itagua in Casanare (46%) (Table 2).

Flood irrigation also showed higher GWP than intermittent irrigation in both regions and seasons, except for F-70 in Season I in Casanare. CH₄ emissions increased GWP values in flooded treatments, while N₂O contributed to higher GWP under intermittent irrigation, except in Casanare during the second season where both emissions were negative. F-2000 and F-70 exhibited the highest accumulated CH₄ emissions under flood treatment in Tolima and Casanare, respectively. Intermittent irrigation in Tolima reduced GWP by 85% for F-67 and 62% for F-2000. In Casanare, reductions were 62% for F-Itagua and 14% for F-70 in the first season, with both varieties achieving 100% reduction in the second season (Table 2).

The F-67 variety under flood treatment demonstrated superior aboveground biomass during the primordium phase, while F-2000, under intermittent irrigation, showed higher biomass during the tillering phase. No discernible trends were observed during the flowering stage. Despite observed variations in biomass and yield, statistical analysis indicated no significant differences in the Tolima region (Table 3).

In Casanare Season I, significant differences in aboveground biomass were recorded between treatments and varieties during the primordium and flowering stages. F-Itagua exhibited higher biomass under flooded conditions during the primordium phase, whereas F-70 showed increased biomass under intermittent irrigation during the flowering phase. No significant yield differences were detected. In Season II, the primordium stage emerged as the most prominent under flooded conditions, particularly in F-Itagua and F-70. Tiller stage biomass varied, with higher values observed in F-Itagua under flooded conditions. F-Itagua under flooded conditions and F-70 under intermittent irrigation displayed elevated biomass during the flowering stage. Although no significant differences in grain yield were recorded, F-Itagua under flood treatment demonstrated a tendency towards higher grain production. Significant differences between planting periods in Casanare were found for tillering, flowering, and yield phases, with F-Itagua excelling in the irrigated planting period and F-70 performing in the rainfed semester (Table 3).

The intermittent irrigation strategy reduced daily CH₄ emissions by 31-96% (Figures 2-4), supporting previous findings on CH₄ emission reduction through intermittent drainage (Cowan et al., 2021; Goto et al., 2004; Meijide et al., 2017; Tirol-Padre et al., 2018). Soil aeration periods during the crop cycle with intermittent irrigation likely influenced water and oxygen dynamics in soil pores, promoting organic carbon oxidation to CO₂ through methanotrophic bacteria under aerobic conditions (Lim et al., 2024; Sun et al., 2016), thus reducing CH₄ emissions (Bo et al., 2022). During brief water replenishment periods, carbon oxidation may continue due to methanotrophic bacteria at the soil-water interface and in the rice rhizosphere (Deppe et al., 2010). Additionally, rice plants supply atmospheric oxygen to roots via aerenchyma (Neue, 1993; Nouchi et al., 1991), facilitating root oxidation and contributing to CH₄ reduction (Bhattacharyya et al., 2016, 2019). Soil aeration and high iron content in both regions may also promote iron oxidation, reducing CH₄ emissions compared to continuously flooded systems (Nishimura et al., 2020).

In some cases, slightly negative CH₄ fluxes were observed under intermittent irrigation. These values are consistent with aerobic soil conditions that suppress methanogenic activity while stimulating methanotrophic populations, resulting in partial methane uptake and an effective reduction of CH₄ emissions to the atmosphere (Ma and Lu, 2011). However, the magnitude of these negative fluxes was small and did not differ significantly from zero, so they should be interpreted as near-complete suppression of CH₄ emissions rather than sustained atmospheric uptake. This justifies reporting the reduction as “up to 100%,” reflecting that intermittent irrigation conditions effectively eliminated CH₄ emissions compared to continuous flooding.

Intermittent irrigation significantly reduced methane emissions to an average of -10 mg m^-2 d^-1, with no regional effect despite climatic differences. Soil aeration favored organic carbon oxidation throughout the cultivation cycle. Previous research on methane emissions in rice focused on correlating environmental parameters like temperature and precipitation with emissions in flooded systems. For instance, Lee et al. (2023) found temperature and sunlight hours positively affect CH₄ emissions, while Hou et al. (2023) noted increased temperature and traditional fertilization contribute to higher CH₄ emissions.

The rice variety tested did not significantly affect daily CH₄ emissions under intermittent irrigation. Previous studies indicate that varietal traits such as aerenchyma development and root oxygen release can influence CH₄ dynamics by supporting methanotrophic activity and reducing methanogenesis (Baruah et al., 2010; Shang et al., 2011). However, in our trials, the observed reductions were primarily associated with irrigation management and soil aeration. CH₄ peaks in flooded treatments coincided with tillering, flowering, and grain-filling stages, when biomass accumulation and aerenchyma development enhanced root exudates and methanogenesis, consistent with previous reports (Feng et al., 2021; Huang et al., 1997; Mariko et al., 1991). In the early vegetative stages, with less biomass and smaller aerenchyma, CH₄ emissions were similar between treatments, consistent with research showing that limited biomass can reduce emissions by up to 27% (Iqbal et al., 2021).

Our results show that the plant phenological stage significantly influences CH₄ emissions in flooded systems, with intermittent irrigation effectively reducing these emissions. Proper water management, especially intermittent drainage, is a viable strategy for lowering CH₄ emissions in rice cultivation, promoting sustainable production.

Daily N₂O emissions peaked at various stages in continuous irrigation across Tolima and Casanare. In Tolima, peaks occurred during the fourth fertilization near maturation, while in Casanare, they were noted during the second fertilization in the rainfed season and between the second and third fertilizations in the irrigation season (Figures 2-4). These peaks likely result from increased nitrogen availability due to urea and ammoniacal fertilizers (Firestone and Davidson, 1989; Xu et al., 2015).

In intermittent irrigation, daily N₂O emissions ranged from 2 to -5 mg N₂O m^-2 d^-1, likely due to controlled soil moisture near or above field capacity, enhancing fertilizer solubilization (Loaiza et al., 2024a). This aligns with Riya et al. (2017) and Islam et al. (2020b), showing that such conditions support nitrification and provide substrates for denitrification.

Compared to previous studies, our emissions were lower. For example, Feng et al. (2021) in Hubei reported 0.02 to 0.03 mg N₂O m^-2 d^-1 under Alternate Wetting and Drying (AWD), while Chirinda et al. (2017) reported 0.55 mg N₂O m^-2 d^-1 in Saldaña, Tolima, under similar conditions. Our trials showed 8 to 15 mg N₂O m^-2 d^-1 in Casanare and 40 mg N₂O m^-2 d^-1 in Tolima under continuous flooding.

Intermittent irrigation reduced N₂O emissions compared to continuous flooding (Figures 2-4), sometimes resulting in net negative emissions. This reduction is due to maintaining soil moisture between saturation and field capacity, which regulates oxygen diffusion and prevents abrupt aerobic-to-anaerobic transitions, thereby avoiding the strong N₂O peaks typically observed in flooded systems (Firestone and Davidson, 1989; Peng et al., 2011; Riya et al., 2017). Under these conditions, controlled soil aeration also facilitated NO₃⁻ to N₂ conversion by anaerobic bacteria during short aerobic phases, maintaining conditions below the threshold for aerobic respiration (Sapkota et al., 2020) and inhibiting N₂O exchange due to low oxygen in soil pores (Suenaga et al., 2018, Pan et al., 2022). Similar mechanisms have also been observed in other cropping systems, such as drip-irrigated cotton, where higher irrigation intensities that kept soil moisture close to or above field capacity led to greater N₂O release, while moderate irrigation maintained lower emissions (Xia and Wander, 2022). This supports the interpretation that moisture regulation around field capacity is critical for mitigating N₂O emissions across production systems. Systems like intermittent irrigation or AWD, which keep soil moisture close to the wilting point or below 15 cm from the surface, can enhance aeration and potentially increase N₂O release (Liang et al., 2022).

No significant emission differences were observed between varieties within regions, though Tolima had higher peaks under continuous flooding, likely due to higher fertilizer application levels suggesting more nitrification substrate (Kim et al., 2021; Yao et al., 2012). Our findings contrast with most studies conducted in Asia and Africa, which report that intermittent irrigation reduces CH₄ emissions but increases N₂O emissions due to greater soil aeration. However, under our experimental conditions, intermittent irrigation-maintained soil moisture close to saturation and did not generate prolonged periods of aeration. This limited nitrification–denitrification processes, resulting in lower N₂O emissions compared with flooded soils. In contrast, prolonged flooding favored the accumulation of mineral nitrogen under reducing conditions, leading to incomplete denitrification and higher N₂O emissions. These particularities highlight the importance of evaluating interactions among water management, soil properties, and rice varieties in local contexts, since results from Asia and Africa cannot be directly extrapolated to Latin American production systems.

Soil conditions in Tolima and Casanare may help explain the differences observed in CH₄ and N₂O emissions. In Tolima, the sandy loam texture, a slightly acidic pH, and higher organic matter favored soil aeration. The elevated iron content also encouraged oxidation processes, which helped reduce CH₄ emissions compared with flooded fields. In Casanare, the sandy and more acidic soils, with lower organic matter and nitrogen, reduced the substrates available for microorganisms. At the same time, rainfall and intermittent irrigation caused strong shifts in soil moisture, which increased the variability of N₂O fluxes (Lesschen et al., 2011). Taken together, these results suggest that soil type and chemical properties play an important role in the microbial processes that generate greenhouse gas emissions under different irrigation practices.

Previous research on intermittent irrigation’s impact on rice yields has been mixed. Some studies report yield reductions (Feng et al., 2021; Islam et al., 2020a), while others note increases (Hassan et al., 2015; Nugroho et al., 2018; Thakur et al., 2018), and some find no change (Bo et al., 2022; de Avila et al., 2015; Haque et al., 2016, 2021; Keiser et al., 2002; Linquist et al., 2015; Loaiza et al., 2024a). In our study, intermittent irrigation-maintained soil moisture near field capacity with replenishment every 3 to 4 days, showing no yield penalties compared to continuous flooding (Table 3). This approach prevented water stress, ensuring optimal conditions for seed formation, panicle and root development, and photosynthesis (Shukla et al., 2013; Yang et al., 2004). Bouman et al. (2007) and Carrijo et al. (2017) found that maintaining water above the permanent wilting point, around -20 kPa, prevents yield reductions. Our study confirmed the optimal moisture threshold and water management strategies, avoiding yield losses.

Yields in Tolima and Casanare remained relatively high in terms of variety or irrigation treatment within regions. However, regional differences were noted, especially in Casanare, likely due to climatic variations and the types of varieties grown. Tolima receives about 431 cal m^-2 d^-1 of solar energy, compared to 358 cal m^-2 d^-1 in Casanare, which may explain regional differences. Higher solar radiation positively affects plant growth and yields (Deng et al., 2015; Quevedo et al., 2019; Peng et al., 2004; Tu et al., 2022). In Tolima, the F-67 and F-2000 varieties showed better tillering than the F-Itagua and F-70 varieties in Casanare (Ospina et al., 2024), indicating that varieties adapted to local conditions perform better. Seasonal yield differences in Casanare are influenced by temperature, precipitation, and water management. The rainy season depends on rainfall, while the dry season relies on irrigation. Increased heat in the dry season can elevate soil tension, slightly affecting yields. Studies show that even minor heat stress can impact growth and yield (Julia and Dingkuhn, 2013; Mittler et al., 2012; Kobayashi et al., 2010).

Our study confirms that well-designed intermittent irrigation systems can maintain crop yields and offer a sustainable alternative to flooded systems, especially in water-scarce regions. However, challenges like inconsistent scheduling and poor moisture control can limit their benefits. Overcoming these challenges will require technological solutions and policy support to promote effective adoption and management of intermittent irrigation.

The balance between methanogenic and methanotrophic activity determines cumulative CH₄ flows (Lee et al., 2014; Nagler et al., 2021; Zhang et al., 2023). Our results show that intermittent irrigation significantly reduces CH₄ emissions compared to continuous flooding, regardless of whether the rice is irrigated or rainfed. This method lowers daily and peak CH₄ emissions, thus reducing cumulative flows during key crop stages. These findings align with studies showing an 82% reduction in CH₄ emissions with controlled irrigation systems (Mazza et al., 2016; Minamikawa et al., 2016; Nie et al., 2023).

Rice varieties affected CH₄ emissions regionally. In Tolima, differences among varieties were notable, with F-67 showing lower emissions. In Casanare, only irrigation treatments influenced emissions, with F-Itagua showing lower accumulated CH₄. These results suggest that certain varieties can help mitigate CH₄ emissions.

Cumulative N₂O flows result from nitrification and denitrification (Firestone and Davidson, 1989; Hassan et al., 2022; Wang et al., 2016). Intermittent irrigation reduced cumulative N₂O flows compared to flooding, with effective moisture control during fertilization being crucial. However, higher N₂O flows in Casanare’s rainfed system, due to dependence on precipitation, align with research suggesting alternating wet and dry conditions increase N₂O emissions (Zhan et al., 2015).

No significant differences were found among rice varieties in either region. In Casanare, intermittent irrigation resulted in lower cumulative N₂O fluxes during the dry season due to better moisture control. This maintained microaerobic conditions, reducing N₂O emissions through improved nitrification and denitrification (Riya et al., 2017; Sapkota et al., 2020).

The Global Warming Potential (GWP) is crucial for assessing environmental impacts (Shang et al., 2011; Tariq et al., 2017). In our trials, N₂O emissions drove the GWP, accounting for over 73% in Tolima and 56–97% in Casanare. While N₂O is significant in GWP, intermittent irrigation showed lower cumulative N₂O emissions. This method can effectively reduce GWP by cutting CH₄ and slightly increasing N₂O, consistent with other studies (Cowan et al., 2021; Sun et al., 2022). Intermittent irrigation can enhance rice production without significantly increasing GHG emissions.

Intermittent irrigation can reduce GHG emissions in flooded rice systems, cutting CH₄ by up to 100% and N₂O by 6-100%. It is easy to implement and more cost-effective than AWD systems, which require expensive infrastructure. Using simple technologies and practical knowledge, intermittent irrigation improves soil drainage and moisture control, promotes wider adoption, and reduces costs. Precise fertilizer management also enhances nutrient uptake and minimizes contamination, supporting sustainable production.