This study employed an indoor disturbed soil incubation experiment, considering three types of nitrogen fertilizers: ammonium sulfate-(NH₄)₂SO₄, potassium nitrate-KNO₃, and sodium nitrite-NaNO₂; two nitrogen application gradients (200, 400 mg N kg⁻¹); and two water quality treatments (CW and RW), as detailed in Table 1. The soil samples for the RW treatment were taken from the 0–30 cm topsoil layer at China Agricultural University Tongzhou Experimental Station in 2020. The sampling area underwent RW and groundwater irrigation trials from 2013 to 2020, as documented in the literature (Chi et al., 2020b; 2020a; 2023). In short-term indoor incubation experiments, the high ion content in groundwater may interfere with the assessment of RW treatments and obscure the differences between treatments, thus affecting the accuracy and interpretability of the experimental results. Therefore, considering the differences in water quality during experimental design, CW should be used as a control treatment to avoid interference caused by complex water quality. Five sampling points were randomly selected and mixed in RW and groundwater irrigated area. Surface soil samples (0–30 cm) for the RW and CW treatments were collected after the final irrigation events with RW and groundwater, respectively, in 2020. The irrigation water quality and soil characteristics for the experiment are detailed in Table 2. According to Table 2, RW contains measurable levels of inorganic nitrogen. Furthermore, soil analysis reveals that long-term RW irrigation markedly elevates soil NO₃ ⁻ content and electrical conductivity (EC). The experiment is divided into two stages: pretreatment phase and experimental phase.

In the data processing stage, quality control was ensured primarily by verifying that repeated tested data did not show significant differences (p > 0.05) based on one-way ANOVA and exhibited a normal distribution trend. Normality tests (e.g., Shapiro-Wilk test) were conducted to confirm the data distribution, and standard deviation (SD) or coefficient of variation (CV%) was calculated to assess data consistency. The CV% values ranged from 2% to 10%, indicating an acceptable level of variability within the dataset.

The main effects analysis was performed using SPSS V25 for Windows (SPSS Inc., Chicago, IL, United States) to assess the effects of treatments. Further analysis of simple effects was conducted upon detecting significant interactions. To identify significant differences between treatment means, the least significant difference procedure (LSD) was applied, with a significance threshold of 0.05. In this context, ‘average’ was defined as the arithmetic mean, and “correlation” was referred to as linear correlation (R²).

After nitrogen application, N₂O emissions peak within 5 days (Figure 1a), while NO emissions reach their peak within 1–2 days after fertilization (Figure 1b), followed by a gradual decline and stabilization. Under high fertilization levels, the N₂O peak appears earlier than under low fertilization levels (Figure 1a), whereas for NO emissions, the peak timing is similar between high and low fertilization treatments (Figure 1b). As illustrated in Figure 1a, peak magnitude analysis showed that under AN and NI treatments, soil N₂O emission peaks increased with rising nitrogen application levels, with emissions consistently higher under RW irrigation compared to CW. Specifically, the peak values were 2,842.48 ng N kg⁻¹ for RAN4, 3,129.30 ng N kg⁻¹ for RNI4, 2,827.87 ng N kg⁻¹ for CAN4, and 2027.72 ng N kg⁻¹ for CNI4. For the KN treatment, peak values were similar between high and low fertilization levels, yet emissions remained higher under RW irrigation than under CW treatment. The specific peak values were 344.75 ng N kg⁻¹ for CKN2, 343.73 ng N kg⁻¹ for CKN4, 423.73 ng N kg⁻¹ for RKN2, and 443.94 ng N kg⁻¹ for RKN4.

Water quality significantly cumulative influenced soil N₂O emissions (p < 0.05), though it had no notable effect on NO cumulative emissions (p > 0.05), as shown in Table 3. The interaction between water quality and fertilizer type, water quality and fertilizer rate, as well as the three-way interaction among water quality, fertilizer type, and fertilizer rate, all had significant effects on soil N₂O emissions. In contrast, for soil NO emissions, only the interaction between water quality and fertilizer rate was significant. Simple effect revealed that under AN treatments, RW irrigation significantly increased soil N₂O and NO emissions (p < 0.05). In contrast, under KN treatments, RW had no significant effect on cumulative N₂O and NO emissions (p > 0.05). Under NI treatments, RW significantly affected only cumulative NO emissions (p < 0.05), with no significant impact on N₂O emissions. As illustrated in Figure 2, at a low nitrogen application rate of 200 mg N kg⁻¹, RW irrigation significantly (p < 0.05) increased cumulative soil NO emissions, with RAN2 showing a 25.04% increase compared to CAN2. Additionally, RNI2 significantly (p < 0.05) increased cumulative N₂O emissions by 27.58% compared to CNI2. At the higher nitrogen application rate of 400 mg N kg⁻¹, RAN4 also significantly (p < 0.05) increased cumulative N₂O emissions by 25.04% compared to CAN4.

As shown in Table 3, both the type and amount of nitrogen applied significantly affected cumulative emissions of soil NO (p < 0.05). As illustrated in Figure 2, cumulative N₂O emissions in the AN and NI treatments were significantly (p < 0.05) higher than those in the KN treatments, with N₂O cumulative emissions increasing by 62.83%–115.49% and 39.08%–52.98%, respectively, and NO cumulative emissions increasing by 97.95%–213%–93% and 30.49%–182.47%, respectively. The nitrogen application rate also significantly (p < 0.05) influenced soil N₂O and NO cumulative emissions. Under the AN treatment, as the nitrogen application rate increased, soil N₂O and NO cumulative emissions significantly rose (p < 0.05) (Figure 2). Specifically, with RW irrigation, N₂O and NO cumulative emissions increased by 45.56% and 48.08%, respectively, while with CW irrigation, they increased by 13.96% and 107.54%, respectively. For the KN treatment, increasing nitrogen application did not significantly affect soil N₂O and NO cumulative emissions (Table 2). In contrast, under the NI treatment, increasing nitrogen application significantly increased soil NO cumulative emissions (p < 0.05) but had no significant effect on N₂O cumulative emissions (p > 0.05) (Figure 2).

During the incubation period, the changes in soil inorganic nitrogen were similar across all treatments, as shown in Figure 3. In the AN treatment, the NH₄ ⁺ content gradually decreased over time. In the NI and AN treatments, the NO₃ ⁻ content gradually accumulated as the incubation time increased. In the AN, KN, and NI treatments, a peak in NO₂ ⁻ content was observed at the beginning of the incubation period, with the peak magnitude increasing with the amount of fertilizer applied.

As shown in Figure 4a, the soil NH₄ ⁺ and NO₃ ⁻ contents directly affected the soil NO₂ ⁻ content. In KN treatments, the R² values for RW and CW treatments were both less than 0.3, and k is less than 0.05. In AN treatments, the R² values for RW and CW treatments were 0.50 and 0.48, respectively, with k greater than 0.1. This indicated that the effect of NO₃ ⁻ application on increasing NO₂ ⁻ content was smaller than that of AN treatment. As shown in Figure 4b, there was a significant linear relationship between NH₄ ⁺ and NO₃ ⁻ application rates and NO₂ ⁻ accumulation intensity. The R² values for RW and CW treatments were 0.99 and 0.99 for NH₄ ⁺, and 0.81 and 0.93 for NO₃ ⁻, respectively. Analysis of the differences between different nitrogen treatments showed that in AN treatments, the linear slopes for RW and CW treatments were 0.89 and 0.86, respectively, while in KN treatments, they were 0.05 and 0.06. Analyzing the impact of water quality on NO₂ ⁻ accumulation intensity, the R² and k values showed very small differences between treatments. This could be due to the experimental setup, where the nitrogen application gradient was relatively large, lacking low nitrogen application treatments. Consequently, the impact of water quality on NO₂ ⁻ accumulation intensity was minimal.

As shown in Figure 4c, there was a strong correlation between soil NO₂ ⁻ accumulation intensity and cumulative emissions of soil NO and N₂O during NH₄ ⁺ treatments. Analyzing the differences in the impact of soil NO₂ ⁻ accumulation intensity on NO and N₂O emissions between different water qualities, the experiment suggested that the effect of NO₂ ⁻ accumulation intensity on soil NO and N₂O emissions was greater under RW treatment compared to CW treatment (k_RW > k_CW). The experiment found that the impact of NO₂ ⁻ accumulation intensity on cumulative NO emissions was greater than on cumulative N₂O emissions (k_NO > k_N2O). In summary, the influence of soil NH₄ ⁺ and NO₃ ⁻ on soil NO and N₂O emissions was likely primarily mediated through changes in NO₂ ⁻ ion content. This highlighted the importance of NO₂ ⁻ ions in the pathways of nitrogen transformations leading to the production of NO and N₂O in the soil.

As shown in Table 3, the main effect analysis indicated that RW significantly affected the gene copy numbers of AOA-amoA and AOB-amoA (p < 0.05), but had no significant impact on the abundance of NOB genes. Similar effects were observed in the interactions between water quality and fertilizer rate, as well as among water quality, fertilizer type, and fertilizer rate. Further simple effect analysis and Figure 5 revealed that under all three nitrogen types, increasing nitrogen application significantly increased the gene copy numbers of AOA-amoA, AOB-amoA and NOB (p < 0.01). Under RW irrigation in AN and NI treatments, the gene copy numbers of AOA-amoA, AOB-amoA, and NOB significantly increased (p < 0.05). In KN treatments, RKN significantly increased the gene copy numbers of AOA-amoA and NOB compared to CKN (p < 0.05), but had no significant effect on AOB-amoA gene copy numbers (p > 0.05). For AN treatments, at a low nitrogen application rate (200 mg N kg⁻¹), RW significantly increased the gene copy numbers of AOA-amoA, AOB-amoA, and NOB compared to CW (p < 0.05). At a high nitrogen application rate (400 mg N kg⁻¹), water quality had no significant effect on AOA-amoA and AOB-amoA gene copy numbers (p > 0.05). For KN treatments, water quality significantly affected AOA-amoA and NOB gene copy numbers at both nitrogen application rates (p < 0.05), but had no significant effect on AOB-amoA (p > 0.05). For NI treatments, water quality significantly affected AOA-amoA and AOB-amoA gene copy numbers at both nitrogen application rates (p < 0.05), but had no significant effect on NOB (p > 0.05).

As shown in Figure 5, the differences in gene copy numbers of denitrification microorganisms nirK and nosZ between different water qualities are minimal. However, under the KN and AN treatments, RW significantly increased the nirS gene copy numbers (p < 0.05). As shown in Table 3, water quality significantly affected the gene copy numbers of nirS (p < 0.05), but had no significant impact on the abundance of nirK and nosZ genes (p > 0.05). Neither fertilizer type nor fertilizer rate significantly influenced the abundance of nirK, nirS, and nosZ genes (p > 0.05). The interaction between water quality and other factors was found to significantly affect the gene copy numbers of nirS (p < 0.05). However, simple effect analysis showed that under the KN-2 treatment, water quality changes had no significant effect on nirS gene copy numbers, and changes in fertilizer rate under the NI treatment also had no significant effect. For nosZ, simple effect analysis revealed that in NI treatments, water quality had no significant effect on nosZ gene copy numbers (p > 0.05). However, at a low nitrogen application rate (200 mg N kg⁻¹), RW significantly increased nosZ gene copy numbers (p < 0.05).

As shown in Figure 6, there was a significant strong correlation between soil N₂O and NO emissions, with an R² value greater than 0.6 (p < 0.05). Soil N₂O emissions were significantly correlated with the abundance of nitrifying microorganisms AOA-amoA and AOB-amoA (p < 0.05), with R² values of 0.48 and 0.71, respectively. Soil NO emissions were significantly correlated with AOB, with an R² value greater than 0.5 (p < 0.05). There was no significant correlation between soil N₂O and NO emissions and the abundance of denitrifying microorganisms nirK, nirS, and nosZ. Additionally, analysis shown in Figure 6 indicated no significant linear relationship between nitrifying and denitrifying genes, suggesting no significant interaction between these two types of microorganisms.

Under conventional agricultural management, the amount of nitrogen fertilizer applied typically far exceeds the nitrogen content found in RW. In this experiment, it was observed that under the AN treatment, RW irrigation significantly increased soil N₂O emissions—a finding consistent with previous studies (Chi et al., 2023; 2024). As NH₄ ⁺ application rates increased, N₂O emissions also rose (Figure 2; Hickman et al., 2014), primarily due to enhanced nitrification. Elevated NH₄ ⁺ levels stimulated the activity of ammonia-oxidizing archaea and ammonia-oxidizing bacteria (Figure 5; Guo et al., 2017; Pei et al., 2022), thereby promoting N₂O production.

In contrast, under the KN treatment, increasing nitrogen application did not lead to a similar rise in N₂O emissions. This may be attributed to the disturbed soil incubation setup, where the availability of oxygen likely inhibited denitrification, even under high NO₃ ⁻ conditions. Among the key intermediates, NO₂ ⁻ plays a crucial role in regulating N₂O emissions. Some studies report an exponential relationship between NO₂ ⁻ concentration and N₂O emissions (Fan et al., 2020), while others indicate a strong linear relationship with NO₃ ⁻ (Chi et al., 2020b). In this study, a strong linear correlation was found between NO₂ ⁻ accumulation intensity and cumulative N₂O and NO emissions, aligning with the majority of previous findings.

Soil NO₂ ⁻ primarily originates from both nitrification and denitrification processes. The addition of NH₄ ⁺ and NO₃ ⁻ significantly increased soil NO₂ ⁻ content (Heil et al., 2016). Our analysis showed that both AN and KN treatments enhanced soil NO₂ ⁻ levels, suggesting that the oxidation of NH₄ ⁺ exceeded the reduction of NO₃ ⁻. The production of N₂O and NO in soil mainly involves the formation of NH₂OH, its oxidation to NO₂ ⁻, and subsequent reduction of NO₂ ⁻ to N₂ (Duan et al., 2019). In the NI treatment, where oxygen was abundant due to the disturbed bottle setup, N₂O production under normal conditions would be minimal, as oxygen tends to oxidize NO₂ ⁻ to NO₃ ⁻. However, the observed N₂O emissions may have originated from nitrifier denitrification, a process in which AOB partially oxidize NH₄ ⁺ to NO₂ ⁻ and then reduce NO₂ ⁻ to N₂O or N₂ under high NO₂ ⁻ concentrations (Wrage et al., 2001), rather than fully oxidizing it to NO₃ ⁻. Another possible mechanism is abiotic, non-biological N₂O formation, RW irrigation can enhance such chemical N₂O production in soils (Zhu et al., 2023). These findings underscore the importance of controlling soil NO₂ ⁻ concentrations in agricultural systems utilizing RW irrigation, as a strategy for mitigating N₂O emissions. The application of nitrification inhibitors or biochar is recommended, as both have been shown to suppress NO₂ ⁻ production and reduce N₂O emissions (Weiske et al., 2001; Di and Cameron, 2011; Fan et al., 2020).

Numerous studies have demonstrated that RW irrigation increases the abundance of soil nitrifying and denitrifying microorganisms (Saha et al., 2010; Ibekwe et al., 2018; Sun et al., 2018; Zhu et al., 2023). In this study, under low nitrogen application rates, RW treatments (RAN and RNI) significantly increased the abundance of microorganisms associated with AOA-amoA and nosZ genes. However, the influence of RW on soil microbial communities is strongly modulated by both the type and amount of nitrogen applied. At higher nitrogen application rates, the quantity of nitrogen becomes the dominant factor shaping the abundance of nitrogen transformation genes, rather than RW. Previous research supports this observation: Tian et al. (2014) found that nitrogen input significantly increased nitrifying microbial populations, while Zhu et al. (2016) reported that higher fertilizer rates enhanced both nitrifying and denitrifying bacteria. In our study, under the KN treatment with nitrogen application rates of 200 and 400 mg N kg⁻¹, RW irrigation did not significantly increase the abundance of denitrifying microbes compared to CW treatment. This may be due to the already high NO₃ ⁻ concentrations, suggesting that shifts in NO₃ ⁻ availability have a more pronounced effect on the growth, reproduction, and community structure of denitrifying bacteria. Therefore, at 200 mg N kg⁻¹, nitrogen input itself becomes the primary driver of microbial proliferation, surpassing the influence of RW irrigation.