The study was carried out on the Soil Science Field Laboratory (24° 71.59′ N, 90° 42.50′ E) of Bangladesh Agricultural University (BAU) in Mymensingh, Bangladesh. The experiment was done with an annual irrigated rice– pre-monsoon rice – monsoon rice cropping sequence, which is a common cropping sequence followed by the farmers of this country. The irrigated rice season, pre- monsoon rice season and monsoon rice season were occupied by Boro, Transplanted Aus (T. Aus), and Transplanted Aman (T. Aman) rice growing seasons, respectively. The field site was characterized as a Non-calcareous Dark Gray Floodplain soil (Aeric Haplaquept in US Soil Taxonomy), and belongs to agro-ecological zone-9, Old Brahmaputra Floodplain soil (FAO/UNDP, 1988). The soil is moderately drained with a silt loam texture and near neutral pH (6.5). The region has a sub-tropical monsoon climate with a mean annual temperature of 26°C, average annual rainfall of 1,800 mm, and relative humidity of 85% (Uddin et al., 2021, Supplementary Data 1).

The experiment was conducted with the same treatment combinations for Boro – T. Aus – T. Aman rice crops in sequence, but with different levels of a nutrient based on the requirements of individual crop and their target yields. That experiment was conducted under integrated plant nutrition system (IPNS; nutrient contents in organic amendments were subtracted from the full recommended fertilizer dose, i.e., RD of chemical fertilizers) where six treatments with four replications were applied in each season. The treatments were: (T₁) no fertilizer (control), (T₂) RD, (T₃) poultry manure biochar + decreased dose of recommended fertilizer (DRD), (T₄) rice husk ash + DRD, (T₅) compost + DRD, and (T₆) compost + biochar + DRD, laid out in a randomized complete block design (RCBD) with four replications. Total plot number was twenty-four for each season and the same plots were used for consecutive rice growing seasons and the unit plot size was 5 m × 4 m, with a 0.75 m inter-plot space, and 1 m inter-block space. The varieties were BRRI dhan28 for Boro, BINA dhan19 for T. Aus and BRRI dhan49 for T. Aman rice, respectively. Boro rice was grown during January–April (winter season), followed by T. Aus rice as a rainfed crop from May to August (pre-monsoon), and then T. Aman rice from August to November (monsoon).

The rate of chemical fertilizer application was based on Fertilizer Recommendation Guide (FRG, 2018) for the test crops. The nutrient contents of used organic amendments are presented in Table 1 while the recommended doses of nutrients for three seasons were presented in Table 2. Urea, triple super phosphate, muriate of potash, gypsum and zinc sulfate were used for N, phosphorus (P), potassium (K), sulfur (S) and zinc (Zn) sources, respectively. Except urea all the nutrients were applied during land preparation. For both Boro and T. Aman rice nitrogenous fertilizer (urea) was applied in equal three splits, followed interval for Boro rice was at 10, 31, and 53 Days After Transplanting (DAT) and for T. Aman that interval was at 9, 24, and 39 DAT. In T. Aus rice two splits of urea application were followed, at 11 and 29 DAT. Compost was collected from Mazim Agro Industries Ltd and rice husk ash from a local rice mill. Biochar was produced using poultry manure by an anaerobic pyrolysis process at 450°C for 4 hr. The organic materials were air dried to 15% moisture content, pulverized and sieved with a 2 mm mesh. In T₃, T₄, and T₅ treatments organic materials were applied at the rate of 3 t ha⁻¹ where T₆ was balanced by applying 1.5 t ha⁻¹ compost and 1.5 t ha⁻¹ biochar, the remaining nutrients were applied from chemical fertilizer based on FRG under IPNS approach. Glyphosate (Round up^®) was sprayed over the field at a rate of 1.85 kg ha⁻¹ 3 days before final land preparation. The field was irrigated to maintain 3 cm standing water throughout rice growing seasons.

Field measurements of NH₃ were conducted during January 2020–November 2021 in the rice field. A low-cost chamber was deployed in field conditions for NH₃ volatilization measurements (Nichols et al., 2018) and used for monitoring NH₃ fluxes in crop fields (Martins et al., 2021a,b; Zaman et al., 2021). The open chamber method was used to measure NH₃ fluxes in the field site on a daily basis. In the laboratory, the amount of NH₃ trapped in acid solution was estimated using the Kjeldahl principle (Keeney and Nelson, 1982). Measurements were done on the soil shortly after urea application and it was carried out until the fluxes were below the detection limit in each case.

The NH₃ fluxes were calculated following Equation 1.

Where, NH₃ flux was measured as mg N m⁻² d⁻¹; FBR, Final Burette Reading (ml); IBR, Initial Burette Reading (ml); molecular weight of N = 14.01 g; normality of H₂SO₄ = 0.01 N; and 1000 = unit conversion factor. The sum of NH₃ fluxes on sampling days across the whole sampling period was used to estimate cumulative NH₃ fluxes.

We derived EF (%) according to Equation 2 (Mazzetto et al., 2020).

Where EF (%) = Emission Factor, in%; Fluxes FT, Fluxes from fertilizer treatment (in kg N ha⁻¹); Fluxes C, Fluxes from control treatment (in kg N ha⁻¹); Applied Fert, Amount of fertilizer applied (in kg N ha⁻¹).

Yield-scaled NH₃ fluxes were determined following the Equation 3.

Before the final harvest of each rice growing season, plants from 1 m² area were collected from each plot, weighed and then oven dried to determine yield and system productivity i.e., pooling together the grain yields of the three rice seasons. For oven drying, 1000 grain samples of each plot were placed in an oven at 65°C until it reached constant weight to determine moisture content. After drying, rice grain samples were weighed and yields were estimated as tonne per hectare.

Composite soil samples were collected with an auger at 0–15 cm soil depth from the sites next to each NH₃ gas sampling chamber and preserved in sealable plastic bags in a cooler box. The field-moist soil was air-dried for 2 weeks in the shade at room temperature (25°C) and processed (2 mm sieved) for analysis of major soil physico-chemical parameters. During the NH₃ loss measurement, the pH of the soil was monitored in the field every seven days using a portable pH meter (HI12923; Hanna Instruments). The Kjeldahl method was used to determine total nitrogen (TN) content in the soil (Fawcett, 1954) and the wet oxidation method (Walkley and Black, 1934) was used for soil organic carbon (SOC) determination. Soil samples were extracted with 2 M KCl (1: 2.5; w/w) and NH4+ and NO3- contents were measured using the method described by Keeney and Nelson (1982).

One-way ANOVA was performed using treatments as fixed factors. The normality test on the NH₃ data was checked before analysis. Post-hoc tests were performed to separate differences among the treatments using the Tukey-Kramer multiple comparison Test. All statistical analyses were considered significant at p ≤ 0.05, unless otherwise mentioned. All the statistical analyses were performed on Statistics 10 and Jamovi1.0.0.0 (R Package). Correlation among the parameters studied was tested by using Pearson's correlation coefficient comparison test.

Ammonia fluxes reached their peaks at 2–3 days after each split of urea application in all seasons. The highest NH₃ fluxes were recorded during the second split application of urea in boro season but that was higher from first split application in both T. Aus and T. Aman seasons (Figure 1). In T. Aman rice, the NH₃ fluxes were 1.5–2.0 times higher in the first split compared to the second and third splits, while the latter two results were almost the same. The NH₃ flux peaks returned to background level at 7–10 days after each split urea application (Figure 1). The highest peak in all splits at each season ranked in the order of chemical fertilizer > RMA > compost > compost + biochar > biochar > control (Figure 1B). On the peak period at 1st, 2nd, and 3rd splits of urea fertilization the NH₃ fluxes from RD treatment were 117, 305, 160 mg N m⁻²d⁻¹ in Boro season and that were 193, 165 mg N m⁻²d⁻¹ in T. Aus and 289, 138 and 136 mg N m⁻²d⁻¹ in T. Aman season, respectively.

The effect of organic and inorganic fertilizers on mean and cumulative ammonia fluxes of all three rice crops was significant. In Boro rice, the highest mean and cumulative NH₃ fluxes were observed in chemical fertilizer treated plot, which was statistically similar to RMA, and the lowest emission was observed in control (Table 3). Integrated use of organic and inorganic fertilizers reduced NH₃ emissions by 6–23% compared to the RD treatment. Either biochar or biochar plus compost reduced N loss via volatilization by 16–23%, while compost alone reduced it by 13%. Likewise, organic and inorganic fertilization also significantly influenced mean and cumulative NH₃ fluxes in T. Aus rice. Mean and cumulative NH₃ fluxes were higher in RD than in other treatments. Reduction in NH₃ fluxes ranged from 10% in rice husk ash to 52% in biochar. Disregarding the control, the highest mean and cumulative NH₃ fluxes were measured in chemical fertilizer treated plots, whereas the lowest emissions were measured in compost plus biochar treated plots. Combined application of organic and inorganic fertilizers reduced NH₃ volatilization by 20–45% compared to the RD application. Pooling the three rice growing seasons together, treatments comprised of biochar or biochar plus compost under IPNS basis reduced N loss via volatilization by 36–37% while biochar alone reduced it by 23% over sole application of full dose of recommended fertilizer as a treatment.

The NH₃ emission factor (EF) was significantly influenced by the application of organic and inorganic fertilizers. The NH₃ EF ranged from 12% in compost + biochar to 16% in chemical fertilizer-treated plots in Boro rice, from 21% in biochar to 29% in sole chemical fertilizer treated plots in T. Aus rice, and from 22% in biochar to 28% in chemical fertilizer-treated plots in T. Aman rice (Table 4). Yield-scaled NH₃ emissions in Boro rice varied from 0.17 kg t⁻¹ in control to 2.88 kg t⁻¹ in chemical fertilizer treated plots (Table 4). Except for the control treatment, yield scaled NH₃ emissions were similar among treatments in Boro rice. Mixture of biochar and compost reduced the NH₃ EF and yield-scaled NH₃ emission in all rice fields. Similarly, yield-scaled NH₃ emissions in T. Aus rice varied between 1.04 kg t⁻¹ in control and 5.83 kg t⁻¹ in chemical fertilizer treated plots (Table 4). Ignoring the control treatment, the highest yield-scaled NH₃ emissions were noted in chemical fertilizer, which was similar to RMA, and the lowest value was in biochar with or without compost treatment. Yield-scaled NH₃ emissions in T. Aus rice were 1 to 6 times higher than that in Boro rice. Similarly, yield-scaled NH₃ emissions in T. Aman rice ranged from 0.47 kg t⁻¹ in control to 4.43 kg t⁻¹ in chemical fertilizer-treated plots (Table 4). Discounting the control treatment, the highest yield-scaled NH₃ emission was recorded in chemical fertilizer, and the lowest value was in biochar with or without compost. Yield-scaled NH₃ emissions in T. Aman rice were 1.0 to 2.7 times higher than that in Boro rice, and 0.5 to 1.0 times that of T. Aman rice (Table 4).

Organic and inorganic fertilizers influenced the grain yield of Boro, T. Aus, and T. Aman rice (p < 0.05, Table 5), and system productivity of Boro – T. Aus – T. Aman rice cropping pattern (p < 0.01, Table 5). All the treatments were similar to each other in term of crop yield except T₁. Treatments under RD or IPNS had no statistical variation for crop yield and system production. In Boro and T. Aman rice, grain yields were the highest for application of compost and likely the system productivity was the highest for compost application and the lowest for rice husk ash, excluding the control treatment.

Organic and inorganic fertilizers had a significant impact on soil organic carbon (SOC) during rice cultivation (p < 0.05, Table 6). Soil organic carbon increased by 6–14% over the control in plots treated with different amendments (Table 6). Biochar and RMA significantly increased soil total nitrogen (TN) content compared to the other treatments including control except in T. Aus season (Table 6). Likewise, organic and inorganic fertilizers significantly influenced soil C:N ratio only in Boro season but not in T. Aus and T. Aman seasons (p < 0.05, Table 6). The highest soil pH was measured in biochar-treated plots, which was similar to other treatments except for compost with biochar and control. Likewise, NH4+ concentrations in soil were significantly influenced by different organic and inorganic fertilizers (p < 0.001, Table 6). The highest NH4+ concentrations were found in plots treated with only chemical fertilizer and were lowest in control.

The relationship between the NH₃ fluxes and soil pH was positive and significant in all rice seasons (Figure 2). NH₃ fluxes had a strong correlation with soil pH in Boro rice (R² = 0.79; p < 0.01). Likewise, NH₃ fluxes had a moderate correlation with soil pH in T. Aus rice (R² = 0.36; p < 0.05) and in T. Aman rice (R² = 0.50; p < 0.05) (Figure 2). Like soil pH, the relationship between NH₃ fluxes and soil NH4+ content was also positive and significant in all rice seasons (Figure 2). Ammonia fluxes had a strong correlation with soil NH4+ content in Boro rice (R² = 0.68; p < 0.01), T. Aus rice (R² = 0.86; p < 0.001), and T. Aman rice (R² = 0.91; p < 0.001) (Figure 2).

The NH₃ flux peak was within 2–3 days after urea application indicating that NH₃ volatilization was a rapid progress that was almost completed within 1 week after each split fertilizer application. The NH₃ volatilization flux patterns were consistent among treatments, suggesting that they were primarily driven by the urea applied. The NH₃ emission patterns were consistent with previous studies in the same (Uddin et al., 2021) and dissimilar geographical areas (Fan et al., 2006) as our experiment. The NH₃ flux from urea hydrolysis usually peaks at 3–7 days after application (Rochette et al., 2009) which is in line with our results but not to Drury et al. (2017) who stated that the peak emissions can take up to 9–15 days if rain occurs after N application. That fluxes were highest from T₂ may be attributed to the highest rate of urea applied which rapidly converted into NH4+ through the ammonification process, which was the first step of ammonia volatilization (Frimpong et al., 2016; Uddin et al., 2021). As NH₃ is in a dynamic equilibrium with NH4+ and H⁺, urea treatment elevates soil pH through urease hydrolysis (Sommer et al., 2004). Following the peak on day 2–4 after urea application, the NH₃ fluxes rapidly declined. While organic amendments did not alter the timing of the peak of NH₃ fluxes, they decreased the magnitude of the peak which could be attributed to the lower rate of chemical N-fertilizer based on the IPNS approach. The decrease in soil NH4+ content and a drop in pH with the organic amendments helps explain the decrease in NH₃ volatilization (Adviento-Borbe et al., 2010). Other processes leading to a decrease in NH₃ volatilization could be the infiltration of mineral N into the crop rooting zone, and increased nitrification over time (Adviento-Borbe et al., 2010). When the NH₃ fluxes for Boro, Aus and Aman seasons were examined in relation to the soil chemical properties, the closest positive correlation was with soil pH followed by soil NH4+ as found in previous studies (Sommer et al., 2004; Rochette et al., 2013).

While NH₃ volatilization is a major N loss from paddy fields, the rate of N loss is dependent on the fertilization type, time of application, environmental conditions and N application rate (Wang et al., 2016). Pan et al. (2016) stated that about 30% of the applied urea was lost through NH₃ fluxes which were consistent with our result that the N loss via volatilization ranged from 16% in Boro to 28% in T. Aman rice season. When N supplied in the urea fertilizer was adjusted based on the N content in the organic amendments, NH₃ fluxes were reduced. In this study biochar alone and with compost reduced the NH₃ loss during three rice growing seasons. The NH₃ fluxes of N fertilizer was higher for Aman rice than for Aus and Boro rice which is most likely due to the seasonal variations in temperature being the lowest in Boro season (15–25°C) and the highest in T. Aman season (25–35°C) while in T. Aus the temperature was moderate (20–30°C). High temperature in standing water in rice fields induces rapid urea hydrolysis and higher ammonia volatilization (Sun et al., 2017). While the Aus season in the Indo-Gangetic plain has high rainfall and moderate temperature, the urea application rate in this season was lower than the other two seasons due to lower yield potential, which may lower volatilization.

Biochar was very effective in reducing NH₃ emissions by reducing chemical N input but may also control the N releases. Our results also showed consistency with Sun et al. (2017) and Asada et al. (2002), where their meta-analysis suggested that NH₃ fluxes were reduced with the application of biochar pyrolyzed at ~400°C. Ammonia adsorbed onto the biochar surface directly reduces the substrate concentration of NH₃ for the volatilization process (Clough et al., 2013). However, the liming effect of alkaline biochar may increase NH₃ fluxes (Sun et al., 2017; Sha et al., 2019). The pH increase in soil amended with biochar in the present study was not high enough to enhance NH₃ fluxes (Kelly et al., 2015). Among the amended plots biochar required the lowest rate of urea fertilizer to equalize total N input with recommended chemical fertilizer dose, which may explain the lower NH₃ fluxes than in compost amended plots. Co-application of biochar with compost has the potential to reduce NH₃ emissions due to high surface area to adsorb NH₄, high internal porosity to trap NH4+ ions but this will depend on N mineralization rate and their inherent N content which varies among biochar and compost products.

All the treatments, except control without N fertilizer applied, had the same yield in all three rice seasons even though the urea application rates were different. Moreover, the N uptake was also the same in each treatment (data not presented). Therefore, questions arise of how biochar-treated soils provided similar N for plant uptake in comparison to a full dose of urea. A moderate substitution (<40%) of N fertilizer by manure has been reported to significantly increase N use efficiency by 14 and 25% for upland crops and rice, respectively (Xu et al., 2016; Zhang et al., 2020). In the present study, rice plants were initially paler green in biochar-treated plots suggesting that it decreased initial N mineralization rate. In addition, N from urea in biochar-treated plots may have been used more efficiently due to better synchronization of N supply and demand.

Rice husk ash had less efficiency in NH4+ retention in all seasons and in controlling NH₃ fluxes than biochar, but still decreased N losses relative to the urea fertilizer alone. While ashes are often alkaline, the present RMA did not alter soil pH and was effective in decreasing NH₄ content in soil except in Aman season and in decreasing NH₃ losses, except in the Boro season. As an abundant biowaste in the Indo-Gangetic Plain, RMA can be used to reduce N fertilizer input and to reduce atmospheric NH₃ emissions. However, as the ashing conditions are likely to vary with farm-produced RMA, more study is needed to determine the consistency of the effects reported here.

In addition to their effects on NH₃ losses, organic amendments had significant effects on some soil properties. In the current research, sole biochar application increased soil pH compared to control treatment, however compost + biochar combination and sole compost application decreased soil pH. Poultry manure biochar may have increased soil pH through its liming effect over the sole compost and chemical fertilizers application but when the mixture of compost and biochar was applied, soil pH decreased relative to the sole biochar application. Herein, soils treated with solitary biochar had the highest pH (7.78) which is not enough to raise NH₃ loss, followed by soils treated with both biochar and compost (7.42). Slight increase in soil pH in biochar treated plots could have increased NH₃ emissions but the lower NH4+ contents in soils resulted in lower NH₃ emissions.

Application of biochar solely or in combination with compost at a rate of 3 t ha⁻¹ has increased SOC in our study, which is in line with previous research (Liu et al., 2021). Biochar is distinguished from compost by its larger proportion of more stable organic carbon molecules (Mahmoud et al., 2018; Eissa, 2019) making it more efficient in enhancing soil physicochemical parameters (Eissa, 2019). Furthermore, Trupiano et al. (2017) also reported that the application of compost and biochar to soils, either alone or in combination, enhanced soil SOC content compared to un-amended soils, implying that biochar and/or compost is a potential source of soil carbon sequestration.