1 Introduction

Nitrogen (N) fertilizers are commonly applied to improve crop yields in agricultural activities (Lv et al. 2022; Wang et al. 2019). However, with the development of agricultural intensification, repeated and excessive N application in agricultural production has resulted in serious N loss, soil acidification, and groundwater nitrate pollution, which are potentially harmful to human health (Wheeler et al. 2015). In addition, various land-use types have differences in fertilization, irrigation, and cropping practices, thus causing various nitrate N (NO3−-N) accumulation patterns in soils (Ju et al. 2006). The annual average N input in the North China Plain was reported to be approximately 1358 kg N ha−1 in a greenhouse vegetable field, 2.5 and 2.1 times higher than that in a grainfield and apple orchard, respectively (Ju et al. 2006). The long-term surplus application of N fertilizer in wheat‒maize fields, orchards, and greenhouse vegetables has accelerated N accumulation in the 0–90 cm, 0–100 cm, and 0–120 cm soil profiles, which is approximately 184, 594, and 881 kg N ha−1, respectively (Fan et al. 2014; Gao et al. 2021; Wang et al. 2014). The accumulated N could be more quickly and easily leached into deep soil, even vadose soil, after heavy rainfall or flood irrigation (Gao et al. 2021).

Soil contains the most abundant species of microorganisms on Earth, and it is one of the most important habitats for microorganisms, providing essential nutrients for microbial colonization and growth (Lian et al. 2022). Recent researches have discovered the response of soil microbial communities to fertilization, especially using high amounts of N (Boot et al. 2016; Fierer et al. 2012; Wang et al. 2018). A previous study revealed that chemical fertilization (NPK) significantly reduced bacterial richness in the 0–10 cm soil layer (Liu et al. 2020b), and N amendment significantly reduced the relative abundance of Acidobacteria and enhanced the relative abundance of Proteobacteria (Zhou et al. 2017). Acidobacteria and Proteobacteria belong to oligotrophic and copiotrophic groups, respectively (Fierer et al. 2012). Apart from N inputs, soil types and anthropogenic long-term production activities such as irrigation and tillage also affect soil bacterial communities (Muhammad et al. 2022; Zhang et al. 2022). Early reports indicated that land-use types could induce significant variations in the soil microbial community structure, alpha- or beta-diversity, and dominant taxa (Lauber et al. 2008; Romdhane et al. 2022). Keystone species are highly related taxa in the microbial community structure and play a uniquely pivotal role in the microbial community (Xun et al. 2021). The removal of keystone species may lead to dramatic changes in the function and composition of the microbial community (Banerjee et al. 2018). Notably, soil microbes have crucial roles in the element biogeochemical cycle around the earth, and variations in soil microbial communities will inevitably affect soil ecological functions, such as carbon and N cycling (Bowles et al. 2014; Lian et al. 2022). Within the N cycle, denitrification, catalyzed by microorganisms, is a process of reducing NO3−-N to gaseous nitrogen, and the functional genes for denitrification (nirK, nirS, and nosZ) usually act as indicators of the denitrifying ability (Li et al. 2019; Wang et al. 2021). There is research demonstrating that long-term N inputs can increase the relative abundances of denitrification genes, thus stimulating the denitrification capacity due to the abundance of the substrate nitrate for denitrifiers (Wang et al. 2018). Previous research has revealed that organic fertilizer amendments improve the efficiency and activity of soil denitrifiers in orchards and reduce N pollution, suggesting that denitrifiers can help alleviate soil nitrate leaching (Kramer et al. 2006). A subsequent study in another microcosmic culture experiment has showed similar results (Chen et al. 2018).

Although studies have highlighted that soil microbial communities can be affected by land-use types, including microbial denitrifiers, most studies only focus on topsoil (Braker et al. 2015; Ding et al. 2022; Lauber et al. 2008). However, the response of soil microbes in deep soil to fertilization under different land-use types remains poorly understood. Soil properties, including pH, soil organic matter, and total carbon, differ considerably with depth (Liu et al. 2020b), which further impact the structure of the microbial community and denitrification activities (Merloti et al. 2019). A previous study found a strong correlation between vertical separation of microbial communities and TN (Liu et al. 2020b). Early reports also found that denitrification gene abundance decreased with increasing soil depth, which was coherent with the change in soil organic matter in winter wheat and summer maize fields (Liu et al. 2020b; Wang et al. 2021). However, in the North China Plain, the impact of fertilization on the vertical distribution of soil microbes and denitrifiers under other typical orchard and vegetable land-use types is less understood. In this study, based on soil samples from 0 to 300 cm under three typical land-use types, a winter wheat‒summer maize field, an apple orchard, and a vegetable field, we aimed to (1) investigate nitrate distribution with depth under the above three typical land-use types, (2) explore the variations in the bacterial communities and denitrifiers with depth under the above three typical land-use types, and (3) clarify the responses between soil properties and the structural and functional variances of bacterial communities under above three typical land-use types.

2 Materials and methods

2.1 Soil sampling

Field sites were situated in Luancheng, Shijiazhuang, Hebei Province, China (37°88′N, 114°65′E), with Cambisol soil and a warm temperate monsoon-type climate. Three typical land-use types with more than 20 years of continuous fertilization were selected: a winter wheat‒summer maize field (WM) (600 kg compound fertilizer ha−1 y−1), an apple orchard (OR) (750 kg potassium phosphate, 1244 kg diammonium hydrogen phosphate, 1244 kg urea, and 10,495 kg pig manure ha−1 y−1) and a tomato vegetable field (VE) (68,216 kg pig manure and 300 kg compound fertilizer ha−1 y−1). Soil samples (0–300 cm depth) were collected in October 2016 from each land-use type using specially designed soil columns (43 mm diameter by 120 cm long) with a Geoprobe (Geoprobe 54DT, USA). Each soil core was split into four layers: 0–30 cm, 30–100 cm, 100–200 cm, and 200–300 cm. In all, 36 samples (three land-use types × four soil layers × three replicates) were collected and delivered to the lab on ice in labeled sterile polythene bags. Each soil sample was passed through a 2-mm sieve and then separated into two parts: one part was used for soil property analysis (stored at 4°C), and the other part was used for molecular analysis (stored at − 80 °C).

2.2 Soil property analysis

The determination of soil properties relied on previous studies (Sun et al. 2018; Wang et al. 2022). Soil samples were dried at 105°C for 12 h to a constant weight to determine the soil water content. Soil pH was assessed at a soil/water (carbon dioxide-free) ratio of 1:2.5 by a pH meter (PHS–3C, Shanghai INESA). The dichromate oxidation method was used to analyze soil organic matter (SOM). An element analyzer (Vario PYRO cube, Elementar, Germany) was used to measure soil total nitrogen (TN) and total carbon (TC). Soil nitrate (NO3−–N) was extracted with 2 M potassium chloride and then determined using a spectrophotometer (UV–6100S, Shanghai METASH).

2.3 Soil DNA extraction and quantification of denitrification genes

The total microbial DNA of each soil sample was extracted with 0.5 g fresh soil according to the manufacturer’s instructions using a FastDNA Spin Kit for Soil (MP Biomedicals, USA). DNA purity and concentration were determined by a Nanodrop spectrophotometer (NanoDrop™ One, Thermo Fisher Scientific, USA). The extracted DNA was stored at − 20°C before further analysis.

Three genes (nirK, nirS, and nosZ), as indicators of the denitrification processes, were quantified using the SYBR Green approach by quantitative polymerase chain reaction (qPCR) based on a former study (Sun et al. 2015). The primers used in this study were FlaCu (5′-ATCATGGTSCTGCCGCG-3′) and R3Cu (5′-GCCTCGATCAGRTTGTGGTT-3′) for the nirK gene, cd3aF (5′-GTSAACGTSAAGGARACSGG-3′) and R3cd (5′-GASTTCGGRTGSGTCTTGA-3′) for the nirS gene, and nosZF (5′-CGYTGTTCMTCGACAGCCAG-3′) and nosZ1622R (5′-CGSACCTTSTTGCCSTYGCG-3′) for the nosZ gene (Throback et al. 2004). The qPCR system (25.0 μL) was as follows: 12.5 μL SYBR Premix Ex Taq™ (Takara Biotech, Dalian, China), 0.5 μL of each primer (10 μM), 1.0 μL DNA (about 20 ng/μL), and 10.5 μL sterilized water. A tenfold serial dilution of the plasmids harboring the target functional genes was used to generate standard curves. The thermal cycling for the reaction of the three denitrification genes was performed according to a previous study (Liu et al. 2020b). Bio-Rad CFX Manager 3.1 (Bio-Rad, Hercules, CA) was used to analyze all qPCRs, and the specificity of the qPCR amplifications was checked based on melting curves.

2.4 16S rRNA gene sequencing and bioinformatics

The primer set 515F (5′-GTGYCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACNVGGGTWTCTAAT-3′) was used to amplify the 16S rRNA gene (V4 region) to explore the soil bacterial communities (Walters et al. 2016). The amplification and purification of the 16S rRNA gene were detailed in our former study (Sun et al. 2018). The amplified products were purified, and sequencing was then performed using an Illumina NovaSeq platform (Shanghai Personalbio Technology Co., Ltd.). QIIME2 (Quantitative Insights Into Microbial Ecology, version 2019.7) was used to process the high-throughput raw sequencing data (Bolyen et al. 2019). DADA2 (Divisive Amplicon Denoising Algorithm) was used for quality filtering, denoising, joining, dechimeric sequences, and clustering of the sequencing data and then identified as amplicon sequence variants (ASVs) (Callahan et al. 2016). The classify-sklearn naïve Bayes taxonomy classifier in the feature-classifier plugin was used to assign the taxonomy of each ASV based on the Silva database (Release 132) (Bokulich et al. 2018). Diversity analyses (alpha- and beta-diversity) were calculated using the q2-diversity plugin (https://github.com/qiime2/q2-diversity). The annotation of the function of bacterial taxa was identified using the Functional Annotation of Prokaryotic Taxa (FAPROTAX) database (Louca et al. 2016). The sequencing data were uploaded to the National Center for Biotechnology Information (NCBI) under accession number PRJNA891031.

2.5 Statistical analysis

The variance analysis of soil properties, Shannon index, functional categories, and denitrifying gene abundances was carried out by one-way analysis of variance or with a nonparametric Kruskal‒Wallis test using IBM SPSS version 20. Principal coordinate analysis (PCoA) was performed to visualize the soil bacterial community structure. Based on the Bray–Curtis distance, permutational multivariate analysis of variance (PERMANOVA) was performed to examine the variations in bacterial communities at different soil depths and land-use types. The relationship between the bacterial communities and soil properties was determined using multivariate regression tree (MRT) analysis in the “mvpart” and “MVPARTwrap” packages. Keystone species were identified by the betweenness centrality value using the R package “igraph” (Chen et al. 2020; Csardi and Nepusz 2006). Spearman’s correlation analysis of the α-diversity, functional categories, denitrification gene abundance, and soil properties was performed using the R package “corrplot.” Heatmaps used for the relative abundance of functional categories under different land-use types with soil depth were generated using the R package “pheatmap.”

3 Results

3.1 Changes in soil properties with depth under different land-use types

The soil properties varied among the different land-use types (Table 1). At the 0–30 cm soil depth, WM treatment had the highest pH, SOM, TC, and NO3−–N contents than that in the VE and OR treatments. For example, the SOM and NO3−–N contents were 21.49 ± 1.80 g/kg and 24.92 ± 3.30 mg/kg in the WM treatment, 15.18 ± 3.51 g/kg and 6.10 ± 0.79 mg/kg in the OR treatment, and 15.69 ± 1.07 g/kg and 13.67 ± 3.44 mg/kg in the VE treatment, respectively. However, the changes in the soil properties with depth were inconsistent among the WM, VE, and OR treatments. In addition, the soil properties presented different trends with soil depth under each land-use type. Soil pH and the C:N ratio had an overall increasing tendency with soil depth, while SOM and TN decreased with soil depth under each land-use type. Notably, the NO3−–N content dramatically increased with soil depth in the OR treatment, from 6.10 ± 0.79 mg/kg at 0–30 cm to 256.31 ± 6.76 mg/kg at 200–300 cm, which was more than 20 times higher than that in the WM and VE treatments. The NO3−–N content first decreased with soil depth and then slightly increased in deeper soil layers in the WM and VE treatments.

Table 1 Soil properties changed with depth under different land-use types (means ± SDs)
Full size table

3.2 Changes in the soil bacterial communities with depth under different land-use types

In total, 2,991,175 sequences were acquired from all 36 samples after quality screening was performed, which clustered into 2,795,348 ASVs. The bacterial community composition varied with different land-use types. In the 0–30 cm soil layer, six phyla Proteobacteria, Actinobacteria, Acidobacteria, Chloroflexi, Planctomycetes, and Gemmatimonadetes accounted for 85.11%, 86.37%, and 77.72% of the bacterial composition in the WM, OR, and VE treatments, respectively (Fig. 1). The relative abundance of Actinobacteria in the VE treatment was only 8.35%, which was much lower than that in the WM (21.51%) and OR (20.67%) treatments. Contrarily, the VE treatment harbored the highest relative abundances of Acidobacteria and Firmicutes, up to 20.33% and 13.12%, respectively. Additionally, the relative abundance of some phyla varied from the topsoil to deep soil layers (Fig. 1). For example, from 0–30 to 200–300 cm in the VE treatment, the relative abundance of Proteobacteria increased from 27.40 to 87.11%, and that of Bacteroidetes increased from 1.10 to 6.49%, while the relative abundances of Gemmatimonadetes, Chloroflexi, and Firmicutes overall decreased with soil depth, which had a similar trend in the WM and OR treatments.

Fig. 1
figure 1

The relative abundance of bacterial phyla in different soil layers under the WM, OR, and VE treatments. WM, winter wheat‒summer maize field; OR, apple orchard field; VE, tomato vegetable field

Full size image

According to the PCoA (Fig. 2a) and PERMANOVA (Table S1) results, there was no significant separation in the bacterial community structure in each soil layer across different land-use types. However, there were significant changes in the bacterial communities with soil depth, as the bacterial structure in the 0–100 cm soil layer was significantly different from that in the 100–200 cm and 200–300 cm layers (Table S1). In addition, the Shannon index, representing the bacterial α-diversity, had a similar trend with soil depth under each land-use type. For example, the Shannon index decreased with soil depth from 0–30 to 200–300 cm, except in the WM treatment, which showed a slight rebound at 200–300 cm compared with 100–200 cm (Fig. 2b). The WM treatment harbored significantly higher bacterial α-diversity than the OR and VE treatments in the 0–30 cm, 30–100 cm, and 200–300 cm soil layers. However, no obvious difference was found between the OR and VE treatments across all horizons, except 100–200 cm.

Fig. 2
figure 2

a Principal coordinate analysis (PCoA) plot of the soil bacterial communities at different soil depths under the three treatments. Each circle represents the same treatment; b bacterial α-diversity in different soil layers under the three treatments. Different letters indicate significant differences among treatments in the same soil layer (p < 0.05). WM, winter wheat‒summer maize field; OR, apple orchard field; VE, tomato vegetable field

Full size image

3.3 Correlations between soil properties and bacterial communities

MRT analysis clarified the correlation between soil properties and the vertical distribution of the bacterial communities, which explained 81.81% of the bacterial community variation (Fig. 3). The C:N ratio divided the distribution of the bacterial communities into two major parts: one part mainly contained 0–100 cm samples with a C:N ratio < 35.00, and the other part contained 100–300 cm samples with a C:N ratio ≥ 35.00 (Fig. 3). In addition, the soil properties had different effects on the α-diversity in each soil layer (Table S2). For instance, α-diversity had a significant and positive correlation with SOM, TC, TN, and NO3−–N in the 0–30 cm soil layer (p < 0.05), but had a negative correlation with NO3−–N in the 30–100 cm soil horizon (p < 0.05) and TC in the 100–200 cm soil horizon (p < 0.05). Soil properties had no significant relationship with α-diversity in the 200–300 cm soil layer.

Fig. 3
figure 3

Multivariate regression tree (MRT) of bacterial communities in soil under the different land-use types. The land-use types and depth are in the rectangle, and the number of samples is in parentheses. The percentages under each split represent the percentage of difference explained by the split. WM, winter wheat‒summer maize field; OR, apple orchard field; VE, tomato vegetable field; SOM, soil organic matter; TC, total carbon

Full size image

3.4 Keystone species and predictive functions under different land-use types

The top twenty ASVs were regarded as keystone species according to the betweenness centrality value (Table S3). The keystone species composition and relative abundance varied under the different land-use types. In the 0–30 cm soil horizon, the most predominant keystone species in the WM, OR, and VE treatments belonged to the Actinobacteria, Actinobacteria, and Chloroflexi phyla, which accounted for 0.49%, 1.37%, and 0.24%, respectively (Fig. S1). The keystone species, affiliated with Firmicutes, only existed in the VE treatment at 0–30 cm and 30–100 cm depths. Additionally, the relative abundance of keystone species varied with soil depth. For example, from 0–30 to 200–300 cm in the WM and OR treatments, the relative abundance of Rokubacteria in keystone species increased from 0 to 1.08% and from 0 to 0.39%, respectively. Moreover, Actinobacteria and Rokubacteria were dominant in the WM and OR treatments at 0–30 cm and 200–300 cm, respectively, while Chloroflexi and Actinobacteria were dominant in the VE treatment at 0–30 cm and 200–300 cm depths.

The relative abundance of the functional categories of N cycling varied under different land-use types and exhibited various trends with soil depth (Fig. 4a). In the 0–30 cm soil layer, the relative abundance of the nitrification and aerobic ammonia oxidation functional categories was the lowest in the WM treatment, accounting for 8.60% and 6.19%, respectively. Additionally, the relative abundance of the denitrification functional category in the WM treatment was slightly higher than that in the OR and VE treatments. In the 30–100 cm soil layer, the OR treatment (7.40%) had a higher relative abundance of the denitrification functional category than that in the WM (3.72%) and VE (3.95%) treatments, but no significant difference was found in the 100–200 cm and 200–300 cm soil layers. The relative abundances of the nitrification, aerobic ammonia oxidation, and denitrification functional categories were significantly and negatively correlated with TC and the C:N ratio, while those of the aerobic ammonia oxidation and denitrification functional categories were positively correlated with SOM and TN (Fig. 4b). However, no significant correlation was found between denitrification and NO3−–N.

Fig. 4
figure 4

a The relative abundances of the nitrogen cycling functional categories in different soil layers under the WM, OR, and VE treatments. Different letters represent significant differences among treatments in the same soil layer (p < 0.05). WM, winter wheat‒summer maize field; OR, apple orchard field; VE, tomato vegetable field; b Spearman’s correlation between the nitrogen cycling functional categories and soil properties. The circle size represents the Spearman correlation coefficient. Red and blue represent positive correlation and negative correlation, respectively, and asterisks represent significant correlation. *: p < 0.05; **: p < 0.01; ***: p < 0.001

Full size image

3.5 Changes in denitrifying genes with depth under different land-use types

The vertical distributions of nirK, nirS, and nosZ were explored under different land-use types (Fig. 5a). In the 0–30 cm soil layer, the nirK, nirS, and nosZ gene abundances in the WM treatment were the highest, reaching 2.51 × 108, 3.11 × 108, and 1.27 × 108 copies per g of dry soil, respectively, followed by those in the OR and VE treatments. From 30 to 300 cm, the OR treatment had the highest abundance of the nirK, nirS, and nosZ genes. In addition, the abundance of the nirK, nirS, and nosZ genes decreased with soil deepness in each land-use type, which decreased by approximately 2–4 orders of magnitude from 0–30 to 200–300 cm deep soil. In addition, the abundances of the nirK, nirS, and nosZ genes had significant and positive correlations with SOM and TN, but had negative correlations with the C:N ratio, TC, and pH (Fig. 5b). Unexpectedly, the soil NO3−–N content had no obvious relation with the denitrification genes.

Fig. 5
figure 5

a The absolute abundances (copies per g dry soil) of the nirK, nirS, and nosZ genes in the WM, OR, and VE treatments. Data were log-transformed. Different letters represent significant differences among treatments in the same soil layer (p < 0.05). WM, winter wheat‒summer maize rotation cropland; OR, apple orchard; VE, tomato vegetable field; b Spearman’s correlation between the denitrification gene abundances and soil properties. The circle size represents the Spearman correlation coefficient. Red and blue represent positive correlation and negative correlation, respectively, and asterisks represent significant correlation. *: p < 0.05; **: p < 0.01; ***: p < 0.001

Full size image

4 Discussion

4.1 Land-use type effects on soil bacterial communities and denitrifying genes

As predicted, the soil bacterial communities were affected partly by land-use types, as indicated by their distinction from one another (Figs. 1 and 2), which is in line with earlier studies (Babin et al. 2019; Jimenez-Bueno et al. 2016). These distinctions were caused by multiple factors, such as fertilization, planting systems, and soil properties. In this study, only chemical fertilizer was applied in the WM treatment, while both chemical fertilizer and pig manure were applied in the OR and VE treatments. Vegetable fields rely heavily on manure fertilizer for regulating soil properties. Numerous studies have shown that the changes in soil properties caused by fertilization can induce differences in soil bacterial communities (Chai et al. 2021; Gu et al. 2019; Liu et al. 2020b). Taking the 0–30 cm soil depth as an example, the WM treatment harbored the most diverse bacterial communities compared with the OR and VE treatments, which is partly caused by the higher contents of SOM, TC, and NO3−–N, as they were significantly and positively correlated with the soil bacterial α-diversity (Table S2). Additionally, Francioli et al. found that the soil bacterial diversity with the combined application of chemical (NPK) and organic (cattle manure) fertilizer was higher than that with chemical fertilizer application alone (2016). However, a previous study reported that the quality and application of organic fertilizer also affect soil bacterial diversity (Hartmann et al. 2015). Therefore, low-quality organic fertilizer or improper application may lead to a decrease in bacterial diversity. In addition, Acosta-Martinez et al. demonstrated that the soil bacterial diversity in the rotation system was higher than that in the single cropping system because rooting systems and different crop residues in the rotation system had a stronger impact on soil bacterial diversity (2008). In contrast, the bacterial communities in the WM, OR, and VE treatments had no significant difference from each other (p > 0.05, Table S1), as all land-use types were geographically close to each other and existed under the same climatic conditions (Barberán et al. 2015). However, the dominant phyla had different relative abundances across these land-use types. Due to the heavy application of pig manure, VE treatment had the highest relative abundance of Firmicutes than that in the WM and OR treatments. Previous studies have reported that Firmicutes is the main phylum in organic fertilizer (Liu et al. 2022a; Semenov et al. 2021). This result indicated that manure-amended bacteria were abundant in these manure-amended soils. In addition, the relative abundance of Actinobacteria in the VE treatment was lower than that in the WM and OR treatments, which may be closely related to the cultivation cycle of greenhouse vegetables (Song et al. 2018). Previous research revealed that the relative abundance of Actinobacteria decreased with planting cycles in greenhouses (Song et al. 2018).

Keystone species had various abundances and compositions across different land-use types. The keystone species belonging to Firmicutes existed only in the VE treatment in the 0–30 cm and 30–100 cm soil layers, which was attributed to Firmicutes as the dominant bacteria in manure fertilizer, and the application of manure fertilizer increased its abundance in manure-amended soils (Liu et al. 2022a). Additionally, a previous study reported that Firmicutes plays an important role in decomposing vegetable litter, which may be one of the reasons why Firmicutes exists only in the 0–100 cm soil layer of vegetable fields (Zhang et al. 2020). Furthermore, the N cycling functional categories revealed that the WM treatment had the lowest relative abundance of the nitrification and aerobic ammonia oxidation functional categories. This may be related to the lower N input, as N fertilizer applied to farmland soil could facilitate nitrification and the aerobic ammonia oxidation capacity (Chu et al. 2008). Rao et al. demonstrated that the changes in soil properties caused by fertilization also affect functional categories (2021). In this study, TC and the C:N ratio significantly affected nitrification, aerobic ammonia oxidation, and denitrification.

The WM treatment had higher abundances of denitrification genes than those in the OR and VE treatments in the 0–30 cm soil layer. A previous study also analyzed denitrification gene abundance under different land-use types and found that wheat‒maize fields had the highest denitrification gene abundance, which was attributed to sufficient NO3− promoting the growth of denitrifying bacteria (He et al. 2020). However, in this study, denitrification genes had no significant correlation with the NO3−–N content but had significant and positive relations with SOM and TN (Fig. 5b). It has been reported that rich soluble SOM in soil, especially water-soluble organic carbon, promotes the soil denitrification process (Surey et al. 2020). Additionally, other researches also found a positive relationship between TN and the abundance of most denitrification genes (Lindsay et al. 2010; Reverchon et al. 2015). The abundant SOM and TN in the WM treatment provided favorable conditions for denitrification.

4.2 Depth effect on soil bacterial communities and denitrifying genes

In this study, soil properties varied with soil depth. For instance, SOM and TN decreased with soil depth under each land-use type, which is in line with other studies (Emadi et al. 2008; Fetene and Amera 2018). Additionally, reduced root exudates and plant litter may result in a decrease in the SOM content with the soil profile (Kramer and Gleixner 2008). Notably, heavy amounts of NO3−–N accumulated in deep soil layers, posing a serious risk of N leaching (Cui et al. 2020). A previous study revealed that frequent irrigation and excessive N inputs led to severe soil nitrate leaching (Liu et al. 2022b).

Apart from soil properties, the influence of soil depth on bacterial richness and bacterial structures was also apparent (Figs. 1 and 2). Previous studies have shown that carbon and nitrogen sources are essential resources for cell growth and have crucial factors affecting bacterial diversity and community structure (Eilers et al. 2012; Seuradge et al. 2017). The NO3−–N content, SOM, and especially the C:N ratio had prominent effects on the vertical distribution of the bacterial communities in this study. Previous study demonstrated that the C:N ratio significantly influenced bacterial growth and reproduction (Gu et al. 2019). On another scale, plant inputs, such as root exudates and litter, may also partly reflect their influence on the variety of bacterial communities between topsoil and deep soil (Seuradge et al. 2017). Previous studies have shown that the various root system distributions account for the wide variations in carbon sources, which might be responsible for the stratification of microbes (Fierer et al. 2003; Goberna et al. 2005). In addition, a substantial portion of microorganisms existed in deep soil layers, and their potential capacities could not lower or even higher than those of microorganisms in surface soil (Fierer et al. 2003). Consistent with previous studies, the relative abundance of Proteobacteria increased with soil depth (Li et al. 2014; Liu et al. 2020a). Proteobacteria exists widely in various environments, including acidic and nutrient-poor conditions, and increases in labile substrate-abundant environments (Curtis et al. 2002; Goldfarb et al. 2011). Soil depth played a strong environmental factor that influenced the bacterial communities (Liu et al. 2020b; Schlatter et al. 2020). The relative abundance of Gemmatimonadetes decreased with soil depth, which was identified as one of the top nine phyla in soil and preferred soil environments (DeBruyn et al. 2011; He et al. 2017). In this study, the prominent keystone species in each soil layer were different in the WM, OR, and VE treatments. The differences in prominent keystone species with soil depth under different land-use types may reflect their viability and life history strategies in distinct habitats (Fierer 2017; Schlatter et al. 2020).

The absolute abundances of the nirK, nirS, and nosZ genes also decreased with soil deepness under these three land-use types (Fig. 5a), which is consistent with other studies (Chen et al. 2018; Wang et al. 2021). This was also reflected by the denitrification functional categories (Fig. 4a). Although NO3−–N, an important substrate for denitrification, has a significant effect on the denitrification potential (Wang et al. 2018), the denitrifier abundance still decreased at 100–300 cm under the OR treatment, which had more than 30–40 times the NO3−–N content of the surface soil. One study noted that denitrifiers are mostly heterotrophic and grow in an environment rich in organic carbon (Gao et al. 2015), indicating that a lower carbon source inhibits the growth of denitrifiers in deep soil. Recent studies have noted that carbon availability is the major hindering factor for denitrification activity in deep soil (Chen et al. 2018; Zhou et al. 2016). Carbon amendments have been considered to sharpen denitrification capacity in deep vadose zone soils (Chen et al. 2018).

5 Conclusion

In this study, the orchard had serious nitrate accumulation at 200–300 cm compared with the other two land-use types, due to the large amount of fertilizer input and the irrigation pattern. In addition, the vertical distribution of the soil bacterial communities under different land-use types was affected by the soil properties, and the C:N ratio was the major driving factor. Land-use types also remarkably influenced the soil bacterial diversity, with the crop field having the highest bacterial diversity in most soil layers. Moreover, the denitrification gene abundance decreased with soil depth under these three land-use types, and the lower denitrification activity partly caused the accumulation of nitration in the orchard soil. This study provided insights into the status quo of the soil nitrate, bacterial community, and denitrification activity in deep soil under three typical land-use types.