Introduction

In recent years, with population growth and ongoing social and economic development of cities, people’s exploitation of natural resources and destruction of the environment have become increasingly critical. Water pollution also becomes an increasingly serious issue, being a serious threat to people’s health and economic development. Non-point source pollution has become the most important source of water pollution in rivers and lakes (Quan and Yan 2002; Chen 2010; Shen 2010). For example, conversion to farmland, agricultural activities, grazing, and discharging industrial and domestic sewage around the river bank can all cause the deterioration of the ecological environment of riparian areas (Jacobs and Gilliam 1985; Tian et al. 2010); moreover, rain or snowmelt runoff can carry non-point source pollutants into rivers, causing a change in the river or lake and reservoir water quality and eventually leading to increased eutrophication, diminished flood control capacity, and other disasters. Eutrophication by large inputs of nitrogen (N), phosphorous (P), and other nutrients occurs most commonly (Klapproth and Johnson 2000; Mankin et al. 2007).

Many studies have shown that a riparian buffer is an important control measure for non-point source pollution (Lowrance et al. 1988; Yan 1999; Jon et al. 2005), especially in watershed restoration aiming at reducing pollutants (Baker et al. 2007). In northeastern China, an excessive amount of fertilizers and pesticides is brought directly into the river by surface runoff, due to lack of riparian vegetation (Mi et al. 2015). Interception is defined as water that is temporarily intercepted and stored on vegetation surfaces (e.g., shrubs, grass, trees, and leaf litter) (Xiang et al. 2017). Thus, one way to reduce non-point source pollution is to utilize the natural purification services provided by the ecosystem (Yan et al. 2018). The water quality and stream habitat can be improved by riparian tree planting (Parkyn et al. 2005). With sediment accumulated, much nutrient load is retained and the discharge can be reduced by evaporation and transpiration. The concentrations of total nitrogen (TN), total P (TP), and total suspended solids (TSS) are also found to be lower (Burchsted et al. 2010, Swanson et al. 2017). Moreover, many technical measures also have been taken (Wang et al. 2009; Wu 2011).

In the early 1980s, research started by United States Department of Agriculture (USDA) researchers in Tifton, Georgia (GA), showed that a riparian deciduous forest buffer can reduce the N content in the runoff by 68.0% (Wang et al. 2009). On the west coast of the Chesapeake Bay, Maryland (MD), scientists have found that riparian buffers can remove 89.0% of the N content (Peterjohn and Correll 1984). However, in terms of the removal of the element P, riparian buffers have less effect than sediment interception and N removal (Parsons et al. 1994).

Dillaha et al. (1989) showed that a 9-m-wide orchard–grass buffer can remove 89.0% of the P in the surface runoff; they also revealed that grass filter strips can only temporarily intercept P particles: once a storm comes, P will be lost. Zhang et al. (2009) found that 20–30 m of adjacent riparian zone can effectively reduce N, P, and other nutrients. Wang and Zhang (2006) analyzed the buffer sampling of Zhalong wetland and found that wetland buffer zones have a strong ability to remove N and P nutrients, and the removal efficiency for TN and TP reached 74.1% and 84.0%, respectively.

However, appropriate vegetation configuration may be also important to reduce the pollutants effectively. The purpose of this study was to determine the effects of a watershed forested riparian buffer system on the surface and subsurface transport of N and P. The basic hydrological and biogeochemical processes in three types of forest riparian buffers (Quercus variabilis*Pinus tabuliformis (SP*YS), Pinus tabuliformis (YS), and Platycladus orientalis (CB)) were explained. Thus, our study can provide theoretical and technical support for the reconstruction of different forest configuration riparian buffers of future water conservation areas.

Materials and methods

Study area

The study was conducted in Banchengzi basin, located on the northern Miyun Reservoir, Beijing, China, at 116° 55′–117° 2′ E and 40° 37′–40° 43′ N. This basin is entirely mountainous and its area is 66.18 km2. There is a reservoir in this basin with a total storage capacity of 1.02 million m3. Banchengzi basin belongs to the warm temperate semi-humid continental monsoon climate, and the annual average temperature is 10.5 °C; the average annual rainfall reaches 669 mm, and is concentrated from June through August inclusively, which account for 75.0% of the annual rainfall. Banchengzi basin has rich vegetation types, with arborvitae forest, pine forest, and acacia forest as the main vegetation types (Fig. 1).

Fig. 1
figure 1

Study area and the locations of the sampling sites

Full size image

Methods

Sampling was collected during summer in 2014. Ensuring that the sample’s nutrient content did not change, thus, the samplings were collected the next day after the rainfall events. The amount of rainfall, throughfall, stemflow, runoff, and subsurface flow of each rainfall was measured. Samples of the rainfall, throughfall, stemflow, runoff, and subsurface flow were collected and brought back to the laboratory. There, the relationship of rainfall, runoff, and water quality as affected by different plant types was analyzed, providing a reference for further study of N transfer.

Throughfall

Three rain barrels with a bunghole diameter of 26 cm were placed uniformly along the diagonal of each of the four 10 × 10-m typical forest plots. After rainfall, the barrels were used to measure the volume of the throughfall collected in the barrels. Rain volume divided by the collection area of the rain barrel is the throughfall of the measuring area; then the average value of three measuring areas is taken as the throughfall of the forest. Thus, the formula of the throughfall is as follows:

$$ \mathrm{TF}=\sum {T}_{\mathrm{i}}/\sum {A}_{\mathrm{i}} $$

where TF is the throughfall (mm), Ti is the collected rainfall volume (mm3), and Ai is the collection area of the rain barrel (mm2).

Stemflow

Three standard trees were selected whose trunk diameter class was representative in the four 10 × 10-m typical forests plots, respectively, and then the rubber tube which was winded on the trunk was employed to measure stemflow. Firstly, prominent parts of the trunk bark were smoothened with a paring knife to prevent puncturing of the rubber tube. Simultaneously, the degree of fit of the rubber tube and the tree bark was increased in order to make the measurement more accurate. Then the rubber tube (1.5 m in length, 3.0 cm in diameter) was cut open and twined around the trunk from a 1.3-m height to the trunk’s base. The rubber tube was affixed on the trunk with small nails, sealing the pores with a thin rubber skin to prevent water leaking. The collection bucket was connected to the rubber tube at the trunk base in order to collect stemflow. After every rainfall, the rain volume collected in the collection bucket was measured with a graduated cylinder. The measured rain volume divided by the standard tree’s canopy area is the stemflow of the standard tree. Therefore, the formula of the stemflow is as follows:

$$ \mathrm{SF}=\frac{1}{M}{\sum}_{i=1}^n\frac{{\mathrm{SF}}_{\mathrm{n}}}{K_{\mathrm{n}}}{M}_{\mathrm{n}} $$

where SF is the stemflow (mm), SFn is the stemflow of each different diameter class (mm), M is the number of trees per unit area, Mn is the number of trees per diameter class, n is the number of diameter classes, and Kn is the average projection area of the canopy in each different diameter class (cm2).

Canopy interception

Canopy interception is a calculated value, and according to the principle of the water balance, the formula for canopy interception is as follows:

$$ I=P-P\hbox{'}-G $$

where I is the canopy interception (mm), P is the rainfall (mm), P' is the throughfall (mm), and G is the stemflow (mm).

Shrub interception

In order to study the rain interception characteristics of the forest’s vertical hierarchy, additional shrub interception observations were made as follows: understory shrubs were dug up and planted in plastic buckets, then a ring-shaped rainwater harvesting device was made and placed at the base of the small shrub’s roots. A rain gauge connected to the water outlet automatically recorded rainfall interception. The setup is shown schematically in the diagrams of Figs. 2 and 3.

Fig. 2
figure 2

Schematic diagram of shrub interception collector

Full size image
Fig. 3
figure 3

Schematic diagram of shrub interception observation

Full size image

Grass interception

To complete the assessment of the rainfall interception characteristics of the forest’s vertical hierarchy, additional grass interception observations were also set up. In a selected plot of grassland in the forest, a steel ring was positioned on the ground in the understory herbaceous vegetation, and was then hammered into the soil layer until the steel ring was at level with the ground surface. The steel ring was dug out with a shovel and placed on a steel ring base. The removed sample of herbal layer was then placed in an adjustable frame, the three parts of which all had connection screws with which their height could be adjusted to correspond with the slope of the soil in the selected plot, ensuring the herbal layer’s natural position. The sampled undisturbed herbal layer was placed on the collector and the automatic recording rain gauge was connected outside the collector; thus, the rainfall could be recorded and the dynamic process of its interception could be observed. A schematic diagram is given in Figs. 4 and 5.

Fig. 4
figure 4

Schematic diagram of herb and litter interception collector

Full size image
Fig. 5
figure 5

Schematic diagram of interception observation for herb and litter

Full size image

Litter interception

Observation of rainfall interception by leaf litter employed the same equipment as that used for the observation of interception by grass in this research, taking the original and undisturbed forest litter and placing it in the rainwater harvesting device. For the specific methods of operation, the reader is referred to the schematic diagram of grass interception observation (Figs. 4 and 5).

Nutrient monitoring

The study area was the Banchengzi Basin, which was on the rural area in Beijing. Crop and forest lands were the primary land types here. Thus, the increase nitrogen, phosphorous, and pesticide concentrations in nearby river water were through agricultural activities (Xiang et al. 2016). Based on the measurement data and using the corresponding statistical analysis method, we analyze and discuss the transportation of nutrients such as PO43−-P, TP, NO3− N, and TN under different plant types in forested riparian buffer, in order to assess their efficiency of interception and transportation of N and P. Water quality indicators measured in this study are PO43−-P, TP, NO3− N, and TN, which were all measured by reference to national standards. The method of PO43−-P and TP is Molybdate spectrophotometry (GB 11893-89) (Environmental Protection Agency of China 1989a, b), which has been widely used to determine phosphorus for fresh water, with a detection limit of 0.01 mg/L. According to the standard curve and spectrophotometer to get the value. The method of NO3− N is UV spectrophotometry (HJ/T 346-2007) (State Environmental Protection Administration of China 2007), and the method of TN is alkaline potassium persulfate digestion ultraviolet spectrophotometry (GB 11894-89) (Environmental Protection Agency of China 1989a, b).

Data analysis

Differences in the measured hydrological processes of the different tree species, i.e., in canopy interception, stemflow, shrub, grass, and litter interceptions, were analyzed statistically and plotted in charts using software such as Sigmaplot 11.0 and SPSS 18.0. In addition, the amounts of various nutrients involved in nutrient transportation in the different hydrological processes were analyzed, and formulae describing these changes were derived. Simultaneously, a theoretical analysis is provided, which supports the different plant type configurations in the forested riparian buffer.

Results

Hydrological processes

There were three types of vegetation such as Quercus variabilis*Pinus tabuliformis (SP*YS), Pinus tabuliformis (YS), and Platycladus orientalis (CB) in this experiment. The hydrological processes in the buffer zone are canopy interception, stemflow, shrub interception, grass interception, litter interception, runoff, and subsurface flow. The data analysis showed that canopy interception was the most important process; SP*YS had the highest canopy interception, and CB had the lowest. Stemflow rates of these three types of forest were similar and all relatively low. The rate of YS was higher than that of SP*YS and CB. The shrub interception rates of these three types were also similar and the rate of YS was higher than that of SP*YS and CB. Grass interception had the lowest interception rate in all three forest types and CB was higher than that of SP*YS and YS. Litter interception was relatively high. YS had the highest litter interception rate, and CB had the lowest. The runoff the subsurface flow showed the low rates. CB had the highest runoff but the lowest subsurface flow while SP*YS had the lowest runoff but the highest subsurface flow (Fig. 6).

Fig. 6
figure 6

Hydrological processes

Full size image

Nutrient transportation processes

PO43− content in rainfall of three types of species was approximately 0.03 mg/L. After canopy interception, the PO43− content measured in the throughfall is improved in all configurations. At this stage, the PO43− content of YS is enriched fastest, and the changes in stemflow also fit well with this. Subsequently, the PO43− content measured in runoff improves significantly, with SP*YS having the highest content. The changes in PO43− content in subsurface flow are opposite because here the PO43− content is significantly reduced by absorption and chemical reactions involving microorganisms inside the soil. The PO43− contents of the three types of vegetation all improve compared with the beginning PO43− content in rainfall outside the forest, and SP*YS has the largest improvement. This shows that PO43− is not absorbed well in the rain process, instead, it rises. The sum of the PO43− content of surface runoff and subsurface flow shows that SP*YS has the most obvious change (Fig. 7).

Fig. 7
figure 7

Nutrient transportation processes

Full size image

The TP content in rainfall in the three types of forest was approximately 0.06 mg/L. As for the PO43−, the TP contents measured in throughfall in the three forest types all improved, but the TP content of SP*YS enriches fastest at this stage; the TP content measured in stemflow is basically the same as in throughfall. Through the runoff, the TP content is increased significantly, but the TP content in subsurface flow declined, as did PO43−, which SP*YS reduced the least, and the data of TP content in subsurface flow of three vegetation types are similar. This also confirms that the element P is enriched significantly in the whole rainfall process.

The NO3− N content in rainfall in the three types of forest was approximately 1.45 mg/L. After canopy interception, the NO3−N content measured in throughfall was reduced in all cases, the decline in CB and YS being big, but the decline in SP*YS being relatively small. The stemflow showed that the NO3−N content for the three forest types is almost identical, approximately 1.5 mg/L. The runoff showed an improvement of the NO3−N content in all three vegetation types, but the magnitudes of change are different. Compared with the rainfall, the increasing proportion in CB is most significant, and the rate of increase in SP*YS is lowest. The subsurface flow shows a decreased NO3−-N content in the three forest types compared with the runoff, and the NO3−N content of the three vegetation types becomes similar. Thus, it can be concluded that the impact of microorganisms from the ground surface and the soil layer on the NO3−-N content is large.

As for TN, the content in rainfall in the three vegetation types was approximately 2.75 mg/L. After canopy interception, the throughfall shows a reduced TN content for YS and CB, but the TN content in SP*YS improves. As for stemflow, the TN content in SP*YS is lower than in YS or CB, YS being highest. The analysis of runoff showed that the TN content in CB is highest, and is lowest in SP*YS. The subsurface flow shows that the TN content in all three types of forest decreases, and is basically the same.

Overall, CB has the maximum transfer ratio and the most obvious enrichment for N and P in the process. Therefore, if practical transportation and absorption of these elements are considered, CB forest is the best choice to plant in riparian buffer zones. Besides, the transfer ratio of TP is higher than that of TN for the three vegetation types, which shows that the enrichment of P is stronger than that of N.

Discussion

The riparian buffer is an important part of the riparian ecosystem that can transform and intercept non-point source pollutants effectively through hydrological and nutrient processes (Tang et al. 2012). This study had revealed the hydrological and nutrient processes in the riparian buffer. It has shown that, starting from the rainfall outside the forest, riparian buffers will intercept the rainfall effectively and are able to intercept and transform N, P, and other nutrients in the rain through a series of hydrological processes such as canopy interception, throughfall, stemflow, shrub interception, grass interception, litter interception, and runoff. Many studies have shown that riparian buffers, such as grass riparian buffer and forest riparian buffer, can intercept N effectively (Lowrance et al. 1984; Cooper 1990). Delgado and Periago (1995) and Heathwaite et al. (1998) have shown that grass riparian buffer has a good effect on non-point source pollution control, and the average removal rates of TN and TP are 70.0–95.0% and 70.0–98.0%, respectively. Dabney (1995) found that establishing rational grass or forest riparian buffers between farmland and water can greatly reduce the content of N and P in the water.

Interception rate

Riparian vegetative buffer is used to intercept groundwater flow and surface runoff to reduce nutrient and other organic pollutants (Lin et al. 2002). The present study shows that a riparian buffer can intercept rainfall effectively, and even more effective than other kinds of plants such as mountain plants. Afforestation may lead to the decrease of annual runoff, but the role of vegetation characteristics on runoff generation is still controversial (Liu et al. 2019). And a study in Southeastern Brazil showed a positive relationship between base flow/runoff and the increases in riparian vegetation (Alvarenga et al. 2017). Both biotic and abiotic factors such as leaf morphology which is related to the differences in leaf area and litter thickness may impact interception. And crown interception storage for needle species is 1.8 and 1.2 times larger than that of broadleaf species (Xiang et al. 2017). Thus, two types of species configurations may contribute to the difference in the interception.

A comparative analysis of interception rate between the riparian buffer and other kinds of plants was made in this study (Table 1). As for YS forest, the interception rate sum of canopy interception and stemflow in this study (48.5%) is higher than those of Taihang Mountain (Hebei, China) (18.9%) (Qian et al. 2012), Jiufeng Mountain (Beijing, China) (31.0%) (Li et al. 2013), Western Mountain (Beijing, China) (39.6%) (Wu et al. 2013), Sierra Madre Oriental Mountain (Nuevo León, Mexico) (19.8%) (Silva and Rodríguez 2001), and the Loess Plateau (Gansu, China) (23.5%) (Fang et al. 2013). As for CB forest, the interception rate sum of canopy interception and stemflow in this study (38.5%) is higher than those of Taihang Mountain (Hebei, China) (24.5%) (Pang 2010), Jiufeng Mountain (Beijing, China) (28.3%) (Li et al. 2013), and Miyun protection forest (Beijing, China) (34.4%) (Zhou and Yang 2008). As for SP*YS forest, the interception rate sum of canopy interception and stemflow in this study (54.4%) is also higher than those of Western Mountain (Beijing, China) (38.1%) (Wu et al. 2013), Sierra Madre Oriental Mountain (Nuevo León, Mexico) (23.0%) (Silva and Rodríguez 2001), and Western Mountain (Beijing, China) (30.2%) (Li et al. 2006a, b) (Table 1). The difference may be because of the different geographical position. For instance, the Taihang mountains are more south than our study and the annual precipitation is higher. And the slope of the study field could also cause the difference (Qian et al. 2012).

Table. 1 Statistical summary of the Interception rate
Full size table

Canopy, shrub, and grass interception basically accounts for about 80.0% of total interception, and therefore, riparian buffer configuration should clearly distinguish three levels of vegetation. From the point of view of interception by different forest types (high interception is superior), canopy, shrub, grass, and litter interception rate of YS forest is highest, totalling up to about 71.1%, when the SP*YS forest is 66.8% and the CB forest is 48.0%. And in YS forest, the canopy, shrub, grass, and litter interception rate is 48.5%, 6.8%, 0.9%, 14.9%, respectively (Table 2).

Table. 2 Interception rate of each species
Full size table

Similarly with this study, for the vegetation of a typical small black-water river watershed in its upstream part, Liu (2013) conducted field observations and manual experiments of the rainfall interception characteristics. The space simulation results of each vegetation rainfall interception in the 2012 growing season show that the average interception rate of Qinghai spruce forest is 44.6%, the average interception rate of shrub is 49.1%, and the average interception rate of mountain meadow in grazed and ungrazed plots is 11.0–16.9% and 16.2–22.1%, respectively. Because of the different study area and different tree species, the research data has much difference from this study, but this research also proves that canopy, shrub, and grass interceptions are really high, which is consistent with our study.

Transportation ratio

Riparian vegetation is necessary to achieve habitat improvement goals as the shade and cover provided by vegetation will result in appreciable benefits in nutrient reduction and sediment (Parkyn et al. 2005). And the nutrient removal effectiveness of riparian buffers is variable because of the different forms of nutrient, slopes, flow pathways, and soils (Parkyn et al. 2005). Li et al. (2006a, b) used two kinds of mixed plant bands (vetiver grass + submerged plants; hydrophytic plants + cattail + reeds) to study a polluted river, and also compared them with a blank band without vegetation. The results show that mixed plants bands are better than the blank band without vegetation in the transportation effect for pollutants. Vetiver grass + submerged plants have the best effect; over the whole process, the transportation ratios for COD, NH4+-N, and TP are 43.5%, 71.1% and 69.3%, respectively. Zhang (2013) studied the removal effect by different riparian plant buffers in Beijing Chaobai River basin of surface runoff N and P pollutants at different pollutant concentrations. In terms of the order of their removal effect for pollutants, the grass buffer riparian vegetation band is best, followed by grass and shrub buffer, shrub buffer, and finally the comparison buffer.

Here, the same comparison method was used to compare the transportation rate for N and P nutrient elements of CB forest, YS forest, and SP*YS forest. Overall, the order of the effect for N and P transportation is as follows: CB > SP*YS > YS. CB has the highest transportation and enrichment for N and P throughout the process. The transportation ratios for PO43−-P, TP, NO3—N, and TN of CB are 81.2%, 80.3%, 58.3%, and 64.2%, respectively (Table 3). However, in the present study, the difference between SP*YS and YS is insignificant. The reason may be that the study design did not include multiple sets of parallel experiments, which should be done in future experiments to improve the accuracy of the results.

Table 3 Nutrient gathering and transference in riparian buffer strips
Full size table

In conclusion, this paper also found that the transportation for TP of the three tree species is higher than that for TN, showing that the enrichment of P is stronger than that of N. According to the results of this study, different forest configurations can be considered to build riparian buffer to remove different kinds of nutrients in the future.

Conclusion

Riparian buffers can intercept the rainfall effectively through a series of hydrological processes and transform N, P, and other nutrients in the rain, thereby reducing the pollution of river water. Canopy, shrub, and grass interception basically accounts for about 80.0% of total interception; therefore, riparian buffer configuration should clearly distinguish three levels of vegetation types. On the one hand, canopy, shrub, grass, and litter interception rate of YS forest is highest, totalling up to about 71.1%; thus, YS is the best configuration considering only the amount intercepted. On the other hand, CB has the highest transportation and enrichment of N and P throughout the process, and the transportation ratios for PO43−-P, TP, NO3−-N, and TN of CB are 81.2%, 80.3%, 58.3%, and 64.2%, respectively; therefore, CB is the best configuration considering the transportation of N and P. Overall, the transportation of TP of the three tree species is higher than the transportation of TN, which shows that the enrichment of P is stronger than that of N.