Water management at fertilization controls urease inhibitor performance in mitigating ammonia volatilization from temperate rice agrosystems

1 Introduction

Ammonia (NH₃) volatilization is a major pathway of nitrogen (N) loss in rice agroecosystems, reducing fertilizer use efficiency and contributing to environmental pollution (Erisman and Schaap, 2004; Behera et al., 2013). Globally, agriculture accounts for approximately 80%–95% of atmospheric NH₃ emissions, with rice paddies contributing about 21% of agricultural emissions. Within rice systems, mineral fertilizers are responsible for approximately 20% of NH₃ emissions (Xu et al., 2019; Skorupka and Nosalewicz, 2021; Zhan et al., 2021). Urea is the most widely used N fertilizer in rice production because of its high N concentration (46%) and relatively low cost. However, it is highly susceptible to NH₃ volatilization, particularly during the first days after application (Bouwman et al., 1997). Reported NH₃ losses from applied urea vary widely depending on soil, climate, and management conditions, ranging from 0.6% to 64% worldwide and from 0.9% to 29.8% in European rice systems, with typical losses under conventional management ranging between 9% and 43% (Fillery and De Datta, 1986; Pan et al., 2016). Overall, NH₃ volatilization from rice systems has been estimated to account for approximately 9.7 million tons of N annually (Matse et al., 2024).

Reducing ammonia emissions from surface-applied urea has become a pressing issue in the context of European and national air quality legislation that must be addressed in the coming months. Reducing NH₃ emissions has also become a regulatory priority. European legislation (Nitrates Directive 91/676/EEC; NEC Directive 2016/2284) and the recent Italian National Air Quality Action Plan promote the replacement of untreated urea with mitigation strategies in intensive agricultural areas, including the Po Valley (Italy, 2025).

One strategy to improve the performance of urea in agriculture and reduce urea-driven pollution is the use of alternative N fertilizers (Swify et al., 2024; Wang et al., 2024; Fang et al., 2026). Ammonium sulfate is known to result in lower NH₃ volatilization losses–less than 10% of applied N (Norman et al., 2009; Marchesan et al., 2011) – compared to urea, due to its acidifying effect, which enhances NH₄ ⁺ retention in soil (Vlek and Craswell, 1979; Freney et al., 1981a and Freney et al., 1981b). However, its use is associated with higher costs (both per unit of N and in terms of logistical management) as well as environmental concerns related to sulfur compounds. Urease inhibitors, such as N-(n-butyl) thiophosphoric triamide (NBPT), are commonly used to mitigate NH₃ emissions from urea-based fertilizers, but their effectiveness may vary substantially depending on environmental conditions. In rice systems, water management at fertilization influences both soil oxygen availability, which is required for NBPT activation, and soil temperature, which strongly affects NH₃ volatilization dynamics. NBPT is the most widely used urease inhibitor to reduce NH₃ volatilization from urea-based fertilizers. After activation under aerobic soil conditions, it temporarily inhibits urease activity, delaying urea hydrolysis and synchronizing N release with crop demand, thereby reducing NH₃ losses and improving N use efficiency (Manunza et al., 1999; Cantarella et al., 2018; Sigurdarson et al., 2018). However, its effectiveness decreases over time and is strongly influenced by environmental conditions such as soil temperature and moisture. However, there is still room for improvement, as NBPT has a relatively short period of effective urease inhibition in soils, especially at high temperatures, and the N release patterns do not always fully match crop requirements throughout the growth period leading to inconsistent yield responses (Suter et al., 2011; Cantarella et al., 2018; Matse et al., 2024). Previous studies reported that NBPT slows urea hydrolysis in flooded rice, but reductions in NH₃ emissions and yield responses remain inconsistent, suggesting that environmental conditions strongly regulate inhibitor performance (Cai et al., 1989; Varel, 1997; Marchesan et al., 2013).

Inhibitor effectiveness is influenced by several factors, including soil temperature, pH, organic matter, and notably, water management. Soil moisture regulates NBPT activity because the molecule must be converted into its active form through an oxygen-mediated process (Sanz-Cobena et al., 2011; 2016). At the same time, lower soil moisture increases air-filled pore space, enhancing NH₃ diffusion from soil to the atmosphere, while intermediate moisture conditions may accelerate urea hydrolysis without providing sufficient water to transport urea deeper into the soil or limit gaseous losses (Sanz-Cobena et al., 2016; Drame et al., 2023). In addition, irrigation events occurring shortly after fertilization can promote the vertical and lateral movement of urea, potentially increasing N losses through leaching (Dawar et al., 2011; Savin et al., 2021). However, most of the current understanding of NBPT behavior derives from studies conducted in aerobic cropping systems (Chien et al., 2009 and references therein; Zhong et al., 2026), where optimizing soil moisture is essential to ensure adequate O₂ diffusion and thus activation of the inhibitor. In contrast, limited attention has been given to rice agro-ecosystems, where water management is fundamentally different and often involves flooded conditions.

In temperate regions water seeding with continuous flooding (WS-CF) remains a widely adopted water management practice as early flooding of the rice area, allowing to recharge the water table and sustain the multiple use of irrigation water (Hoek et al., 2001; Peyron et al., 2016; Mayer et al., 2019; Gharsallah et al., 2023). Fertilization under flooded conditions is also considered technically easier to manage compared to more complex water-saving strategies such as alternate wetting and drying (AWD) or drainage-based systems (Han et al., 2021; Mir et al., 2025), although some studies suggest that these practices may increase ammonia volatilization (Zhao et al., 2010). Rice is inherently adapted to flooded environments and continuous submergence contributes to yield stability across many production contexts. Under these conditions, the role of water management at fertilization becomes crucial. In water-seeded systems, fertilization occurs directly in the presence of water, requiring careful timing and management to balance N availability and environmental losses (Daigh et al., 2014; Dempsey et al., 2017). Moreover, in temperate environments, soil moisture at fertilization can be highly variable due to fluctuations in air temperature, further complicating the dynamics of NBPT effectiveness. This is particularly relevant considering that readily available mineral N at topdressing is essential for achieving high rice yields, especially when pre-sowing fertilization is insufficient, but must be managed carefully to minimize environmental impacts (Zhou et al., 2022a and Zhou et al., 2022b; Aryan et al., 2025).

Meta-analyses and reviews consistently report average NH₃ losses of about 31% of applied N from conventional urea versus roughly 15% from urea + NBPT, corresponding to ∼50–60% mitigation of volatilization across a wide range of soils, climates and crops, and with modest mean yield gains of ∼5% (Silva et al., 2017; Wang et al., 2020; Klimczyk et al., 2021). Nonetheless, the variability observed in rice systems highlights the need for a better understanding of the interaction between inhibitor performance and water management at fertilization time. Although the effectiveness of NBPT has been widely investigated, little attention has been paid to how short-term drainage immediately before fertilization influences inhibitor activation under temperate water-seeded rice conditions. This aspect is particularly relevant in temperate rice systems, where moderate temperatures may favor NBPT activation, provided that sufficient oxygen is available before reflooding. To the best of our knowledge, no previous field study has specifically evaluated the interaction between short-term pre-fertilization drainage, NBPT activation, prompt resubmersion and NH₃ mitigation under temperate water-seeded rice conditions.

To address this knowledge gap, this study evaluated whether precise water management at fertilization enhances the effectiveness of urea + NBPT under temperate water-seeded rice conditions, by assessing NH₃ emissions, soil N dynamics, apparent N recovery, and crop performance. In order to establish the efficiency of the urease inhibitor urea and urea + NBPT were applied and compared with ammonium sulfate during 2 N topdressing applications under two contrasting water management regimes at fertilization (drained and flooded soil). Specifically, we hypothesized that: i) the effectiveness of NBPT in mitigating NH₃ emissions in temperate rice cropping systems depends on water management at fertilization; ii) significant interaction effects exist between N fertilization and soluble N pools along the soil profile; iii) NH₃ emission mitigation has also agronomic impacts as it is associated with enhanced plant performance and higher apparent N recovery, ultimately leading to increased rice yields.

2 Materials and methods

2.1 Experimental site and crop seeding

The research was carried out in 2024 in an experimental site located within the Rice Research Centre of Ente Nazionale Risi at Castello d’Agogna (Pavia, Italy) (45°14′48″ N, 8°41′52″ E, NW Italy). The site is located in the low section of the river Po plain, which includes the distal part of the glacial alluvial Würmian flat and is characterized by the presence of bumps of Holocene fluvial dynamics and leveling due to the more recent agricultural processes (Miniotti et al., 2016). The climate is temperate, characterized by hot summers and two main rainy periods in spring and autumn. The mean annual precipitation in 2024 was 1,057 mm, while the mean annual temperature was +14,7 °C (Figure 1), slightly higher with respect to the previous 20-year mean (+13,3 °C). The first 0–15 cm soil layer was characterized before fertilizer application resulting in a loamy texture, a pH of 5.5 (1:2.5 soil-to-water ratio), an organic carbon (C) content of 13.4 g kg^-1 and total N content of 1.3 g kg^-1. Average C/N ratio was 10.2, while cation exchange capacity (CEC) was 11.8 cmol₍₊₎ kg^-1. Soil properties were statistically similar among the treatment plots.

Water seeding was carried out in the second half of May (May 22nd) using the Selenio variety (150 kg ha^-1). Selenio is a round grain variety with a medium cycle (about 135 days) which shows high yield performance, due to strong tillering capacity, and number of spikelets per panicle, but it is susceptible to lodging because of its medium height (Miniotti et al., 2016).

2.2 Experimental design and treatments

The experimental design consisted of a non-randomized block design with thirty-two plots (6.8 × 6.5 m). A fully randomized allocation of treatments was not feasible because the experiment required independent control of water supply and drainage to impose the two contrasting water management regimes (drained and flooded soil) at each fertilization event. Therefore, plots were arranged according to the irrigation and drainage infrastructure to ensure precise water management. To minimize potential positional effects, the experimental area was laser-levelled before sowing, soil physicochemical properties were verified to be homogeneous among plots prior to treatment application, all plots received identical agronomic management apart from the experimental treatments, and three replicates plus one more for better statistic robustness for agronomic parameters statement were established for each treatment.

Three N fertilizers were evaluated: (a) Urea, (b) Urea + Urease Inhibitor (NBPT), (c) Ammonium Sulphate (AS), and (d) a non-fertilized control. AS was chosen as reference fertilizer, due to notoriously extremely low ammonia emissions after its application. Nitrogen fertilization followed the typical fertilization protocol for the Selenio variety (120 kg N ha^-1). Each fertilizer treatment was equally split and applied at two fertilizer topdressing events (tillering and panicle differentiation) under two contrasting water management regimes at the time of fertilization (WMF): (i) drained soil (DF) and (ii) flooded soil (WF). In DF treatments, N fertilizer was applied 2 days after complete drainage while reflooding occurred 2 days after fertilization. In WF treatments fertilizers were applied in standing water, closing the water recycling for 5 days.

Water management consisted of water seeding and continuous flooding for most of the growing season, except for a 16-day drainage period after sowing to allow plant establishment. Diversification of water management has occurred at the time of fertilization. In the DF treatment, two additional mid-season drainage events (4 days each) were applied at tillering (from June 15th) and panicle differentiation (from July 13th) to enable fertilizer applications. In the WF treatment, fertilization was conducted under continuous flooding. For both treatments, drainage was then allowed at the ripening stage (August 27th), when grain moisture was between 26% and 28%, approximately 30 days before harvest (September 30th).

Ammonia emissions were measured in three of the four replicate plots. Each plot row was bounded by two levees (50 cm above the soil surface) and flanked by two 3.5 m wide channels; one used for water supply and the other for drainage. Within each row, the eight plots were separated with corrugated polyester sheets to ensure strict control of water management. This setup allowed independent regulation of water levels and separated drainage for each plot.

2.3 Sampling and measurements

2.3.1 Ammonia emission

Ammonia emissions following fertilizer application were monitored for a 15-day period by means of a semi-open static chamber system similar to that originally described by Beyrouty et al. (1988). Off-white PVC tubes with a diameter of 12.5 cm and a total height of 65 cm were inserted into the soil to a depth of 15 cm. Each chamber was protected from rainfall and animal droppings by a plastic cover suspended 5 cm above the top of the tube by metal rods, allowing air circulation while preventing external contamination. Inside each chamber, two polyurethane foam sponges (density: 30 kg m^-3, 2 cm thick) were installed. The sponges had a slightly larger diameter than the chambers to ensure a tight fit and stable positioning. Each sponge was impregnated with 8 mL of a 33% (v/v) phosphoric acid (H₃PO₄)-glycerol solution (0.73 M), suitable for ammonia capture and conversion to ammonium ion. This volume was sufficient to fully saturate the sponge while preventing dripping from the sorbent or leaching along the chamber walls. The first sponge was placed 10 cm above the emission source (soil surface or the water surface under flooded conditions) and was used to capture volatilized ammonia. A second sponge, placed 15 cm above the first, served to prevent potential atmospheric interferences. This monitoring system did not allow the presence of plants within the chamber; therefore, only emission from the soil and water layers was measured. Prior to use, new sponges were pre-conditioned by adding 8 mL of H₃PO₄ followed by addition of 100 mL of 2 M KCl. The sponges were then rinsed with deionized water to eliminate KCl residues and dried at 50 °C for 16–24 h. Ammonia volatilization was monitored by periodically replacing the lower sorbent in each chamber 1, 2, 3, 7, and 14 days after fertilization. The duration of the NH₃ emission monitoring period was set based on preliminary studies carried out during the previous growing season showing that NH₃ emissions became negligible 14 days after fertilization. In the WF treatments, chambers were filled carefully by hand to match the water level outside the chamber. After exposure, the sponges were placed in sealed polyethylene bags containing 100 mL of 0.5 M KCl to initiate ammonium extraction, which was completed in the laboratory. Simultaneously, a new sponge impregnated with the trapping solution was installed in each chamber. After overnight extraction on horizontal shakers, an aliquot of the KCl extract was analyzed to determine ammonium concentration by the modified Berthelot method (Crooke and Simpson, 1971).

Preliminary laboratory tests were carried out to assess the potential risk of N release from the polyurethane sponge material. For this purpose, new sponges were washed, dried and conditioned with an acidic solution of 0.73 M H₃PO₄ and 33% v/v glycerol. After a 24-h conditioning period, ammonium extraction was carried out using 0.5 M KCl. No ammonium was detected in the extracts. Additional tests were conducted to evaluate the capacity of sponges to capture NH₃. For this purpose, Petri dishes containing known amounts of ammonium chloride (NH₄Cl)_s were placed beneath the chambers containing the sampling sponges. Gaseous ammonia was generated by adding sodium hydroxide (NaOH)_l to the NH₄Cl and subsequently captured by the conditioned sorbents. Ammonium concentration in the 0.5 M KCl extracts was determined spectrophotometrically as described above. Results demonstrated complete ammonia uptake by the sponges across the range of concentrations assessed.

2.3.2 Soil nitrogen forms

Soil samples were collected in replicate around the NH₃ measuring chamber from the same plots where ammonia emissions were monitored, at 2 and 7 days after fertilization (DAF) and at 0–7 cm and 7–15 cm of depth. Fresh aliquots of these soils were used to quantify urea, exchangeable ammonium (NH₄ ⁺), and nitrate (NO₃ ⁻) contents in order to assess soil N incorporation and stratification following fertilizer application. The fresh soil samples were manually homogenized and coarse fragments were removed. Inorganic N forms (NH₄ ⁺ and NO₃ ⁻) were extracted using 1 M KCl, while deionized water was used for the extraction of water-soluble urea. Extractions were performed by shaking the samples for 60 min, followed by centrifugation at 3,000 rpm for 6 min and filtration of the supernatant through a <0.45 μm filter. Soil moisture content was determined gravimetrically for each sample analyzed.

Ammonium (N-NH₄ ⁺) in the extracts was determined as described in the modified Berthelot method (Crooke and Simpson, 1971), while nitrate (N-NO₃ ⁻) concentration in the same extracts was determined following -Miranda et al. (2001). Urea content was determined using a modified colorimetric method based on Douglas and Bremner (1970). A set of urea samples at different concentrations has been prepared and evaluated for the colorimetric response. Afterwards, by combining solutions from Mulvaney and Bremner (1978) and Greenan et al. (1995) method adjustments, a water bath at 87 °C for 55 min was used instead of the oven (120 °C for 30 min) for the heating of samples. Then the cooling took place in a thermostatically controlled bath at 15 °C instead of using running water.

2.3.3 Yield, yield components and agronomic traits

Crop yields, yield components, agronomic parameters and N contents at flowering stage and harvest were determined as a function of fertilization application under the two water regimes. Grain yield was determined by harvesting a 25 m² area within each plot and expressed at a standardized moisture content of 14%. Yield components, including panicle density (number of panicles m^-2), number of spikelets per panicle and sterility (%), were also determined. Panicle density was assessed using three replicated 0.25 m² sampling areas per plot. The number of spikelets per panicle and the sterility were determined from a sample of 20 panicles per plot. Lodging was visually assessed and expressed as the percentage of the plot surface affected. Milled rice yield was determined using a G390/R dehuller (Colombini and Co. Srl, Abbiategrasso, Milano, Italy) and a TM-05 grain testing mill (Satake Engineer-ing Co., Tokyo, Japan), operating within a milling range of 11.25%–12.75%. Broken kernels were separated by a rice length grader (TRG, Satake Engineering Co., Tokyo, Japan) equipped with cylinders appropriate for the Selenio variety. Milling quality was further evaluated by quantifying the proportion of unripe kernels. Leaf chlorophyll content was estimated using a chlorophyll meter (SPAD-502, Minolta Camera Co., Osaka, Japan). SPAD readings were acquired weekly on the Y leaf throughout the rice season until the maturation stage.

Nitrogen content in plant biomass at the flowering stage, as well as in the grain and straw at harvest, was determined by elemental analysis (UNICUBE®, Elementar Analysensysteme GmbH, Langenselbold, Germany). Apparent N recovery was calculated as the difference between aboveground plant N uptake in fertilized and control plots, divided by the amount of applied fertilizer N.

2.4 Statistical analyses

All the statistical analyses were carried out using R Studio (2024.12.1 + 563 version). A two-way analysis of variance (ANOVA) was performed with a mixed effect model using the nlme package (Pinheiro et al., 2007) on NH₃ emission and agronomic data, with Fertilizers (F) and WMF (water management at fertilization) as fixed factors, while the plot as random effect. Soil N data were subjected to a four-way ANOVA, with F, WMF, Time (T), and Depth (D) as fixed factors, while the plot as random effect of the mixed model. Data were log-transformed prior to analysis. Model diagnostics were performed through graphical inspection of residuals to assess normality, using Q–Q plots, whereas homogeneity of variance was assessed by plotting normalized residuals against fitted values. When minor deviations were observed, the robustness of the linear mixed model approach was considered adequate given the experimental design and the inclusion of random effects accounting for repeated measurements and block variability. Post-hoc comparisons were then performed using Bonferroni adjustment. Significant differences between the levels of the factors were indicated by different letter groups, with levels sharing the same letter not significantly differing from each other. A level of alpha = 0.05 (95% confidence level) was used to determine statistical significance. A principal component analysis (PCA) was also carried out to analyze agronomic data variability and identify the main relationships between plant parameters and F and WMF factors.

3 Results

3.1 Ammonia emissions

The cumulative NH₃ losses from different applied fertilizers under the two water regimes are presented in Figure 2. Urea application was always responsible for the highest NH₃ emissions, most of which occurred within the first hours following fertilizer application.

At tillering, when urea was applied on drained soils (DF), NH₃ emission rates were high but sharply decreased after submersion, resulting in cumulative emissions equal to 10% of the applied N. Under submerged conditions (WF), urea application caused slightly higher losses that persisted for a longer period, reaching 11% of the applied N. However, the differences in cumulative NH₃ emissions after urea application under WF or DF were not statistically significant (Supplementary Figure S1a). The presence of the urease inhibitor (NBPT) significantly (p < 0.01) reduced NH₃ emissions by approximately 72% compared to urea, but only under DF, with values only slightly higher than those observed with ammonium sulphate and control. Conversely, when fertilization occurred on submerged soil, the inhibitor effect of NBPT was not observed, resulting in NH₃ losses statistically comparable to those observed with conventional urea. AS was comparable to the control even under WF condition.

During the second topdressing fertilization at the panicle differentiation stage, NH₃ emissions increased markedly compared to those observed at the tillering stage. Among all treatments, the highest losses under the drained soil conditions were observed with urea, reaching 30% of the applied N. As in the first fertilization, emissions sharply decreased after submersion, and the presence of the urease inhibitor significantly reduced NH₃ losses similar to those observed with ammonium sulfate. No significant effects in cumulative emissions between urea and urea + NBPT treatment were observed when fertilization was conducted under submerged conditions (WF) (Supplementary Figure S1b). Ammonium sulphate resulted in mean NH₃ losses of 4 kg N ha^-1 under flooded conditions and 3 kg N ha⁻¹ under DF that were not significantly different from the unfertilized control. The lowest values were recorded in control treatment, confirming that N losses were minimal in the absence of fertilization.

3.2 Soil nitrogen forms

Soil urea contents (Figure 3) in the 0–7 cm layer of urea + NBPT fertilized plots (both DF and WF) were significantly higher than non-treated urea fertilized plots 2 days after N fertilization at tillering, irrespective of water management. Across all treatments concentrations decreased significantly 7 days after fertilization. Statistical analysis revealed significant effects of both time (p < 0.0001) and type of fertilizer (p < 0.0001), as well as a significant F × T interaction (p = 0.0001). In the 7–15 cm layer no significant difference was detected except for the urea content in urea WF fertilized plot at 2 DAF that was significantly higher than control. This significant difference disappears at 7 DAF. At panicle differentiation urea contents were low in all treatments and depths. In the 0–7 cm layer the NBPT effect is significant (p < 0.0001) at 2 DAF for DF. In the 7–15 cm layer no significant difference was detected except from the control plot.

The soil NH₄ ⁺ content (Figure 4) was affected by fertilizer performance, showing variations not only with depth and water management at the time of fertilization, but also with time. Statistical analysis revealed significant effects on NH₄ ⁺ concentration from the interaction of soil depth, water management, fertilizer type, and time. In general, NH4+ content was higher in the 0–7 cm layer than in the deeper one. At tillering, the highest mean NH₄ ⁺ content was observed 2 DAF with urea + NBPT in the 0–7 cm soil layer under dry conditions (44.7 ± 5.7 mg N kg^-1). This value showed a significant difference from both urea and AS fertilized plots. As expected, the lowest values were associated with the control plots, particularly at 0–7 cm across for both DF and WF conditions and time. When urea was applied, NH₄ ⁺ content was higher in WF plots compared to DF ones at 0–7 cm, 2 DAF. Meanwhile, AS showed an unusual temporal pattern in NH₄ ⁺ dynamics at 0–7 cm, especially under WF, where it was the only fertilizer that caused an increase in NH₄ ⁺ content over time at the soil surface. A slight increase in NH₄ ⁺ was also observed in the 7–15 cm layer in all treatments, from 2 to 7 DAF. In the upper soil layer, significant differences in NO₃ ⁻ content (Supplementary Figure S2) were detected only for DF plots with AS at 2 DAF, which showed slightly higher concentrations than both the control and the other fertilizers. However, no statistically significant effects of time or water management were observed. In the 7–15 cm layer, NO₃ ⁻ concentrations were low and comparable across treatments, except for the WF control at 7 DAF.

At panicle differentiation, in the 0–7 cm layer NH₄ ⁺ content was similar for all treatments with no significant differences between them, but just from the control plots, both in DF and WF management (Figure 4c). A marked decrease in NH₄ ⁺ content was observed in the topsoil layer from 2 to 7 DAF, with values approaching those of the control. NH₄ ⁺ concentrations in the 7–15 cm layer remained relatively low across all fertilizer treatments and were only slightly higher for the urea and urea + NBPT with respect to the AS and control in the drained soil at 2 DAF (Figure 4d). Slightly higher but significant soil NO₃ ⁻ contents were only detected 2 DAF in the topsoil layer (0–7 cm) of plots fertilized with urea (both DF and WF), while in all the other plots, NO₃ ⁻ contents were negligible and comparable to control plots. In the deeper layer NO₃ ⁻ concentrations remained low and comparable across all treatments.

3.3 Yield, yield components and agronomic traits

SPAD values did not differ among treatments at 28 DAS (day after seeding) under either drained or flooded WMF (Table 1), with no significant fertilization × water management interaction detected at this stage. At 61 DAS, SPAD values were higher in all fertilized plots than in the unfertilized control under both water management regimes, increasing in fertilized treatments with respect to the control plots. At this stage, SPAD responses were consistent across DF and WF, and no F × WMF interaction was observed. At 105 DAS, differences between water management regimes became evident, with fertilized plots maintaining higher SPAD values than the control under both DF and WF but showing contrasting absolute values between the two conditions. SPAD values ranged from 31.4 (urea) to 33.3 (AS) under DF and from 26.5 (urea) to 30.1 (AS) under WF, reflecting a significant fertilization × WMF interaction at the end of the cropping cycle.

TABLE 1. Chlorophyll content (SPAD) in the rice leaves at different times during the cropping season as affected by water management at fertilizer topdressing application in relation to the four N fertilizers, and significance of the different analyzed effects. Means followed by different letters are significantly different at the 0.05 probability level according to the Bonferroni test.

Abbreviations: ns: not significant; WMF: water management at fertilization; DF: drained soil; WF: flooded soil; N = nitrogen; NBPT: N-(n-butyl) thiophosphoric triamide; AS: Ammonium sulphate; DAS: day after seeding.

Rice yields and associated traits showed N treatment-dependent responses, with fertilized treatments consistently differing from the unfertilized control. However, these patterns varied between drained and flooded WMF. Under DF, grain yield increased from the unfertilized control to higher values in all fertilized treatments, with urea + NBPT (10.1 t ha^-1) exceeding urea (9.6 t ha^-1) and reaching values comparable to AS (10.2 t ha^-1) (Table 2). Straw biomass showed a similar pattern, increasing from 5.7 t ha^-1 in the control to 9.2 t ha^-1 with urea and to even higher values with urea + NBPT and AS. Under WF, grain yield increased from the unfertilized control with respect to plots fertilized with urea and urea + NBPT, while still higher values were recorded with AS. Straw biomass showed comparable values for urea and urea + NBPT, which were lower than those observed with AS. Yield components responded consistently to N fertilization. Panicle density increased from 432 (DF) and 490 (WF) m^-2 in the unfertilized control to values above 600 m^-2 in all fertilized treatments under both WMF. Both under DF and WF, not significant differences among fertilized treatments were observed. Spikelets per panicle increased from the control to fertilized plots, with no significant differences among N sources under either WMF.

TABLE 2. Performance of the two soil conditions at fertilizer topdressing application (DF: on drained soil; WF: on flooded soil) in relation with the four N fertilizers in terms of grain yield, straw biomass, and yield components (panicle density and spikelets). Means followed by different letters are significantly different at the 0.05 probability level according to the Bonferroni test.

Abbreviations: ns: not significant; WMF: water management at fertilization; N = nitrogen; NBPT: N-(n-butyl) thiophosphoric triamide; AS: Ammonium sulphate.

Reproductive traits showed trait-specific responses to fertilization and WMF (Table 3). For DF and WF, spikelet sterility was significantly higher in all fertilized treatments than in the unfertilized control. The percentage of immature kernels shows no significant differences between treatments, but only a significant effect of WMF. Lodging showed the widest variability, ranging from 0% in the control to 68.8% with AS under DF; under WF, lodging remained low with urea and urea + NBPT and increased with AS (43.8%).

TABLE 3. Performance of the two soil conditions at fertilizer topdressing application (DF = on drained soil; WF = on flooded soil) in relation with the four N fertilizers in terms of reproductive and lodging traits. Means followed by different letters are significantly different at the 0.05 probability level according to the Bonferroni test.

Abbreviations: ns: not significant; WMF: water management at fertilization; N = nitrogen; NBPT: N-(n-butyl) thiophosphoric triamide; AS: Ammonium sulphate.

PCA analysis of yield and yield components showed that the first two principal components explained 65.2% of the total variance (Figure 5). Grain yield, straw biomass, panicle density, and spikelets per panicle loaded mainly on the negative side of the first component (Dim1), whereas SPAD values loaded positively on the second component (Dim2). Sterility, unripe kernels, and lodging were located in the upper-left region of the biplot. Fertilized treatments were separated from the unfertilized control mainly along Dim1, while differences between DF and WF water management were expressed along Dim2.

3.4 Nitrogen uptake and apparent recovery

Nitrogen uptake by rice plants, measured at flowering and harvest in both grain and straw, was significantly affected by fertilizers (p < 0.01), whereas soil conditions during fertilization showed no influence on most variables (Table 4). At flowering, the highest N uptake was recorded with AS under DF (151 kg N ha^-1), although this value did not differ significantly from those of the other fertilized treatments under DF or from AS applied under WF. All fertilized plots showed higher N uptake than the unfertilized control, but no significant differences emerged among fertilizers in terms of N concentration (%N) at flowering.

TABLE 4. Performance of the two soil conditions at fertilizer topdressing application (DF = on drained soil; WF = on flooded soil) in relation with the crop management without or with the three levels of N fertilization in terms of the flowering biomass, grain and straw N contents, and the apparent N recovery. Means followed by different letters are significantly different at the 0.05 probability level according to the Bonferroni test.

Abbreviations: ns: not significant; WMF: water management at fertilization; N = nitrogen; NBPT: N-(n-butyl) thiophosphoric triamide; AS: Ammonium sulphate.

At harvest, grain N uptake responded strongly to fertilizer type (p < 0.01). The highest values were obtained under DF with urea + NBPT and AS (114 and 115 kg N ha^-1, respectively), which were significantly greater than those observed for urea under both soil conditions and for urea + NBPT under WF. However, when expressed as grain N concentration (%N), these differences were no longer significant, indicating that variations in N uptake reflected differences in grain yield rather than shifts in grain N content. Two-way ANOVA confirmed a significant effect of fertilizer type on grain N%, and a significant interaction WMF x F (p < 0.01), suggesting that fertilizer-driven differences in N accumulation were only modestly modulated by WMF. Consistent with this pattern, water management exerted no significant effect on most N-related variables, except for apparent N recovery (p < 0.01), which was markedly higher under DF, reaching up to 86% with AS.

Straw N content showed the lowest absolute values, with limited variation among fertilizers and no significant effect of soil condition. Nonetheless, straw N uptake followed the same hierarchy observed in grain, with fertilized treatments performing similarly within each soil condition and consistently exceeding the control.

4 Discussion

4.1 Ammonia emissions

Ammonia volatilization represents a crucial pathway of N loss in rice systems. As widely reported in the literature, fertilization consistently increases NH₃ emissions compared with unfertilized conditions (Larsson et al., 1998; Liu et al., 2015; Chuong et al., 2020). In these control plots, the small amount of volatilized NH₃ likely originated from mineralized N released through microbial decomposition of soil organic matter (Beare et al., 1994). Across treatments, emissions followed a rapid and transient pattern, with peaks occurring shortly after fertilization, as consistently reported for N fertilizers in paddy soils (Tian et al., 2001; Hayashi et al., 2006; Chen et al., 2015; Lu et al., 2022; Dillon et al., 2012; Chuong et al., 2020). This indicates that the first days after application represent a critical window for N loss, although the magnitude of emissions differed among fertilizers. Urea consistently caused the highest volatilization, due to the localized increase of pH associated with its hydrolysis, which enhances the conversion of NH₄ ⁺ to gaseous NH₃ even under acidic soil conditions (pH 5.5) (Overrein and Moe, 1967; Abalos et al., 2014).

4.1.1 WMF impact on ammonia emissions

Water management at fertilization markedly influenced the volatilization kinetics. Under dry conditions, losses were concentrated within a short time frame, whereas the presence of a water layer delayed and prolonged emissions. This indicates that water acts not only as a physical barrier to gas diffusion and a regulator of temperature and solute distribution, but also indirectly affects the chemical conditions controlling ammonia volatilization (Abalos et al., 2012; Ni et al., 2014; Pan et al., 2016; Drame et al., 2023). In particular, flooded conditions are generally associated with an increase in soil pH due to reductive processes, whereas drained conditions may promote partial re-acidification. These shifts can alter the equilibrium between NH₄ ⁺ and NH₃, potentially enhancing volatilization under wetter conditions and limiting it under drier ones (Ding et al., 2019; Rupngam and Messiga, 2024). Despite these contrasting dynamics, cumulative emissions remained comparable, indicating that water regime primarily affects short-term kinetics rather than the overall magnitude of NH₃ emissions from urea.

Conversely, substantial differences were observed with the application of urease inhibitor. As widely demonstrated (Cantarella et al., 2018 and references therein), NBPT strongly decreased NH₃ volatilization. In our experiment, its mitigation effect was particularly pronounced compared with previous reports (Silva et al., 2017). However, its performance was strongly dependent on soil moisture at fertilization, since the inhibitor was effective only under dry conditions, while its mitigation effect was largely suppressed in flooded soils. The high abating effect of NBPT may reflect conditions known to favor NBPT performance, such as the mild climate of temperate areas, soils with low C content (Abalos et al., 2012) and pH close to neutrality (Engel et al., 2015; Silva et al., 2017). In contrast, acidic soils accelerate NBPT degradation and reduce its efficacy (Hendrickson and Douglass, 1993; Engel et al., 2013; 2015; Cantarella et al., 2018; Lasisi and Akinremi, 2022).

In accordance with the early study of Cai et al. (1989), application in water limited NBPT activity, resulting in NH₃ losses comparable to urea alone. This corroborates the fact that the soil surface must be sufficiently dry for NBPT to be oxidized into its active form, as largely demonstrated in the literature (McCarty et al., 1989; McCarty and Bremner, 1989; Christianson et al., 1990; Creason et al., 1990; Hendrickson and Douglass, 1993; Keerthisinghe and Blakeley, 1995; Sigurdarson et al., 2018). Recent work by Peters and Thiele-Bruhn (2022) further reported that NBPT degrades rapidly in soil, with degradation active products such as NBPTO, NBPD (N-(n-butyl)thiophosphoric diamide), DAP (diamido phosphoric acid), and DATP (diamido thiophosphoric acid) suggesting the need of synchronization between fertilizer application and O₂ availability in rice paddies. The dry fertilization strategy adopted in this study, involving a short drying period before fertilizer application followed by prompt reflooding, optimized the effectiveness of NBPT. Our findings demonstrate that the benefit of short-term drainage extends beyond the direct effect of water management on NH₃ volatilization. Under temperate field conditions, the temporary increase in soil aeration before fertilization likely provided sufficient oxygen for the activation of NBPT, while the moderate temperatures limited its degradation. This combination of climatic and management factors may explain the marked reduction in NH₃ emissions observed under drained conditions and highlights the importance of considering both water management and climatic context when optimizing the use of urease inhibitors in temperate rice systems. The initial dry conditions favored inhibitor activation, while subsequent flooding helped limit NH₃ losses by reducing gas diffusion and redistributing N available forms within the soil.

4.1.2 Environmental factors and soil condition impact on ammonia emissions

The second fertilization event (panicle differentiation) showed patterns consistent with those observed at tillering, but with overall higher emission intensity. This increase can be attributed to the higher temperatures occurring during this phenological stage, which are known to enhance ammonia volatilization rates (Sommer et al., 2004). In addition, prolonged flooding between fertilizations plausible increased soil pH, maintaining conditions favorable for urea hydrolysis and NH₃ emissions (Fan et al., 2022; Lu et al., 2022). Under these conditions, the presence of standing water partially mitigated peak emissions, probably through its buffering effect on soil temperature.

The effectiveness of NBPT was again evident but remained strongly dependent on soil conditions. Its mitigation effect was observed under drained conditions, whereas it was reduced under flooding, due to temperature-driven changes in inhibitor stability and performance (Fan et al., 2022; Soares and Cantarella, 2023). Despite this temperature constraint, cumulative 14-day emissions were lower in dry than in flooded treatments, indicating that well-drained, oxidized conditions are essential for increasing the potential of NBPT to mitigate ammonia losses, regardless of temperature effects (Sanz-Cobena et al., 2011; Xu et al., 2012; Sanz-Cobena et al., 2016). This suggests that proper water management at fertilization is a key factor modulating the abatement achieved with NBPT.

4.1.3 AS: a different case

Ammonium sulphate showed a more stable mitigation effect, largely independent of water management, and consistently reduced NH₃ volatilization. This behavior is related to its high solubility and its direct supply of nitrogen in NH₄ ⁺ form, without undergoing urea hydrolysis and without increasing soil pH. In addition, its slight acidifying effect helps maintain N in the NH₄ ⁺ form, thereby limiting the conversion to NH₃ and reducing volatilization losses. Nevertheless, its practical adoption may be constrained by economic costs (Fryer et al., 2019).

4.2 Soil nitrogen forms

Urea applied to the soil surface was rapidly hydrolyzed, confirming its low persistence in paddy soils, as widely reported in previous studies (Dawar et al., 2011; Daigh et al., 2014; Savin et al., 2021). This rapid transformation explains the transient nature of urea in soil and the strong temporal dynamics observed, with N quickly transitioning from urea to ammonium and subsequently being subject to plant uptake, volatilization, or retained in the soil. Despite the high solubility of urea, its vertical movement remained limited, indicating that most transformations occurred in the upper soil layer where fertilizer was applied, as also reported by Dawar et al. (2011).

In contrast, the application of urea + NBPT effectively delayed urea hydrolysis, resulting in the greatest amounts of soluble urea remaining in the upper soil layer. NBPT-treated plots showed a delayed accumulation of ammonium compared with untreated urea, reflecting the slower hydrolysis of urea in the presence of the inhibitor. This pattern is consistent with previous studies indicating that ammonium formation occurs rapidly following urea application, whereas it is temporally shifted when NBPT is applied (Dawar et al., 2011; Sanz-Cobena et al., 2011). The more gradual release of ammonium contributes to reduced ammonia emissions in the initial stages after fertilization. Over time, ammonium concentrations declined and decreased with soil depth, reflecting the combined effects of plant uptake and N losses, as also reported by Savin et al. (2021). These dynamics became more pronounced at later growth stages when crop N demand increases, and volatilization losses are higher.

These findings align with earlier studies in which rapid urea hydrolysis promoted substantial NH₃ volatilization in surface-applied, especially when flood establishment was delayed (Norman et al., 2009; Dillon et al., 2012). Ammonium distribution in the soil profile reflected the progressive transformation of applied nitrogen and its subsequent redistribution, with similar vertical patterns observed for both urea + NBPT and urea despite differences in hydrolysis rates (Savin et al., 2021). Water management influenced these dynamics through its effect on inhibitor performance. Under flooded conditions, lower ammonium retention in the surface layer was associated with higher NH₃ emissions, supporting the reduced effectiveness of NBPT in water. Over time, differences between treatments diminished, indicating the transient nature of urease inhibition, as also highlighted by previous studies (Fan et al., 2022).

Ammonium sulphate provided a more stable source of ammonium, maintaining higher NH₄ ⁺ levels in the topsoil and reducing volatilization losses, while nitrate concentrations remained consistently low, confirming the limited role of nitrification under these conditions (Motasim et al., 2024).

Fertilizer type and time after application were the main factors controlling soil N forms, whereas water management at fertilization had a limited direct effect on soil urea content. At later growth stages, differences among treatments became less pronounced due to increased temperature and crop N demand, which accelerated N turnover. Higher temperatures enhanced urea hydrolysis and reduced inhibitor effectiveness (Moyo et al., 1989; Fan et al., 2022; Lu et al., 2022), while increased plant uptake and volatilization losses limited ammonium accumulation. As a result, soil N availability became more transient, and the expected increase in ammonium was not observed, as N was rapidly either lost or assimilated by plants during this critical reproductive stage (Fageria, 2003; Fageria and Baligar, 2005; Yu et al., 2013; Fageria et al., 2015).

4.3 Alternative N loss

Although the present study focused on NH₃ volatilization, changes in water management around fertilization may also influence other nitrogen loss pathways. Short-term drainage can temporarily increase soil aeration, potentially promoting nitrification and subsequent nitrate formation. Following reflooding, the accumulated nitrate may become available for denitrification, increasing the risk of N₂O emissions under favorable environmental conditions. Likewise, nitrate formed during the drained phase could be susceptible to leaching if intense rainfall or irrigation events occur before crop uptake. However, under the conditions of this study, nitrate concentrations remained consistently low, suggesting that these alternative N loss pathways were limited. Future studies simultaneously quantifying NH₃ volatilization, N₂O emissions and nitrate leaching would provide a more comprehensive evaluation of nitrogen management strategies in temperate rice systems.

4.4 Agronomic performance, nitrogen uptake, and apparent recovery

Water management at fertilization played a decisive role in determining the agronomic effectiveness of urea-based fertilization, particularly when urea was combined with NBPT. Treatments with higher apparent N recovery were all DF and were also characterized by greater plant N accumulation both at flowering and at harvest, supporting the hypothesis that improved nitrogen conservation enhanced fertilizer utilization by the crop. As widely reported, the agronomic benefits of N fertilization strongly depend on the effectiveness of mitigating early N losses following application (Li et al., 2022; Chen et al., 2023; Islam et al., 2022). The higher apparent N recovery observed under DF suggests that short-term drainage before fertilizer application created more favorable conditions for NBPT to act, allowing a better synchronization between fertilizer N availability and crop demand. Under these conditions, the delayed urea hydrolysis induced by NBPT reduced early N losses and increased the fraction of applied N retained in the soil–plant system. This resulted in higher grain yield, straw biomass and grain N uptake compared with untreated urea. In this context, the agronomic advantage of NBPT was mainly related to improved N retention and availability during key growth stages, promoting biomass accumulation rather than changes in tissue N concentration (Zhou et al., 2023; Hu et al., 2025). Conversely, when urea + NBPT was applied under flooded conditions, its effectiveness was largely reduced, and there were no clear differences in yield, biomass production and N uptake were observed compared with conventional urea. This indicates that, under WF, the expected improvement in N use efficiency was not achieved, potentially due to higher early N losses. As also reported, when N losses are not effectively mitigated, improvements in crop performance remain limited despite similar N inputs (Zhou et al., 2022a). At harvest, differences in grain N uptake closely reflected yield responses, indicating that enhanced N recovery was primarily translated into grain production rather than into increased grain N concentration (Padhan et al., 2023; Sung et al., 2023; Zi et al., 2025). Therefore, the effectiveness of NBPT in rice systems appears to be strongly dependent on the soil conditions at the time of application. These findings indicate that combining urea + NBPT with short-term drainage before application, followed by timely reflooding, represents a key strategy to enhance apparent N recovery and sustain rice productivity in temperate paddy systems.

From a practical perspective, the short-term drainage strategy evaluated in this study appears technically feasible in temperate rice-growing areas where irrigation infrastructure already allows water levels to be regulated before and after fertilizer application. Since drainage was limited to a few days and was immediately followed by reflooding, the additional water management required is expected to remain operationally manageable where irrigation systems allow flexible regulation of water levels. Nevertheless, adoption of this practice may depend on farm-specific factors, including irrigation scheduling, labor availability, water resources during the growing season and the operational costs associated with additional water management. Although an economic assessment was beyond the scope of this study, the reduction in NH₃ losses together with the improved apparent N recovery and grain yield observed under drained conditions suggest that this strategy has the potential to improve fertilizer use efficiency under commercial rice production. Future studies integrating agronomic, environmental, and economic assessments would help quantify its overall feasibility and cost-effectiveness.

5 Conclusion

The present study provided evidence that in water seeded temperate rice cropping systems, ammonia volatilization from urea + NBPT is strongly modulated by soil conditions at fertilization, particularly through the effects of water management on soil aeration and temperature. While urea consistently produced the highest NH₃ emissions, with rapid emission peaks occurring shortly after application, NBPT effectively reduced ammonia volatilization only under water management regimes that included rapid soil drying prior to fertilization followed by prompt submergence, and when ambient temperatures did not exceed the buffering capacity of ponding water. Ammonia emission reduction was reflected in higher rice N uptake and productivity. Importantly, the consistency between emission patterns, soil N dynamics and agronomic responses show that yield differences among treatments were not merely the result of fertilizer type but rather driven by the interaction between fertilizer type and water management at fertilization.

Overall, these results demonstrate that, under the environmental conditions of this study, optimizing water management at fertilization represents a key strategy for improving NUE, reducing NH₃ losses and sustaining rice productivity. Although the findings provide strong evidence for the benefits of combining urea + NBPT with short-term drainage before fertilization in temperate water-seeded rice systems, they are based on a single growing season and a single experimental site. Therefore, further multi-year and multi-location studies are needed to validate the robustness of these results under different climatic, soil and management conditions. Nevertheless, the mechanisms identified in this study highlight the potential of integrating optimized water management and fertilizer strategies into agronomic practices to support more sustainable temperate rice production.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvs.2026.1911870/full#supplementary-material

Boxplot of cumulative NH₃ emissions (kg N ha^-1) during the 14 days following fertilizer application for the different treatments at tillering, a) (DF, left and WF, right side), and panicle differentiation, b) (DF, left and WF, right side). Treatments consisted of unfertilized control, urea, urea + NBPT, and ammonium sulphate fertilizer applied. Different letters in graphs denote significant differences according to two-way ANOVA followed by Bonferroni post hoc test (p < 0.05). Error bars represent standard deviations. Abbreviations: NBPT: N-(n-butyl) thiophosphoric triamide; AS: Ammonium sulphate; DF: drained soil; WF: flooded soil. Supplementary material at the source

N-NO₃ ^- concentration (mg N kg^-1 of fresh soil) at soil depth 0–7 cm: a) and c) and 7–15 cm: b) and d); at 2 (black) and 7 (grey) days after fertilization (DAF), for both tillering: a) and b) and panicle differentiation: c) and d) topdressing fertilization events (DF, left and WF, right side). Treatments consisted of unfertilized control, urea, urea + NBPT, and ammonium sulphate fertilizer applied. Different letters in graphs denote significant differences according to four-way ANOVA followed by Bonferroni post hoc test (p < 0.05). Error bars represent standard deviations. Abbreviations: NBPT: N-(n-butyl) thiophosphoric triamide; AS: Ammonium sulphate; DF: drained soil; WF: flooded soil. Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Isabeli Pereira Bruno, Instituto de Desenvolvimento Rural do Paraná (IDR), Brazil

Reviewed by: Muhammad Akhtar, Institute of Crop Sciences (CAAS), China

Prabu Padanillay Chidambaram, Tamil Nadu Agricultural University, India