Removal effect of pollutants from stormwater runoff in shallow bioretention system with gramineous plants

The urbanization and industrialization of China have resulted in a concerning issue of water eutrophication, whereby many urban water bodies have become contaminated to varying degrees (Ahmed et al. 2019; Wang et al. 2022). The above changes ultimately lead to a significant increase in the volume of urban stormwater runoff and a significant reduction in the convergence time, causing great impact and pressure on the existing urban drainage system (Sharma et al. 2021). Stormwater runoff carries pollutants from urban surfaces including the atmosphere, rooftops, sidewalks, green spaces, and pavements (Fan et al. 2022; Pochodyła-Ducka et al. 2023). The polluted stormwater runoff is eventually collected and discharged into the urban receiving waters through the drainage network, resulting in serious pollution of the urban river and lake systems, greatly damaging the aquatic environment of the city, and seriously affecting the daily lives of residents around the receiving waters.

Traditional stormwater runoff management measures have focused solely on reducing peak flow rates. In recent years, methods have emerged that utilize the natural processes of vegetation and soils to improve runoff quality, with common measures including permeable pavements, constructed wetlands, bioretention, and green roofs (Barbosa et al. 2012; Rentachintala et al. 2022; Zhao et al. 2023). As one of the most effective methods, bioretention is also known as biofiltration or bioswales and has been widely used due to the potential to improve runoff quality and control the quantity (Vijayaraghavan et al. 2021; Beral et al. 2023; Nazarpour et al. 2023). Domestic and foreign researchers have researched the type of filler, filler layer thickness, plant species, water purification effect, etc. Xiong et al. (2019) evaluated bioretention cells with iron-coated biochar (ICB)-amended media and found that this experimental column showed the best performance for chemical oxygen demand (COD), ammonia nitrogen (⁠-N), and total phosphorus (TP) removal (94.6, 98.3, and 93.70%, respectively).

Previous research showed that the traditional bioretention system has a good removal effect on COD, suspended solids (SS), TP, etc. However, the efficiency of nitrogen removal in the bioretention system is unstable, owing to the different chemical properties of various forms of nitrogen and the limitations of the current bioretention system for nitrogen transformation (Zhang et al. 2021). Plants are an essential part of bioretention cells, which can directly or indirectly treat runoff pollutants. Most researchers have focused on different plant species or combinations of them. Técher & Berthier (2023) comprehensively compared plant species, functional types, and traits associated with the infiltration process in biofilters, and the results converged to a positive impact of plants on water infiltration and percolation. Zuo et al. (2020) found that ammonia nitrogen removal in the bioretention cell with Lythrum salicaria L. was the highest (88.1%), which proved aquatic plants could improve nitrogen removal from stormwater in bioretention cells. Stagge et al. (2012) found that green swales planted with a mixture of three gramineous plants significantly reduced pollutant mass and mean concentrations in stormwater. Meanwhile, it has been confirmed that some aquatic plants have obvious seasonal variations in nitrogen uptake and the absorption reached its maximum value from June to August and gradually decreased to zero from September to October (Yu & Qin 2023). Compared to aquatic plants, gramineous plants can adapt to temperature changes in different seasons, making them a favored choice for treating stormwater runoff from pavements (Lange et al. 2020; Zuo et al. 2020). Moreover, gramineous plants have been increasingly employed in the agricultural industry due to their exceptional durability, versatility, affordability, and low maintenance requirements. However, fewer studies have systematically compared the effects of seasonal gramineous plants on bioretention systems and selected optimal plants for comparison with common aquatic plants.

Therefore, a shallow bioretention system was constructed with two cool-season plants (ryegrass and bermuda), and two warm-season plants (bahiagrass and green grass). The growth traits and nutrient absorption ability of four gramineous plants were analyzed and their tolerance, enrichment, and transportation capacity were also evaluated to make a comparison between plant species and their absorptive capacity of heavy metals (Cu, Pb, and Zn). Finally, the optimal grass was selected and three groups of bioretention systems were set up to compare the purification effect of bioretention cells under different rainfall intensities, providing technical references for the practical application of gramineous grasses and exploring more appropriate grass configurations in bioretention systems.

Gramineous plants have the advantages of short growth periods, many optional species, and dense roots, which can be intercropped with other lawn plants in different seasons to realize the rapid change of plants and enhance the retention efficiency of pollutants simultaneously. Therefore, four kinds of gramineous plants (ryegrass, bermuda, bahiagrass, and green grass) were selected as the common plants in bioretention cells, we researched the germination rate, growth condition, and the kinetic characteristics of the root system on the uptake of nitrogen, phosphorus, and heavy metals, and then optimized the plants planted in the bioretention cells.

The experimental influent is typically classified as natural rainwater and synthetic rainwater (Jiang et al. 2019). However, the collection of natural rainwater poses challenges as it cannot be stored for long periods and does not ensure consistency of other indicators such as influent concentration (Sharma & Malaviya 2021). To reduce the high variability and uncertainty caused by the actual rainwater quality, this experiment selected synthetic rainwater to ensure the consistency of the influent concentration based on the results of monitoring the water quality of rainwater runoff in Nanjing.

The rainwater was simulated by potassium dihydrogen phosphate (KH₂PO₄), potassium phosphate (KNO₃), ammonia chloride (NH₄Cl), lead chloride (PbCl₂), zinc chloride (ZnCl₂), and copper chloride (CuCl₂). The main pollutant concentrations in simulated rainwater were 1 mg/L -P, 3 mg/L -N, 5 mg/L -N, 0.5 mg/L Pb, 0.5 mg/L Zn, and 0.6 mg/L Cu with 0.1 mmol/L CaSO₄ solution as the electrolyte, an appropriate amount of four plants after 20 days of growth was transferred into a plastic tray containing 2 L rainwater, and samples were taken at the 0, 0.5, 1, 2, 4, 6, 8, 10, 20, 26, 32, 38, 44, and 48 h, respectively. The pollutant indexes of water samples tested in this experiment included -N, total nitrogen (TN), TP, Cu, Pb, and Zn. The measurement methods for each pollutant index referred to standard methods for the examination of water and wastewater (Rice et al. 2012).

The experiment was divided into the test group and the control group. We recorded plant height, root length, and biomass to determine the growth condition. The above and underground structural morphology of the plants were observed, and plant height and root length were measured; plant biomass was calculated by the drying method, in which the dry mass of plants was weighed after drying for 1 h at 105 °C; the growth quality of the experimental plants was assessed by color, density, and texture, with the higher values representing the better growth condition.

Suppose that X → 0 in Equation (5), at this point Y' = b, i.e., the maximum rate of change in concentration. Then the maximum absorption rate per unit root mass is I_max = bV/W_d, where V is the volume of the absorbing solution, and W_d is the root dry weight. Substitute Y' = b/2 into Equation (5), X is obtained when the absorption rate reaches half of the maximum absorption rate, and then substitute X into Equation (4), at this point, the value of Y is K_m.

The data were analyzed by SPSS 21.0 and plotted by Origin 2021. One-way analysis of variance (ANOVA) was used to test the significance of each experimental indicator, and the significance of difference was defined as P < 0.05. The statistical values are expressed as mean values in sections 3.1–3.3 and mean-standard values in section 3.4.

Among them, system 1# was unplanted, and systems 2# and 3# were planted with calamus and ryegrass, respectively. In this paper, the pollutant purification efficiency of the three groups of bioretention systems under different rainfall intensities (light, moderate, and heavy rainfall) was investigated through the simulated rainfall test, which referred to the model of You et al. (2019).

The plant height, biomass, and growth quality of the four grasses after 30 days of growth are shown in Table 1. The plant heights of ryegrass and bermuda were significantly higher than that of the control group, indicating that the growth of these two grasses was not inhibited by heavy metals in stormwater runoff, while the plant heights of bahiagrass and green grass showed the opposite trend of growth inhibition compared to those of ryegrass and bermuda. It was also reported that the inhibitory effect became more pronounced as heavy metal pollution levels in stormwater increased (Jhonson et al. 2023). The biomass of various plants (in terms of overall fresh weight) was also tested under stormwater runoff stress on the 30th day of plant growth, and the overall biomass of bahiagrass and green grass decreased compared to the control group. The biomass of both ryegrass and bermuda showed similar trends and was higher than the control group at the initial runoff concentration.

The absorption kinetics of -N by four gramineous plants in this research differed from the results of other wetland vegetables and plants. The maximum absorption rate of -N by hollow cabbage in Table 3 was higher than 700 μg/(g·h), which was 2–4 times that of other vegetables, 6–8 times that of other grasses, and 6–12 times that of this paper and had a strong absorption capacity. Gong et al. (2019) found bluegrass and tall fescue had a slightly higher absorption capacity than that of bahiagrass and bermuda in this research, which might be related to the absorbed water quality. Based solely on the removal of -N, it has been determined that ryegrass is the most effective bioretained plant in comparison to others.

Table 3

Absorption kinetic characteristics of different plants

Plants .	Imax (μg/(g·h)) .		Km (mg/L) .		Reference .
	TP .	-N .	TP .	-N .
Romaine lettuce	28.21	162.71	1.39	10.28	Han & Lyu (2017)
California burclover	36.61	331.81	1.54	11.35	Han & Lyu (2017)
Water spinach	81.21	706.82	0.47	8.03	Han & Lyu (2017)
Potherb mustard	2.63	183.38	1.44	13.56	Han & Lyu (2017)
Taro	3.06	23.49	1.62	12.69	Han & Lyu (2017)
Leek	74.6	246.23	1.53	10.69	Han & Lyu (2017)
Water bamboo	32.48	/	1.40	/	Xie et al. (2016)
Canna generalis	116.4	/	3.56	/	Tang et al. (2011)
Cyperus papyrus	58.2	/	5.63	/	Tang et al. (2011)
Colocasia tonoimo	145.5	/	4.77	/	Tang et al. (2011)
Phragmites australis	37.78	/	6.18	/	Zhang et al. (2014)
Typha orientalis	29.61	/	6.26	/	Zhang et al. (2014)
Scripus triqueter	64.31	/	6.81	/	Zhang et al. (2014)
Agrostis stolonifera	48.86	89.47	3.15	12.61	Xie et al. (2016)
Annual bluegrass	62.71	101.85	2.59	11.81	Xie et al. (2016)
Festuca arundinacea	63.35	102.67	3.05	11.19	Xie et al. (2016)

Plants .	Imax (μg/(g·h)) .		Km (mg/L) .		Reference .
	TP .	-N .	TP .	-N .
Romaine lettuce	28.21	162.71	1.39	10.28	Han & Lyu (2017)
California burclover	36.61	331.81	1.54	11.35	Han & Lyu (2017)
Water spinach	81.21	706.82	0.47	8.03	Han & Lyu (2017)
Potherb mustard	2.63	183.38	1.44	13.56	Han & Lyu (2017)
Taro	3.06	23.49	1.62	12.69	Han & Lyu (2017)
Leek	74.6	246.23	1.53	10.69	Han & Lyu (2017)
Water bamboo	32.48	/	1.40	/	Xie et al. (2016)
Canna generalis	116.4	/	3.56	/	Tang et al. (2011)
Cyperus papyrus	58.2	/	5.63	/	Tang et al. (2011)
Colocasia tonoimo	145.5	/	4.77	/	Tang et al. (2011)
Phragmites australis	37.78	/	6.18	/	Zhang et al. (2014)
Typha orientalis	29.61	/	6.26	/	Zhang et al. (2014)
Scripus triqueter	64.31	/	6.81	/	Zhang et al. (2014)
Agrostis stolonifera	48.86	89.47	3.15	12.61	Xie et al. (2016)
Annual bluegrass	62.71	101.85	2.59	11.81	Xie et al. (2016)
Festuca arundinacea	63.35	102.67	3.05	11.19	Xie et al. (2016)

View Large

The relationship between TN concentration in stormwater runoff and absorption time is shown in Figure 3(b), with the prolongation of the absorption time, the TN concentration in rainwater gradually decreased and the absorption rate gradually slowed down, but there was no elevated trend during the absorption cycle compared to the absorption of P by grasses. As shown in Table 2, I_max of TN ranged from 61.6 to 131.9 μg/(g·h), and the order of TN absorptive capacity of the four grasses was as follows: Ryegrass > Bermuda > Green grass > Bahiagrass, and there was no significant difference between the I_max of Bahiagrass and Green grass (P > 0.05); the K_m ranged from 17.5 to 27.3 mg/L with the following order: Green grass > Ryegrass > Bermuda > Bahiagrass. The N absorption of the four gramineous plants examined in this experiment had no significant edge over vegetable wetland plants and aquatic flowering plants, but the continuous decrease in rainfall concentration indicated that the application of gramineous plants to the bioretention system could reduce the N concentration. Therefore, appropriate plants should be selected for the configuration in practical engineering based on the characteristics of the catchment subsurface.

The relationship between TP concentration and absorption time in rainwater is shown in Figure 3(c), and results showed that TP concentration gradually decreased with the growth of absorption time, the concentration reached equilibrium at about 65 h, and then slowly increased. Ryegrass showed an obvious advantage in absorbing P, which was the faster absorption rate of grass in the test cycle. The absorption kinetic parameters of P by plants are shown in Table 2. There were large differences in the absorption kinetic characteristics of P among the four grasses, I_max ranged from 22.1 to 42.4 μg/(g·h), and the absorptive capacity of the four grasses was in the following order: Ryegrass > Green grass > Bermuda > Bahiagrass. K_m ranged from 3.81 to 5.19 mg/L and the affinity of the four grasses was in the order of Ryegrass > Green grass > Bermuda > Bahiagrass. According to I_max and K_m values proposed by the Michaelis–Menten equation to evaluate the purification ability of grasses for water nutrients, the larger I_max indicated that grasses had a greater potential for absorbing certain ions, and the smaller K_m indicated that grasses had a stronger affinity for certain ions. Ryegrass had the largest I_max and the smallest K_m, so it had a better absorption ability for water with high and low levels of pollution, and it was adapted to the wastewater containing a variety of conditions, while I_max of bermuda was the lowest, indicating its weak absorption capacity for P. Stormwater runoff treated with permeable pavements contained low concentrations of phosphorus and was suitable for treatment with retention facilities planted with green grass and ryegrass.

In this research, the absorption kinetics of P by four gramineous plants differed from those of other plants. Table 3 shows the absorptive characteristics of N and P by plants commonly used in ecological facilities. The results of Han & Lyu (2017) showed that the water spinach was suitable for wastewater with a greater variation in concentration, leeks, and potherbs were better suited to treating wastewater with higher concentrations, and taro was the weakest in terms of absorptive capacity, and that planting aquatic vegetables in an artificial wetland could achieve good environmental and economic benefits. Zhang et al. (2014) found that large landscape plants such as Phragmites australis, Typha orientalis, and Scirpus triqueter were more suitable for water bodies with higher pollution concentrations despite their large uptake rates (>50 μg/(g·h)), but also relatively large K_m values.

The TC is calculated to evaluate the ability of grasses to transfer and enrich different heavy metals from the roots to the stem and leaves (Zarek et al. 2020), the number of heavy metals transferred from the lower to the upper part of the grass increases when the transfer coefficient of the grass is higher. The TC of the heavy metals Cu, Pb, and Zn in stormwater runoff by gramineous plants are shown in Figure 5(a), the TC of four gramineous plants for the three heavy metals were all less than 1, indicating that gramineous plants had poor ability to transfer the heavy metals from the roots in stormwater runoff, with the lowest for bermuda and the highest for bahiagrass; the transport coefficients of four gramineous plants were the lowest for Zn (<0.55) and the highest for Cu (>0.6). Hussain et al. (2021) found that clover had a transport value of greater than 1 for Cd added to the soil at a concentration of 5–10 mg/kg. Simek et al. (2018) found that Chlorophytum comosum had a TC of less than 1 and a poorer transfer ability under Cd concentration of 10 mg/kg. These findings have proved that the tolerance and transport ability of plants to heavy metals is not only related to the heavy metal content in the medium, but also closely related to the plant species, climatic conditions, and the physicochemical properties of the culture medium. Therefore, further research is necessary to investigate the mechanisms behind the influence of these factors.

Figure 7(b) shows the removal effect of the bioretention system for TN in runoff rainwater, the removal rate of systems 2# and 3# showed a positive trend with rainfall intensity, and system 1# showed a negative correlation, but the difference in the removal rate of the different systems at different rainfall intensities was small, only about 16%. With the gradual increase of rainfall intensity, the TN removal rate of 1# decreased from 57.6 to 42.8%, mainly due to the shorter contact time between rainwater and substrate when the rainfall was higher, and the removal effect was poor. The complex roots of the plants in 2# and 3# improved the permeability performance of the filler layer to make it have the effect of water retention and storage and thus promoted the denitrification process. In Figure 7(c), due to the complex system of plants and microorganisms on -N removal, the removal rate of -N in systems 2# and 3# increased with the increase of rainfall intensity, and the average removal rate increased from 31.2 to 48.6% and 34.6 to 58.7%, respectively. The effluent concentration of system 1# was the highest under different rainfall intensities, and the average removal rate decreased from 30.6 to 22.9%, mainly because microorganisms and plants play a dominant role in the ammonia oxidation process, and the filler has little effect on ammonia nitrogen removal (Xiong et al. 2019).

The removal effects of three groups of bioretention systems on heavy metals (Cu, Pb, and Zn) are shown in Figure 7(d)–7(f). As can be seen from the figure, under different rainfall intensities, the differences in the removal effects of the three systems on Cu were more obvious, with the removal rates ranging from 30.1 to 60.2% in heavy rainfall, and about 61.2 to 80.3% and 68.6 to 85.1% in moderate and light rainfall, respectively, and the effluent concentrations were all lower than 0.31 mg/L. The overall removal effect of system 3# was significantly higher than that of systems 1# and 2#, mainly due to the short-term uptake effect by plants. For Pb, the removal rates ranged from 35.6 to 61.1% in heavy rainfall, and about 52.2–75.9% and 58.3–85.4% in moderate rainfall and light rainfall, respectively, and the order of the removal effect of the three groups of systems on Pb was: 3# > 2# > 1#. The removal efficiency and removal amount of Zn by the three bioretention systems under the same rainfall intensity were better than that of Cu and Pb, and the removal rate of Zn in heavy rainfall ranged from 41.2 to 61.3%, and the removal rates in moderate rainfall and light rainfall were about 46.4 ∼ 79.1% and 54.7 ∼ 87.2%, respectively. Overall, the removal effects of the three systems on Zn were in the order of 3# > 2# > 1#. The degradation ability of heavy metals was positively correlated with the rainfall intensity, the higher the rainfall, the higher the pollution load and the faster the flow rate, the faster the rainwater runoff would pass through the substrate layer, which reduced the hydraulic retention time of the rainwater runoff, decreased the rapid adsorption and filtration of the substrate for heavy metals, and cut down the treatment effect (Mei et al. 2020).

• (1)

  The results of plant optimization in the bioretention system indicated that the order of TN absorptive capacity of the four grasses was as follows: Ryegrass > Bermuda > Green grass > Bahiagrass; the order of TP absorptive capacity of the four grasses was as follows: Ryegrass > Green grass > Bermuda > Bahiagrass; ryegrass grew better under higher concentration runoff and had a high tolerance, enrichment, and transport capacity to three heavy metal pollutants, which could be used in retention facilities.

• (2)

  By analyzing the kinetic characteristics, I_max of P and -N ranged from 22.1 to 42.4 μg/(g·h) and 65.49 to 104.8 μg/(g·h), respectively, and the absorptive capacity of the four grasses was both in the following order: Ryegrass > Green grass > Bermuda > Bahiagrass; K_m of P ranged from 3.81 to 5.19 mg/L and the affinity of the four grasses was in the order of Ryegrass > Green grass > Bermuda > Bahiagrass; the K_m ranged from 15.81 to 35.85 mg/L, and the order of affinity of the four grasses was: Bermuda > Ryegrass > Green grass > Bahiagrass. Overall, ryegrass showed the strongest absorptive capacity for P and -N and bahiagrass had the weakest absorptive capacity for P and -N.

• (3)

  According to the absorptive ability of different grasses to heavy metals in stormwater runoff, ryegrass, green grass, bermuda, and bahiagrass removed 78.4, 46.1, 53.2, and 56.1% of Cu, 59.4, 50.1, 47.1, and 39.1% of Pb, and 51.3, 35.6, 34.6, and 32.4% of Zn, respectively; the absorption capacity of grasses for Pb was ranked as follows: Ryegrass > Bermuda > Green grass > Bahiagrass, and that for Cu was ranked as follows: Ryegrass > Green grass > Bermuda > Bahiagrass. The results revealed that ryegrass had the strongest absorption capacity for heavy metals in rainwater.

• (4)

  According to the simulated rainfall test in bioretention cells, the removal effects of three groups of bioretention systems on heavy metals Cu, Pb and Zn were in the order of 3# > 2# > 1#; the purification ability of system 3# planted with gramineous plants for various pollutants was significantly better than that of system 1# without plants.

This research was supported by the Xinjiang Biomass Solid Waste Resources Technology and Engineering Center of China (KSUGCZX2022), Lianyungang Key Research and Development Plan (Social Development) project of China (SF2130), Postgraduate Research & Practice Innovation Program of Jiangsu Province (SJCX23_1814, SJCX23_1816, and KYCX2023-24), and The Natural Science Foundation of the Jiangsu Higher Education Institutions of China (22KJB560001).

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

The authors declare there is no conflict.