- Email: [email protected]
Is returning farmland to forest an effective measure to reduce phosphorus delivery across distinct spatial scales?
Journal of Environmental Management 252 (2019) 109663
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: http://www.elsevier.com/locate/jenvman
Research article
Is returning farmland to forest an effective measure to reduce phosphorus delivery across distinct spatial scales? Wenzhuo Wang, Lei Chen *, Yingxin Zhu, Kai Wang, Shibo Chen, Zhenyao Shen State Key Laboratory of Water Environment Simulation, School of Environment, Beijing Normal University, Beijing, 100875, PR China
A R T I C L E I N F O
A B S T R A C T
Keywords: Nonpoint source pollution Land use change Returning farmland to forests Spatial scales The soil and water assessment tool The three gorges reservoir region
As one typical land use change, the mechanism of returning farmland to forests (RFF) on nonpoint source pollution (NPS) is not clear, especially at multiple spatial scales. In this study, by using the Soil and Water Assessment Tool (SWAT), the changes in several flow-related and NPS-related indicators across several nested catchments were quantified and compared in the Three Gorges Reservoir Region, China. The results indicated that RFF could reduce the total flow and total phosphorus (TP), which are higher in the dry season (41% and 79%, respectively) than in the wet season (21% and 47%, respectively) at the watershed with a total area of 2423.74 km2. In comparison, RFF has a larger impact on the baseflow index during the wet season (367.02%) than during the dry season (166.54%). The results also indicated that a spatial scaling effect did exist, while the reduction in TP increased from 24.57% to 48.46% as the drainage area increased from 65.92 km2 to 2104.35 km2. Specific thresholds of RFF efficiency were also observed (approximately 2000 km2 for the study area). It is suggested that other source control measures could supplement RFF by stabilizing the efficiency of RFF across different spatial scales. The results of this study could provide valuable suggestions for land use development and water quality protection, especially for large, complex watersheds.
1. Introduction With rapid global change, the impacts of land use change on hy drological processes and related water quality have become a hot topic (Zhang et al., 2018; Cameron et al., 2019). In recent years, to feed the ever-growing population, more fertilizers and pesticides have been used, so agricultural nonpoint source (NPS) pollution has become a key threat to the nearby water bodies (Dubrovsky and Hamilton, 2010; Cho et al., 2016). Zhang et al. (2018) indicated that dry land was major source for both NPS pollution, with the contributed proportions of 81.3 and 81.8% of total nitrogen (N) and phosphorus (P) respectively, in Nansi Lake Basin, China. Du et al. (2016) studied that conservation tillage and vegetative buffers were effective measures by reducing 14.2% and 12.5% of N and P loads, in Liu river watershed, China. Farmland, a human-impacted landscape, often produces serious soil erosion and a large amount of NPS pollutants (Yun et al., 2015; Kerr et al., 2016). Zhang et al. (2017) proved that dry land farming practices on steep slopes would lead to large exports of sediments and total P (TP). In contrast, forestland is generally identified as a sink for sediment and pollutants due to its high vegetation coverage, strong soil consolidation
ability, high soil porosity and good infiltration capacity (Shangguan and Zheng, 2006; Abdulkareem et al., 2018). To date, returning farmland to forests (RFF) has been adopted as a policy by many countries as a measure for controlling soil erosion and NPS pollution (Cheng et al., 2016; Strehmel et al., 2016). After the implementation of RFF, the related NPS pollutants are usually reduced significantly because of the canopy interception, soil infiltration and evapotranspiration processes (Qiu et al., 2017; Wang et al., 2017). Souza-Filho et al. (2016) indicated that 52% deforestation of an area is responsible for an 85% increase in observed discharge. At the same time, most of the existing studies focus on the rate of reduction of NPS pollution due to RFF. However, the mechanism of RFF on NPS pollution is still not clear due to the lack of a proper evaluation indicator system. The scaling effect is another key barrier for the comprehensive assessment of RFF. The scale problem in hydrological studies originated from the surface parameterization of atmospheric circulation simulation in the 1970s (Deardorff, 1972). With the enhancement of public awareness, scale issues have become a focus of hydrological studies and has also become a focus of NPS pollution studies (Larsen et al., 2016; Ouyang et al., 2018). Cerdan et al. (2004) indicated that a significant
* Corresponding author. E-mail addresses: [email protected], [email protected] (L. Chen). https://doi.org/10.1016/j.jenvman.2019.109663 Received 10 May 2019; Received in revised form 1 September 2019; Accepted 30 September 2019 Available online 14 October 2019 0301-4797/© 2019 Elsevier Ltd. All rights reserved.
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
decrease in the runoff coefficient is observed as the drainage area in creases. Zhang et al. (2016) showed that the event-based suspended sediment and flow–sediment relationships remained spatially constant. Xu et al. (2010) showed that flow and TP in the Heihe River Watershed (a drainage area of 1481 km2) would be reduced by 50.64% and 45.44%, respectively, after RFF. Liu et al. (2013) indicated that RFF would result in a reduction of 9.77% and 20.61% of runoff and TP loads in the Xiangxi River (a drainage area of 3099 km2). Large-scale studies show that RFF on 15 degree- and 25 degree-slope lands would reduce NPS-TP by 14.92% and 44.96%, respectively, for the whole Three Gorges Reservoir Region (Chen et al., 2013). However, Wu et al. (2016) demonstrated that 81.75% of TP would be reduced if RFF were performed on lands with slopes above 25� in the same area. It is clear that there are significant differences in the effect of RFF across different spatial scales, but few studies have compared these differences from the perspective of the scaling effect. Thus, the objectives of this study are as follows: 1) establishing a
novel evaluation system for exploring the mechanism of RFF on NPS pollution; 2) characterizing the spatial scaling effect of RFF; and 3) improving the effect of RFF by introducing other complementary measures. 2. Materials and methods 2.1. Watershed description and data collection In this study, the Daning River Watershed (108� 440 –110� 110 E, 31� 040 –31� 440 N) in the Three Gorges Reservoir Region, China, with a total area of 2423.74 km2, was selected as the study area (Fig. 1a). Forest land and farmland account for 63.1% and 23.98% of land use, respec tively, with widely distributed sloping farmland (Wang et al., 2016). The topography is dominated by mountainous areas, and soil erosion is identified as a serious problem in this region (Shen et al., 2014). Due to an increase in fertilizer inputs, the P concentrations have increased
Fig. 1. (a) Location, (b) Sub-Watershed Delineation, (c) Land Use in 2010 as an Example of the Daning River Watershed. Note. RFF: returning farmland to forests. FRST: forestland; ORCD: orchard; PAST: pasture land; WATR:water land; URMD: land used for building; RICE: rice land; AGRL: farmland. 2
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
greatly over the last 10 years, which has already resulted in significant phytoplankton growth in the region. Thus, TP is selected as the target NPS pollutant in this study. In this study, daily precipitation and temperature were derived from the websites of the National Meteorological Information Center (htt p://data.cma.cn/), State Meteorological Administration of China. These weather data were processed to meet the required format of ArcGIS, and missing data were simulated by the Weather Generator. A detailed DEM, at a scale of 1:50,000, was obtained from a photogram metric aerial survey by the National Geomatics Center of China. These data were used for the calculation of slope degree and for the delineation of subwatersheds. A digital polygon map of the local river system was also used to adjust the generated digital river network. The soil map was downloaded from the website of the Nanjing Institute of Soil Science (http://www.issas.ac.cn/), whereas the related soil attribute properties were collected from the local soil database of the municipality of Chongqing. The data for agricultural operations, such as planting, tillage, irrigation, and harvest, were collected based on our field in vestigations with local farmers. The measured flow and sediment monthly data were obtained from the Changjiang Water Resources Commission, and the measured monthly TP data were collected from the Wuxi Environmental Protection Bureau.
forestland types is consistent with the previous ones. Five independent models were set up based on these five RFF land use maps, while the flow and NPS-TP data at multiple scales were simulated (Ding et al., 2016). In this study, the SWAT model was calibrated using the measured flow and TP at Wuxi Station, which is located at the outlet of sub watershed No. 13. The years 1998 and 1999 were used as the model warm-up period, and the calibration and validation were performed from 2000 to 2009 and from 2010 to 2015. respectively. The SUFI al gorithm of the SWAT-CUP was used for calibration, while the correlation coefficient (R2) and Nash-Sutcliffe coefficient (Ens) were selected for goodness-of-fit comparison between simulation and measurement (Table 1). 2.4. Construction of the evaluation system In this study, a new evaluation system was developed based on the Indicators of Hydrologic Alteration (IHA) methods developed by Richter et al. (2010). Both the mean value, extreme value and frequency of change of hydrological indicators and NPS exports were considered to accurately describe the mechanism of RFF on NPS pollution. The details of this evaluation system can be found in Table 2. For the study area, the means value of total flow, baseflow and TP were calculated based on the simulations from 2000 to 2015. The maximum and minimum value was calculated using the following equations, while the frequency of change is estimated as the mean of all positive (or negative) differences between consecutive monthly means. Pn Mi Meani ¼ i¼1 � 100% (2) n
2.2. Delineation of nested watersheds Based on geographic theory, a watershed can be broken down into a distinct stream network and a corresponding number of subwatersheds. A subwatershed can be defined by the variations in hill slope and valley morphology, whereas stream reaches can be outlined by the landscape, cross section, channel relief and the surrounding terrain (Wolock and Jr, 1995). In this task, the nested watershed method was used by dividing the Daning watershed into 6 drainage areas (Fig. 1a) with nested upstream-downstream relationships (Mcnamara et al., 1998). The cor responding drainage area of each nested catchment was quantified as 65.92 km2, 349.43 km2, 777.82 km2, 825.22 km2, 2012.52 km2 and 2423.74 km2, respectively.
m �X � Extrememax ¼ max Mi m
m �X
Extrememin ¼ min m
In this study, the Soil and Water Assessment Tool (SWAT), developed by the U.S. Department of Agriculture, was used during the baseline and RFF scenarios (Arnold et al., 1998). As a semi-distributed model, the SWAT incorporates a submodule Soil Conservation Service (SCS) method to simulate surface runoff and a modified rational method to quantify peak runoff rates (USDA-SCS, 1972). The available evapo transpiration approaches are the Penman-Monteith, Hargreaves and Priestley-Taylor equations. The Muskingum and variable storage equa tions are used for routing channel flow (Arnold et al., 2012). Besides, the major forms of P, such as soluble P, insoluble mineral P and organic P associated with humus, were considered, and their movements and transformations were simulated (Das et al., 2008). These soluble and attached P forms would be transported into the nearby reach segment and delivered within the river channel. out
¼ Po out þ PM out
�
(4)
Mi i¼1
where Meani is the mean value of the perennial monthly mean value in a specific month; n is a year from 2000 to 2015; Mi is the monthly value in a specific month; Extrememax is the maximum value; Extrememin is the minimum value; and m is one-month, three-month and seven-month. The baseflow index (BFI) is then defined as the ratio of long-term mean baseflow to total streamflow (Eckhardt, 2008; Beck et al., 2013). The formula for calculation is as follows: � Pn ðGWi � Ai Þ A � GWOUT ¼ FLOW OUT � Pn i¼1 (5) i¼1 ðWYLDi � Ai Þ A
2.3. The generation of flow and TP data at multiple spatial scales
PT
(3)
i¼1
BFI ¼
GWOUT � 100% FLOW OUT
(6)
where GWOUT is the baseflow simulation value (m3/s) for the export of the watershed; FLOW_OUT is the total flow (m3/s) for the export of the watershed; GW is the groundwater (m3/s) for inflowing into the watershed; WYLD is the water yield (m3/s) from subwatersheds into the river way; A is the area (km2) of all subwatersheds before assessment points; and i is the subwatershed coding controlled by assessment points.
(1)
where PT out is the TP load (kg) for the export of the watershed; PO out is the load (kg) of organic P load for the export of the watershed; and PM out is the load (kg) of mineral P for the export of the watershed. Land use data for 1995, 2000, 2005, 2010 and 2015 were interpreted from the Landsat5 TM dataset with a resolution of 30 m processed in this study (Fig. 1). The RFF scenario was generated using ArcGIS by changing the farmland over 15� in slope to forestland (RFF15� ). Using the Raster Analysis module, the slope layer and land use layer were subjected to overlay analysis, and the previous land use maps were modified through the Raster Calculator. The new configuration of
Table 1 Simulated evaluation results of flow and TP. Period
Date
Data type
R2
ENS
calibration
2001.01–2009.12
verification
2010.01–2015.12
Flow TP Flow TP
0.79 0.51 0.77 0.59
0.74 0.54 0.72 0.63
Note. TP: total phosphorus. 3
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
Table 2 Evaluation system of measures for RFF. Indicator
Flow (m3)
TP (kg)
Baseflow (m3)
Monthly mean Extreme value
Mean flow from January to December (MF-Month) Minimum value for one-month/three -month/seven -month (MinFL-1/3/7) Maximum value for one-month/three -month/seven -month (MaxFL-1/3/7) Average increase rate of monthly flow (IRMF) Average reduction rate of monthly flow (DRMF)
Mean TP load from January to December (TP-Month) Minimum value for one-month/three -month/seven -month (MinTP-1/3/7) Maximum value for one-month/three -month/seven -month (MaxTP-1/3/7) Average increase rate of TP load (IRTP) Average reduction rate of TP load (DRTP)
Baseflow index (BFI) Mean baseflow from January to December (BFI-Month)
Change Frequency
Note. RFF: returning farmland to forest; TP: total phosphorus.
2.5. The evaluation of the scaling effects of RFF
more obvious spatial scaling effects (Kuzuha et al., 2010).
According to Ichter and Summar (1997), we defined the change of each indicator by formula:
CV ¼ � 100%
CRm ¼
Rj
Ri Ri
� 100%
σ μ
(8)
where CV is the coefficient of variation of the NPS indicator across spatial scales; σ is the standard deviation of the NPS indicator across spatial scales; and μ is the mean value of the NPS indicator across spatial scales. The data preprocessing and statistical analysis were completed by Excel, 2016) (Microsoft Corporation, USA), Origin 9.0 (Origin Lab Corporation, USA) and ArcGIS 10.2 (ESRI, USA) software.
(7)
where CRm is the change rate of the mth NPS index; and Ri and Rj are the indicator value before and after the implementation of RFF, respectively. The spatial effect of each indicator was based on the coefficient of €schl and variation (CV) value of CRm at multiple spatial scales (Blo Sivapalan, 1997). If the CV is large for a specific indicator, it should be concluded that this indicator would have greater spatial variability and
Fig. 2. The Variation of Monthly Rainfall, Runoff and TP before and after RFF15� during 2000–2015. Note. RFF: returning farmland to forests. Total P: total phosphorus. 4
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
3. Results
6% during RFF, while the flow in other months showed a downward trend. Among these indicators, the reduction rates of the MF-Feb and the MF-Mar were estimated to be 41% and 37%, respectively, while the reduction rates of the MF-Jun, the MF-Jul, the MF-Aug and the MF-Sep were approximately 25%. This indicated that RFF showed less impact on the mean flow during the wet season than during the dry season. The MaxFL-1 and the MinFL-1 decreased from 374.6 mm/D to 326.6 mm/D and increased from 0.8 mm/D to 1.7 mm/D, respectively, during RFF. After RFF, the MaxFL-3, the MaxFL-7 and the MinFL-7 decreased by 17.0%, 18.6% and 26.3%, respectively. This indicated that RFF has a lower relative impact on extreme flow than on mean flow. In addition, the change rate of the BFI-Apr, the BFI-Jul, the BFI-Sep and the BFI-Oct was estimated to be approximately 300%. This indicated that RFF showed larger impacts on the baseflow compared to the total flow or extreme flow at the watershed scale.
3.1. Effects of RFF on flow/NPS results Fig. 1b shows the land use change during RFF15� , and the changes in flow and NPS-TP during RFF are given in Fig. 2. Farmland area decreased sharply from 23.98% to 6.09% during RFF, while the area of forestland changed from 1512.00 km2 to 1977.53 km2. The peak flow during the wet season decreased significantly, and the low flow during the dry season increased accordingly. Overall, the temporal variability of annual flow became more uniform during RFF. In comparison, both the peak and low NPS-TP values decreased, which was different from the results for flow. This indicated that RFF has different impacts on flow and NPS pollution. Regarding different evaluation indicators, the MF-Dec increased by
Fig. 3. (a)Spatial Variability Analysis of Characteristic Indexes; Fitting Relationship between (b) Extreme Value of Flow, (c) Extreme Value of TP, (d) BFI and Drainage Area. Note. Total P: total phosphorus; AVE-Jan: average montly mean value; Max-1: the maximum one-month value; Min-1: the minimum one-month value; BFI-Jan: Baseflow index in January. 5
Journal of Environmental Management 252 (2019) 109663
W. Wang et al.
After the implementation of RFF, the NPS-TP loss decreased from 85 tons to 57 tons annually, with the reduction of NPS-TP being approxi mately 59%. The reduction rates of the TP-Feb and the TP-Jul were overserved to be the largest (79%) and the lowest (47%), respectively. The reduction rate of NPS-TP load was higher in the wet season and lower in the dry season, which was inconsistent with of the results for flow. The MaxTP-3 and the MaxTP-7 decreased by 50.9% and 53.8%, respectively, while the MinTP-3 and the MinTP-7 decreased by 71.7% and 80.2%, respectively. This indicated that RFF had a greater impact on NPS pollution than on flow.
Fig. 3a, while the relationships between drainage area and indicators are given in Fig. 3b, c and d. As shown in Fig. 3a, the CV value showed that RFF has a significant impact on the mean flow at different spatial scales. Greater spatial variability was observed in the total flow indicators, such as the MF-Jan and the MF-Dec, with CV values of 78.66% and 49.89%, respectively. In comparison, RFF showed little impact on the TP in dicators (with CV values ranging from 20.11% to 4.1%) and baseflow indicators (with CV values ranging from 7.07% to 14.71%). And the frequency-related indicators, such as the MinFL-1, the MaxTP-3, the MaxTP-7, the MinTP-1, the MinTP-3, the MinTP-7, the IRMF, the DRMF and the IRTP, showed higher spatial variability than the other indicators. Fig. 4a shows the overall trend that the mean flow in most months showed a decreasing trend after RFF, except for the MF-Dec. The MF-Dec showed an increasing trend at multiple spatial scales after RFF, but its growth rate showed a downward trend with the increase in the drainage
3.2. The spatial scaling effect of RFF In this section, the spatial scaling effects were further explored by comparing the changes in flow and NPS-TP due to RFF in different nested catchments. The CV of different evaluation indicators is shown in
Fig. 4. Fitting relationship between (a) monthly mean flows, (b) Monthly mean TP and drainage area. 6
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
area. It is worth mentioning that the larger the drainage area, the smaller the change in mean flow. The MF-Jan also showed a decreasing trend at different spatial scales after RFF. Its reduction rate decreased if the drainage area increased to 2000 km2 and then stabilized with any further increase in the drainage area. Both the MF-Feb and the MF-Mar showed a significantly decreasing trend in each nested catchment, and the decreasing rate varied between 30% and 45%, which changed sharply compared with other months. Changes in these indicators increased with increasing drainage area (spatial scale). Fig. 4a shows that the change rate of the MF-Jul, the MF-Sep and the MF-Oct showed similar linear increases with the increase in spatial scale. For those extreme value-related indicators, the change rates of the MaxRL-7, the MaxRL-1 and the MaxRL-3 showed increasing trends with increasing spatial scale to 2000 km2. This indi cated that scale effects of RFF did exist, especially for the drainage area when changed under a specific threshold. With the increase in drainage area (spatial scale), the reduction rate of the NPS-TP showed a significant upward trend after RFF. Unlike flow, RFF showed greater impacts on the NPS-TP than on flow, which ranged from 21% to 90% (Fig. 4b). This was because NPS-TP is affected by both runoff and other processes, such as the soil erosion process, which is another important carrier of NPS-TP. At the same time, the change rate of flow between the flood season (from 13% to 21%) and the dry season (from 35% to 43%) was similar across the spatial scale. However, the change rate of TP in the flood season (from 26% to 49%) was signifi cantly higher than that in the dry season (from 72% to 88%). Most TP indicators changed greatly if the drainage area increased, and the maximum reduction rate of NPS-TP was found in the drainage area (spatial scale) of 2000 km2, indicating the greatest efficiency of the RFF on the reduction of NPS pollution at this spatial scale. Then, the reduction rate of NPS-TP decreased slightly with the drainage area reaching 2400 km2. For example, a trend was also observed for the TPFeb, with a maximum reduction rate of 93% at the drainage area of 2000 km2. The fitting relationship between BFI and drainage area is shown in Fig. 3d. After RFF, BFI increased significantly. The change rates of the BFI-Apr and the BFI-Sep were larger across spatial scales. In comparison, the change rate of the BFI-Jan was relatively flat, but the change rate of the BFI-Feb decreased slightly with the increase in drainage area. This indicated that the BFI shows a greater response to RFF in the wet season than in the dry season across spatial scales.
2005). At the same time, the change rate of BFI presents seasonal characteristics during RFF; BFI was lower in the dry season (166.54%) than in the wet season (367.02%). This indicated that BFI rose sharply in the wet season after RFF because the increased infiltration led to a high proportion of baseflow to total flow because of the well-grown vegeta tive root system during the wet season. It could be inferred that the larger underground root system would have a significant effect on baseflow after RFF, and this effect was more obvious during the wet season. The NPS-TP load also showed a sharp decrease in all months (the reduction rate varied from 47% to 79%), and the reduction rate was remarkably larger than that of flow. This was because that the NPS-TP is caused by both rainfall-runoff processes and other processes, such as soil erosion (Neris et al., 2013; Djodjic et al., 2018). The vegetation canopy will intercept some precipitation and reduce the direct soil erosion with the increase in soil stability from the vegetative root system, thus resulting in decreased NPS-TP export (Xia et al., 2014). During the wet season, the reduction rates of flow and TP reached 21% and 47%, and these were estimated to be 41% and 79% for flow and NPS-TP reduction in the dry season. This was because the faster water velocity in the wet season limited the effect of vegetation interception. Thus, it could be inferred that the effect of RFF on NPS pollution would be lower in wa tersheds located in rainfall-prone areas compared to those arid water sheds such as those in Northwest China (Chen et al., 2016a, 2016b). 4.2. Spatial scaling effect of RFF In this study, the spatial scaling effect was explored and shown in Table 3. The results showed that polynomial fitting relationships be tween most indicators and drainage area were good (R2 > 0.5). From Fig. 4a and Table 3, the change rate of the MF-Feb showed a gentle in crease from 43.46% to 41.85% in the drainage area from 1744.19 km2 to 2423.74 km2 and a slow downward trend from 35.55% to 43.46% in the drainage area from 65.92 km2 to 1744.19 km2. The change rate of the MF-Jul decreased linearly (from 16.60% to 20.63%) Table 3 The indicators of higher spatial variability. Category
Indicator
Function Type
Function
R2
Runoff
MF-Jan MF-Feb MF-Mar MF-Apr MF-Jul MF-Sep MF-Oct MF-Dec MaxFL-1 MaxFL-3 MaxFL-7 TP-Feb TP-Mar TP-Apr TP-May TP-Jun TP-Jul TP-Aug TP-Sep TP-Oct TP-Nov MaxTP-1 MaxTP-3 MaxTP-7 BFI-Jan BFI-Feb BFI-Apr BFI-Sep
Power Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Power Polynomial Polynomial Polynomial
y ¼ 0.35x 0.24 y ¼ 3.44E-8x2-1.2E-4x-0.33 y ¼ 1.09E-8x2-4.67E-5x-0.33 y ¼ 3.04E-9x2-2.97E-5x-0.22 y ¼ 5.97E-9x2-3.28E-5x-0.16 y ¼ 8.28E-10x2-2.00E-5x-0.16 y ¼ 1.55E-9x2-1.88E-5-0.17 y ¼ 3.6E-8x2-1.29E-4xþ0.17 y ¼ 9.21E-9x2-5.1E-5x-0.06 y ¼ 9.78E-9x2-2.35E-5x-0.10 y ¼ 3.88E-8x2-1.64E-4x-0.06 y ¼ 7.73E-8x2-2.36E-4x-0.69 y ¼ 3.84E-8x2-1.76E-4x-0.51 y ¼ 7.26E-8x2-2.71E-4x-0.44 y ¼ 7.01E-8x2-2.85E-4x-0.29 y ¼ 5.66E-8x2-2.24E-4x-0.34 y ¼ 5.75E-8x2-2.42E-4x-0.23 y ¼ 4.42E-8x2-1.92E-4x-0.37 y ¼ 8.71E-8x2-3.23E-4x-0.33 y ¼ 5.90E-8x2-2.22E-4x-0.50 y ¼ 7.14E-8x2-2.47E-4x-0.53 y ¼ 4.40E-8x2-2.06E-4x-0.17 y ¼ 6.46E-8x2-5.82E-4x-0.06 y ¼ 1.06E-7x2-7.46E-4x-0.08 y ¼ 1.25 x 0.37 y ¼ 2.86E-7x2-6.79E-4xþ1.71 y ¼ 6.38E-7x2-0.00136xþ3.12 y ¼ 2.77E-7x2-2.15E-4xþ2.75
0.79 0.75 0.61 0.58 0.99 0.99 0.78 0.96 0.83 0.76 0.96 0.90 0.76 0.99 0.99 0.91 0.98 0.91 0.97 0.99 0.99 0.99 0.94 0.92 0.50 0.63 0.56 0.72
4. Discussion 4.1. The influence mechanism of RFF Based on the results, the mean flow decreased in almost all months (the reduction rate varied from 21% to 41%) after RFF, which was consistent with the results of previous studies both in the Loess Plateau of China and the Amazon River watershed (Moraes et al., 2010; Yu et al., 2015). This could be because RFF is related to high, dense vegetation coverage, so after RFF, the substantial growth and interpolation of the underground root system makes the soil looser and increases soil porosity. At the same time, the ratio of evaporation to gross rainfall increased with increasing forest cover (Marin et al., 2000). Thus, a large amount of precipitation would be infiltrated and evaporated through reforestation, and the portion of surface runoff would be reduced greatly (Burch et al., 2010). Many scholars believe that as forest vegetation coverage increases, runoff and nutrient discharge decrease (Chen et al., 2016a, 2016b). However, we found that the MF-Dec and the MinFL-3 indicators increased by 5.9% and 58.5% after the increase in forest coverage. This indicated that the flow would increase in the dry season after RFF. This could be because forests have a great capacity to regulate hydrological processes and store water (Zhang et al., 2010). Forest vegetation would supply underground flow by increasing soil infiltration (Brown et al.,
TP
Baseflow
Note. MF-Jan: the mean flow in January; MaxFL-1: the maximum one-month flow; TP-Feb: total phosphorus loads in February; MaxTP-1: the maximum one-month total phosphorus loads; BFI-Jan: Baseflow index in January. 7
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
with the increase in drainage area from 65.92 km2 to 2423.74 km2. According to the fitting function, the change rate will attain a minimum value ( 20.51%) if the drainage area reaches 2747.07 km2. On the whole, there were two kinds of spatial scaling effects of RFF. After RFF, the MF changed and then tended to be flat with increasing spatial scale. Another rule was that the MF decreased after RFF, but its change rate decreased with increasing spatial scale. This was due to the smaller runoff and slower velocity in the dry season, as well as the large per centage of farmland in downstream watersheds (Dunne et al., 1991). As shown in Table S1, the proportion of vegetation in those areas increased significantly after RFF. However, flow in the wet season was less affected by spatial scale due to the larger quantity and faster runoff velocity. The MF-Dec showed an increasing trend after RFF, and the change rate declined sharply from 16.58% to 5.44% with the increase in drainage area from 65.92 km2 to 1791.67 km2. Finally, this indicator changed slightly from 5.44% to 6.00% with the increase in drainage area from 1791.67 km2 to 2423.74 km2. This could be caused by the increased contribution of surface runoff, and the response of total flow to groundwater input would reduce with the increase in drainage area (Arnold et al., 2000). At the same time, the total flow was less affected by the inflow of tributaries due to the lower flow in the dry season. The change rates of the BFI-Sep and the BFI-Feb were the maximum (from 270.83% to 385.61%) and the minimum (from 130.70% to 174.44%), respectively. The BFI in the wet season was greatly influenced by spatial scale due to the stronger root system with the increased proportion of forest area. The NPS-TP decreased after RFF, and the change rate showed a significant upward trend with increasing spatial scale. For example, the change rate of the TP-Jul varied from 24.57% to 48.46% in the drainage area from 65.92 km2 to 2104.35 km2. This indicated that P export (kg/ha/a) decreases as contributing areas increase because the amount of P leached from local farmland is trapped during transport to watershed outlets (Almendinger and Ulrich, 2017). However, when the drainage area reached approximately 2000 km2, the change rate of TP showed a downward trend with the change rate of the TP-Jul varying from 48.46% to 47.88% in the drainage area from 2104.35 km2 to 2423.74 km2. This could be due to the similar vegetation coverage and the gentler slopes with the increase in drainage area leading to lower interception of NPS-TP (Jensco and McGlynn, 2011). In the previous section, RFF showed greater impacts on NPS-TP (the reduction rate
varied from 47% to 79%) than on the flow (the reduction rate varied from 21% to 41%). The difference remained, and there was a greater difference between NPS-TP and flow across spatial scales. The change rate of NPS-TP (varying from 26.16% to 89.52%) is larger than that of flow (from 38.06% to 21.34%), which indicated that NPS-TP was more affected than flow by the spatial scale effect. This was also because the NPS-TP is also caused by soil erosion. Due to high forest coverage after RFF, the reduction rate of TP would increase with the increase in spatial scales. Mineral P was the main P in the study area. And the rate of change of the percent of mineral P in TP were stabilized with any further increase in the drainage area after RFF. We find that the forms of P had little effect on NPS pollution in this study. 4.3. The implications for NPS control considering the scaling effect From the above analysis, it could be inferred that the efficiency of NPS control would be impacted by spatial scale. For example, although RFF showed the greatest impacts on the NPS-TP around the drainage area of approximately 2000 km2, its efficiency might be much smaller at other spatial scales. Fig. 5 shows the relationship between the efficiency of RFF (mean value) and its spatial scale effect (CV). RFF reduced the NPS-TP in the dry season with a stable reduction rate. However, most of the indicators, such as the mean TP in the wet season and the extreme flow value, showed smaller reduction and higher CV values. This indi cated that RFF might show unstable efficiency across spatial scales, probably because of underlying surface conditions of NPS-TP during the wet season (Sun et al., 2018). After RFF15� , the TP loss decreased by 59% but the CV of NPS-TP reduction was 0.16. This is more obvious for those upstream watersheds, i.e. only 38% for subwatershed No. 1. This may be due to the smaller area of farmland in the upstream watershed. Thus, we further considerd other control measures as a supplement to RFF by considering spatial scales. Based on previous studies, fertil ization management was considered the most important factor affecting the NPS-TP (Dan and Brown, 2010). In this study, 30% fertilization reduction (FR30) was also simulated by setting up to the. mgt file of the SWAT model (Shen et al., 2013). By this combination, the results showed that reduction rates of NPS-TP were stable across spatial scales (Table 4). After FR30 þ RFF15� , the annual TP loss decreased by 77%, and the CV values changed to 0.06, indicating that the efficiency of NPS control was stable across spatial scales (Table 4). Based on the results
Fig. 5. Mean and CV of Change Rate of Different TP Indicators. Note. AVE-Jan: average montly mean of total phosphorus loads; Max-1: the maximum one-month total phosphorus loads; Min-1: the minimum one-month total phosphorus loads. 8
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
References
Table 4 TP load reduction rate after different operations. Drainage Area (km2) 65.92 349.43 777.82 825.22 2012.52 2423.74 CV
Abdulkareem, J.H., Wan, N.A.S., Pradhan, B., Jamil, N.R., 2018. Long-term hydrologic impact assessment of non-point source pollution measured through land use/land cover (LULC) changes in a tropical complex catchment. Earth Syst. Environ. 1–18. https://doi.org/10.1007/s41748-018-0042-1. Almendinger, J.E., Ulrich, J.S., 2017. Use of SWAT to estimate spatial scaling of phosphorus export coefficients and load reductions due to agricultural BMPS. J. Am. Water Resour. Assoc. 53, 547–561. https://doi.org/10.1111/1752-1688.12523. Arnold, J., Srinivasan, R., Muttiah, R., Williams, J.R., 1998. Large area hydrologic modeling and assessment Part I: model development. J. Am. Water Resour. Assoc. 34, 73–89. https://doi.org/10.1111/j.1752-1688.1998.tb05961.x. Arnold, J.G., Moriasi, D.N., Gassman, P.W., Abbaspour, K.C., White, M.J., 2012. SWAT: model use, calibration, and validation. Trans. Asabe 55, 1345–1352. https://doi. org/10.13031/2013.42246. Arnold, J.G., Muttiah, R.S., Srinivasan, R., Allen, P.M., 2000. Regional estimation of base flow and groundwater recharge in the Upper Mississippi river basin. J. Hydrol. 227, 21–40. https://doi.org/10.1016/S0022-1694(99)00139-0. Beck, H.E., Dijk, A.I.J.M., Miralles, D.G., Jeu, R.A.M.D., Bruijnzeel, S.L.A., Mcvicar, T.R., Schellekens, J., 2013. Global patterns in base flow index and recession based streamflow observations from 3394 catchments. Water Resour. Res. 49 https://doi. org/10.1002/2013WR013918. Bl€ oschl, G., Sivapalan, M., 1997. Process controls on regional flood frequency: coefficient of variation and basin scale. Water Resour. Res. 33, 2967–2980. https://doi.org/ 10.1029/97WR00568. Brown, A.E., Zhang, L., Mcmahon, T.A., Western, A.W., Vertessy, R.A., 2005. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. J. Hydrol. 310, 28–61. https://doi.org/10.1016/j. jhydrol.2004.12.010. Burch, G.J., Moore, I.D., Burns, J., 2010. Soil hydrophobic effects on infiltration and catchment runoff. Hydrol. Process. 3, 211–222. https://doi.org/10.1002/ hyp.3360030302. Cameron, C., Hutley, L.B., Friess, D.A., Munksgaard, N.C., 2019. Hydroperiod, soil moisture and bioturbation are critical drivers of greenhouse gas fluxes and vary as a function of landuse change in mangroves of Sulawesi, Indonesia. Sci. Total Environ. 654, 365–377. https://doi.org/10.1016/j.scitotenv.2018.11.092. Cerdan, O., Bissonnais, Y.L., Govers, G., Lecomte, V., Oost, K.V., Couturier, A., King, C., Dubreuil, N., 2004. Scale effect on runoff from experimental plots to catchments in agricultural areas in Normandy. J. Hydrol. 299, 4–14. https://doi.org/10.1016/j. jhydrol.2004.02.017. Chen, L., Sela, S., Svoray, T., Assouline, S., 2016. Scale dependence of Hortonian rainfallrunoff processes in a semiarid environment. Water Resour. Res. 52 https://doi.org/ 10.1002/2015WR018315. Chen, L., Wei, G.Y., Shen, Z.Y., 2016. Incorporating water quality responses into the framework of best management practices optimization. J. Hydrol. 541, 1363–1374. https://doi.org/10.1016/j.jhydrol.2016.08.038. Chen, Y., Cheng, S.Y., Liu, L., Guo, X.R., Wang, Z., Qin, C.H., Hao, R.X., Lu, J., Gao, J.J., 2013. Assessing the effects of land use changes on non-point source pollution reduction for the three Gorges watershed using the SWAT model. J. Environ. Inf. 22, 13–26. https://doi.org/10.3808/jei.201300242. Cheng, J.Z., Lee, X.Q., Tang, Y., Pan, W.J., Gao, W.C., Chen, Y., Wang, B., Cheng, H.G., 2016. Changes in above- and below-ground nitrogen stocks and allocations following the conversion of farmland to forest in rocky desertification regions. Agric. Ecosyst. Environ. 232, 9–16. https://doi.org/10.1016/j.agee.2016.07.019. Cho, K.H., Pachepsky, Y.A., Oliver, D.M., Muirhead, R.W., Park, Y., Quilliam, R.S., Shelton, D.R., 2016. Modeling fate and transport of fecally-derived microorganisms at the watershed scale: State of the science and future opportunities. Water Res. 100, 38–56. https://doi.org/10.1016/j.watres.2016.04.064. Dan, B., Brown, T.C., 2010. Forest practices as nonpoint sources of pollution in North America. Jawra J. Am. Water Resour. Assoc. 29, 729–740. https://doi.org/10.1111/ j.1752-1688.1993.tb03233.x. Das, T., Bardossy, A., Zehe, E., He, Y., 2008. Comparison of conceptual model performance using different representations of spatial variability. J. Hydrol. 356, 106–118. https://doi.org/10.1016/j.jhydrol.2008.04.008. Deardorff, J.W., 1972. Parameterization of the planetary boundary layer for use in general circulation models. Rev. Geophys. 14, 215–226. https://doi.org/10.1175/ 1520-0493(1972)1002.3.CO;2. Ding, J., Jiang, Y., Liu, Q., Hou, Z.J., Liao, J.Y., Fu, L., Peng, Q.Z., 2016. Influences of the land use pattern on water quality in low-order streams of the Dongjiang River basin, China: a multi-scale analysis. Sci. Total Environ. 551–552, 205–216. https://doi.org/ 10.1016/j.scitotenv.2016.01.162. Djodjic, F., Elmquist, H., Collentine, D., 2018. Targeting critical source areas for phosphorus losses: evaluation with soil testing, farmers’ assessment and modelling. Ambio 47, 45–56. https://doi.org/10.1007/s13280-017-0935-5. Du, X., Su, J., Li, X., Zhang, W., 2016. Modeling and evaluating of non-point source pollution in a semi-arid watershed: implications for watershed management. Clean. Soil, Air, Water 44, 247–255. https://doi.org/10.1002/clen.201400773. Dubrovsky, N.M., Hamilton, P.A., 2010. Nutrients in the nation’s streams and groundwater: national findings and implications. Fact Sheet. U.S. Geol. Surv. https://doi.org/10.3133/fs20103078. Dunne, T., Zhang, W., Aubry, B.F., 1991. Effects of rainfall, vegetation, and microtopography on infiltration and runoff. Water Resour. Res. 27, 2271–2285. https://doi.org/10.1029/91WR01585. Eckhardt, K., 2008. A comparison of baseflow indices, which were calculated with seven different baseflow separation methods. J. Hydrol. 352, 168–173. https://doi.org/ 10.1016/j.jhydrol.2008.01.005.
TP load reduction rate FR30
RFF15�
FR30 þ RFF15�
18.42% 16.82% 25.11% 26.27% 32.80% 32.72% 24.50%
38.30% 42.51% 51.70% 52.99% 61.38% 59.01% 16.21%
66.26% 65.20% 71.47% 72.24% 76.39% 76.90% 6.28%
Note. TP: total phosphorus; FR30: 30% fertilization reduction; RFF15� : return ing farmland to forests above 15� ; FR30 þ RFF15� : the combination of returning farmland to forests above 15� and 30% fertilization reduction; CV: coefficient of variation.
obtained from this study, it could be inferred that the combination of RFF and other source control measures would be more efficient if the spatial scaling effect were considered. 5. Conclusion In this study, a novel indicator system was developed for the eval uation of RFF on NPS pollution and the spatial scaling effect of the RFF efficiency was explored. The results of study indicated that RFF can effectively reduce total flow and NPS pollution, while the effect of RFF on baseflow, is stronger during the wet season. The results also indicated that a spatial scaling effect did exist, and specific thresholds of drainage area could be used as good reference of RFF efficiency. To reach a stable efficiency, other source control measures could be used as a supplement of the RFF. The results from this study can serve as a guide for similar research on other watersheds with different underlying surface condi tions and climates and provide a basis for more efficient and accurate implementation of RFF across spatial scales. However, much remains to be explored. First, the static land use input was used to avoid local or seasonal effects. Dynamic land use scenario should be considered to provide more detailed mechanism of the RFF. Second, the spatial scaling effect was found in this study. More studies are suggested in the future to explore the mechanism of this effect by considering other models and catchments. Declaration of competing interest We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled, ‘Is returning farmland to forest an effective measure to reduce phosphorus delivery across distinct spatial scales?‘. Acknowledgements Authors thank the anonymous referees for their helpful comments and suggestions on this paper. This work was supported by the National Natural Science Foundation of China (Nos. 51779010 and 51579011), the Fund for the Innovative Research Group of the National Natural Science Foundation of China (No. 51721093), Key Laboratory of Nonpoint Source Pollution Control, Ministry of Agriculture, P.R. China and the Interdiscipline Research Funds of Beijing Normal University. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.jenvman.2019.109663. 9
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663 land-use and hydroclimatology changes in the Itacaiúnas River watershed, southeastern Amazon. J. Environ. Manag. 167, 175–184. https://doi.org/10.1016/j. jenvman.2015.11.039. Strehmel, A., Schmalz, B., Fohrer, N., 2016. Evaluation of land use, land management and soil conservation strategies to reduce non-point source pollution loads in the three Gorges region, China. Environ. Manag. 58, 906–921. https://doi.org/10.1007/ s00267-016-0758-3. Sun, G.H., Hu, Z.Y., Wang, J.M., Ma, W.Q., Gu, L.L., Sun, F.L., Xie, Z.P., Yan, X.Q., 2018. The spatial heterogeneity of land surface conditions and its influence on surface fluxes over a typical underlying surface in the Tibetan Plateau. Theor. Appl. Climatol. 1–15. https://doi.org/10.1007/s00704-018-2369-9. USDA-SCS, 1972. Hydrology Sect. 4. Soil Conservation Service National Engineering Handbook, Washington, DC. Wang, G.B., Chen, L., Huang, Q., Xiao, Y.C., Shen, Z.Y., 2016. The influence of watershed subdivision level on model assessment and identification of non-point source priority management areas. Ecol. Eng. 87, 110–119. https://doi.org/10.1016/j. ecoleng.2015.11.041. Wang, H., Sun, F.B., Xia, J., Liu, W.B., 2017. Impact of LUCC on streamflow based on the SWAT model over the wei river basin on the Loess Plateau in China. Hydrol. Earth Syst. Sci. 21, 1–30. https://doi.org/10.5194/hess-21-1929-2017. Wolock, D.M., Jr, M.C., 1995. Comparison of single and multiple flow direction algorithms for computing topographic parameters in TOPMODEL. Water Resour. Res. 31, 1315–1324. Wu, D., Huang, Z.L., Xiao, W.F., Zeng, L.X., 2016. Land use structure change and its control effect of nitrogen output in a small watershed of three Gorges Reservoir area: a case study of lanlingxi watershed. Environ. Sci. 37, 2940–2946. https://doi.org/ 10.13277/j.hjkx.2016.08.015. Xia, Z., Guo, Q.Y., Zhan, B.L., Peng, L., 2014. Experimental study on slope runoff, erosion and sediment under different vegetation types. Water Resour. Manag. 28, 2415–2433. https://doi.org/10.1007/s11269-014-0603-5. Xu, Y.L., Wang, S.J., Ni, Y.M., 2010. Research on the influence of land use changing on non-point source pollution in Heihe River basin. J. Environ. Sci. Eng. 26, 72–77. https://doi.org/10.1111/j.1468-5965.1987.tb00312.x. Yu, H.B., Zhang, L., Hou, W., Wang, Y., Wu, D., Zhang, S.W., Bu, K., 2015. Cumulative effect of land use and cover changes in Naoli River basin in Sanjiang Plain on agricultural nonpoint source pollution load. Sustain. Water Resour. Manag. 1, 355–362. https://doi.org/10.1007/s40899-015-0028-1. Yun, S.L., Ahn, J.H., Min, K.S., Chu, K.H., Um, C.Y., Ko, K.B., 2015. Non-point sources of pollution from cultivated lands in river districts and their contribution to water bodies along the North Han River Basin in Korea. Desalin. Water Treat. 53, 2301–2311. https://doi.org/10.1080/19443994.2014.927124. Zhang, B., Li, W., Xie, G., Xiao, Y., 2010. Water conservation of forest ecosystem in Beijing and its value. Ecol. Econ. 69, 1416–1426. https://doi.org/10.1016/j. ecolecon.2008.09.004. Zhang, B.L., Cui, B.H., Zhang, S.M., Wu, Q.Y., Yao, L., 2018. Source apportionment of nitrogen and phosphorus from non-point source pollution in Nansi Lake Basin, China. Environ. Sci. Pollut. Res. 25, 19101. https://doi.org/10.1007/s11356-0181956-8. Zhang, L.T., Li, Z.B., Wang, S.S., 2016. Spatial scale effect on sediment dynamics in basin-wide floods within a typical agro-watershed: a case study in the hilly loess region of the Chinese Loess Plateau. Sci. Total Environ. 572, 476–486. https://doi. org/10.1016/j.scitotenv.2016.08.082. Zhang, Q., Chen, S., Dong, Y., Liu, D., Yang, X., Yang, Z., 2017. Controllability of phosphorus losses in surface runoff from sloping farmland treated by agricultural practices: P losses from sloping farmland treated by Agro-practices. Land Degrad. Dev. https://doi.org/10.1002/ldr.2702.
Ichter, B.D.R., Summar, Y., 1997. How much water does a river need? Freshw. Biol. 37, 231–249. https://doi.org/10.1046/j.1365-2427.1997.00153.x. Jensco, K.G., McGlynn, B.L., 2011. Hierarchical controls on runoff generation: topographically driven hydrologic connectivity, geology, and vegetation. Water Resour. Res. 47, 431–435. https://doi.org/10.1029/2011WR010666. Kerr, J.M., Depinto, J.V., Mcgrath, D., Sowa, S.P., Swinton, S.M., 2016. Sustainable management of Great Lakes watersheds dominated by agricultural land use. J. Gt. Lakes Res. 42, S038013301630185X https://doi.org/10.1016/j.jglr.2016.10.001. Kuzuha, Y., Tomosugi, K., Kishii, T., Komatsu, Y., Yamashiki, Y., Tsujimura, M., 2010. Coefficient of variation of annual flood peaks: variability of flood peak and rainfall intensity. Hydrol. Process. 23, 546–558. https://doi.org/10.1002/hyp.7184. Larsen, M.A., Christensen, J.H., Drews, M., Butts, M.B., Refsgaard, J.C., 2016. Local control on precipitation in a fully coupled climate-hydrology model. Sci. Rep. 6, 22927. https://doi.org/10.1038/srep22927. Liu, R.M., Zhang, P.P., Wang, X.J., Chen, Y.X., Shen, Z.Y., 2013. Assessment of effects of best management practices on agricultural non-point source pollution in Xiangxi River watershed. Agric. Water Manag. 117, 9–18. https://doi.org/10.1016/j. agwat.2012.10.018. Marin, C.T., Bouten, W., Sevink, J., 2000. Gross rainfall and its partitioning into throughfall, stemflow and evaporation of intercepted water in four forest ecosystems in western Amazonia. J. Hydrol. 237, 40–57. https://doi.org/10.1016/s0022-1694 (00)00301-2. Mcnamara, J.P., Kane, D.L., Hinzman, L.D., 1998. An analysis of streamflow hydrology in the Kuparuk River Basin, Arctic Alaska: a nested watershed approach. J. Hydrol. 206, 39–57. https://doi.org/10.1016/S0022-1694(98)00083-3. Moraes, J.M.D., Schuler, A.E., Dunne, T., Figueiredo, R.D.O., Victoria, R.L., 2010. Water storage and runoff processes in plinthic soils under forest and pasture in Eastern Amazonia. Hydrol. Process. 20, 2509–2526. https://doi.org/10.1002/hyp.6213. Neris, J., Tejedor, M., Rodríguez, M., Fuentes, J., Jim�enez, C., 2013. Effect of forest floor characteristics on water repellency, infiltration, runoff and soil loss in Andisols of Tenerife (Canary Islands, Spain). Catena 108, 50–57. https://doi.org/10.1016/j. catena.2012.04.011. Ouyang, W., Yang, W.X., Mats, T., Xu, Y.X., Lin, C.Y., Gao, X., Hao, Z.C., 2018. Using river sediments to analyze the driving force difference for non-point source pollution dynamics between two scales of watersheds. Water Res. 139, 311–320. https://doi. org/10.1016/j.watres.2018.04.020. Qiu, L.J., WU, Y.P., Wang, L.J., Lei, X.H., Liao, W.H., Hui, Y., Meng, X.Y., 2017. Spatiotemporal response of the water cycle to land use conversions in a typical hilly–gully basin on the Loess Plateau, China. Hydrol. Earth Syst. Sci. 21, 1–30. https://doi.org/10.5194/hess-2017-343. Richter, B.D., Baumgartner, J.V., Powell, J., Braun, D.P., 2010. A method for assessing hydrologic alteration within ecosystems. Conserv. Biol. 10, 1163–1174. https://doi. org/10.2307/2387152. Shangguan, Z.P., Zheng, S.X., 2006. Ecological properties of soil water and effects on forest vegetation in the Loess Plateau. Int. J. Sustain. Dev. World Ecol. 13, 307–314. https://doi.org/10.1080/13504500609469682. Shen, Z.Y., Chen, L., Hong, Q., Qiu, J.l., Xie, H., Liu, R.M., 2013. Assessment of nitrogen and phosphorus loads and causal factors from different land use and soil types in the Three Gorges Reservoir Area. Sci. Total Environ. 454–455, 383–392. https://doi. org/10.1016/j.scitotenv.2013.03.036. Shen, Z.Y., Qiu, J.l., Hong, Q., Chen, L., 2014. Simulation of spatial and temporal distributions of non-point source pollution load in the Three Gorges Reservoir Region. Sci. Total Environ. 493, 138–146. https://doi.org/10.1016/j. scitotenv.2014.05.109. Souza-Filho, P.W.M., Souza, E.B.D., Júnior, R.O.S., Jr, W.R.N., Mendonça, B.R.V.D., Guimar~ aes, J.T.F., Dall’Agnol, R., Siqueira, J.O., 2016. Four decades of land-cover,
10
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: http://www.elsevier.com/locate/jenvman
Research article
Is returning farmland to forest an effective measure to reduce phosphorus delivery across distinct spatial scales? Wenzhuo Wang, Lei Chen *, Yingxin Zhu, Kai Wang, Shibo Chen, Zhenyao Shen State Key Laboratory of Water Environment Simulation, School of Environment, Beijing Normal University, Beijing, 100875, PR China
A R T I C L E I N F O
A B S T R A C T
Keywords: Nonpoint source pollution Land use change Returning farmland to forests Spatial scales The soil and water assessment tool The three gorges reservoir region
As one typical land use change, the mechanism of returning farmland to forests (RFF) on nonpoint source pollution (NPS) is not clear, especially at multiple spatial scales. In this study, by using the Soil and Water Assessment Tool (SWAT), the changes in several flow-related and NPS-related indicators across several nested catchments were quantified and compared in the Three Gorges Reservoir Region, China. The results indicated that RFF could reduce the total flow and total phosphorus (TP), which are higher in the dry season (41% and 79%, respectively) than in the wet season (21% and 47%, respectively) at the watershed with a total area of 2423.74 km2. In comparison, RFF has a larger impact on the baseflow index during the wet season (367.02%) than during the dry season (166.54%). The results also indicated that a spatial scaling effect did exist, while the reduction in TP increased from 24.57% to 48.46% as the drainage area increased from 65.92 km2 to 2104.35 km2. Specific thresholds of RFF efficiency were also observed (approximately 2000 km2 for the study area). It is suggested that other source control measures could supplement RFF by stabilizing the efficiency of RFF across different spatial scales. The results of this study could provide valuable suggestions for land use development and water quality protection, especially for large, complex watersheds.
1. Introduction With rapid global change, the impacts of land use change on hy drological processes and related water quality have become a hot topic (Zhang et al., 2018; Cameron et al., 2019). In recent years, to feed the ever-growing population, more fertilizers and pesticides have been used, so agricultural nonpoint source (NPS) pollution has become a key threat to the nearby water bodies (Dubrovsky and Hamilton, 2010; Cho et al., 2016). Zhang et al. (2018) indicated that dry land was major source for both NPS pollution, with the contributed proportions of 81.3 and 81.8% of total nitrogen (N) and phosphorus (P) respectively, in Nansi Lake Basin, China. Du et al. (2016) studied that conservation tillage and vegetative buffers were effective measures by reducing 14.2% and 12.5% of N and P loads, in Liu river watershed, China. Farmland, a human-impacted landscape, often produces serious soil erosion and a large amount of NPS pollutants (Yun et al., 2015; Kerr et al., 2016). Zhang et al. (2017) proved that dry land farming practices on steep slopes would lead to large exports of sediments and total P (TP). In contrast, forestland is generally identified as a sink for sediment and pollutants due to its high vegetation coverage, strong soil consolidation
ability, high soil porosity and good infiltration capacity (Shangguan and Zheng, 2006; Abdulkareem et al., 2018). To date, returning farmland to forests (RFF) has been adopted as a policy by many countries as a measure for controlling soil erosion and NPS pollution (Cheng et al., 2016; Strehmel et al., 2016). After the implementation of RFF, the related NPS pollutants are usually reduced significantly because of the canopy interception, soil infiltration and evapotranspiration processes (Qiu et al., 2017; Wang et al., 2017). Souza-Filho et al. (2016) indicated that 52% deforestation of an area is responsible for an 85% increase in observed discharge. At the same time, most of the existing studies focus on the rate of reduction of NPS pollution due to RFF. However, the mechanism of RFF on NPS pollution is still not clear due to the lack of a proper evaluation indicator system. The scaling effect is another key barrier for the comprehensive assessment of RFF. The scale problem in hydrological studies originated from the surface parameterization of atmospheric circulation simulation in the 1970s (Deardorff, 1972). With the enhancement of public awareness, scale issues have become a focus of hydrological studies and has also become a focus of NPS pollution studies (Larsen et al., 2016; Ouyang et al., 2018). Cerdan et al. (2004) indicated that a significant
* Corresponding author. E-mail addresses: [email protected], [email protected] (L. Chen). https://doi.org/10.1016/j.jenvman.2019.109663 Received 10 May 2019; Received in revised form 1 September 2019; Accepted 30 September 2019 Available online 14 October 2019 0301-4797/© 2019 Elsevier Ltd. All rights reserved.
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
decrease in the runoff coefficient is observed as the drainage area in creases. Zhang et al. (2016) showed that the event-based suspended sediment and flow–sediment relationships remained spatially constant. Xu et al. (2010) showed that flow and TP in the Heihe River Watershed (a drainage area of 1481 km2) would be reduced by 50.64% and 45.44%, respectively, after RFF. Liu et al. (2013) indicated that RFF would result in a reduction of 9.77% and 20.61% of runoff and TP loads in the Xiangxi River (a drainage area of 3099 km2). Large-scale studies show that RFF on 15 degree- and 25 degree-slope lands would reduce NPS-TP by 14.92% and 44.96%, respectively, for the whole Three Gorges Reservoir Region (Chen et al., 2013). However, Wu et al. (2016) demonstrated that 81.75% of TP would be reduced if RFF were performed on lands with slopes above 25� in the same area. It is clear that there are significant differences in the effect of RFF across different spatial scales, but few studies have compared these differences from the perspective of the scaling effect. Thus, the objectives of this study are as follows: 1) establishing a
novel evaluation system for exploring the mechanism of RFF on NPS pollution; 2) characterizing the spatial scaling effect of RFF; and 3) improving the effect of RFF by introducing other complementary measures. 2. Materials and methods 2.1. Watershed description and data collection In this study, the Daning River Watershed (108� 440 –110� 110 E, 31� 040 –31� 440 N) in the Three Gorges Reservoir Region, China, with a total area of 2423.74 km2, was selected as the study area (Fig. 1a). Forest land and farmland account for 63.1% and 23.98% of land use, respec tively, with widely distributed sloping farmland (Wang et al., 2016). The topography is dominated by mountainous areas, and soil erosion is identified as a serious problem in this region (Shen et al., 2014). Due to an increase in fertilizer inputs, the P concentrations have increased
Fig. 1. (a) Location, (b) Sub-Watershed Delineation, (c) Land Use in 2010 as an Example of the Daning River Watershed. Note. RFF: returning farmland to forests. FRST: forestland; ORCD: orchard; PAST: pasture land; WATR:water land; URMD: land used for building; RICE: rice land; AGRL: farmland. 2
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
greatly over the last 10 years, which has already resulted in significant phytoplankton growth in the region. Thus, TP is selected as the target NPS pollutant in this study. In this study, daily precipitation and temperature were derived from the websites of the National Meteorological Information Center (htt p://data.cma.cn/), State Meteorological Administration of China. These weather data were processed to meet the required format of ArcGIS, and missing data were simulated by the Weather Generator. A detailed DEM, at a scale of 1:50,000, was obtained from a photogram metric aerial survey by the National Geomatics Center of China. These data were used for the calculation of slope degree and for the delineation of subwatersheds. A digital polygon map of the local river system was also used to adjust the generated digital river network. The soil map was downloaded from the website of the Nanjing Institute of Soil Science (http://www.issas.ac.cn/), whereas the related soil attribute properties were collected from the local soil database of the municipality of Chongqing. The data for agricultural operations, such as planting, tillage, irrigation, and harvest, were collected based on our field in vestigations with local farmers. The measured flow and sediment monthly data were obtained from the Changjiang Water Resources Commission, and the measured monthly TP data were collected from the Wuxi Environmental Protection Bureau.
forestland types is consistent with the previous ones. Five independent models were set up based on these five RFF land use maps, while the flow and NPS-TP data at multiple scales were simulated (Ding et al., 2016). In this study, the SWAT model was calibrated using the measured flow and TP at Wuxi Station, which is located at the outlet of sub watershed No. 13. The years 1998 and 1999 were used as the model warm-up period, and the calibration and validation were performed from 2000 to 2009 and from 2010 to 2015. respectively. The SUFI al gorithm of the SWAT-CUP was used for calibration, while the correlation coefficient (R2) and Nash-Sutcliffe coefficient (Ens) were selected for goodness-of-fit comparison between simulation and measurement (Table 1). 2.4. Construction of the evaluation system In this study, a new evaluation system was developed based on the Indicators of Hydrologic Alteration (IHA) methods developed by Richter et al. (2010). Both the mean value, extreme value and frequency of change of hydrological indicators and NPS exports were considered to accurately describe the mechanism of RFF on NPS pollution. The details of this evaluation system can be found in Table 2. For the study area, the means value of total flow, baseflow and TP were calculated based on the simulations from 2000 to 2015. The maximum and minimum value was calculated using the following equations, while the frequency of change is estimated as the mean of all positive (or negative) differences between consecutive monthly means. Pn Mi Meani ¼ i¼1 � 100% (2) n
2.2. Delineation of nested watersheds Based on geographic theory, a watershed can be broken down into a distinct stream network and a corresponding number of subwatersheds. A subwatershed can be defined by the variations in hill slope and valley morphology, whereas stream reaches can be outlined by the landscape, cross section, channel relief and the surrounding terrain (Wolock and Jr, 1995). In this task, the nested watershed method was used by dividing the Daning watershed into 6 drainage areas (Fig. 1a) with nested upstream-downstream relationships (Mcnamara et al., 1998). The cor responding drainage area of each nested catchment was quantified as 65.92 km2, 349.43 km2, 777.82 km2, 825.22 km2, 2012.52 km2 and 2423.74 km2, respectively.
m �X � Extrememax ¼ max Mi m
m �X
Extrememin ¼ min m
In this study, the Soil and Water Assessment Tool (SWAT), developed by the U.S. Department of Agriculture, was used during the baseline and RFF scenarios (Arnold et al., 1998). As a semi-distributed model, the SWAT incorporates a submodule Soil Conservation Service (SCS) method to simulate surface runoff and a modified rational method to quantify peak runoff rates (USDA-SCS, 1972). The available evapo transpiration approaches are the Penman-Monteith, Hargreaves and Priestley-Taylor equations. The Muskingum and variable storage equa tions are used for routing channel flow (Arnold et al., 2012). Besides, the major forms of P, such as soluble P, insoluble mineral P and organic P associated with humus, were considered, and their movements and transformations were simulated (Das et al., 2008). These soluble and attached P forms would be transported into the nearby reach segment and delivered within the river channel. out
¼ Po out þ PM out
�
(4)
Mi i¼1
where Meani is the mean value of the perennial monthly mean value in a specific month; n is a year from 2000 to 2015; Mi is the monthly value in a specific month; Extrememax is the maximum value; Extrememin is the minimum value; and m is one-month, three-month and seven-month. The baseflow index (BFI) is then defined as the ratio of long-term mean baseflow to total streamflow (Eckhardt, 2008; Beck et al., 2013). The formula for calculation is as follows: � Pn ðGWi � Ai Þ A � GWOUT ¼ FLOW OUT � Pn i¼1 (5) i¼1 ðWYLDi � Ai Þ A
2.3. The generation of flow and TP data at multiple spatial scales
PT
(3)
i¼1
BFI ¼
GWOUT � 100% FLOW OUT
(6)
where GWOUT is the baseflow simulation value (m3/s) for the export of the watershed; FLOW_OUT is the total flow (m3/s) for the export of the watershed; GW is the groundwater (m3/s) for inflowing into the watershed; WYLD is the water yield (m3/s) from subwatersheds into the river way; A is the area (km2) of all subwatersheds before assessment points; and i is the subwatershed coding controlled by assessment points.
(1)
where PT out is the TP load (kg) for the export of the watershed; PO out is the load (kg) of organic P load for the export of the watershed; and PM out is the load (kg) of mineral P for the export of the watershed. Land use data for 1995, 2000, 2005, 2010 and 2015 were interpreted from the Landsat5 TM dataset with a resolution of 30 m processed in this study (Fig. 1). The RFF scenario was generated using ArcGIS by changing the farmland over 15� in slope to forestland (RFF15� ). Using the Raster Analysis module, the slope layer and land use layer were subjected to overlay analysis, and the previous land use maps were modified through the Raster Calculator. The new configuration of
Table 1 Simulated evaluation results of flow and TP. Period
Date
Data type
R2
ENS
calibration
2001.01–2009.12
verification
2010.01–2015.12
Flow TP Flow TP
0.79 0.51 0.77 0.59
0.74 0.54 0.72 0.63
Note. TP: total phosphorus. 3
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
Table 2 Evaluation system of measures for RFF. Indicator
Flow (m3)
TP (kg)
Baseflow (m3)
Monthly mean Extreme value
Mean flow from January to December (MF-Month) Minimum value for one-month/three -month/seven -month (MinFL-1/3/7) Maximum value for one-month/three -month/seven -month (MaxFL-1/3/7) Average increase rate of monthly flow (IRMF) Average reduction rate of monthly flow (DRMF)
Mean TP load from January to December (TP-Month) Minimum value for one-month/three -month/seven -month (MinTP-1/3/7) Maximum value for one-month/three -month/seven -month (MaxTP-1/3/7) Average increase rate of TP load (IRTP) Average reduction rate of TP load (DRTP)
Baseflow index (BFI) Mean baseflow from January to December (BFI-Month)
Change Frequency
Note. RFF: returning farmland to forest; TP: total phosphorus.
2.5. The evaluation of the scaling effects of RFF
more obvious spatial scaling effects (Kuzuha et al., 2010).
According to Ichter and Summar (1997), we defined the change of each indicator by formula:
CV ¼ � 100%
CRm ¼
Rj
Ri Ri
� 100%
σ μ
(8)
where CV is the coefficient of variation of the NPS indicator across spatial scales; σ is the standard deviation of the NPS indicator across spatial scales; and μ is the mean value of the NPS indicator across spatial scales. The data preprocessing and statistical analysis were completed by Excel, 2016) (Microsoft Corporation, USA), Origin 9.0 (Origin Lab Corporation, USA) and ArcGIS 10.2 (ESRI, USA) software.
(7)
where CRm is the change rate of the mth NPS index; and Ri and Rj are the indicator value before and after the implementation of RFF, respectively. The spatial effect of each indicator was based on the coefficient of €schl and variation (CV) value of CRm at multiple spatial scales (Blo Sivapalan, 1997). If the CV is large for a specific indicator, it should be concluded that this indicator would have greater spatial variability and
Fig. 2. The Variation of Monthly Rainfall, Runoff and TP before and after RFF15� during 2000–2015. Note. RFF: returning farmland to forests. Total P: total phosphorus. 4
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
3. Results
6% during RFF, while the flow in other months showed a downward trend. Among these indicators, the reduction rates of the MF-Feb and the MF-Mar were estimated to be 41% and 37%, respectively, while the reduction rates of the MF-Jun, the MF-Jul, the MF-Aug and the MF-Sep were approximately 25%. This indicated that RFF showed less impact on the mean flow during the wet season than during the dry season. The MaxFL-1 and the MinFL-1 decreased from 374.6 mm/D to 326.6 mm/D and increased from 0.8 mm/D to 1.7 mm/D, respectively, during RFF. After RFF, the MaxFL-3, the MaxFL-7 and the MinFL-7 decreased by 17.0%, 18.6% and 26.3%, respectively. This indicated that RFF has a lower relative impact on extreme flow than on mean flow. In addition, the change rate of the BFI-Apr, the BFI-Jul, the BFI-Sep and the BFI-Oct was estimated to be approximately 300%. This indicated that RFF showed larger impacts on the baseflow compared to the total flow or extreme flow at the watershed scale.
3.1. Effects of RFF on flow/NPS results Fig. 1b shows the land use change during RFF15� , and the changes in flow and NPS-TP during RFF are given in Fig. 2. Farmland area decreased sharply from 23.98% to 6.09% during RFF, while the area of forestland changed from 1512.00 km2 to 1977.53 km2. The peak flow during the wet season decreased significantly, and the low flow during the dry season increased accordingly. Overall, the temporal variability of annual flow became more uniform during RFF. In comparison, both the peak and low NPS-TP values decreased, which was different from the results for flow. This indicated that RFF has different impacts on flow and NPS pollution. Regarding different evaluation indicators, the MF-Dec increased by
Fig. 3. (a)Spatial Variability Analysis of Characteristic Indexes; Fitting Relationship between (b) Extreme Value of Flow, (c) Extreme Value of TP, (d) BFI and Drainage Area. Note. Total P: total phosphorus; AVE-Jan: average montly mean value; Max-1: the maximum one-month value; Min-1: the minimum one-month value; BFI-Jan: Baseflow index in January. 5
Journal of Environmental Management 252 (2019) 109663
W. Wang et al.
After the implementation of RFF, the NPS-TP loss decreased from 85 tons to 57 tons annually, with the reduction of NPS-TP being approxi mately 59%. The reduction rates of the TP-Feb and the TP-Jul were overserved to be the largest (79%) and the lowest (47%), respectively. The reduction rate of NPS-TP load was higher in the wet season and lower in the dry season, which was inconsistent with of the results for flow. The MaxTP-3 and the MaxTP-7 decreased by 50.9% and 53.8%, respectively, while the MinTP-3 and the MinTP-7 decreased by 71.7% and 80.2%, respectively. This indicated that RFF had a greater impact on NPS pollution than on flow.
Fig. 3a, while the relationships between drainage area and indicators are given in Fig. 3b, c and d. As shown in Fig. 3a, the CV value showed that RFF has a significant impact on the mean flow at different spatial scales. Greater spatial variability was observed in the total flow indicators, such as the MF-Jan and the MF-Dec, with CV values of 78.66% and 49.89%, respectively. In comparison, RFF showed little impact on the TP in dicators (with CV values ranging from 20.11% to 4.1%) and baseflow indicators (with CV values ranging from 7.07% to 14.71%). And the frequency-related indicators, such as the MinFL-1, the MaxTP-3, the MaxTP-7, the MinTP-1, the MinTP-3, the MinTP-7, the IRMF, the DRMF and the IRTP, showed higher spatial variability than the other indicators. Fig. 4a shows the overall trend that the mean flow in most months showed a decreasing trend after RFF, except for the MF-Dec. The MF-Dec showed an increasing trend at multiple spatial scales after RFF, but its growth rate showed a downward trend with the increase in the drainage
3.2. The spatial scaling effect of RFF In this section, the spatial scaling effects were further explored by comparing the changes in flow and NPS-TP due to RFF in different nested catchments. The CV of different evaluation indicators is shown in
Fig. 4. Fitting relationship between (a) monthly mean flows, (b) Monthly mean TP and drainage area. 6
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
area. It is worth mentioning that the larger the drainage area, the smaller the change in mean flow. The MF-Jan also showed a decreasing trend at different spatial scales after RFF. Its reduction rate decreased if the drainage area increased to 2000 km2 and then stabilized with any further increase in the drainage area. Both the MF-Feb and the MF-Mar showed a significantly decreasing trend in each nested catchment, and the decreasing rate varied between 30% and 45%, which changed sharply compared with other months. Changes in these indicators increased with increasing drainage area (spatial scale). Fig. 4a shows that the change rate of the MF-Jul, the MF-Sep and the MF-Oct showed similar linear increases with the increase in spatial scale. For those extreme value-related indicators, the change rates of the MaxRL-7, the MaxRL-1 and the MaxRL-3 showed increasing trends with increasing spatial scale to 2000 km2. This indi cated that scale effects of RFF did exist, especially for the drainage area when changed under a specific threshold. With the increase in drainage area (spatial scale), the reduction rate of the NPS-TP showed a significant upward trend after RFF. Unlike flow, RFF showed greater impacts on the NPS-TP than on flow, which ranged from 21% to 90% (Fig. 4b). This was because NPS-TP is affected by both runoff and other processes, such as the soil erosion process, which is another important carrier of NPS-TP. At the same time, the change rate of flow between the flood season (from 13% to 21%) and the dry season (from 35% to 43%) was similar across the spatial scale. However, the change rate of TP in the flood season (from 26% to 49%) was signifi cantly higher than that in the dry season (from 72% to 88%). Most TP indicators changed greatly if the drainage area increased, and the maximum reduction rate of NPS-TP was found in the drainage area (spatial scale) of 2000 km2, indicating the greatest efficiency of the RFF on the reduction of NPS pollution at this spatial scale. Then, the reduction rate of NPS-TP decreased slightly with the drainage area reaching 2400 km2. For example, a trend was also observed for the TPFeb, with a maximum reduction rate of 93% at the drainage area of 2000 km2. The fitting relationship between BFI and drainage area is shown in Fig. 3d. After RFF, BFI increased significantly. The change rates of the BFI-Apr and the BFI-Sep were larger across spatial scales. In comparison, the change rate of the BFI-Jan was relatively flat, but the change rate of the BFI-Feb decreased slightly with the increase in drainage area. This indicated that the BFI shows a greater response to RFF in the wet season than in the dry season across spatial scales.
2005). At the same time, the change rate of BFI presents seasonal characteristics during RFF; BFI was lower in the dry season (166.54%) than in the wet season (367.02%). This indicated that BFI rose sharply in the wet season after RFF because the increased infiltration led to a high proportion of baseflow to total flow because of the well-grown vegeta tive root system during the wet season. It could be inferred that the larger underground root system would have a significant effect on baseflow after RFF, and this effect was more obvious during the wet season. The NPS-TP load also showed a sharp decrease in all months (the reduction rate varied from 47% to 79%), and the reduction rate was remarkably larger than that of flow. This was because that the NPS-TP is caused by both rainfall-runoff processes and other processes, such as soil erosion (Neris et al., 2013; Djodjic et al., 2018). The vegetation canopy will intercept some precipitation and reduce the direct soil erosion with the increase in soil stability from the vegetative root system, thus resulting in decreased NPS-TP export (Xia et al., 2014). During the wet season, the reduction rates of flow and TP reached 21% and 47%, and these were estimated to be 41% and 79% for flow and NPS-TP reduction in the dry season. This was because the faster water velocity in the wet season limited the effect of vegetation interception. Thus, it could be inferred that the effect of RFF on NPS pollution would be lower in wa tersheds located in rainfall-prone areas compared to those arid water sheds such as those in Northwest China (Chen et al., 2016a, 2016b). 4.2. Spatial scaling effect of RFF In this study, the spatial scaling effect was explored and shown in Table 3. The results showed that polynomial fitting relationships be tween most indicators and drainage area were good (R2 > 0.5). From Fig. 4a and Table 3, the change rate of the MF-Feb showed a gentle in crease from 43.46% to 41.85% in the drainage area from 1744.19 km2 to 2423.74 km2 and a slow downward trend from 35.55% to 43.46% in the drainage area from 65.92 km2 to 1744.19 km2. The change rate of the MF-Jul decreased linearly (from 16.60% to 20.63%) Table 3 The indicators of higher spatial variability. Category
Indicator
Function Type
Function
R2
Runoff
MF-Jan MF-Feb MF-Mar MF-Apr MF-Jul MF-Sep MF-Oct MF-Dec MaxFL-1 MaxFL-3 MaxFL-7 TP-Feb TP-Mar TP-Apr TP-May TP-Jun TP-Jul TP-Aug TP-Sep TP-Oct TP-Nov MaxTP-1 MaxTP-3 MaxTP-7 BFI-Jan BFI-Feb BFI-Apr BFI-Sep
Power Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Polynomial Power Polynomial Polynomial Polynomial
y ¼ 0.35x 0.24 y ¼ 3.44E-8x2-1.2E-4x-0.33 y ¼ 1.09E-8x2-4.67E-5x-0.33 y ¼ 3.04E-9x2-2.97E-5x-0.22 y ¼ 5.97E-9x2-3.28E-5x-0.16 y ¼ 8.28E-10x2-2.00E-5x-0.16 y ¼ 1.55E-9x2-1.88E-5-0.17 y ¼ 3.6E-8x2-1.29E-4xþ0.17 y ¼ 9.21E-9x2-5.1E-5x-0.06 y ¼ 9.78E-9x2-2.35E-5x-0.10 y ¼ 3.88E-8x2-1.64E-4x-0.06 y ¼ 7.73E-8x2-2.36E-4x-0.69 y ¼ 3.84E-8x2-1.76E-4x-0.51 y ¼ 7.26E-8x2-2.71E-4x-0.44 y ¼ 7.01E-8x2-2.85E-4x-0.29 y ¼ 5.66E-8x2-2.24E-4x-0.34 y ¼ 5.75E-8x2-2.42E-4x-0.23 y ¼ 4.42E-8x2-1.92E-4x-0.37 y ¼ 8.71E-8x2-3.23E-4x-0.33 y ¼ 5.90E-8x2-2.22E-4x-0.50 y ¼ 7.14E-8x2-2.47E-4x-0.53 y ¼ 4.40E-8x2-2.06E-4x-0.17 y ¼ 6.46E-8x2-5.82E-4x-0.06 y ¼ 1.06E-7x2-7.46E-4x-0.08 y ¼ 1.25 x 0.37 y ¼ 2.86E-7x2-6.79E-4xþ1.71 y ¼ 6.38E-7x2-0.00136xþ3.12 y ¼ 2.77E-7x2-2.15E-4xþ2.75
0.79 0.75 0.61 0.58 0.99 0.99 0.78 0.96 0.83 0.76 0.96 0.90 0.76 0.99 0.99 0.91 0.98 0.91 0.97 0.99 0.99 0.99 0.94 0.92 0.50 0.63 0.56 0.72
4. Discussion 4.1. The influence mechanism of RFF Based on the results, the mean flow decreased in almost all months (the reduction rate varied from 21% to 41%) after RFF, which was consistent with the results of previous studies both in the Loess Plateau of China and the Amazon River watershed (Moraes et al., 2010; Yu et al., 2015). This could be because RFF is related to high, dense vegetation coverage, so after RFF, the substantial growth and interpolation of the underground root system makes the soil looser and increases soil porosity. At the same time, the ratio of evaporation to gross rainfall increased with increasing forest cover (Marin et al., 2000). Thus, a large amount of precipitation would be infiltrated and evaporated through reforestation, and the portion of surface runoff would be reduced greatly (Burch et al., 2010). Many scholars believe that as forest vegetation coverage increases, runoff and nutrient discharge decrease (Chen et al., 2016a, 2016b). However, we found that the MF-Dec and the MinFL-3 indicators increased by 5.9% and 58.5% after the increase in forest coverage. This indicated that the flow would increase in the dry season after RFF. This could be because forests have a great capacity to regulate hydrological processes and store water (Zhang et al., 2010). Forest vegetation would supply underground flow by increasing soil infiltration (Brown et al.,
TP
Baseflow
Note. MF-Jan: the mean flow in January; MaxFL-1: the maximum one-month flow; TP-Feb: total phosphorus loads in February; MaxTP-1: the maximum one-month total phosphorus loads; BFI-Jan: Baseflow index in January. 7
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
with the increase in drainage area from 65.92 km2 to 2423.74 km2. According to the fitting function, the change rate will attain a minimum value ( 20.51%) if the drainage area reaches 2747.07 km2. On the whole, there were two kinds of spatial scaling effects of RFF. After RFF, the MF changed and then tended to be flat with increasing spatial scale. Another rule was that the MF decreased after RFF, but its change rate decreased with increasing spatial scale. This was due to the smaller runoff and slower velocity in the dry season, as well as the large per centage of farmland in downstream watersheds (Dunne et al., 1991). As shown in Table S1, the proportion of vegetation in those areas increased significantly after RFF. However, flow in the wet season was less affected by spatial scale due to the larger quantity and faster runoff velocity. The MF-Dec showed an increasing trend after RFF, and the change rate declined sharply from 16.58% to 5.44% with the increase in drainage area from 65.92 km2 to 1791.67 km2. Finally, this indicator changed slightly from 5.44% to 6.00% with the increase in drainage area from 1791.67 km2 to 2423.74 km2. This could be caused by the increased contribution of surface runoff, and the response of total flow to groundwater input would reduce with the increase in drainage area (Arnold et al., 2000). At the same time, the total flow was less affected by the inflow of tributaries due to the lower flow in the dry season. The change rates of the BFI-Sep and the BFI-Feb were the maximum (from 270.83% to 385.61%) and the minimum (from 130.70% to 174.44%), respectively. The BFI in the wet season was greatly influenced by spatial scale due to the stronger root system with the increased proportion of forest area. The NPS-TP decreased after RFF, and the change rate showed a significant upward trend with increasing spatial scale. For example, the change rate of the TP-Jul varied from 24.57% to 48.46% in the drainage area from 65.92 km2 to 2104.35 km2. This indicated that P export (kg/ha/a) decreases as contributing areas increase because the amount of P leached from local farmland is trapped during transport to watershed outlets (Almendinger and Ulrich, 2017). However, when the drainage area reached approximately 2000 km2, the change rate of TP showed a downward trend with the change rate of the TP-Jul varying from 48.46% to 47.88% in the drainage area from 2104.35 km2 to 2423.74 km2. This could be due to the similar vegetation coverage and the gentler slopes with the increase in drainage area leading to lower interception of NPS-TP (Jensco and McGlynn, 2011). In the previous section, RFF showed greater impacts on NPS-TP (the reduction rate
varied from 47% to 79%) than on the flow (the reduction rate varied from 21% to 41%). The difference remained, and there was a greater difference between NPS-TP and flow across spatial scales. The change rate of NPS-TP (varying from 26.16% to 89.52%) is larger than that of flow (from 38.06% to 21.34%), which indicated that NPS-TP was more affected than flow by the spatial scale effect. This was also because the NPS-TP is also caused by soil erosion. Due to high forest coverage after RFF, the reduction rate of TP would increase with the increase in spatial scales. Mineral P was the main P in the study area. And the rate of change of the percent of mineral P in TP were stabilized with any further increase in the drainage area after RFF. We find that the forms of P had little effect on NPS pollution in this study. 4.3. The implications for NPS control considering the scaling effect From the above analysis, it could be inferred that the efficiency of NPS control would be impacted by spatial scale. For example, although RFF showed the greatest impacts on the NPS-TP around the drainage area of approximately 2000 km2, its efficiency might be much smaller at other spatial scales. Fig. 5 shows the relationship between the efficiency of RFF (mean value) and its spatial scale effect (CV). RFF reduced the NPS-TP in the dry season with a stable reduction rate. However, most of the indicators, such as the mean TP in the wet season and the extreme flow value, showed smaller reduction and higher CV values. This indi cated that RFF might show unstable efficiency across spatial scales, probably because of underlying surface conditions of NPS-TP during the wet season (Sun et al., 2018). After RFF15� , the TP loss decreased by 59% but the CV of NPS-TP reduction was 0.16. This is more obvious for those upstream watersheds, i.e. only 38% for subwatershed No. 1. This may be due to the smaller area of farmland in the upstream watershed. Thus, we further considerd other control measures as a supplement to RFF by considering spatial scales. Based on previous studies, fertil ization management was considered the most important factor affecting the NPS-TP (Dan and Brown, 2010). In this study, 30% fertilization reduction (FR30) was also simulated by setting up to the. mgt file of the SWAT model (Shen et al., 2013). By this combination, the results showed that reduction rates of NPS-TP were stable across spatial scales (Table 4). After FR30 þ RFF15� , the annual TP loss decreased by 77%, and the CV values changed to 0.06, indicating that the efficiency of NPS control was stable across spatial scales (Table 4). Based on the results
Fig. 5. Mean and CV of Change Rate of Different TP Indicators. Note. AVE-Jan: average montly mean of total phosphorus loads; Max-1: the maximum one-month total phosphorus loads; Min-1: the minimum one-month total phosphorus loads. 8
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663
References
Table 4 TP load reduction rate after different operations. Drainage Area (km2) 65.92 349.43 777.82 825.22 2012.52 2423.74 CV
Abdulkareem, J.H., Wan, N.A.S., Pradhan, B., Jamil, N.R., 2018. Long-term hydrologic impact assessment of non-point source pollution measured through land use/land cover (LULC) changes in a tropical complex catchment. Earth Syst. Environ. 1–18. https://doi.org/10.1007/s41748-018-0042-1. Almendinger, J.E., Ulrich, J.S., 2017. Use of SWAT to estimate spatial scaling of phosphorus export coefficients and load reductions due to agricultural BMPS. J. Am. Water Resour. Assoc. 53, 547–561. https://doi.org/10.1111/1752-1688.12523. Arnold, J., Srinivasan, R., Muttiah, R., Williams, J.R., 1998. Large area hydrologic modeling and assessment Part I: model development. J. Am. Water Resour. Assoc. 34, 73–89. https://doi.org/10.1111/j.1752-1688.1998.tb05961.x. Arnold, J.G., Moriasi, D.N., Gassman, P.W., Abbaspour, K.C., White, M.J., 2012. SWAT: model use, calibration, and validation. Trans. Asabe 55, 1345–1352. https://doi. org/10.13031/2013.42246. Arnold, J.G., Muttiah, R.S., Srinivasan, R., Allen, P.M., 2000. Regional estimation of base flow and groundwater recharge in the Upper Mississippi river basin. J. Hydrol. 227, 21–40. https://doi.org/10.1016/S0022-1694(99)00139-0. Beck, H.E., Dijk, A.I.J.M., Miralles, D.G., Jeu, R.A.M.D., Bruijnzeel, S.L.A., Mcvicar, T.R., Schellekens, J., 2013. Global patterns in base flow index and recession based streamflow observations from 3394 catchments. Water Resour. Res. 49 https://doi. org/10.1002/2013WR013918. Bl€ oschl, G., Sivapalan, M., 1997. Process controls on regional flood frequency: coefficient of variation and basin scale. Water Resour. Res. 33, 2967–2980. https://doi.org/ 10.1029/97WR00568. Brown, A.E., Zhang, L., Mcmahon, T.A., Western, A.W., Vertessy, R.A., 2005. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. J. Hydrol. 310, 28–61. https://doi.org/10.1016/j. jhydrol.2004.12.010. Burch, G.J., Moore, I.D., Burns, J., 2010. Soil hydrophobic effects on infiltration and catchment runoff. Hydrol. Process. 3, 211–222. https://doi.org/10.1002/ hyp.3360030302. Cameron, C., Hutley, L.B., Friess, D.A., Munksgaard, N.C., 2019. Hydroperiod, soil moisture and bioturbation are critical drivers of greenhouse gas fluxes and vary as a function of landuse change in mangroves of Sulawesi, Indonesia. Sci. Total Environ. 654, 365–377. https://doi.org/10.1016/j.scitotenv.2018.11.092. Cerdan, O., Bissonnais, Y.L., Govers, G., Lecomte, V., Oost, K.V., Couturier, A., King, C., Dubreuil, N., 2004. Scale effect on runoff from experimental plots to catchments in agricultural areas in Normandy. J. Hydrol. 299, 4–14. https://doi.org/10.1016/j. jhydrol.2004.02.017. Chen, L., Sela, S., Svoray, T., Assouline, S., 2016. Scale dependence of Hortonian rainfallrunoff processes in a semiarid environment. Water Resour. Res. 52 https://doi.org/ 10.1002/2015WR018315. Chen, L., Wei, G.Y., Shen, Z.Y., 2016. Incorporating water quality responses into the framework of best management practices optimization. J. Hydrol. 541, 1363–1374. https://doi.org/10.1016/j.jhydrol.2016.08.038. Chen, Y., Cheng, S.Y., Liu, L., Guo, X.R., Wang, Z., Qin, C.H., Hao, R.X., Lu, J., Gao, J.J., 2013. Assessing the effects of land use changes on non-point source pollution reduction for the three Gorges watershed using the SWAT model. J. Environ. Inf. 22, 13–26. https://doi.org/10.3808/jei.201300242. Cheng, J.Z., Lee, X.Q., Tang, Y., Pan, W.J., Gao, W.C., Chen, Y., Wang, B., Cheng, H.G., 2016. Changes in above- and below-ground nitrogen stocks and allocations following the conversion of farmland to forest in rocky desertification regions. Agric. Ecosyst. Environ. 232, 9–16. https://doi.org/10.1016/j.agee.2016.07.019. Cho, K.H., Pachepsky, Y.A., Oliver, D.M., Muirhead, R.W., Park, Y., Quilliam, R.S., Shelton, D.R., 2016. Modeling fate and transport of fecally-derived microorganisms at the watershed scale: State of the science and future opportunities. Water Res. 100, 38–56. https://doi.org/10.1016/j.watres.2016.04.064. Dan, B., Brown, T.C., 2010. Forest practices as nonpoint sources of pollution in North America. Jawra J. Am. Water Resour. Assoc. 29, 729–740. https://doi.org/10.1111/ j.1752-1688.1993.tb03233.x. Das, T., Bardossy, A., Zehe, E., He, Y., 2008. Comparison of conceptual model performance using different representations of spatial variability. J. Hydrol. 356, 106–118. https://doi.org/10.1016/j.jhydrol.2008.04.008. Deardorff, J.W., 1972. Parameterization of the planetary boundary layer for use in general circulation models. Rev. Geophys. 14, 215–226. https://doi.org/10.1175/ 1520-0493(1972)1002.3.CO;2. Ding, J., Jiang, Y., Liu, Q., Hou, Z.J., Liao, J.Y., Fu, L., Peng, Q.Z., 2016. Influences of the land use pattern on water quality in low-order streams of the Dongjiang River basin, China: a multi-scale analysis. Sci. Total Environ. 551–552, 205–216. https://doi.org/ 10.1016/j.scitotenv.2016.01.162. Djodjic, F., Elmquist, H., Collentine, D., 2018. Targeting critical source areas for phosphorus losses: evaluation with soil testing, farmers’ assessment and modelling. Ambio 47, 45–56. https://doi.org/10.1007/s13280-017-0935-5. Du, X., Su, J., Li, X., Zhang, W., 2016. Modeling and evaluating of non-point source pollution in a semi-arid watershed: implications for watershed management. Clean. Soil, Air, Water 44, 247–255. https://doi.org/10.1002/clen.201400773. Dubrovsky, N.M., Hamilton, P.A., 2010. Nutrients in the nation’s streams and groundwater: national findings and implications. Fact Sheet. U.S. Geol. Surv. https://doi.org/10.3133/fs20103078. Dunne, T., Zhang, W., Aubry, B.F., 1991. Effects of rainfall, vegetation, and microtopography on infiltration and runoff. Water Resour. Res. 27, 2271–2285. https://doi.org/10.1029/91WR01585. Eckhardt, K., 2008. A comparison of baseflow indices, which were calculated with seven different baseflow separation methods. J. Hydrol. 352, 168–173. https://doi.org/ 10.1016/j.jhydrol.2008.01.005.
TP load reduction rate FR30
RFF15�
FR30 þ RFF15�
18.42% 16.82% 25.11% 26.27% 32.80% 32.72% 24.50%
38.30% 42.51% 51.70% 52.99% 61.38% 59.01% 16.21%
66.26% 65.20% 71.47% 72.24% 76.39% 76.90% 6.28%
Note. TP: total phosphorus; FR30: 30% fertilization reduction; RFF15� : return ing farmland to forests above 15� ; FR30 þ RFF15� : the combination of returning farmland to forests above 15� and 30% fertilization reduction; CV: coefficient of variation.
obtained from this study, it could be inferred that the combination of RFF and other source control measures would be more efficient if the spatial scaling effect were considered. 5. Conclusion In this study, a novel indicator system was developed for the eval uation of RFF on NPS pollution and the spatial scaling effect of the RFF efficiency was explored. The results of study indicated that RFF can effectively reduce total flow and NPS pollution, while the effect of RFF on baseflow, is stronger during the wet season. The results also indicated that a spatial scaling effect did exist, and specific thresholds of drainage area could be used as good reference of RFF efficiency. To reach a stable efficiency, other source control measures could be used as a supplement of the RFF. The results from this study can serve as a guide for similar research on other watersheds with different underlying surface condi tions and climates and provide a basis for more efficient and accurate implementation of RFF across spatial scales. However, much remains to be explored. First, the static land use input was used to avoid local or seasonal effects. Dynamic land use scenario should be considered to provide more detailed mechanism of the RFF. Second, the spatial scaling effect was found in this study. More studies are suggested in the future to explore the mechanism of this effect by considering other models and catchments. Declaration of competing interest We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled, ‘Is returning farmland to forest an effective measure to reduce phosphorus delivery across distinct spatial scales?‘. Acknowledgements Authors thank the anonymous referees for their helpful comments and suggestions on this paper. This work was supported by the National Natural Science Foundation of China (Nos. 51779010 and 51579011), the Fund for the Innovative Research Group of the National Natural Science Foundation of China (No. 51721093), Key Laboratory of Nonpoint Source Pollution Control, Ministry of Agriculture, P.R. China and the Interdiscipline Research Funds of Beijing Normal University. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.jenvman.2019.109663. 9
W. Wang et al.
Journal of Environmental Management 252 (2019) 109663 land-use and hydroclimatology changes in the Itacaiúnas River watershed, southeastern Amazon. J. Environ. Manag. 167, 175–184. https://doi.org/10.1016/j. jenvman.2015.11.039. Strehmel, A., Schmalz, B., Fohrer, N., 2016. Evaluation of land use, land management and soil conservation strategies to reduce non-point source pollution loads in the three Gorges region, China. Environ. Manag. 58, 906–921. https://doi.org/10.1007/ s00267-016-0758-3. Sun, G.H., Hu, Z.Y., Wang, J.M., Ma, W.Q., Gu, L.L., Sun, F.L., Xie, Z.P., Yan, X.Q., 2018. The spatial heterogeneity of land surface conditions and its influence on surface fluxes over a typical underlying surface in the Tibetan Plateau. Theor. Appl. Climatol. 1–15. https://doi.org/10.1007/s00704-018-2369-9. USDA-SCS, 1972. Hydrology Sect. 4. Soil Conservation Service National Engineering Handbook, Washington, DC. Wang, G.B., Chen, L., Huang, Q., Xiao, Y.C., Shen, Z.Y., 2016. The influence of watershed subdivision level on model assessment and identification of non-point source priority management areas. Ecol. Eng. 87, 110–119. https://doi.org/10.1016/j. ecoleng.2015.11.041. Wang, H., Sun, F.B., Xia, J., Liu, W.B., 2017. Impact of LUCC on streamflow based on the SWAT model over the wei river basin on the Loess Plateau in China. Hydrol. Earth Syst. Sci. 21, 1–30. https://doi.org/10.5194/hess-21-1929-2017. Wolock, D.M., Jr, M.C., 1995. Comparison of single and multiple flow direction algorithms for computing topographic parameters in TOPMODEL. Water Resour. Res. 31, 1315–1324. Wu, D., Huang, Z.L., Xiao, W.F., Zeng, L.X., 2016. Land use structure change and its control effect of nitrogen output in a small watershed of three Gorges Reservoir area: a case study of lanlingxi watershed. Environ. Sci. 37, 2940–2946. https://doi.org/ 10.13277/j.hjkx.2016.08.015. Xia, Z., Guo, Q.Y., Zhan, B.L., Peng, L., 2014. Experimental study on slope runoff, erosion and sediment under different vegetation types. Water Resour. Manag. 28, 2415–2433. https://doi.org/10.1007/s11269-014-0603-5. Xu, Y.L., Wang, S.J., Ni, Y.M., 2010. Research on the influence of land use changing on non-point source pollution in Heihe River basin. J. Environ. Sci. Eng. 26, 72–77. https://doi.org/10.1111/j.1468-5965.1987.tb00312.x. Yu, H.B., Zhang, L., Hou, W., Wang, Y., Wu, D., Zhang, S.W., Bu, K., 2015. Cumulative effect of land use and cover changes in Naoli River basin in Sanjiang Plain on agricultural nonpoint source pollution load. Sustain. Water Resour. Manag. 1, 355–362. https://doi.org/10.1007/s40899-015-0028-1. Yun, S.L., Ahn, J.H., Min, K.S., Chu, K.H., Um, C.Y., Ko, K.B., 2015. Non-point sources of pollution from cultivated lands in river districts and their contribution to water bodies along the North Han River Basin in Korea. Desalin. Water Treat. 53, 2301–2311. https://doi.org/10.1080/19443994.2014.927124. Zhang, B., Li, W., Xie, G., Xiao, Y., 2010. Water conservation of forest ecosystem in Beijing and its value. Ecol. Econ. 69, 1416–1426. https://doi.org/10.1016/j. ecolecon.2008.09.004. Zhang, B.L., Cui, B.H., Zhang, S.M., Wu, Q.Y., Yao, L., 2018. Source apportionment of nitrogen and phosphorus from non-point source pollution in Nansi Lake Basin, China. Environ. Sci. Pollut. Res. 25, 19101. https://doi.org/10.1007/s11356-0181956-8. Zhang, L.T., Li, Z.B., Wang, S.S., 2016. Spatial scale effect on sediment dynamics in basin-wide floods within a typical agro-watershed: a case study in the hilly loess region of the Chinese Loess Plateau. Sci. Total Environ. 572, 476–486. https://doi. org/10.1016/j.scitotenv.2016.08.082. Zhang, Q., Chen, S., Dong, Y., Liu, D., Yang, X., Yang, Z., 2017. Controllability of phosphorus losses in surface runoff from sloping farmland treated by agricultural practices: P losses from sloping farmland treated by Agro-practices. Land Degrad. Dev. https://doi.org/10.1002/ldr.2702.
Ichter, B.D.R., Summar, Y., 1997. How much water does a river need? Freshw. Biol. 37, 231–249. https://doi.org/10.1046/j.1365-2427.1997.00153.x. Jensco, K.G., McGlynn, B.L., 2011. Hierarchical controls on runoff generation: topographically driven hydrologic connectivity, geology, and vegetation. Water Resour. Res. 47, 431–435. https://doi.org/10.1029/2011WR010666. Kerr, J.M., Depinto, J.V., Mcgrath, D., Sowa, S.P., Swinton, S.M., 2016. Sustainable management of Great Lakes watersheds dominated by agricultural land use. J. Gt. Lakes Res. 42, S038013301630185X https://doi.org/10.1016/j.jglr.2016.10.001. Kuzuha, Y., Tomosugi, K., Kishii, T., Komatsu, Y., Yamashiki, Y., Tsujimura, M., 2010. Coefficient of variation of annual flood peaks: variability of flood peak and rainfall intensity. Hydrol. Process. 23, 546–558. https://doi.org/10.1002/hyp.7184. Larsen, M.A., Christensen, J.H., Drews, M., Butts, M.B., Refsgaard, J.C., 2016. Local control on precipitation in a fully coupled climate-hydrology model. Sci. Rep. 6, 22927. https://doi.org/10.1038/srep22927. Liu, R.M., Zhang, P.P., Wang, X.J., Chen, Y.X., Shen, Z.Y., 2013. Assessment of effects of best management practices on agricultural non-point source pollution in Xiangxi River watershed. Agric. Water Manag. 117, 9–18. https://doi.org/10.1016/j. agwat.2012.10.018. Marin, C.T., Bouten, W., Sevink, J., 2000. Gross rainfall and its partitioning into throughfall, stemflow and evaporation of intercepted water in four forest ecosystems in western Amazonia. J. Hydrol. 237, 40–57. https://doi.org/10.1016/s0022-1694 (00)00301-2. Mcnamara, J.P., Kane, D.L., Hinzman, L.D., 1998. An analysis of streamflow hydrology in the Kuparuk River Basin, Arctic Alaska: a nested watershed approach. J. Hydrol. 206, 39–57. https://doi.org/10.1016/S0022-1694(98)00083-3. Moraes, J.M.D., Schuler, A.E., Dunne, T., Figueiredo, R.D.O., Victoria, R.L., 2010. Water storage and runoff processes in plinthic soils under forest and pasture in Eastern Amazonia. Hydrol. Process. 20, 2509–2526. https://doi.org/10.1002/hyp.6213. Neris, J., Tejedor, M., Rodríguez, M., Fuentes, J., Jim�enez, C., 2013. Effect of forest floor characteristics on water repellency, infiltration, runoff and soil loss in Andisols of Tenerife (Canary Islands, Spain). Catena 108, 50–57. https://doi.org/10.1016/j. catena.2012.04.011. Ouyang, W., Yang, W.X., Mats, T., Xu, Y.X., Lin, C.Y., Gao, X., Hao, Z.C., 2018. Using river sediments to analyze the driving force difference for non-point source pollution dynamics between two scales of watersheds. Water Res. 139, 311–320. https://doi. org/10.1016/j.watres.2018.04.020. Qiu, L.J., WU, Y.P., Wang, L.J., Lei, X.H., Liao, W.H., Hui, Y., Meng, X.Y., 2017. Spatiotemporal response of the water cycle to land use conversions in a typical hilly–gully basin on the Loess Plateau, China. Hydrol. Earth Syst. Sci. 21, 1–30. https://doi.org/10.5194/hess-2017-343. Richter, B.D., Baumgartner, J.V., Powell, J., Braun, D.P., 2010. A method for assessing hydrologic alteration within ecosystems. Conserv. Biol. 10, 1163–1174. https://doi. org/10.2307/2387152. Shangguan, Z.P., Zheng, S.X., 2006. Ecological properties of soil water and effects on forest vegetation in the Loess Plateau. Int. J. Sustain. Dev. World Ecol. 13, 307–314. https://doi.org/10.1080/13504500609469682. Shen, Z.Y., Chen, L., Hong, Q., Qiu, J.l., Xie, H., Liu, R.M., 2013. Assessment of nitrogen and phosphorus loads and causal factors from different land use and soil types in the Three Gorges Reservoir Area. Sci. Total Environ. 454–455, 383–392. https://doi. org/10.1016/j.scitotenv.2013.03.036. Shen, Z.Y., Qiu, J.l., Hong, Q., Chen, L., 2014. Simulation of spatial and temporal distributions of non-point source pollution load in the Three Gorges Reservoir Region. Sci. Total Environ. 493, 138–146. https://doi.org/10.1016/j. scitotenv.2014.05.109. Souza-Filho, P.W.M., Souza, E.B.D., Júnior, R.O.S., Jr, W.R.N., Mendonça, B.R.V.D., Guimar~ aes, J.T.F., Dall’Agnol, R., Siqueira, J.O., 2016. Four decades of land-cover,
10