Introduction

Nonpoint source (NPS) pollutants (phosphorus, nitrates, pesticides, pathogens, organic matter, colloids, and metals) severely threaten ecological environments and water resources (Arora et al. 1996; Withers and Haygarth 2007; Rasmussen et al. 2011; Yu 2012; Wu et al. 2014; Muñoz-Carpena et al. 2018). Rivers and streams in China are deteriorating because of the extensive use of fertilizers and pesticides, which is severely affecting human health. Thus, the Chinese government and researchers have focused on the management of water resources to control NPS pollutants.

Vegetative filter strips (VFSs), defined as vegetative areas situated downslope of agricultural lands, have been recommended as the best management practice for controlling NPS pollutants (Abu-Zreig et al. 2004; Dosskey et al. 2007; Kuo and Muñoz-Carpena 2009; Fox et al. 2018). VFSs are widely used in the United States and Europe, and they are relatively cost effective and require little maintenance other than biomass harvesting. VFSs reduce the transport of sediment and pollutant loading to nearby surface water bodies primarily through interception, infiltration, deposition, adsorption, and absorption (Muñoz-Carpena et al. 1999; USDA 2000; Dorioz et al. 2006; Blanco and Lal 2008; Kuo and Muñoz-Carpena 2009; Giri et al. 2010; Muñoz-Carpena et al. 2010; Yang et al. 2015).

The effectiveness of VFSs in reducing NPS pollutants transport depends on the characteristics of the VFSs (width, slope, vegetation, height, and density), inflow (runoff volume and velocity), rainfall intensity, and soil properties (pH, moisture, and particle size) (Reichenberger et al. 2007; Liu et al. 2008; Kuo and Muñoz-Carpena 2010; Lambrechts et al. 2014; Rahman et al. 2014). Sediment retention efficiency increases logarithmically with an increase in vegetation density (Abu-Zreig et al. 2004). A plant’s morphology and growth dynamics also influence the sediment and total phosphorus (TP) retention efficiencies (Krutz et al. 2005; Xiao et al. 2011). A VFS with a slope between 3 and 12% was reported to reduce sediment by 79–56% (Wang et al. 2008), and a lower slope resulted in a higher retention effect of total nitrogen and TP (Pan and Deng 2003). Researchers have reported that the effect of rainfall intensity was considerably greater than that of the inflow rate on the sediment transport rate (Guy et al. 1987; Wen et al. 2015).

Riparian VFSs are located along river networks or streams, where shallow groundwater tables are frequently present (Reichenberger et al. 2007; Lacas et al. 2012; Carluer et al. 2017). A shallow groundwater table affects the surface hydrological response of a VFS by generating soil water saturation during rainfall, which limits the infiltration capacity of a VFS (Lauvernet and Muñoz-Carpena 2018). Without a shallow groundwater table in fine soils, the effectiveness of a VFS in reducing NPS pollutants is typically controlled by excess rainfall (Shin and Gil 2014). With a shallow groundwater table, the length of a VFS and the soil characteristics dominate the VFS’ surface hydrological response (Carluer et al. 2017). Simpkins et al. (2002) reported that the position of the water table in thin-sand aquifers underlying a VFS was probably the most pertinent factor for determining VFS performance. Dosskey et al. (2001, 2006) also reported that runoff pollutants can interact with a VFS root zone in a shallow groundwater table (< 1.8 m). Lauvernet and Muñoz-Carpena (2018) found that the runoff and pesticide retentions of a VFS were sensitive to a shallow water table depth (WTD) within a certain range (0–1.5 m from the surface based on the soil characteristics in their study). Moreover, variations in shallow WTD, which influence how the zones of aeration and saturation are distributed in subsurface soil, play a critical role in controlling nutrient cycling in soils and influencing the qualities of groundwater and adjacent streams (DeVito et al. 2000; Hill et al. 2004; Schilling and Jacobson 2014). Fox et al. (2018) called for further research on VFS effect on sediment and nutrient trapping efficiency under shallow water tables, indicating the relevance of quantifying the effect of shallow groundwater tables on the effectiveness of VFSs.

Infiltrated water in a VFS may accumulate and divert laterally downslope to rivers when it meets a shallow groundwater table (Newman et al. 1998; Zhang et al. 2011; Carluer et al. 2017). This lateral diversion of water flow is known as subsurface lateral flow. Numerous studies have confirmed that subsurface lateral flow is vital to hydrological response and chemical transport across the soil profile (van der Velde et al. 2010; Peyrard et al. 2016). Subsurface lateral flow may affect surface water quality when the adsorbed nutrients leach from the subsurface soil inducing by a rise of groundwater table. Thus, the pollution of subsurface lateral flow toward a stream should also be considered.

This study sought to quantify the effect of a shallow groundwater table on VFS efficiency for retarding the transport of runoff, sediment, phosphorus, and bromide in different grass spacings and slopes. A controlled vegetated soil tank with silt loam soil was placed under a laboratory rainfall simulator to (1) assess the retention efficiency of surface runoff, sediment, phosphorus, and bromide, (2) analyze both the similarities and differences in the transport behavior of phosphorus and bromide in the surface runoff and subsurface lateral flow, and (3) study the transport mechanisms of phosphorus and bromide in the subsurface lateral flow.

Materials and Methods

Laboratory Experimental System

A vegetated soil tank (2 m long, 1 m wide, and 0.6 m deep), placed on a metal shelf with an adjustable slope, was designed to evaluate the effect of a shallow groundwater table on VFS efficiency for retarding the transport of runoff, sediment, phosphorus, and bromide in different grass spacings and slopes (Fig. 1). The 2.0-m length fell within the range of VFS flow lengths reported in previous field and laboratory studies: 0.5 m to 29 m VFS lengths were reported by Sabbagh et al. (2009) and Poletika et al. (2009); 2.0 m length was used by Fox et al. (2018) in laboratory experiment. A water tank (0.2 m × 1 m × 0.5 m) on the lower side of the soil chamber was used to control the groundwater table. Silt loam soil (15.08% clay, 62.47% silt, and 22.45% sand based on the hydrometer method) was packed into the chamber to a bulk density of 1.28 g cm−3. The tested soil was collected at a 0–20 cm depth from a field near the Hanbei River, Hubei Province, China. To represent a typical riparian vegetation in the Jianghan Plain, Shortleaf Lilyturf was established in the laboratory soil tank.

Fig. 1
figure 1

Schematic of the laboratory runoff experimental system containing the collectors for the surface and subsurface lateral flow

Full size image

A peristaltic pump system and a polytetrafluoroethylene tube with uniformly distributed small holes were placed at the upper end of the soil tank to deliver runoff with given concentrations of sediment, phosphorus, and bromide. The VFS experienced a step inflow hydrograph with typical inflow rates of 4.02–4.56 L min−1. Such step inflow hydrographs are commonly used to simplify the experimental procedures in VFS studies. For example, see Sabbagh et al. (2009) for a review of several such studies and Fox et al. (2018) for controlled laboratory experiments.

A rainfall simulator, set at 3.2 m above ground, used a peristaltic pump to supply a constant rainfall intensity of 28 mm h−1 for all the experiments, which represented the average rainfall on the Jianghan Plain, Hubei Province, China (Zhou 2009; Kuo et al. 2019). A rain gauge (TE525 tipping bucket rain gauge, Texas Electronics, Dallas, TX, USA) was installed between the rainfall simulator and the soil box to measure the rainfall intensity.

Experimental Design

The study contained nine experimental treatments (three experimental groups × three WTDs). The three experimental groups were performed with different grass spacings, which were obtained using Eq. (1) (Muñoz-Carpena 1993):

$$G_{s}=100\sqrt{\frac{1}{D_{s}},}$$
(1)

where Gs is the grass spacing (cm) and Ds is the grass density (stems m−2),

and slopes (grass spacing and the slope for G7S05 were 6.69 cm and 5%, grass spacing and the slope for G4S05 were 4.18 cm and 5%, and grass spacing and the slope for G4S15 were 4.18 cm and 15%). Experiments with 0.08-m, 0.22-m, and 0.36-m WTDs were performed for each experimental group. For each treatment, the prepared nutrient solution (the concentrations of sediment, phosphorus, and bromide were 0.20 g s−1, 0.70 mg L−1, and 10.00 mg L−1, respectively) was introduced to the inflow spreader (the running time depends on the results of the adsorption kinetic experiment), and then, the inflow was switched to tap water at the same inflow rate for another 30 min. Water samples from the surface runoff outflow and subsurface lateral flow were collected every 5 min. The volumes of the water samples were used to calculate the surface runoff and subsurface lateral flow rates, and the sediment concentration was measured using the evaporation method at 105 °C. The TP concentration (ppm) was analyzed using the potassium persulfate oxidation–molybdenum antimony colorimetric method (UV-1800 spectrophotometer, SHJH Company, Shanghai, China). Bromide (Br−) was applied to all the runs as a conservative solute, and the bromide concentration (ppm) in the water samples was measured using an ion chromatograph (CIC100, SHINEHA, Qingdao, China). The retention efficiency is the ratio (%) between the trapped mass and the total mass in the inflow. Duplicated samples were collected and average breakthrough concentrations are reported.

Adsorption Experiment

The soil samples were collected from soil chamber. All samples were air dried and then passed through a 0.15 mm sieve. Phosphorus adsorption kinetics analyses were conducted by mixing the dried soil samples (~ 1.0 g) and 20-mL 0.01 M KCl solutions containing 0.5, 1.0, and 2.0 mg[P] L−1, which were added to a series of 50-mL acid-washed centrifuge tubes. The samples were taken at different time intervals (0, 2, 5, 10, 15, 20, 30, 45, and 60 min from the start of the experiment, respectively). After centrifugation at 5000 rpm for 20 min, the supernatants were filtered through a 0.45-μm membrane filter immediately to analyze the phosphorus concentration. The phosphorus adsorbed in the soil samples was calculated using the difference between the initial and final phosphorus concentrations in the solution.

Statistical Analysis

Spearman’s rank correlation coefficient was used to evaluate the statistical dependence between the sediment yield, surface runoff volume, and peak flow rate, as well as between the bromide and phosphorus in the outflow. Regression analysis was used to study the relationships of normalized bromide and phosphorus concentrations with time in the subsurface lateral flow.

Results and Discussion

Figure 2 shows the phosphorus adsorption of the experimental soil included quick and slow adsorption steps (Zhang et al. 2011). The quick adsorption rate mainly occurred within 30 min, and it was then followed by a slow adsorption reaction. The adsorption and desorption of phosphorus between solid and liquid reached dynamic equilibrium within first 60 min. According to the results, the prepared nutrient solution was introduced to the inflow spreader for 60 min, and then was switched to tap water.

Fig. 2
figure 2

The measured adsorption kinetics for different initial phosphorus concentrations

Full size image

Table 1 shows the effects of the different shallow WTDs on VFS retention efficiencies in surface outflow, sediment, phosphorus, and bromide for the nine experimental treatments. The retention efficiencies varied greatly from 35%, 70%, 64%, and 55% to 96%, 98%, 96%, and 95% for surface runoff, sediment, phosphorus, and bromide, respectively. The lowest retention occurred at a 0.08-m WTD in the group with high grass spacing (G7S05), whereas the highest retention occurred at a 0.36-m WTD in the group with low grass spacing and a low slope (G4S05). The deep WTD, low grass spacing, and low slope had high VFS retention efficiencies. For example, the surface runoff retentions were 35%, 44%, and 64% in G7S05 and 71%, 76%, and 96% in G4S05 for WTDs of 0.08 m, 0.22 m, and 0.36 m, respectively. The large variations in the VFS retention efficiencies among the different WTDs revealed the importance of the groundwater table on VFS performance. When compared with the experimental group (G4S05), the steeper slope in G4S15 (at the same grass spacing) increased the effects of the WTD on VFS performance.

Table 1 Measured (a) surface runoff (Qret), (b) sediment (Sedret), (c) phosphorus (TPret), and (d) bromide (\({\text{Br}}_{{{\text{ret}}}}^{ - }\)) retention efficiencies (%) in G7S05, G4S05, and G4S15 (grass spacing and the slope for G7S05 were 6.69 cm and 5%, grass spacing and the slope for G4S05 were 4.18 cm and 5%, and grass spacing and the slope for G4S15 were 4.18 cm and 15%)
Full size table

We hypothesized that the phosphorus sorption is not inhibited by the saturation of the soil media and its adsorbed phosphorus would not accumulate in the VFS soil. The predictive equation for the TP and bromide retention efficiencies (TPret, \({\text{Br}}_{{{\text{ret}}}}^{ - }\)) was based on the runoff retention efficiencies (Qret, %) and sediment retention efficiencies (Sedret, %) (Eqs. 2 and 3) in the study treatments without high initial phosphorus and bromide concentrations in the VFS soil. Figure 3 shows the relationship between the predicted and observed retentions of TP and bromide:

$${\text{TP}}_{{{\text{ret}}}} = 54.5 + 0.76 \, \times \, Q_{{{\text{ret}}}} {-}0.32 \, \times {\text{ Sed}}_{{{\text{ret}}}} R^{2} = \, 0.98,$$
(2)
$${\text{Br}}_{{{\text{ret}}}}^{ - } = - 12.4 + 0.73 \, \times \, Q_{{{\text{ret}}}} + 0.40 \, \times {\text{ Sed}}_{{{\text{ret}}}} R^{2} = \, 0.95,$$
(3)
Fig. 3
figure 3

Predicted versus observed retentions (%) of TP and bromide when using a multiple linear regression model based on the runoff retention efficiencies (Qret, %) and sediment retention efficiencies (Sedret, %) in the study treatments

Full size image

where Qret (%) and Sedret (%) were defined as the total mass input to the VFS minus the mass in the runoff efflux from the VFS divided by the total mass input. A multiple linear regression model, similar to that proposed by Fox and Penn (2013), was developed for quantifying the TP trapping efficiency of the VFSs. The proposed equation provided an opportunity to evaluate the performance of a VFS installed in areas with shallow groundwater tables and to design a VFS for specific retention efficiency objectives.

Shallow WTDs Affected the Hydrologic and Sedimentological Responses

The 0.08-m, 0.22-m, and 0.36-m WTDs distinctly influenced the surface runoff flow and subsurface lateral flow (Fig. 4). The average surface runoff outflow rate was less than the inflow rate, and was typically between 37 and 69% of the inflow rate in the high grass spacing (G7S05) (Fig. 4a; Table 2). The reason why the normalized surface runoff outflow decreased with WTD might be related to the capillary action above the shallow WTD in the soil, which moved groundwater from wet areas to dry areas in the soil, filled the soil pore space, and decreased the runoff infiltration rate (Carluer et al. 2017). However, the relatively low capillary zone and unsaturated soil condition in the deep WTD (0.36 m) resulted in a high infiltration rate and low surface runoff outflow rate. Conversely, the moving average subsurface lateral flow rate increased from 0.27% to 0.53% of the inflow rate when the WTD decreased from 0.08 to 0.36 m (Fig. 4d; Table 2).

Fig. 4
figure 4

Hydrological response of the VFS in the study treatments with the normalized surface runoff outflow rate (q/q0) for (a) G7S05, (b) G4S05, and (c) G4S15, and the normalized subsurface lateral flow rate (q/q0) for (d) G7S05, (e) G4S05, and (f) G4S15. Values are means, standard errors as error bars

Full size image
Table 2 Start time and the average flow rate of the surface runoff and subsurface lateral flow for the experimental treatments
Full size table

The surface runoff flows for the 0.08-m and 0.22-m WTDs occurred within 4 and 5 min, respectively, after the start of the nutrient solution application, which were much earlier than the occurrence time for the 0.36-m WTD, which took more than 15 min in G7S05 (Table 2). This phenomenon indicated that the WTDs influenced the occurrence time of the surface runoff by affecting the infiltration capacity. The results may also be related to the initial soil water content. The initial soil water content at the WTD of 0.36 m was 28.8% in G7S05, which was lower than those of the 0.08-m and 0.22-m WTDs (36.4% and 34.6% in G7S05, respectively). At the early stages of the experiments under low initial soil water content conditions, the infiltration rates were equal to or exceeded the inflow rate. As the experiments progressed and the soil became wetter, the infiltration rates decreased and the surface runoff occurred (Cardoso et al. 2012).

In addition to WTD, the grass spacing and slope also influenced the hydrology responses and sediment retention efficiencies (Table 1). For the same WTD, the occurrence time of the surface runoff outflow in G4S05 was later than that in G7S05, and the surface runoff outflow rate in G4S05 was lower than that in G7S05, whereas the subsurface lateral flow rate in G4S05 was higher than that in G7S05 (Fig. 4b, e; Table 2). The plant roots in the VFSs increased the soil surface Manning’s roughness and soil permeability, which enhanced the runoff infiltration rate, reduced the surface runoff velocity, and promoted sediment deposition (Schiettecatte et al. 2008; Zhang et al. 2011; Cardoso et al. 2012; Wang et al. 2014; Wu et al. 2014; Zhao et al. 2016). Compared with the hydrologic response in G4S05, for the same WTD, the surface runoff outflow in G4S15 occurred sooner and the average surface runoff outflow rate in G4S15 was higher, whereas the average subsurface lateral flow rate in G4S15 was lower (Fig. 4c, f; Table 2). A steep slope, which decreases runoff infiltrating time and sediment depositing time, causes a large transport capacity for overland flow and reduces residence time of surface runoff on soils (Gumiere et al. 2011). Therefore, at the same WTD, the low grass spacing and low slope had high surface runoff retentions (a low surface runoff outflow rate and high subsurface lateral flow rate).

Runoff sediment retention efficiencies in the VFSs increased with a deep WTD (e.g., 70%, 72%, and 79% at WTDs of 0.08 m, 0.22 m, and 0.36 m, respectively, in G7S05) for the relatively low flow conditions experienced, which resulted in a low flow transport capacity and a high sediment deposition. The low grass spacing and low slope had high sediment retention (Table 1). The sediment yields in the surface runoff can be estimated using Eq. (4), as shown in Fig. 5. The study was restricted to a VFS width of 1 m:

$${\text{Sed}} = 7E + 06\left( {Q \, \times \, Q_{{\text{p}}} } \right)^{0.74} R^{2} = \, 0.96,$$
(4)

where Sed is the sediment yield (kg/ha), Q is the surface runoff volume (m3), and Qp is the peak flow rate (m3/s). A power equation, similar to the results proposed by Kuo (2007), was used to describe the relationship among the sediment yields, runoff volume, and peak flow rate for each runoff treatment.

Fig. 5
figure 5

Relationship between sediment yield, Q, and Qp in the VFS

Full size image

Shallow WTDs Affected Bromide and Phosphorus Transport

After the inflow (prepared solution) application, the normalized bromide concentrations (C/C0 Br−) in the surface runoff increased rapidly and then were relatively stable in the first 60 min (Fig. 6b, d, f), whereas the normalized phosphorus concentrations (C/C0 TP) in the surface runoff fluctuated slightly (Fig. 6a, c, e), which may have been caused by the phosphorus in the surface runoff interacting with the soil. When the inflows were switched to tap water, the C/C0 Br− dropped dramatically to 0, and the C/C0 TP dropped but exhibited persistent concentration tailing, which suggested a slow release of phosphorus from the surface soil to runoff. For each experimental group, the average normalized concentrations of bromide and phosphorus increased with the WTD (Table 3), and the average C/C0 Br− was higher than the C/C0 TP at the same WTD.

Fig. 6
figure 6

Transport of (a) phosphorus and (b) bromide for G7S05, (c) phosphorus and (d) bromide for G4S05, and (e) phosphorus and (f) bromide for G4S15 in the surface runoff under different shallow water table depths. Values are means, standard errors as error bars

Full size image
Table 3 Average normalized concentrations of bromide and phosphorus in the surface runoff and maximum normalized concentrations of bromide and phosphorus in the subsurface lateral flow
Full size table

The transport of bromide and phosphorus in the subsurface lateral flow, however, exhibited typical breakthrough behaviors during each runoff experiment (Fig. 7). The maximum normalized concentrations (C/C0 max) for phosphorus were lower than those for bromide (Table 3; Fig. 7) because of the phosphorus adsorbed by the mediums (e.g., soil colloids).

Fig. 7
figure 7

Transport of (a) phosphorus and (b) bromide for G7S05, (c) phosphorus and (d) bromide for G4S05, and (e) phosphorus and (f) bromide for G4S15 in the subsurface lateral flow under different shallow water table depths. Values are means, standard errors as error bars

Full size image

For each experimental group, the C/C0 TP and C/C0 Br− in the increasing limbs can be described by power equations (C/C0 = W × TE) (Table 4). The coefficient (W) in the power function indicates the infiltration rate of surface runoff, which was positively influenced by the decreased WTD. The coefficient (W) can be obtained from the empirical equations [Eqs. (5) and (6)] as follows:

$$W = a \times q/q_{0 - Ave} + b,$$
(5)
$$q/q_{{0 - {\text{Ave}}}} = c \times {\text{WTD}} + d,$$
(6)

where q/q0-Ave is the average flow rate of the subsurface lateral flow; WTD is the shallow water table depth (m); and a, b, c, and d are the regression coefficients. Therefore, the decreased WTD increased the infiltration rate, which resulted in a more obviously increase rate of the normalized phosphorus and bromide concentrations.

Table 4 Power equations of normalized phosphorus and bromide concentrations in the increasing limbs in the subsurface lateral flow
Full size table

The exponent (E) in the power function indicates that phosphorus and bromide are adsorbed or intercepted by clay or organic matter, and a relatively high E value represents a relatively low adsorption or interception ability of soil mediums. For all the aforementioned experiments, the bromide positively correlated with phosphorus (R2 = 0.85, p < 0.01) in the subsurface lateral flow, and the slope of the equation presented in Fig. 8 was smaller than one (0.52), which showed a higher adsorption ability of phosphorus than that of bromide. The phenomenon was also reflected on the exponent (e.g., E was 0.943 × a for phosphorus and 0.943 for bromide in G7S05, where a was 0.521/2).

Fig. 8
figure 8

Correlation between the bromide and phosphorus in subsurface lateral flow

Full size image

Different grass spacings and slopes also influenced the transport of phosphorus and bromide in the subsurface lateral flow. For example, compared with the experimental group G7S05, the low grass spacing in G4S05 (the same slope as G7S05) increased the infiltration of surface runoff, the adsorption of organic acids secreted by plant roots, and the interception by plant roots, resulting in high W and low E (e.g., for phosphorus, E was 0.943 × a for G7S05 and 0.943 × a × b for G4S05, where a was 0.521/2 and b was (G4/G7)1/2) in the power function. Moreover, compared with the experimental group G4S05, the high slope of the experimental group G4S15 (the same grass spacing as G4S05) increased the surface runoff outflow and transport rate of infiltrated water in the subsurface soil, which resulted in a low W and high E (e.g., for phosphorus, E was 0.943 × a × b for G4S05 and 0.943 × a × c for G4S15, where a was 0.521/2, b was (G4/G7)1/2, and c was (G4/G7)1/6) in the power function.

Mass balance calculations based on this study showed that the highest total recovery rate of bromide (both in the surface runoff outflow and subsurface lateral flow) was almost 75% in the aforementioned experimental conditions. This incomplete recovery rate may be attributed to a certain amount of bromide remaining in the soil at the end of each runoff treatment. However, the highest total recovery rate of phosphorus was only approximately 61%. Such low recovery rates may be attributed to the phosphorus adsorption of the mediums, including soil colloids (Edzwald et al. 1976; Gérard 2016), organic acids secreted by the plant roots (Wang et al. 2012), and iron oxides in the soil (Tomer et al. 2010; Fink et al. 2016). The weaker adsorption of bromide in the porous media resulted in C/C0 Br− decreasing quicker than C/C0 TP when the inflows were switched to tap water (Fig. 7). Therefore, if we extended the experimental time, the slow release of phosphorus from the soil to the subsurface lateral flow would obviously decrease.

The effect of a seasonal shallow groundwater table on VFS retention efficiencies should be considered, because the high concentrations of pollutants in subsurface lateral flow may also affect adjacent water bodies or groundwater. When the groundwater level rises in the rainy season, phosphorus adsorbed in the soil is desorbed into groundwater, which then pollutes the groundwater.

Conclusions

Laboratory-scale VFS runoff experiments were conducted to study the effects of a shallow groundwater table on VFS efficiency for retarding the transport of runoff, sediment, phosphorus, and bromide in different grass spacings and slopes, and to examine the transport dynamics of phosphorus and bromide in the surface runoff outflow and subsurface lateral flow. We found that a deep WTD, low grass spacing, and low slope were beneficial for reducing NPS pollutants. The transport of phosphorus in the surface runoff almost resembled that of bromide, and the average concentration (C/C0 Ave) of phosphorus and bromide decreased with a deep WTD. The transport of phosphorus in the subsurface lateral flow was lower than that of bromide, which was due to the adsorption of phosphorus by the soil mediums. Both normalized phosphorus and bromide concentrations in the increasing limbs could be described by the adopted power equations, and the increased concentration rate was positively influenced by the decreased WTD. High concentrations of pollutants in the subsurface lateral flow may also affect adjacent water bodies and groundwater. Future research should further examine the transport dynamic of nitrogen in VFSs with a shallow groundwater table. Although this paper focuses on effects of shallow water table depth on vegetative filter strips retarding transport of nonpoint source pollution, only phosphorus is discussed in this paper. As an important part of nonpoint source pollution, nitrogen will also be studied in the further research. Moreover, controlled in-field VFS experiments should be conducted to explore the effects of a deep groundwater depth (more than 1.0 m) on VFS performance.