Nitrogen process in stormwater bioretention: effect of the antecedent dry days on the relative abundance of nitrogen functional genes

Urban runoff is the main source of nitrogen pollutants in downstream basins (Zhang et al. 2022), and its nitrogen mainly exists in dissolved forms, making it difficult to remove by precipitation or filtration. Bioretention, also known as biofiltration or stormwater biofilters, have been widely used in different scale landscapes because of their ability to improve water quality and hydrologic condition (Zhang et al. 2021a). Bioretention is a plant–soil–microbe-based system that remove pollutants in stormwater runoff through filtration, adsorption, plant uptake, biotransformation, and several other unknown naturally occurring mechanisms (Zhang et al. 2021a). The microorganisms have a unique role in controlling runoff pollutants in bioretention systems (Skorobogatov et al. 2020). However, bioretention showed a variability in nitrogen removal (Valenca et al. 2021), and nitrogen leaching may occur under certain circumstances, instead, it became a ‘nitrogen source’ (Wan et al. 2018). Nitrogen removal in bioretention systems was mainly achieved through assimilation by plants and microorganisms, as well as nitrification and denitrification by microbes (Skorobogatov et al. 2020).

Nitrification refers to the process in which ammonium nitrogen (NH₄⁺-N) was oxidized to nitrate nitrogen (NO₃⁻-N) by ammonia-oxidizing bacteria (AOB) or ammonia-oxidizing archaea (AOA) and nitrite-oxidizing bacteria (NOB) under aerobic conditions. Ammonia oxidation is a rate-limiting and critical step in the nitrification process (Tu et al. 2019), and the key enzyme for this step is encoded by the amoA gene of AOA and AOB (Fan et al. 2015). Denitrification is a process in which nitrate nitrogen is reduced to N₂ under the catalysis of nitrate reductase (Nar), nitrite reductase (Nir), nitric oxide reductase (Nor) and nitrous oxide reductase (Nos). Among them, the process of reduction from nitrite to NO is the symbolic reaction that distinguishes denitrification from other nitrate metabolism, and is also the most important rate-limiting step in denitrification (Zhao et al. 2019). One study found that nitrite reductase (Nir) was the rate-limiting enzyme to perform this step (Tosha et al. 2021). Therefore, the corresponding nitrite reductase genes nirK and nirS were typically used as effective marker genes for characterizing the abundances and community composition of denitrifying bacteria in ecosystems (Kou et al. 2021).

Microbial communities in bioretention soils are sensitive to environmental changes. Wet and dry alternations expose microbes to irregular water stress by affecting soil moisture content in bioretention (Chen et al. 2015), and directly and indirectly affect the soil microbial community composition and function. Studies have shown that the nitrogen removal efficiency of bioretention systems can be significantly affected by ADD. Cho et al. (2011) found that when ADD was increased from 5 d to 20 d, the effluent concentration of NH₄⁺-N gradually increased, and NO₃⁻-N was affected by media types. However, with the increase in ADD, NO₃⁻-N leaching occurred. Similarly, Hatt et al. (2007) found that the effluent concentrations of total nitrogen (TN) and NO₃⁻-N increased with the increase in ADD.

One study showed that the change in soil moisture can regulate the abundance of functional N-cycling genes, nifH (N fixers), amoA, amoB (nitrifiers), nirK, nirS and nosZ (denitrifiers) in soil (Morugán-Coronado et al. 2019). The abundance of nitrifying and denitrifying genes can be significantly increased by moderately prolonging the duration of drought (Chen et al. 2017). However, there one study showed that the increase in soil moisture can increase the abundance of denitrification functional genes (Liu et al. 2014). Obviously, there were no unified conclusions about the effect of soil moisture on nitrogen functional genes. However, the community structure of nitrogen functional bacteria in bioretention system soils was more complex (Chen et al. 2018). This makes its unsuitable to fully refer to the above findings for bioretention system. In addition, there have been few studies focused on the microorganisms in the bioretention system on the nitrogen cycle (Morse et al. 2018; Waller et al. 2018) and how the environmental factors affected the soil microbes (Li et al. 2021), especially under alternating dry and wet conditions.

Therefore, in this study, a series of laboratory experiments was designed to investigated the effect of different ADD on the nitrogen functional genes and soil microorganisms in bioretention systems. The main objective of this study was to analyze the change in nitrogen functional genes (amoA, nirS and nirK) and the soil microorganisms under different ADD. The results will clarify the relationship between nitrogen functional genes and environmental factors under alternating wet and dry conditions in bioretention, which may help to improve the nitrogen removal in bioretention systems.

For each filter column, the mature Cyperus alternifolius L. was both roughly the same root density and stem length. After cultivation, the plants were irrigated with tap water for a period of time to eliminate the interference of background on the experiment. After the adaptation, simulated runoff rainwater was added according to the inflow cycle.

According to the statistical data for rainfall interval with daily rainfall greater than 2 mm in Chongqing, China during 2011–2017 (typical year), 1 d, 2 d, 3 d and 5 d were the most frequent rainfall intervals, accounting for 60.98%, 10.59%, 7.94% and 6.47%, respectively (Supplementary Material). Considering the effects of general drought conditions and long-term drought conditions on the bioretention system, 7 d, 12 d and 22 d were added as the experimental rainfall interval days.

The simulated stormwater runoff consisting of actual sediment and dechlorinated tap water was prepared before each experiment, the sediment was collected from a stormwater pond in Chongqing, China. The target concentrations for contaminants matched those found in typical stormwater runoff from Chongqing's municipal roads in urban areas (Chen et al. 2021a), as shown in Table 1. The experiment lasted for 4 months (September 26, 2018–January 15, 2019), and the sample collection information is shown in Table 2.

When not sampling, the dosing volumes (V) were 4.1 L based on 75% ATRCR. When sampling, these were 4.6 L based on 80% ATRCR, and is slightly greater than the sampling volume.

Water samples were collected in the middle of each month. Clean collecting buckets were used to collect the outflow when the water outlet was stable, and a plastic needle syringe with volume of 25 mL was used to collect pore water through the sampling points of each filter column at depths of 10, 20, 30, 40, 50 cm. About 100 mL of water samples were taken to determine the contents of TN, NH₄⁺-N and NO₃⁻-N in the pore water. The TN, NH₄⁺-N and NO₃⁻-N in water samples were measured in accordance with the standard methods (SEPA 2002).

Soil samples were collected after the 4-month experiment, and were divided into three portions and packaged in separate sterile sealed bags for the determination of soil physicochemical properties and microbial-related parameters. Soil moisture content was determined by measuring the weight of the soil before and after drying at 105 °C for 48 h (Yin et al. 2019). Soil pH was measured using a pH meter (Hach HQ11D meter) in a solution, while volume ratio of soil to water was 1:2.5 (Zheng et al. 2014). Soil organic matter (SOM) was measured by HACH DR6000 (Hach, USA) after potassium dichromic digestion, and soil TN was determined by semi-micro Kjeldahl digestion (Lu et al. 2015). Soil NH₄⁺-N and NO₃⁻-N were extracted by potassium chloride solution-spectrophotometry (Chen et al. 2015; Lu et al. 2015).

After the soil samples were collected, the soil layer was gently washed with tap water. Then the stems, leaves and roots of plants were separated after cleaning and air-drying, finally, the root length (cm) and the plant root biomass, expressed as fresh weight (g), were measured.

The DNA was extracted using a DNA extraction kit (MP Biomedicals, USA) by following the manufacturer's instruction. The DNA which was extracted from the soil microbes was thawed on the ice, thoroughly mixed and centrifuged, and the purity and relative concentration of DNA were detected using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). The integrity of DNA was detected by agarose gel electrophoresis with 1% concentration (5 V/cm, 20 min). The diluted genomic DNA was used as the template and the 16S rDNA was amplified with the primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 553R (5′-TTACCGCGGCTGCTGGCAC-3′) (Ludwig 2007). The amplified soil genomic DNA was sent to Shanghai Majorbio Bio-pharm Technology Co., Ltd for high-throughput sequencing using the Illumina MiseQ PE300 system (Illumina, United States USA).

The abundance of amoA and denitrification genes was detected using an ABI 7500 Thermocycler System (Applied Biosystems, United States). The amoA gene amplification used bamoA1F (5′-GGGGTTTCTACTGGTGGT-3′) and bamoA2R (5′-CCCCTCKGSAAAGCCTTCTTC-3′) (Rotthauwe et al. 1997) as primers. The nirS gene amplification used cd3aF (5′-GTSAACGTSAAGGARACSGG-3′) and R3cdR (5′-GASTTCGGRTGSGTCTTGA-3′) as primers (Throbck et al. 2004). The nirK gene amplification used 1aCuF (5′-ATCATGGTSCTGCCGCG-3′) and R3Cu (5′-GCCTCGATCAGRTTGTGGTT-3′) as primers (Hallin et al. 1999). The master mixture was as follows: 10 μL ChamQ SYBR Color qPCR Master Mix (2×) (Vazyme Biotech Co., Ltd, China), 0.8 μL forward primer F (5 μmol/L) (Hongxun Biotechnologies Co., Ltd, China), 0.8 μL reverse primer R (5 μmol/L) (Hongxun Biotechnologies Co., Ltd, China), 1 μL template DNA, 7.4 μL ddH₂O, 20 μL in total. The thermal cycling protocol for amoA was as follows: initial denaturation at 95 °C for 5 min for one cycle, and 95 °C for 30 s, annealing at 55 °C for 30 s, 72 °C for 1 min, all for 35 cycles. The thermal cycling protocol for nirS and nirK were as follows: initial denaturation at 95 °C for 5 min for one cycle, and 95 °C for 30 s, annealing at 50 °C for 30 s, 72 °C for 1 min, all for 35 cycles.

The constructed plasmids were diluted by 10 times gradient (90 μL dilution + 10 μL plasmid) after sequencing identification. Then, 10⁻²–10⁻⁹ diluents of amoA (7.5 × 10⁹ copies/μL) and nirS (1.2 × 10¹⁰ copies/μL) samples and 10⁻²–10⁻⁷ diluents of nirK (1.7 × 10¹⁰ copies/μL) samples were selected to prepare the standard curve. When performing PCR detection, the 96-well plates with the added samples were placed in an ABI 7500 fluorescence quantitative PCR instrument for reaction. The copy number per unit volume was calculated according to Ct value and standard curve, and the copy number per unit mass was calculated according to DNA mass and volume.

In this study, Microsoft Excel 2016 was used to process the test data. Microsoft Excel 2016 and Origin Pro 9.0 were used to make the chart. Redundancy Analysis (RDA) and Canonical Correspondence Analysis (CCA) were used to analyze the relationship between environmental factors on bacterial communities and copy number of three nitrogen functional genes by using R (R Programming Language) and RStudio program.

Figure 2 also indicated that the contents of TN, NH₄⁺-N, NO₃⁻-N and SOM in the soil decreased with the increase in depth, The order of their concentrations at different depths was as follows: surface layer > planting layer > submerged layer. The moisture content increased with depth, except for ADD12 and ADD22 treatment groups, the free water in the submerged layer was completely lost due to the long dry period, part of the reason was that the sand in the submerged layer was poor in holding water (Smagin et al. 2021), while the top soil lost most of its water due to the high temperature in summer. Therefore, the moisture content of the planting layer was the highest in the treatment group with long drought periods. The contents of TN, NH₄⁺-N, NO₃⁻-N and SOM in the soil decreased with the increase in depth, which was caused by the retention of nutrients in the bioretention system (Mehmood et al. 2021). Due to the use of wood chips as an additional carbon source in submerged layer, SOM content in partial submerged layer was higher than in the planting layer and surface layer.

The frequent wetting and drying conditions led to bioretention and became an anoxic or anaerobic state, while organic nitrogen can be converted into NH₄⁺-N through mineralization in an anaerobic environment. Therefore, even when ADD = 1, the nitrogen removal efficiency in the system was still not high. The removal of NH₄⁺-N in a bioretention system mainly depended on the assimilation of plants and soil adsorption (Skorobogatov et al. 2020). The long-term drought reduced plant activity and prevented bacteria reactivating quickly, this reduced the removal efficiency of NH₄⁺-N in bioretention. It can be seen that the change of the ADD can affect the removal efficiency of NH₄⁺-N by changing the soil moisture content in the system.

The results showed that when the ADD was less than 5 days, the TN removal efficiency was increased with the increasing ADDs, the average removal efficiency reached 72.51%. With the increasing in drought days, the removal efficiency gradually decreased, and the worst TN removal efficiency was at ADD22.

In the experiment on the spatial distribution of pollutants in the bioretention system (supplementary material), when ADD = 1, 2 and 3, NH₄⁺-N was mainly removed in the surface layer and planting layer of the system. Results showed that NH₄⁺-N concentration first decreased continuously in the planting layer (0–30 cm) and the upper part of the submerged layer (30–40 cm), when ADD = 5, 7, 12 and 22. Compared with ADD = 1, 2 and 3, the soil moisture content was lower in the groups ADD = 5, 7, 12 and 22, so the oxygen content was higher at the upper part of the submerged layer (30–40 cm), and nitrification reaction occurred at this soil depth. The removal efficiency of NH₄⁺-N of the system was low at this time, because of the increase of drought time. Meanwhile, the plant roots withered and the death of microorganisms increased the SOM and promoted the process of DNRA (Huygens et al. 2007).

When ADD = 1, 2 and 3, NO₃⁻-N was mainly removed in the planting layer due to the high oxygen content in the surface layer. When ADD = 5, NO₃⁻-N concentration continued to decrease in the upper part of the planting layer and submerged layer, which indicated that the denitrification process mainly occurred in this interval. When ADD = 7 and 12, due to the death of microorganisms in plant roots caused by high temperature in summer, NO₃⁻-N concentration increased first and then decreased, indicated that denitrification mainly occurred in the submerged layer. When ADD = 22, NO₃⁻-N might be mainly removed in the drainage layer.

In conclusion, the different ADD can affected the removal efficiency of nitrogen pollutants by changing the distribution of soil moisture and oxygen in bioretention system.

In the vertical direction of the system, the copy number of amoA in the surface layer ranged from 5.16 × 10² to 1.34 × 10⁵ copies/g, and 4.35 × 10³–1.37 × 10⁵ copies/g in the planting layer, in the submerged layer, the maximum was 1.88 × 10⁴ copies/g. nirK copy numbers ranged from 7.83 × 10⁵–6.33 × 10⁶ copies/g in the surface layer and 8.43 × 10⁵–2.61 × 10⁷ copies/g in the planting layer. The nirK copies of the submerged layer ranged from 6.91 × 10⁴ to 2.32 × 10⁷ copies/g. The copy number of nirS ranged from 1.15 × 10⁶ to 3.22 × 10⁸ copies/g in the surface layer, and 4.95 × 10⁶ to 1.95 × 10⁸ copies/g in the planting layer, 3.02 × 10⁴ to 5.1 × 10⁷ copies/g in the submerged layer. The results showed that the copy number of nirS gene in each layer was more than the nirK, combined with the nitrogen removal efficiency, it can be seen that the removal efficiencies of TN and NO₃⁻-N were about the same as the nirS gene copy number, but not for the nirK gene. Therefore, it can be speculated that the nirS-type denitrifiers may have a greater impact on the performance of nitrogen removal in bioretention. Some studies have also found that the nirS-type denitrifiers were the main functional groups controlling the nitrogen removal (Zeng et al. 2016; Zhang et al. 2021b). In addition, some studies have shown that the copy numbers of both nitrification and denitrification genes in the bioretention system decreased with the increasing soil depth (Chen et al. 2013), however, the results of this study showed a trend of increasing first and then decreasing with the increasing depth. This may be because the experiment was carried out in summer, which led to the rapid evaporation of the soil water in the surface layer and caused the death of bacteria. Meanwhile, according to the proportion of the three genes in each layer, with the increase in ADD, the bacteria corresponding to the three functional genes all showed a tendency to migrate down in the system (supplementary material).

Clostridium_sensu_stricto_1 and Anaerolinea most exist in the submerged layer, and decreased with ADD increasing within 5 days. Clostridium has good NO₃⁻-N reduction ability and is were strongly correlated with the nitrogen removal efficiency of the system (Subedi et al. 2017). Meanwhile, it has been reported that Clostridium can also occur in the DNRA process in an anaerobic environment (Shengjie et al. 2018). Saccharimonadales which was related to the ammonia oxidation process and amoA gene, had a trend of gradually transferring to the lower part of the system with the increase in ADD, and its proportion of bacteria gradually decrease, which can also explain the change in ammonia nitrogen removal efficiency in the system to a certain extent.

The results showed that different ADD not only affected the physicochemical properties of soil, pore water and plant growth, but also changed the distribution and abundance of nitrogen functional bacteria in the bioretention system. The overall copy numbers of the three most important nitrogen functional genes (amoA, nirK and nirS) in the bioretention system gradually decreased with the increase in ADD (supplementary material), and their spatial distribution showed a first increasing and then decreasing trend. The three functional genes (amoA, nirK and nirS) were mainly distributed in the planting layer, and gradually migrated to the submerged layer with the increase in ADD. The copy number of amoA was closely related to the content of NH₄⁺-N in soil and pore water, while the copy number of nirK was closely related to plant root length and biomass. The copy number of nirS was most affected by soil TN, NH₄⁺-N and SOM. Therefore, different ADDs can change the spatial distribution and abundance of nitrogen functional genes by influencing the physicochemical properties of soil, pore water and the growth of plants in the bioretention system, thus affecting the removal efficiency of nitrogen pollutants in the bioretention system.

The authors are grateful for the financial support from the National Natural Science Foundation of China (51709024), the Construction Science and Technology Project of Chongqing (CKZ2020 No. 5-7), the Natural Science Foundation of Chongqing Municipal Science and Technology Commission (cstc2020jcyj-msxmX1000), the Natural Science Foundation Project of CQ CSTC (cstc2020jcyj-msxmX0716), and Venture & Innovation Support Program for Chongqing Overseas Returnees (cx2017065).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.