The field experiment was located at the Southern Anhui Tobacco Experiment Station of the Chinese Academy of Agricultural Sciences in Xuancheng City, Anhui Province (118°45′E, 30°56′N). The regional climate is classified as subtropical monsoon, characterized by mild, humid conditions with precipitation concentrated in the warm season. Experiments were conducted in local farmlands with flat terrain and stable soil. Critically, the field had a confirmed history of tobacco root rot, ensuring natural pathogen pressure for the experiment; the land was fallow before the current trial. Before establishing the experimental treatments, initial soil sampling was conducted to assess the field’s spatial homogeneity. Three independent composite samples were collected from across the field using the five-point sampling method, with soil taken from a depth of 0–20 cm after clearing surface debris. These samples were air-dried, homogenized, and analyzed separately, confirming a consistent baseline for the subsequent experiment.

The experiment included four pesticide treatments and a control, with all application concentrations selected based on the manufacturers’ recommended field application rates: T1: prothioconazole (250 g/L EC, 4400× dilution, Shandong Qingdao Kaiyuanxiang Chemical Co., Ltd.); T2: pyrisoxazole (25% EW, 2200×, Jiangsu Yangnong Chemical Co., Ltd.); T3: kasugamycin combined with Paenibacillus polymyxa, (3% SC, 1800×, Wuhan Kernel Bio-tech Co, Ltd.); T4: cyclobutrifluram (450 g/L SC, 6000×, Syngenta Crop Protection Co., Ltd.). A randomized complete block design featuring three replicates was implemented, accommodating a total of 15 individual plots, each covering an area of 100 m². Tobacco was transplanted at a density of 18,000 plants per hectare with a row spacing of 1.2 m and plant spacing of 0.5 m, yielding an aboveground dry biomass of approximately 2,470 kg per hectare at harvest. All plots received identical management except for pesticide application. Pesticides were applied via root irrigation at transplanting and again 20 days later, with 200 ml of diluted solution per plant per application. An equivalent volume of water was applied to the CK group in place of the pesticide solutions.

Two months after tobacco transplanting, soil samples were obtained from every treatment plot via the five-point method. After removing surface debris such as leaves, weeds, and gravel, Sampling was conducted at a depth of 0–20 cm with alcohol-sterilized samplers, and approximately 0.3 kg of soil was obtained from each sampling point. Concurrently, relevant information including plot number and sampling date was recorded. Soil from each plot was thoroughly mixed and divided into two aliquots. A 0.5 kg aliquot was snap-frozen in liquid nitrogen for microbial analysis, while a 1 kg aliquot was air-dried for physicochemical analysis.

NH₄⁺-N and NO₃^–-N from soil are extracted using potassium chloride solution (Saha et al., 2018). NH₄⁺-N reacts with phenol to form blue indophenol dye, while NO₃^–-N reacts with N(1-naphthyl) -ethylenediamine hydrochloride to produce red dye. Soil pH was determined potentiometrically using a 1:5 (w/v) soil-water suspension (Faria et al., 2023). The soil moisture content is calculated based on the quality difference by using the drying method (Astm International, 2010). Soil bulk density was determined by the ring knife method (Jia et al., 2025). A certain volume of unprocessed soil was collected, dried and weighed. Soil organic matter was detected by the potassium dichromate oxidation method (Xu et al., 2025). The activities of urease, dehydrogenase, nitrate reductase, and nitrite reductase were evaluated with commercial assay kits (Beijing Boxbio Science & Technology Co., Ltd.), in accordance with the manufacturer’s instructions (Page, 1982).

Metagenomic DNA was isolated from soil samples using the E.Z.N.A.^® Soil DNA Kit (Omega Bio-tek, USA). Following paired-end sequencing on an Illumina NovaSeq platform (Majorbio, China), raw sequencing data underwent quality control with FASTP (v 0.20.0). High-quality reads were subsequently assembled into contigs via MEGAHIT (v 1.1.2) (Li et al., 2015). Contigs meeting a length threshold of ≥ 300 bp were retained for subsequent gene prediction using MetaGene (Noguchi et al., 2006). A non-redundant gene set was then constructed with CD-HIT (v 4.6.1) (Fu et al., 2012), applying thresholds of 90% for both sequence identity and coverage. For taxonomic classification, the representative sequences from this gene set were aligned to the NCBI NR database using DIAMOND (v 0.8.35) (Buchfink et al., 2015) with a maximum e-value of 1e-5. Furthermore, nitrogen cycle-related genes were identified through KEGG annotation, also performed with DIAMOND (v 0.8.35) (Pal et al., 2014; Metch et al., 2018). All obtained raw sequence datasets have been uploaded to the NCBI Sequence Read Archive (SRA) with the accession number PRJNA1346056.

Microbial community composition was analyzed using R (v3.3.1). Specifically, at the phylum level, the prevalent microbial communities were profiled using bar plots. The overall microbial structure was assessed through Principal Coordinates Analysis (PCoA) based on Bray-Curtis distances. To determine the significance of differences in community structure, Permutational Multivariate Analysis of Variance (PERMANOVA) was performed with 999 permutations (Ezeokoli et al., 2020). Alpha diversity were estimated with the mothur package (v1.30.2) (Schloss et al., 2009), as measured by the Shannon–Wiener index {H′ = −Σ[p_i × ln(p_i)]} (Shannon, 1948) and Chao1 richness estimator [Chao1 = S_obs + (F₁²/2F₂)] (Chao, 1984). LEfSe (Linear Discriminant Analysis Effect Size) was employed to identify differentially abundant taxa, discerning features and associated categories that exhibited significant differences. Xenobiotic biodegradation pathways were annotated using the Tax4Fun software (Aßhauer et al., 2015) with reference to the KEGG pathway database. Network analysis was based on Spearman correlation analysis (p < 0.05, r = 0.5) on Gephi platform software (Huang et al., 2021). Key genes involved in nitrogen cycle processes were identified by querying the nitrogen metabolism pathway within the KEGG database. A schematic diagram of the nitrogen cycle metabolic pathway was created using Adobe Illustrator. SPSS 23.0 software was employed to conduct the statistical analysis. To assess statistical significance, differences among samples were evaluated by one-way ANOVA followed by Duncan’s multiple range test.

This study assessed the impact of four pesticide treatments on the concentrations of NH₄⁺-N and NO₃^–-N, alongside the activities of four key soil enzymes: urease, dehydrogenase, nitrate reductase, and nitrite reductase. Before the start of the experimental treatment, the basic soil properties of the entire experimental field were determined. The soil was weakly acidic (pH 6.16 ± 0.12), with moderate organic matter content (25.28 ± 2.80 g/kg) and suitable physical structure (bulk density 1.14 ± 0.09 g/cm³, water content 14.17 ± 0.31%). The coefficients of variation for key physicochemical parameters among the three sampling sites were relatively low (e.g., 1.9% for pH and 2.2% for water content), providing direct evidence of the initial homogeneity of the entire experimental field.

Pesticide treatments significantly affected (P < 0.05) soil NH₄⁺-N and NO₃^–-N. As illustrated in Figures 1A, B, T1 and T2 markedly increased the soil NH₄⁺-N content by 2.38 times and 7.34 times, respectively. After T2 and cyclobutrifluram (T4) treatments, the soil NO₃^–N content increased by 12.83 times and 9.36 times (P < 0.05), respectively. As shown in Figures 1C–F, soil urease activity was significantly increased by 1.36 times in the T3 (kasugamycin combined with Paenibacillus polymyxa) treatment and by 1.58 times in T4, whereas the T2 treatment resulted in significant inhibition (Figure 1C). After the T4 treatment, soil dehydrogenase and nitrate reductase activities were significantly reduced by 2.36 times and 2.03 times, respectively (Figures 1D, E).

The results of species-level alpha diversity analysis showed no significant effect of pesticide application on the richness or diversity of soil microbial communities. Both the Chao1 and Shannon indices revealed no statistically significant differences between the CK group and the four pesticide treatment groups (Figures 2A, B). The PCOA plot, generated from Bray-Curtis distances, revealed that the variance explained by PC1 and PC2 was 28.92% and 17.08%, respectively. The PCoA results revealed a clear separation among the T2, T4 group, and the CK group, suggesting that the root irrigation treatment of T2 and T4 significantly altered the soil microbial community structure (Figure 2C).

Metagenomic sequencing was employed to assess the impact of pesticide application on soil microbial community structure. The dominant phyla across all groups were Pseudomonadota (with proportions of 27.09, 26.55, 31.63, 27.10, and 27.65% in CK, T1, T2, T3, and T4, respectively), Actinomycetota (19.86, 20.26, 16.37, 20.13, and 18.42%), and Chloroflexota (9.24, 8.81, 9.42, 9.77, and 9.43%) (Figure 3A). In comparison to the CK group, Pseudomonadota exhibited an increased relative abundance in the T2, T3, and T4 treatments following pesticide application, while a decrease was observed in the T1 treatment (Figure 3A).

Analysis of significant differences at the phylum level showed that, among the top 10 most abundant phyla, Myxococcota and Gemmatimonadota exhibited significant variations across treatments (Figure 3B). The T2 treatment significantly increased the abundance of Myxococcota, while the T4 treatment group significantly reduced the abundance of Myxococcota. Both T2 and T4 treatment groups significantly increased the abundance of Gemmatimonadota. Actinomycetota was higher in the T1 and T3 treatments but decreased in T2 and T4 (Figure 3A). The relative abundance of Chloroflexota decreased in the T1 treatment but increased in T2, T3, and T4 (Figure 3B).

LEfSe analysis revealed that the T1 treatment group exhibited significant enrichment of the genera Dechloromonas and Trebonia (Figure 4A). The T2 treatment group showed significant enrichment of Bellilinea, Enhydrobacter, Gemmatimonas, Mizugakiibacter, Nitrosospira, Paraburkholderia, Trinickia, Asticcacaulis, Burkholderia, Croceibacterium, Fluviicola, Gemmatirosa, Ignavibacterium, Mesorhizobium, Nitrosovibrio, Pedobacter, Reyranella, Sphaerobacter, Anaerolinea, Bacillus, Fulvimonas, Hanamia, Longilinea, Neobacillus, Pseudomonas, Rhodanobacter, Dyella, Hypericibacter, Luteibacter, Nitrobacter, Panacibacter, and Roseisolibacter (Figure 4B). The T3 treatment group demonstrated significant aggregation of Bacillus, Candidatus Udaeobacter, Desulfobacca, Neobacillus, Oleiagrimonas, Trebonia, Edaphobacter, Hypericibacter, Syntrophorhabdus, Alicyclobacillus, Candidatus Sulfotelmatomonas, Enhydrobacter, Occallatibacter, and Porphyrobacter (Figure 4C). The T4 treatment group exhibited significant enrichment of Enhydrobacter, Frateuria, Hanamia, Mizugakiibacter, Nannocystis, Nitrobacter, Nitrospira, Oxalicibacterium, Vampirococcus, Acetitomaculum, Dyella, Haliangium, Neorickettsia, Nitrosovibrio, Occallatibacter, Pseudogulbenkiania, and Rhodanobacter (Figure 4D).

Calculation of Spearman correlations between environmental factors and microbial genera showed that the abundance of 5 genera was significantly positively correlated with physicochemical properties, while the abundance of 9 genera was significantly negatively correlated with soil physicochemical factors. NH₄⁺-N was positively correlated with the genus Rhodanobacter and negatively correlated with Methyloceanibacter, Conexibacter, Gaiella, Candidatus Gaiellasilicea, Anaeromyxobacter, and Mycobacterium. NO₃^–-N was positively correlated with Gemmatimonas and Rhodanobacter and negatively correlated with Methyloceanibacter, Conexibacter, Gaiella, Candidatus Gaiellasilicea, Kouleothrix, Anaeromyxobacter, Mycobacterium, and Singulisphaera. Urease (S-UE) was negatively correlated with Gemmatimonas. Dehydrogenase (S-DHA) was positively correlated with Singulisphaera. Nitrite reductase (S-NIR) was positively correlated with Methyloceanibacter and Kouleothrix (Figure 4E).

Metagenomic analysis identified a total of 15 xenobiotic degradation pathways (Figure 5). The results indicated that pesticide treatments inhibited the degradation capacity of soil microorganisms, with the extent of inhibition varying by pesticide type. First, all treatments collectively suppressed six degradation pathways, including Ethylbenzene degradation, Bisphenol degradation, Caprolactam degradation, Chloroalkane and chloroalkene degradation, Aminobenzoate degradation, and Benzoate degradation.

The suppressive effects of the pesticide treatments on degradation pathways varied in scope. The T2 treatment exhibited the broadest impact, inhibiting all 15 pathways. Among these, six pathways—Atrazine degradation, Caprolactam degradation, Chloroalkane and chloroalkene degradation, Aminobenzoate degradation, Nitrotoluene degradation, and Benzoate degradation—showed significant differences (P < 0.05). The T1 treatment suppressed 10 pathways. T3 suppressed 11 pathways, among which Bisphenol degradation was significantly decreased (P < 0.05). The T4 treatment inhibited 9 pathways, with significant suppression observed in Bisphenol degradation and Caprolactam degradation (P < 0.05).

In summary, pesticide treatments generally suppressed the microbial degradation capacity for exogenous substances. The T2 treatment had the most extensive and significant impact, while certain pathways, such as Bisphenol degradation, were significantly inhibited across multiple treatments.

To evaluate the effects of the four pesticides on soil nitrogen transformation processes, we analyzed the key functional genes associated with the nitrogen cycle (Figure 6A). The genes involved included those for nitrogen fixation (nifD, nifH, nifK), nitrification (amoA, amoB, amoC, hao, nxrA, nxrB), denitrification (nirK, nirS, norB, norC, nosZ), assimilatory nitrate reduction to ammonium (ANRA: nasB, narB, nirA), dissimilatory nitrate reduction to ammonium (DNRA: narG, narH, narI, napA, napB, napC, nrfA, nrfH), and organic nitrogen metabolism (ureC, gdh, glnA). The results revealed that different pesticide treatments significantly affected nitrogen cycling pathways: The T1 treatment enhanced the nitrogen fixation pathway. The T2 treatment significantly strengthened the nitrification, denitrification, and nitrogen metabolism pathways, with levels reaching 1.14 times, 1.27 times, and 1.18 times that of the control, respectively (Figures 6C, D). The T3 and T4 treatments showed no significant impact on the soil nitrogen cycle processes (Figures 6B–G).

The co-occurrence network analysis revealed 70 nodes in total, comprising 27 nitrogen cycle gene nodes and 43 microbial genus nodes (Figure 7). Correlation analysis indicated that the abundance of nitrogen fixation genes (nifD, nifH, nifK) was significantly and positively correlated with several microbial genera, including Bradyrhizobium, Nitrobacter, Nostoc, Desulfovibrio, Nitrospina, Nitrosococcus, Synechococcus, and Beijerinckia. The abundance of key nitrification genes (hao, nxrA, nxrB) was positively correlated with a range of genera, such as Anaeromyxobacter, Geobacter, Nitrosomonas, Thiobacillus, Aeromonas, Anabaena, Trichodesmium, Campylobacter, Burkholderia, Nitrobacter, Nitrosovibrio, Escherichia, and Salmonella.

Pesticide treatments differentially affected the transformation of soil nitrogen forms and key enzyme activities. Prothioconazole and Pyrisoxazole treatments significantly increased the soil NH₄⁺-N content, likely from the inhibition and promotion of specific microbial populations that altering nitrogen transformation pathways (Castaldi and Smith, 1998). Similarly, Rose et al. (2018) also found that the herbicides Metsulfuron-methyl and 2,4-dichlorophenoxyacetic acid promoted NH₄⁺-N content. Notably, pyrisoxazole simultaneously enhanced both NH₄⁺-N and NO₃^–-N, presumably by partially activating nitrification. This aligns with the NO₃^–-N increase reported in hexaconazole-treated soils (Ju et al., 2017). Chen et al. (2001) suggested that the accumulation of both NH₄⁺-N and NO₃^–-N by captan might relate to a lower nitrification rate coupled with higher urease activity.

Regarding soil enzyme activity, the kasugamycin combined with Paenibacillus polymyxa and cyclobutrifluram treatments significantly enhanced urease activity, while pyrisoxazole markedly inhibited this enzyme’s activity. Satapute et al. (2019) reported that propiconazole application to soil initially enhanced urease activity, but significantly inhibited it with increasing application rates, suggesting that these changes are influenced by both pesticide concentration and treatment duration. Bacmaga et al. (2020) found that tebuconazole inhibited urease activity by 15.6% to 59.9% in a dose-dependent manner, an effect they attribute to the fungicide’s toxicity toward the soil microbial community and its consequent impact on microbial enzyme secretion. Singh (2005) found that triadimefon application in soils reduced dehydrogenase activity by 70% and 50%, respectively, which might delay the degradation process of the fungicide in the soil. The changes in nitrogen and enzyme activities suggest that the metabolic activity and community structure of soil microorganisms may have altered.

Pesticide application alters the soil environment and influences the structure of microbial communities. In this study, Pseudomonadota, Actinobacteria, and Chloroflexi were the dominant phyla, contributing, respectively, to biocontrol, pathogen suppression, and soil aggregation (Ren et al., 2024; Barka et al., 2015; Luan et al., 2020; Wei et al., 2022). Additionally, Myxococcota, known for its predatory characteristics, was significantly enriched under the T2 treatment, potentially due to the release of biomass resources following pesticide-induced microbial mortality (Muñoz-Dorado et al., 2016). The widespread enrichment of Gemmatimonadota under both T2 and T4 treatments suggests its high tolerance to environmental stress (Hao et al., 2025; Yuan et al., 2024).

The T2 treatment notably enriched Gemmatimonas. Research indicates that Gemmatimonas can enhance soil nitrogen content (Zhang L. et al., 2025) and promote organic nitrogen mineralization (Hui et al., 2020), which aligns with the observed rise in NH₄⁺-N content under the T2 treatment. Concurrently, this study found a negative correlation between Gemmatimonas and urease activity, indicating that urea hydrolysis was not the dominant pathway for nitrogen transformation under this treatment. The T2 treatment likely first enriched functional genera such as Gemmatimonas, driving intense organic nitrogen mineralization, thereby substantially increasing soil NH₄⁺-N content. Subsequently, this NH₄⁺-N was converted into NO₃^–-N through nitrification, facilitated by the enrichment of ammonia-oxidizing bacteria such as Nitrosospira and Nitrosovibrio, ultimately resulting in the simultaneous accumulation of both nitrogen forms (Brochado et al., 2023).

In contrast, the T4 treatment enriched multiple key nitrifying genera, including the ammonia-oxidizing genus Nitrosovibrio and the nitrite-oxidizing genera Nitrobacter and Nitrospira (Ayiti et al., 2022). These taxa participate in ammonia and nitrite oxidation processes, forming an efficient nitrification chain that rapidly converts NH₄⁺-N into NO₃^–-N (Matsuba et al., 2003; Sánchez et al., 2004).

Soil microbial communities degrade various xenobiotic pollutants to maintain soil health. Studies confirm the crucial role of specific functional microorganisms. For instance, Zhang et al. (2020) demonstrated that the bacterial strain Brevundimonas naejangsanensis J3 can degrade the harmful xenobiotic dimethachlon. Similarly, Wang X. et al. (2025) found that Bacillus subtilis ZW plays a significant role in the bioremediation of p-cresol and other aromatic compounds.

In this study, metagenomic data revealed that pesticide treatments exhibited inhibitory trends across multiple xenobiotic degradation pathways. Concurrently, fungicide root irrigation significantly inhibited the overall metabolic activity of microorganisms, also reflecting potential disruption to the microbial community structure and functional diversity (Xu et al., 2020). Therefore, we speculate that the four fungicides may have interfered with or weakened the activity of key functional microorganisms, leading to concurrent functional impairments in several associated degradation pathways. This broad attenuation of soil detoxification potential, even if mostly not statistically significant, may further increase the potential ecological risks of persistent existence, accumulation, and migration of residual pollutants in the environment.

Metagenomic analysis reveals how pesticides regulate nitrogen transformation pathways at the functional gene level. In the T1 treatment, NH₄⁺-N accumulation was primarily linked to upregulation of the ANRA pathway gene nirA. Prothioconazole likely inhibited nitrifying microorganisms or altered microbial community structure, thereby relieving the potential suppression on assimilatory nitrate reduction and ultimately inducing nirA expression, which promoted NH₄⁺-N accumulation (Rahman et al., 2021; Zhang et al., 2021). In contrast, T2 treatment accumulates NH₄⁺-N by reshaping the nitrogen cycling pathways. It significantly suppressed the urease activity, thereby suppressing the urea hydrolysis pathway for NH₄⁺-N production, but activated organic nitrogen mineralization pathway via microbial necromass (reflected by gdh upregulation). This pathway remodeling led to robust mineralization that outweighed the suppressed urea hydrolysis, becoming the main driver of NH₄⁺-N accumulation. This may result from substantial microbial mortality caused by pyrisoxazole (Bünemann et al., 2006), which rapidly released cellular nitrogen pools via mineralization (Jenkinson and Parry, 1989). This reveals the resilience of the microbial community in maintaining nitrogen transformation functionality through compensatory responses in functional pathways (Allison and Martiny, 2008). Concurrently, enhanced nitrification gene (nxrAB) expression promoted NO₃^–-N accumulation. In T4, NO₃^–-N increase was primarily due to nxrAB upregulation.

Pesticides can affect the transformation and accumulation of nitrate and NH₄⁺-N in soil by regulating the expression of key functional genes involved in the nitrogen cycle within the soil microbial community. This finding aligns with several previous studies: Cao et al. (2023) found that the herbicide mesosulfuron-methyl could influence nitrification and denitrification processes by altering the abundance of soil nitrogen cycle functional genes, thereby regulating NO₃^–-N and NH₄⁺-N content. Lyu et al. (2024) reported that the herbicide Acetochlor inhibited NO₃^–-N transformation and reduced potential nitrification and denitrification rates, mechanisms associated with changes in enzyme activity and microbial communities. Du et al. (2018) also observed that the application of the herbicide mesotrione induced changes in soil microbial community composition, subsequently affecting NO₃^–-N and NH₄⁺-N content.

By constructing a correlation network between nitrogen cycling genes and microbial genera, this study revealed the key microbial taxa driving functional changes in soil nitrogen cycling under pesticide stress. Notably, the network revealed both established and unexpected links. Expected correlations included positive links between nitrification genes (hao, nxrAB) and known nitrifying microorganisms such as Nitrobacter and Nitrosovibrio (Farges et al., 2012), confirming nitrification as the source of nitrate increases in T2 and T4. We also observed atypical patterns: nitrogen-fixing genes correlated not only with known diazotrophs (e.g., Bradyrhizobium, Nostoc), but also with the nitrifier Nitrobacter. These unusual connections may reflect functional reorganization of microbes under pesticide stress or new interspecies partnerships (Wertz et al., 2012; Dobrojan et al., 2016; Zhong et al., 2024).

Moreover, the associations between microbial genera and N-cycle genes highlights how microbial community shifts drive nitrogen cycling, aligning with reports on Massilia and Arthrobacter by Shi et al. (2021). In conclusion, the fungicide treatments not only affected the abundance of nitrogen cycle genes but also reconfigured the functional network of the nitrogen cycle by altering the microbial community structure.