Environ Sci Pollut Res DOI 10.1007/s11356-017-9733-7
RESEARCH ARTICLE
Spatial scale and seasonal dependence of land use impacts on riverine water quality in the Huai River basin, China Jianfeng Liu 1,2 & Xiang Zhang 1,2 & Bi Wu 1,2 & Guoyan Pan 1,2 & Jing Xu 1,2 & Shaofei Wu 1,2
Received: 25 March 2017 / Accepted: 6 July 2017 # Springer-Verlag GmbH Germany 2017
Abstract Land use pattern is an effective reflection of anthropic activities, which are primarily responsible for water quality deterioration. A detailed understanding of relationship between water quality and land use is critical for effective land use management to improve water quality. Linear mixed effects and multiple regression models were applied to water quality data collected from 2003 to 2010 from 36 stations in the Huai River basin together with topography and climate data, to characterize the land use impacts on water quality and their spatial scale and seasonal dependence. The results indicated that the influence of land use categories on specific water quality parameter was multiple and varied with spatial scales and seasons. Land use exhibited strongest association with dissolved oxygen (DO) and ammonia nitrogen (NH3-N) concentrations at entire watershed scale and with total phosphorus (TP) and fluoride concentrations at finer scales. However, the spatial scale, at which land use exerted strongest influence on instream chemical oxygen demand (COD) and biochemical oxygen demand (BOD) levels, varied with seasons. In addition, land use composition was responsible for the seasonal pattern observed in contaminant concentrations. COD, NH3-N, and fluoride generally peaked during dry Responsible editor: Kenneth Mei Yee Leung Electronic supplementary material The online version of this article (doi:10.1007/s11356-017-9733-7) contains supplementary material, which is available to authorized users. * Xiang Zhang
[email protected]
1
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China
2
Hubei Provincial Collaborative Innovation Center for Water Resources Security, Wuhan University, Wuhan 430072, China
seasons in highly urbanized regions and during rainy seasons in less urbanized regions. High proportion of agricultural and rural areas was associated with high nutrient contamination risk during spring. The results highlight the spatial scale and seasonal dependence of land use impacts on water quality and can provide scientific basis for scale-specific land management and seasonal contamination control. Keywords Land use . Water quality . Spatial scale . Season . Mixed effect model . Relationship
Introduction In general, river water serves as the principal water supplier for humans and offers a wide range of aquatic habitats. Watershed characteristics such as land use, climate, and physiographic features can greatly affect the riverine water quality (Meynendonckx et al. 2006; Teixeira and Marques 2015). Specifically, water quality degradation often occurs because of land development causing increase in flash runoff and nutrient loading (Sangani et al. 2015). In the strategic action plans developed by the US National Ocean Council, Water Quality and Sustainable Practices on Land has been listed as one of the nine priority objectives to enhance surface water quality via promoting and implementing sustainable practices on land (US National Ocean Council 2013). To predict pollution potential and make effective watershed management plans, a detailed understanding of the linkage between land use and water quality is critical. For this purpose, the following two key aspects of information are necessary: (1) effects of land use categories on water quality variables and (2)
Environ Sci Pollut Res
the spatial scales at which the strongest linkages exist (Uriarte et al. 2011). Land use itself does not have any influence on river water quality. However, land use categories can reflect the spatial variation of anthropogenic activities, which are responsible for the generation and transport of terrestrial contaminants. Numerous studies have described the influence of different land uses on river water quality (Amiri et al. 2012; Dosskey et al. 2010; Taka et al. 2016). For example, urban expansion generally increases instream organic pollutants, nutrients, and heavy metals (Ahearn et al. 2005; Tran et al. 2010), while agricultural land use is usually associated with high level of nutrients, especially nitrogen and phosphorus (Kröger et al. 2007; Taranu and Gregory-Eaves 2008). In addition, the arbitrary emission of decentralized domestic and livestock wastewater can cause serious river water quality deterioration in many rural areas as a result of inadequate sewage treatment facilities (Liang et al. 2010). Furthermore, vegetated land is characterized by low human disturbance and can play an important role in mitigating soil erosion and filtering contaminants from overland flow (Haidary et al. 2013). Instream water is subject to accumulative disturbance depending on the land use characteristics at different distance to rivers. Thus, the spatial scale at which land use exerts strongest influence on water quality has been a research hotspot for developing cost-effective land management practices (Maillard and Santos 2008; Zhou et al. 2012; Huang et al. 2016). However, mixed results regarding the scale dependence of land use impacts have been reported in previous studies. For instance, some studies suggested that land use within the entire watershed is appropriate for predicting water quality since nutrient loading and retention occur at a larger spatial extent (Dodds and Oakes 2008; Meynendonckx et al. 2006), while others emphasized the role of riparian land use in regulating nutrients and sediments (Connolly et al. 2015; Tran et al. 2010) for that pollutant accumulation in receiving water depends on the proximity of the pollutant sources to watercourse (Dosskey et al. 2010). The impacts of land use on water quality can vary seasonally due to monsoon climate and irregular agricultural activities (Kröger et al. 2007; Wilkes et al. 2009). Specifically, land use can be used to explain the seasonal variation in water quality to some extent. In a highly disturbed river in China, Liu et al. (2016) found that land use composition differs in drainage sites with different seasonal patterns of water quality. Land use plays an important role in pollutant yield, while the transport of pollutants into receiving waters can be affected by non-land use factors, especially climate and topography features (Delpla and Rodriguez 2014; Hurley and Mazumder 2013). Rainfall runoff can serve as the carrier of terrestrial pollutants and influence surface water quality through flushing and dilution (Park et al. 2011). In Puerto Rico, which has a distinct precipitation gradient, Uriarte et al. (2011) established
regression models to explore the influence of land use on water quality and found that it was modulated by hydrological regime. Topography is also an underlying influential factor to water quality. Yu et al. (2016) indicated that instream water quality is more susceptible to land use in steeper areas than in flatter areas within Wei River basin, China. In addition, due to the hierarchical characteristic of river system and spatial heterogeneity of natural characteristics, geographically adjacent sites in the same reach typically exhibit similar contamination levels (Chang 2008). Thus, the mediating effects of non-land use factors and spatial correlation should be considered when analyzing the land use and water quality relationship. The Huai River basin is located in one of the most densely populated regions in China, and surface flow plays a critical role in watershed eco-environment and human water consumptions. However, water security within the basin has been threatened by increasing land development in the headwater regions. Understanding the linkage between land use and water quality in the Huai River basin can provide scientific reference for watershed water pollution prevention and land use management. The main aims of this research were (1) to investigate the influence of land use categories on river water quality, (2) to develop the relationship between land use and river water quality in the context of varying seasons and spatial scales, and (3) to evaluate whether land use is responsible for seasonal pattern in river water quality.
Materials and methods Study area Huai River, which is one of the seven key rivers in China, is located in the eastern China and serves as the traditional border between the north and south China in terms of climate and vegetation distribution (Fig. 1). It flows across Hubei, Henan, Anhui, Shandong, and Jiangsu provinces of China and has a drainage area of 270,000 km2. Annual mean temperature is 11–16 °C, and the mean annual precipitation is 920 mm, more than 60% of which occurs from June to September, leading to a cycle of wet and dry seasons. The Huai River basin has many water projects, and the total storage capacity of dams and sluices accounts for approximately 51% of the annual runoff (Zhang et al. 2010). The terrain of Huai River basin generally descends from west to east with the landform transferring from mountains to plain land. As the most densely inhabited area and key breadbasket of China, Huai River basin has a population density of 614 persons/km2 and primarily consists of agricultural and urban areas that are distributed throughout the basin, whereas forested area is located in the western headwater regions. The gradual increase in anthropogenic activities has resulted in
Environ Sci Pollut Res Fig. 1 Study area and location of sampling sites. Labels represent the site ID
serious deterioration of water environment, for instance the occurrence of five serious water pollution incidents in 1989, 1994, 1995, 2000, and 2004, which caused disastrous impact on aquatic ecology and environment (Zhang et al. 2015). The Huai River Pollution Control Project has been successful in improving sewage treatment rate and reducing pollutant quantity inlets into river from its inception in 1994. Nevertheless, diffuse pollution remains serious and even shows an increasing trend. In this study, the upstream region of the Huai River basin was selected as the study area due to its predominant contribution to watershed hydrology and ecoenvironmental conditions. Sampling sites and water quality data Monthly water chemistry data for the period of 2003–2010 were collected at 36 monitoring stations (Fig. 1 and Table 1), with 6 stations at the mainstream of Huai River, 7 stations at the Guo River, 5 stations at the Ying River, 6 stations at the Sha
River, 8 stations at the Ru River, and 4 stations at other tributaries. According to the Environmental Quality Standards for Surface Water (GB3838-2002) in China, six water quality parameters, viz., dissolved oxygen (DO), chemical oxygen demand (COD), biochemical oxygen demand (BOD), ammonia nitrogen (NH3-N), total phosphorus (TP), and fluoride, were selected for analysis because they were identified as the primary pollution indicators monitored routinely and could reflect the main concerns of water quality in the Huai River basin (Dou et al. 2016; Shi et al. 2016). DO is a measure of instream aeration and photosynthetic activity and strongly influenced by instream oxygen demanding substances, including algal biomass, dissolved organic matters, and volatilesuspended solids (Sánchez et al. 2007). BOD and COD are used as the measure of organic pollution level, while NH3-N and TP can reflect the nutrient inputs from draining watershed (Uriarte et al. 2011). Fluoride is an important indicator of drinking water quality and also used as a chemical tracer of sewage inputs into rivers (Kaushal et al. 2011). All the parameters were measured following the Environmental Quality
Environ Sci Pollut Res Table 1
Main characteristics of drainage area and water quality for selected sampling sites TP (mg/L) Fluoride (mg/L) (mean ± SD) (mean ± SD)
Number of observations
44.35 ± 25.67 21.75 ± 16.44 15.35 ± 13.56
0.29 ± 0.89
1.42 ± 0.35
54
33.55 ± 11.44
5.60 ± 3.89
0.13 ± 0.24
1.05 ± 0.24
96
39.20 ± 52.36 11.75 ± 16.99 31.10 ± 38.61 3.70 ± 8.99
1.22 ± 4.59 1.00 ± 5.58
0.30 ± 0.86 0.17 ± 0.44
1.12 ± 0.61 1.00 ± 0.34
54 96
1.09 7.50 ± 1.79
33.65 ± 17.93
3.80 ± 4.15
0.42 ± 1.51
0.15 ± 0.21
1.10 ± 0.45
96
0.99 5.21 ± 2.77 1.01 7.10 ± 1.42
45.65 ± 51.22 24.90 ± 12.94
5.65 ± 3.90 4.00 ± 3.02
6.00 ± 5.27 2.40 ± 4.79
0.23 ± 0.34 0.10 ± 0.14
1.00 ± 0.37 1.09 ± 0.28
96 96
859 1764
14.75 6.20 ± 2.05 12.72 7.85 ± 1.63
33.35 ± 35.67 20.00 ± 7.65
9.40 ± 24.59 2.20 ± 1.24
2.24 ± 4.28 0.15 ± 0.17
0.24 ± 0.39 0.06 ± 0.06
0.85 ± 0.16 0.47 ± 0.17
66 70
Y3
2112
10.83 7.15 ± 2.84
28.00 ± 10.65
2.65 ± 1.66
0.25 ± 0.38
0.09 ± 0.05
0.60 ± 0.15
54
Y4
2374
9.72 8.10 ± 2.44
21.40 ± 13.68
2.40 ± 1.38
Y5 S1 S2
3215 1233 5220
7.26 6.00 ± 2.52 31.11 9.05 ± 1.66 13.35 8.05 ± 1.41
37.40 ± 59.36 13.75 ± 31.07 11.15 ± 1.15 0.70 ± 0.44 16.00 ± 3.90 1.20 ± 0.58
S3
2746
14.65 7.65 ± 1.81
12.70 ± 4.27
S4 S5 S6 R1
9632 11.72 7.90 ± 1.95 1281 5.86 7.40 ± 0.64 12,599 10.04 7.95 ± 2.11 1335 5.05 6.10 ± 1.96
40.60 ± 30.39 15.90 ± 3.53 31.00 ± 24.71 36.50 ± 57.83
R2 R3 R4
3784 2002 1836
2.53 6.00 ± 1.52 9.45 8.00 ± 1.40 7.44 7.50 ± 1.61
23.80 ± 12.69 11.00 ± 3.03 15.20 ± 7.75
R5 R6 R7
1969 1020 7458
7.01 7.05 ± 1.40 1.58 5.40 ± 2.28 3.80 5.25 ± 2.46
Slope DO (mg/L) COD (mg/L) (‰) (mean ± SD) (mean ± SD)
River
Site Area ID (km2)
Guo River
G1
1455
0.87 3.00 ± 1.44
G2
4180
0.96 7.60 ± 2.11
G3 G4
1778 3969
0.86 5.85 ± 2.65 0.92 7.80 ± 2.83
G5
1939
G6 G7
10,756 13,817
Y1 Y2
Ying River
Sha River
Ru River
BOD (mg/L) (mean ± SD)
3.25 ± 1.99
NH3-N (mg/L) (mean ± SD)
0.17 ± 0.24
0.05 ± 0.05
0.56 ± 0.16
54
1.95 ± 2.77 0.11 ± 0.16 0.14 ± 0.31
0.34 ± 0.51 0.03 ± 0.10 0.05 ± 0.04
1.01 ± 0.23 1.07 ± 0.32 0.51 ± 0.18
54 70 54
1.55 ± 1.30
0.11 ± 0.07
0.03 ± 0.33
0.49 ± 0.14
60
2.45 ± 1.45 2.75 ± 0.77 2.90 ± 1.29 8.60 ± 14.77
0.82 ± 1.01 0.18 ± 0.06 0.45 ± 1.02 1.62 ± 1.94
0.14 ± 0.15 0.05 ± 0.23 0.11 ± 0.07 0.19 ± 0.13
0.58 ± 0.15 0.42 ± 0.06 0.56 ± 0.15 0.60 ± 0.18
54 54 54 66
4.50 ± 4.26 2.35 ± 1.53 2.30 ± 1.90
1.00 ± 0.67 0.30 ± 0.18 0.65 ± 1.23
0.32 ± 0.21 0.03 ± 0.06 0.08 ± 0.38
0.62 ± 0.10 0.45 ± 0.15 0.48 ± 0.29
54 71 69
15.85 ± 9.32 2.50 ± 1.84 42.40 ± 43.66 11.20 ± 16.04 29.20 ± 28.70 6.10 ± 11.38
0.49 ± 0.93 3.17 ± 5.36 1.53 ± 10.08
0.08 ± 0.24 0.55 ± 0.71 0.43 ± 0.65
0.54 ± 0.26 0.52 ± 0.16 0.53 ± 0.14
54 70 72
R8 Main stream H1 H2 H3 H4
11,424 3.33 6.60 ± 1.51 753 12.97 8.65 ± 2.06 3152 9.32 7.85 ± 1.75 1593 15.39 6.05 ± 1.64 1433 12.00 8.40 ± 1.40
19.65 ± 27.32 18.50 ± 7.70 15.15 ± 8.72 17.85 ± 5.42 14.05 ± 5.20
2.75 ± 6.22 1.90 ± 2.08 2.00 ± 1.37 3.00 ± 2.91 1.20 ± 1.23
0.60 ± 0.64 0.35 ± 0.65 0.42 ± 0.68 1.99 ± 2.82 0.28 ± 0.18
0.20 ± 0.12 0.43 ± 1.07 0.10 ± 0.10 0.22 ± 0.28 0.06 ± 0.16
0.50 ± 0.12 0.80 ± 2.01 0.50 ± 0.23 0.36 ± 0.14 0.29 ± 0.13
96 54 71 54 54
H5 H6 Other 1 tributaries 2 3 4
1348 15,687 3898 1976 1523 2210
14.15 ± 4.78 14.65 ± 11.52 14.80 ± 4.68 7.85 ± 2.85 7.75 ± 2.42 10.75 ± 5.99
1.70 ± 0.84 2.70 ± 1.59 1.90 ± 0.90 0.80 ± 0.58 0.60 ± 0.41 0.90 ± 1.06
0.25 ± 0.11 0.51 ± 0.93 0.47 ± 0.35 0.14 ± 0.10 0.08 ± 0.04 0.17 ± 1.02
0.08 ± 0.17 0.13 ± 0.09 0.10 ± 0.39 0.04 ± 0.03 0.02 ± 0.01 0.05 ± 0.05
0.28 ± 0.13 0.49 ± 0.14 0.37 ± 0.12 0.12 ± 0.03 0.12 ± 0.04 0.12 ± 0.04
60 96 60 65 64 66
13.08 6.72 16.94 23.91 26.26 27.34
7.85 ± 1.76 7.10 ± 1.30 7.30 ± 1.47 7.50 ± 1.02 7.50 ± 0.87 7.60 ± 0.21
Standards for Surface Water (GB3838-2002) in China by the monitoring center of Huai River Basin Water Resources Protection Bureau. To account for the seasonal variations in water quality, the data were grouped into four seasons including winter (December–February), spring (March–May), summer (June–August), and fall (September–November) and were then averaged by season for statistical analysis. The correlation between land use and river water quality was analyzed for four seasons separately.
Meteorological data Daily total rainfall and mean air temperature for the study period were obtained from the meteorological database of the Chinese Meteorological Administration (http://data.cma.cn). Weather stations closest to sampling sites were considered as the representative measures of onsite climate conditions. The distance between sampling sites and weather stations generally ranged between 1.8 and 34.9 km with a mean value of 11.9 km. Mean 3-day
Environ Sci Pollut Res
temperature and cumulative 3-day rainfall prior to the sampling were computed for each sampling sites based on the work of Wilkes et al. (2009), who found that these meteorological factors showed better correlation with riverine microbial activity. Considering the intense flow regulation by dams and floodgates, the response of river discharge to antecedent rainfall might be weak in a short term. Therefore, seasonal cumulative rainfall and mean temperature were also calculated as longer term measures of local hydrological condition.
Land use and topography Land use data extracted from Landsat TM images (30 m × 30 m resolution) in 2005 were obtained from Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (RESDC). Original 25 land use categories were further aggregated into five groups: (1) forested land; (2) agricultural land, including both dry land and paddy field; (3) rural land, including all rural residential areas; (4) urban land, including industrial and residential areas; and (5) water body, including rivers, lakes, and wetlands. The area proportions of land use categories 1–4 were used for statistical analysis. DEM data at a resolution of 30 m obtained from Geospatial Data Cloud (http://www.gscloud.cn/) were used for watershed delineation with each sampling site set as a watershed outlet. Since the focus of the study was on the spatial scale dependence of the water quality and land use relationship, land use was delineated by the Spatial Analyst Tools in ArcGIS software (ESRI, 2008) for seven scales: the entire area upstream of sampling sites, the circle buffers (2, 5, and 10 km) upstream of the sampling sites, and the riparian buffers (200, 500, and 1000 m), which are shown in Fig. 2. The buffers were set based on previous research (Delpla and Rodriguez Fig. 2 Schematic diagram of seven spatial scales including entire upstream watershed (a), upstream circle buffers (b–d), and riparian buffers (e–g)
2014; Maillard and Santos 2008) and by considering the sampling positioning error and resolution of land use data. Both land use and topography data were first extracted at seven scales for all draining sites, and then, land use and topography metrics were calculated by ArcGIS.
Statistical analysis The normality of water quality data was evaluated using onesample Kolmogorov-Smirnov (K-S) test. All water quality parameters except DO were found non-normal and consequently log-transformed to obey the normal distribution. One-way ANOVA test with the Welch’s correlation accounting for unequal variance was performed on processed data to determine whether there was an overall variation in water quality across seasons (Teixeira et al. 2014). In addition, the Games-Howell post hoc test based on Welch’s correction was used to identify the seasons where the variation existed. Considering the hierarchical structure of river system, it was assumed that sites within the same watershed (Table 1) were nested and exhibited spatial correlation, which violated the independence of water quality samples. The linear mixed effect model was used to identify the relationship between land use and water quality. In particular, a spatial correlation structure was incorporated, in which uncorrelated sites were treated independently, while the degree of correlation for nested sites was assumed to vary inversely with the Euclidian distance between them (Dodds and Oakes 2006). Five spatial correlation structures, viz., linear correlation, exponential correlation, Gaussian correlation, rational quadratic correlation, and spherical correlation, were employed for optimization. To account for the differences in baseline contamination level across regions, a flexible intercept term was
Environ Sci Pollut Res
included in the model (Taranu and Gregory-Eaves 2008). The selection of model type and spatial correlation structure were conducted following the methodology outlined by Zuur et al. (2009). Meanwhile, topography and local climate characteristics (Table 2) were used as covariates to control for influential non-land use factors. The stepwise backward selection strategy was adopted to determinate optimal fixed component (Zuur et al. 2009). Finally, constructed models were used to characterize the effects of land use on water quality, while the contribution of non-land use characteristics to water quality was not explored beyond a superficial analysis. Furthermore, multiple linear regression analysis was performed to characterize the strength of relationship between significant land use categories and water quality at seven scales. The regression models were intended to be descriptive with R2 reflecting the strength of relationships due to the violation of independence (Dodds and Oakes 2006). The nlme package in R (Pinheiro et al. 2014) was used for above statistical analysis. To explore the potential determining effect of land use on seasonal water quality variation, sample sites were divided into groups (groups W, SP, SU, and AU) for each water quality variable according to the seasonal pattern, and then, the mean land use composition of each group was statistically calculated (Liu et al. 2016). The seasonal pattern was defined according to the occurrence of peak seasonal contamination level, which is usually the focus of water pollution prevention. Thus, for a specific water quality variable, sample sites with peak seasonal concentration in winter, spring, summer, and autumn were grouped into groups W, SP, SU, and AU, respectively.
Results Description of water quality and watershed characteristics The watershed area encompassed by the study sites varied by several orders of magnitude (753–15,687 km2). The slope was steep in western mountainous regions and got gentler Table 2 Description of watershed characteristics used as predictors of water quality
gradually in the downstream direction. Generally, land use composition varied greatly across watersheds with agricultural land being the dominant land use category. Specifically, the percentage ranges for forested land, agricultural land, rural land, and urban land were 0.13–90.31, 7.40–83.38, 0.11– 18.48, and 0.00–6.31%, respectively (Online Resource 1). Forested land was distributed mostly in western headwater regions, while agricultural land was dispersed over the study area. According to the ANOVA analysis outcomes (Table 3), only DO displayed a significant seasonal cycle (p < 0.01), with generally higher mean concentration in winter and lower mean concentration in summer at investigated sites, which could be ascribed to its susceptibility to seasonal temperature change (Akkoyunlu et al. 2011). Land use impact on spatial water quality variation Mixed effect models were established for each water quality variable, which differed greatly across seasons and spatial scales (Fig. 3). Specifically, significant effects of land use categories on water quality variables were identified and further analyzed (Fig. 4). Across seasons, DO was significantly associated with land use characteristics at most spatial scales except the upstream 10-km scale. As shown in Fig. 4a, forested land had a positive association with DO at watershed scale, while the rural land was negatively associated with DO generally at finer scales. DO was insusceptible to agricultural land at most spatial scales except the negative association between spring DO and agricultural land delineated within upstream 5-km scale. Urban land had a negative influence on DO at both watershed and buffer scales in summer and autumn and only at watershed scale in winter and spring. Land use within the entire watershed could better explain seasonal DO level than that within finer scales (Fig. 4b), which was attributed to the contribution of forested and urban lands. Meanwhile, the association between DO and land use characteristics was found stronger in summer and autumn seasons than in winter and spring seasons.
Variable type
Variable name
Investigated scale
Unit
Topography
Mean slope
Upstream watershed
‰
Circle buffers of 2, 5, and 10 km Land use
Local climate
Forested land Agricultural land Rural land Urban land Antecedent rainfall Antecedent mean air temperature Seasonal accumulative rainfall Seasonal mean air temperature
Riparian buffers of 200, 500, and 1000 m Ditto Ditto Ditto Ditto 3 days for four seasons 3 days for four seasons Season Season
% % % % mm °C mm °C
Environ Sci Pollut Res Table 3 One-way ANOVA with Welch’s correction and pairwise comparisons to assess seasonal difference Variable
F value
Pr (>F)
Pairwise comparison (Games-Howell t test)
DO
15.16
0.00
Winter > spring > summer, winter > autumn > summer
COD
0.51
0.68
–
BOD NH3-N
0.92 1.21
0.43 0.31
– –
TP Fluoride
0.70 0.08
0.55 0.97
– –
The results of pairwise comparison are significant at the 0.05 level
In the mixed effect models, significant association between seasonal COD and land use was observed at all investigated spatial scales. As shown in Fig. 4c, COD was positively associated with urban land within watershed and upstream buffer extents throughout the year, while it was influenced by urban land characterized within riparian buffer zones only in winter and summer. Rural land exhibited positive association with COD over seasons, which existed at all finer scales. In contrast, COD was positively correlated with agricultural land within
Fig. 3 Scatter plots of observed vs. predicted a DO, b COD, c BOD, d TP, e NH3-N, and f fluoride for constructed mixed effect models. All water quality variables except DO were log-transformed. R2 represents
entire watershed in winter and summer, and within both watershed and upstream 5-km extents in spring. In addition, forested land described as a proportion of riparian 200-m buffer area had a significant reducing effect on COD in spring. Across spatial scales, land use characteristics generally provided good predictive power for seasonal COD level (Fig. 4d), and COD variation was better explained at different scales for different seasons. Specifically, the explained variance R 2 of winter and spring COD was the largest at entire watershed scale with a decreasing trend in the direction from entire watershed scale to narrow 1000-m buffer to 200-m buffer. In summer and autumn, land use better explained COD at the 1000-m riparian buffer scale. As shown in Fig. 4e, BOD was negatively associated with forested land within the entire watershed in summer and within the delineated 200-m riparian buffer in winter. The influence of other land use categories on BOD mostly existed in watershed or upstream circle buffers. Specifically, a positive association between BOD and rural land within upstream 2-km scale was observed in all seasons. Agricultural land had a positive association with winter and spring BOD level at watershed, and upstream 10 and 5-km scales. Similarly, urban land was positively correlated
the determination coefficient of mixed effect models. BS1^–BS4^ mean winter, spring, summer, and autumn, respectively
Environ Sci Pollut Res
Fig. 3 (continued)
with BOD at the watershed, and upstream 5 and 2-km scales across seasons. The watershed land use characteristics well explained BOD variability to a certain extent with R2 > 0.5 and R2 > 0.6 in winter and spring, respectively, in comparison to the land use within finer buffers (Fig. 4f). However, in spring and autumn, the predictive power of land use for instream BOD level was generally low (R2 < 0.31) and differed weakly across spatial scales. In spring and summer seasons, instream TP concentration was susceptible to land use within finer extents, suggesting a local TP and land use relationship. As shown in Fig. 4g, agricultural land had a significant positive effect on summer TP level at most buffer scales. Besides, spring TP was also positively associated with urban land distributed in the narrow 200-m riparian zone. Nevertheless, land use alone explained only a small fraction of variance in spring and summer TP with the R2 less than 0.20 (Fig. 4h). Land use quantified at the riparian buffer scales had a better predictive power toward summer TP than land use within upstream buffer scales. Across all scales, there was no significant relationship between TP and land use identified for winter and autumn season. Within upstream watershed extent, NH3-N was negatively correlated with forested and agricultural land during winter, while positively correlated with rural and urban land in other seasons (Fig. 4i). In addition, forested land
within riparian 500 m zone also had a negative effect on spring NH3-N. Across seasons, NH3-N was positively associated with rural land within riparian buffer zones, while positive correlation of NH3-N with urban land was observed at upstream 2 and 5-km scales. Throughout the whole year, variation in NH3-N was best explained by land use quantified within the entire watershed with relatively higher R2 > 0.40 (Fig. 4j). Interestingly, the explained variance of NH3-N by land use decreased gradually as the spatial scale narrowed from watershed to riparian 1000 m and further 200-m scales. In the mixed effect models, the association between fluoride and land use varied significantly across spatial scales (Fig. 4k). Forested land within the quantified 500 and 200-m riparian zones had a significant negative association with summer and autumn fluoride. There was no significant fluoride and agricultural land relationship observed at all investigated scales. Rural land was negatively correlated with fluoride in summer at watershed scale, while in winter at upstream 2-km scale. Besides, summer fluoride was positively influenced by urban land characterized at upstream 2-km scale, while spring and autumn fluoride was closely related to urban land within riparian 200-m zone. Overall, land use delineated within riparian buffer extents provided better explanatory power toward seasonal fluoride level in summer and autumn seasons (Fig. 4l).
Environ Sci Pollut Res
a
b
DO
c
d
COD
e
f
BOD
g
h
TP
Fig. 4 Significant coefficients (significant level = 0.05) in the mixed effect models describing the associations between land uses and water quality (a, c, e, g, i, k) and the adjusted R2 of multiple regression models
describing the strength of associations between land use and water quality across spatial scales and seasons (b, d, f, h, j, l)
Environ Sci Pollut Res
i
j
NH3-N
k
l
Fluoride
Fig. 4 (continued)
Land use impact on seasonal water quality variation The absence of overall significant seasonal variation in water quality variables except DO by ANOVA does not suggest that these water quality variables are independent of season, but instead that the seasonal patterns of them are not consistent across investigated sites. It has been well documented that the effect of seasonal change on instream contamination level differs in different underlying surface conditions (Park et al. 2011). Consequently, the potential determining effect of land use on seasonal water quality variation was investigated. As discussed in BLand use impact on spatial water quality variation^ section, the spatial scale at which strongest land use impacts existed differed for water quality variables. Thus, the mean land use compositions of group W, SP, SU, and AU sites were calculated for COD, BOD, and NH3-N at entire watershed scale, while for TP and fluoride at riparian 500 and 1000-m buffer zones, respectively. As shown in Fig. 5, for COD and NH3-N, group W sites generally had higher proportion of urban area, while group SU sites generally had lower proportion of urban area. The sites belonging to group SP were characterized by higher proportion of agricultural and rural areas. However, for BOD, the proportion of urban area was generally higher at sites belonging to group SU, while it was lower at sites belonging to group SP. In particular, the proportion of agricultural area was lower at group W sites. The proportion of urban area varied slightly among groups
for TP, while agricultural and rural areas accounted for higher proportion at group SP and SU sites. For fluoride, the difference in land use composition among groups was unexpected with lower and higher proportion of urban area at group SU and AU sites, respectively. In addition, group SP and SU sties were characterized by lower proportion of agricultural and rural areas.
Discussion Contribution of land use to spatial water quality Previous studies (Huang et al. 2016; Zhou et al. 2015) have concluded that urban and rural land uses are generally associated with significant surface water quality deterioration due to the release of domestic and industrial sewage, livestock wastewater, storm runoff, etc. Furthermore, municipal wastewater is often characterized by high level of fluoride, which could act as tracer to detect the contribution of sewage inputs to river flow (Kaushal et al. 2011). The good explanation of summer and autumn fluoride variation by urban and rural lands suggests the damage of wastewater discharge to instream water quality. However, the variance explained for the winter and spring fluoride level was relatively low, indicating that urban and rural wastewater was less influential to river water quality in winter and spring. In fact, the contribution of urban land to
Environ Sci Pollut Res Fig. 5 Mean land use composition of site groups classified according to seasonal pattern in water quality variables. For each water quality variable, group W, SP, SU, and AU represent draining sites with peak seasonal concentration in winter, spring, summer, and autumn, respectively. Agricultural land, forested land, and water are on the primary axis, and the others are on the secondary axis
instream water quality can be continuous throughout the year, and stronger in dry season because urban land generally acts as a proxy for point source (Bu et al. 2014; Yu et al. 2016). This could be exemplified by the significant positive/negative effects on COD, BOD, NH3-N, and DO observed for urban land across seasons with few exceptions. Ahearn et al. (2005) argued that land cover in urban areas is mostly impervious and drainage is directly routed into rivers as point sources. Therefore, the weak explanation of fluoride by urban land in dry season could be attributed to the contribution of underground water discharge along with the geochemical process (Chen et al. 2012). Agricultural land contributed positively to TP in rainy summer, which agrees with previous findings (Fuchs et al. 2009; Hoorman et al. 2008), suggesting that surface runoff and erosion from cropland were the main source to instream phosphorus during storm period. However, agricultural land also exerted positive influence on COD and BOD in dry winter and spring, which could be attributed to the injection of contaminated subsurface flow from agricultural land (Connolly et al. 2015). In a farmland-dominated watershed in the Netherlands, Rozemeijer and Broers (2007) found shallow groundwater in agricultural areas to be the primary contributor of surface water contamination. However, agricultural land showed significant reducing effect on winter NH3-N level. Since point source contamination is generally more prominent in dry season, while non-point source contamination is typically rainfall
dependent, it appears that instream NH3-N was primarily contributed by urban sewage in winter, with limited contribution from agricultural land. As such, agricultural land was not a source; instead, it acted as a sink of NH3-N during winter. The seasonal variation in the contribution of agricultural land to river NH3-N can be analogous to the spatial variation in the effect of agricultural land on surface water quality across urbanization gradient (Huang et al. 2013; Tu 2011). Forested land has been widely reported to have a prominent role in improving surface water quality (Dosskey et al. 2010), and similar result was found in this study. Forested land exhibited a reducing effect on COD, BOD, NH3-N, and fluoride within buffer scales, which could be attributed to its role in conserving water, mitigating soil erosion, and filtering contaminants from overland flow (Haidary et al. 2013). Specifically, the reducing effect of forested land on COD, BOD, and NH3-N was variable and existed in sporadic seasons, suggesting a change in the seasonal functional mechanism. For instance, Bu et al. (2014) highlighted the fixation and absorption effect of vegetation on contaminants in wet season, while Ding et al. (2015) considered that stronger effect of forested land exists during dry season due to low retention capacity of vegetation in high flow condition. Although the benefit of forested land on water quality was emphasized mostly at riparian buffer zones in the literature (Maillard and Santos 2008), the outcomes of this study suggest a reducing effect on BOD and NH3-N and increasing
Environ Sci Pollut Res
effect on DO within the entire watershed extent. This could be explained by the fact that forested areas are generally characterized by less human disturbances and year-round low contaminant loading (Ding et al. 2015). In addition, streams flowing through highly vegetated areas generally exhibit lower water temperature, which contributes to high DO level (Allan 2004). Buffer zones vs. entire watershed scale Conflicting findings have been reported regarding the dependency of relationship between water quality and land use on spatial scale (Hurley and Mazumder 2013; Zhou et al. 2012). This study found that land use explained DO and NH3-N levels well at the watershed scale. This is in contrast with the findings of Uriarte et al. (2011), who found that land use within finer scale better explains the water quality variance due to its higher heterogeneity. In BResults^ section, urban land was found to have the primary influence on seasonal DO and NH3-N concentration (Fig. 3). Since urban drainage is generally directed to sewage facilities and can be transported into streams over a long distance, the urban nutrient loading and retention tend to occur at larger spatial scales (Dodds and Oakes 2008). In contrast, diffuse contaminants in close proximity are more likely to reach watercourse than those far away from rivers, since the migration of diffuse (mainly from rural and agricultural lands) source contaminants is under the combined interference of surface runoff, infiltration, and subsurface flow processes (Tran et al. 2010). As such, riparian land use could play a predominate role in the migration of rural and agricultural source contaminants, which could be demonstrated by the significant linkage between TP and land use within the riparian zones. The spatial scale with better relationship between land use and water quality could vary with seasonal changes in hydrological regime and contaminant migration pathways. Specifically, riparian land use plays a significant role in governing contaminate transportation and retention during rainy season as demonstrated by the better explanation of riparian land use for summer and autumn COD. In contrast, the better explanation for winter and spring COD exists at the watershed scale, indicating the predominant contribution of urban land in rainless condition. These results are consistent with previous conclusions, that is urban point source pollution is the primary cause for water quality degradation during dry seasons, while pollution from mixed point and non-point sources is responsible for water quality degradation during wet seasons (Yu et al. 2016; Zhou et al. 2015). Therefore, the scale dependency of land use impacts could be interpreted as the relative contribution of point and non-point source contamination to a certain extent. As spatial scale narrowed from entire watershed to riparian 200 m, land use showed non-monotonically varying explanation
power for most quality variables except for NH3-N. Similar finding was reported by Guo et al. (2010), in which the strength of land use impacts was found to fluctuate irregularly within 0– 100% buffer zones. This can be attributed to the uneven spatial distribution of land use, which has been rarely discussed in previous studies (Buck et al. 2004; Hurley and Mazumder 2013). In addition, most water quality variables were found to be less susceptible to land use within upstream circle buffers, suggesting the cumulative contribution of headwater land use. Specifically, headwater streams provide predominant hydrologic contribution to the watershed, and the migration and retention processes of headwater terrestrial contaminants can exert predominant influence on downstream water quality (Dodds and Oakes 2008). In Sha, Ru, and other southern tributaries, less contaminated water from headwater mountainous regions could effectively mitigate the disturbance of downstream land use, while it was the opposite in Guo and Ying tributaries where headwater streams was highly contaminated. Contribution of land use to seasonal contamination risk Land use composition could be responsible for seasonal variation in water quality in addition to determining the spatial distribution of contamination level. The results of this study suggest a distinct land use composition for sites characterized by different seasonal water quality patterns. As mentioned previously, urban land is regarded as the sign of point source load, while non-urban land, especially agricultural land, acted as the primary diffuse load (Bu et al. 2014; Yu et al. 2016). The monsoon climate can influence the instream water quality through rainfall dilution or flushing effect (Park et al. 2011), resulting in seasonally varying contribution of point and nonpoint source contaminants (Ye et al. 2014). Although the separation of dilution and flushing effect from each other is difficult, COD, TP, NH3-N, and fluoride were more vulnerable to diffuse source load by flushing effect in less urbanized areas, where their concentrations peaked during rainy season. On the contrary, sites in highly urbanized areas generally exhibited elevated COD, TP, and NH3-N in rainless winter/autumn as a result of reduced dilution effect. Interestingly, sites with relatively higher proportion of agricultural and rural areas generally had peak COD, BOD, and TP level in spring months, probably as a result of spring agricultural activities. In the study area, agricultural production is characterized by the wheat-maize double cropping system, and fertilization is common in March and April as the main growing period of winter wheat. Due to the scarcity of precipitation, irrigation is necessary and frequent for wheat transpiration during this period (Fang et al. 2010). Since the efficiency of agricultural water by traditional broad irrigation is low, the irrigation return flow via surface and subsurface runoff along with soil nutrient loss can reach adjacent rivers, resulting in elevated nutrient concentrations (Yu et al. 2016).
Environ Sci Pollut Res Table 4 Significant effects of topography and local climate characteristics on water quality variables in the mixed effect models
Water quality variable DO
Spatial scale Watershed 10 km
Slope
Seasonal rainfall
3-day rainfall
− (S1, S2, S4)
− (S2, S4) − (S4) − (S4)
2 km 1000 m
+ (S3)
− (S1)
500 m 200 m
+ (S3) + (S3, S4)
− (S1)
Watershed
− (S4)
− (S4)
+ (S2)
10 km
+ (S2, S4)
5 km
+ (S2, S4) − (S1, S4)
2 km
BOD
TP
NH3-N
Fluoride
3-day temperature
− (S4)
+ (S3, S4)
5 km
COD
Seasonal temperature
1000 m
− (S1)
500 m 200 m
− (S1) − (S1)
Watershed 10 km 5 km
− (S4) − (S3) − (S3)
− (S1) − (S1)
+ (S1, S4) + (S1, S4)
2 km 1000 m
− (S1)
500 m 200 m
− (S1) − (S1)
Watershed 10 km 5 km 2 km 1000 m
− (All)
− (S1)
− (S1, S4)
− (S1)
500 m 200 m Watershed 10 km 5 km
− (S1, S4)
2 km 1000 m 500 m 200 m Watershed 10 km 5 km 2 km 1000 m 500 m 200 m
+ (S2) + (S2) + (S2)
+ (S3)
+ (S1, S2) + (S1, S2)
− (All) − (All) − (All) − (All) − (All) − (All) − (All)
B+^ and B−^ represent positive and negative coefficients (significant level = 0.05), respectively, in the mixed effect models. BS1^–BS4^ mean that the significant coefficients with water quality variables exist in winter, spring, summer, and autumn, respectively. BAll^ means the significant coefficients with water quality variables exist throughout the year
Environ Sci Pollut Res
Furthermore, the seasonal first flushing effect by initial rainfall in pre-flood season is another potential cause (Liu et al. 2016). As initial significant rain events occurs in late spring after a long dry period, substantial pesticides and nutrients on croplands and contaminants accumulated in rural areas can be flushed into watercourse, leading to aggravated deterioration of river water quality (Banks and E. 2006; Kröger et al. 2007; Olsson et al. 2013).
Other potential influential factors Natural variability including topography and climate characteristics could partly account for the unexplained variation in water quality, especially in DO, TP, and fluoride (Table 4). Topography is a crucial factor as it can influence the degree of contact between organic-rich soil and surface flow. For example, land use in a steeper slope can exert stronger influence on river water quality (Yu et al. 2016). In this study, watershed slope acted as the sinks of COD, BOD, and TP. Similar results have also been reported in the studies of Chang (2008) and Meynendonckx et al. (2006), although steeper slope was hypothesized to increase the risk of soil erosion and nutrient loss. Furthermore, fast-moving river flow in steeper areas could be conductive for quick re-aeration of water resulting in higher DO level. Rainfall runoff carrying various oxygen-demanding contaminants can lead to faster oxygen depletion in streams, while higher river flow could effectively reduce the concentration of dissolved ions due to dilution effect (Park et al. 2011). Though some potential influencers like geology and soil were not considered in the analysis due to the lack of available data, their direct or indirect contribution to instream water quality needs further investigation. As a primary sink and source of terrestrial contaminants, soil can affect river water quality through subsurface and soil water, which might provide some insight into micro-scale migration process of contaminates (Hurley and Mazumder 2013; Taka et al. 2016). Geology has been shown to be a primary influencing factor of river chemical composition in the Huai River basin (Zhang et al. 2011). Specifically, geology can ply an important role in the spatial distribution of instream fluoride level (Chen et al. 2012). Furthermore, river water quality is generally influenced by both point and non-point source contaminants. Since point source pollution is generally not proportional to land use area (Zhou et al. 2015), investigating the influence of land use on water quality without considering the point source load might cause great uncertainty in the result. Therefore, data regarding point source load can be particularly informative for future investigations.
Conclusions This study investigated the driving mechanism for water quality evolution in upper Huai River basin from the perspective of land use. The effects of land use categories on water quality are multiple and scale dependent. The study results suggest that DO and NH3-N are more susceptible to land use characterized at watershed scale, while the influence of land use on TP and fluoride is limited to finer extents. Besides, land use impacts on river water quality are also seasonal dependent due to natural process dynamics, for instance, hydrological regime. Specifically, the relative importance of land use within different spatial scales in explaining BOD and COD levels is found varying with seasons. Since the dependency of land use impacts on spatial scale differs for different pollutants, the management of certain pollutants may be more efficient in scale-specific zones. However, the assessment of river water quality generally focuses on multiple water quality variables. Therefore, a comprehensive land conservation and management implemented within entire watershed is necessary to address multiple concerns of water quality. Seasonal variation in the relationship between land use and water quality provides land use the ability to explain seasonal pattern in water quality. Specifically, COD, NH3-N, and fluoride level generally peak during dry seasons in highly urbanized regions, while it is the opposite in less urbanized regions. High proportion of agricultural and rural areas is generally associated with high risk of nutrient contamination during spring. The seasonal dependence of land use impacts can be beneficial for managing the seasonal contamination risk. Acknowledgements This research was supported by the Natural Science Foundation of China (No. 51409191, No. 51279143), the Major Science and Technology Program for Water Pollution Control and Treatment of China (No. 2014ZX07204-006), and the National K ey R e se ar ch a n d D e v e l o p m e n t Pr o g r a m o f Ch i n a ( N o. 2016YFC0401301). We would like to thank the Huai River Basin Water Resources Protection Bureau, Chinese Meteorological Administration, and Chinese Academy of Sciences for access to data. Compliance with ethical standards Conflict of interest The authors declare that they have no conflict of interest.
References Ahearn DS, Sheibley RW, Dahlgren RA, Anderson M, Johnson J, Tate KW (2005) Land use and land cover influence on water quality in the last free-flowing river draining the western Sierra Nevada, California. J Hydrol 313:234–247 Akkoyunlu A, Altun H, Cigizoglu HK (2011) Depth-integrated estimation of dissolved oxygen in a lake. J Environ Eng 137:961–967 Allan JD (2004) Landscapes and riverscapes: the influence of land use on stream ecosystems. Annu Rev Ecol Evol S 35:257–284
Environ Sci Pollut Res Amiri BJ, Sudheer KP, Fohrer N (2012) Linkage between in-stream total phosphorus and land cover in Chugoku District, Japan: an ANN approach. J of Hydrol Hydromech 60:33–44 Banks PFH, E K (2006) Compositions of first flush and composite storm water runoff in small urban and rural watersheds, north-central Texas. Urban Water J 3:43–49 Bu H, Wei M, Yuan Z, Wan J (2014) Relationships between land use patterns and water quality in the Taizi River basin, China. Ecol Indic 41:187–197 Buck O, Niyogi DK, Townsend CR (2004) Scale-dependence of land use effects on water quality of streams in agricultural catchments. Environ Pollut 130:287–299 Chang H (2008) Spatial analysis of water quality trends in the Han River basin, South Korea. Water Res 42:3285–3304 Chen H, Mei Y, Yang X, Zhong C, Wang G, Schmidt-Vogt D, Xu Y, Xu J (2012) Spatial distribution and temporal variation of high fluoride contents in groundwater and prevalence of fluorosis in humans in Yuanmou County, Southwest China. J Hazard Mater 235-236:201– 209 Connolly NM, Pearson RG, Loong D, Maughan M, Brodie J (2015) Water quality variation along streams with similar agricultural development but contrasting riparian vegetation. Agric Ecosyst Environ 213:11–20 Delpla I, Rodriguez MJ (2014) Effects of future climate and land use scenarios on riverine source water quality. Sci Total Environ 493: 1014–1024 Ding J, Jiang Y, Fu L, Liu Q, Peng Q, Kang M (2015) Impacts of land use on surface water quality in a subtropical river basin: a case study of the Dongjiang River basin, Southeastern China. Water 7:4427–4445 Dodds WK, Oakes RM (2006) Controls on nutrients across a prairie stream watershed: land use and riparian cover effects. Environ Manag 37:634–646 Dodds WK, Oakes RM (2008) Headwater influences on downstream water quality. Environ Manag 41:367–377 Dosskey MG, Vidon P, Gurwick NP, Allan CJ, Duval TP, Lowrance R (2010) The role of riparian vegetation in protecting and improving chemical water quality in streams. J Am Water Resour As 46:261– 277 Dou M, Zhang Y, Li G (2016) Temporal and spatial characteristics of the water pollutant concentration in Huaihe River basin from 2003 to 2012, China. Environ Monit Assess 188:522 ESRI (2008) ArcGIS 9.3. Environmental Systems Research Institute, Redlands Fang QX, Ma L, Green TR, Yu Q, Wang TD, Ahuja LR, Green TR, Yu QA, Ma LW, Wang TD (2010) Water resources and water use efficiency in the North China Plain: current status and agronomic management options. Agric Water Manag 97:1102–1116 Fuchs JW, Fox GA, Storm DE, Penn CJ, Brown GO (2009) Subsurface transport of phosphorus in riparian floodplains: influence of preferential flow paths. J Environ Qual 38:473–484 GB3838–2002 (2002) The state environmental protection standards of the People’s Republic of China: environmental quality standards for surface water quality [in Chinese] Guo QH, Ma KM, Yang L, He K (2010) Testing a dynamic complex hypothesis in the analysis of land use impact on lake water quality. Water Resour Manag 24:1313–1332 Haidary A, Amiri BJ, Adamowski J, Fohrer N, Nakane K (2013) Assessing the impacts of four land use types on the water quality of wetlands in Japan. Water Resour Manag 27:2217–2229 Hoorman J, Hone T, Sudman T, Dirksen T, Iles J, Islam KR (2008) Agricultural impacts on lake and stream water quality in grand lake St. Marys, Western Ohio. Water, Air, Soil Pollut 193:309–322 Huang J, Li Q, Pontius RG, Klemas V, Hong H (2013) Detecting the dynamic linkage between landscape characteristics and water quality in a subtropical coastal watershed, Southeast China. Environ Manag 51:32–44
Huang Z, Han L, Zeng L, Xiao W, Tian Y (2016) Effects of land use patterns on stream water quality: a case study of a small-scale watershed in the Three Gorges Reservoir Area, China. Environ Sci and Pollut Res 23:3943–3955 Hurley T, Mazumder A (2013) Spatial scale of land-use impacts on riverine drinking source water quality. Water Resour Res 49:1591– 1601 Kaushal SS, Groffman PM, Band LE, Elliott EM, Shields CA, Kendall C (2011) Tracking nonpoint source nitrogen pollution in humanimpacted watersheds. Environ Sci Technol 45:8225–8232 Kröger R, Holland MM, Moore MT, Cooper CM (2007) Hydrological variability and agricultural drainage ditch inorganic nitrogen reduction capacity. J Environ Qual 36:1646–1652 Liang H, Gao M, Liu J, Wei Y, Guo X (2010) A novel integrated step-feed biofilm process for thet reatment of decentralized domestic wastewater in rural areas of China. J Environ Sci 22:321–327 Liu J, Zhang X, Xia J, Wu S, She D, Zou L (2016) Characterizing and explaining spatio-temporal variation of water quality in a highly disturbed river by multi-statistical techniques. SpringerPlus 5:1171 Maillard P, Santos NAP (2008) A spatial-statistical approach for modeling the effect of non-point source pollution on different water quality parameters in the Velhas river watershed–Brazil. J Environ Manag 86:158–170 Meynendonckx J, Heuvelmans G, Muys B, Feyen J (2006) Effects of watershed and riparian zone characteristics on nutrient concentrations in the river Scheldt Basin. Hydrol Earth Syst Sc 10:913–922 Olsson O, Khodorkovsky M, Gassmann M, Friedler E, Schneider M, Dubowski Y (2013) Fate of pesticides and their transformation products: first flush effects in a semi-arid catchment. CLEAN - Soil Air Water 41:134–142 Park JH, Inam E, Abdullah MH, Agustiyani D, Lei D, Hoang TT, Kim KW, Sang DK, Nguyen MH, Pekthong T (2011) Implications of rainfall variability for seasonality and climate-induced risks concerning surface water quality in East Asia. J Hydrol 400:323– 332 Pinheiro J, Bates D, DebRoy S, Sarkar D (2014) R Core team. Nlme: linear and nonlinear mixed effects models. R package version 3.1117. R Foundation for Statistical Computing, Vienna Rozemeijer JC, Broers HP (2007) The groundwater contribution to surface water contamination in a region with intensive agricultural land use (Noord-Brabant, The Netherlands). Environ Pollut 148:695–706 Sánchez E, Colmenarejo MF, Vicente J, Rubio A, García MG, Travieso L, Borja R (2007) Use of the water quality index and dissolved oxygen deficit as simple indicators of watersheds pollution. Ecol Indic 7:315–328 Sangani MH, Amiri BJ, Shabani AA, Sakieh Y, Ashrafi S (2015) Modeling relationships between catchment attributes and river water quality in southern catchments of the Caspian Sea. Environ Sci Pollut Res 22:4985–5002 Shi W, Xia J, Zhang X (2016) Influences of anthropogenic activities and topography on water quality in the highly regulated Huai River basin, China. Environ Sci Pollut Res 23:21460–21474 Taka M, Aalto J, Virkanen J, Luoto M (2016) The direct and indirect effects of watershed land use and soil type on stream water metal concentrations. Water Resour Res 52:7711–7725 Taranu ZE, Gregory-Eaves I (2008) Quantifying relationships among phosphorus, agriculture, and lake depth at an inter-regional scale. Ecosystems 11:715–725 Teixeira Z, Teixeira H, Marques JC (2014) Systematic processes of land use/land cover change to identify relevant driving forces: implications on water quality. Sci Total Environ 470:1320–1335 Teixeira Z, Marques JC (2015) Relating landscape to stream nitrate-N levels in a coastal eastern-Atlantic watershed (Portugal). Ecol Indic 61:693–706
Environ Sci Pollut Res Tran CP, Bode RW, Smith AJ, Kleppel GS (2010) Land-use proximity as a basis for assessing stream water quality in New York state (USA). Ecol Indic 10:727–733 Tu J (2011) Spatially varying relationships between land use and water quality across an urbanization gradient explored by geographically weighted regression. Appl Geogr 31:376–392 Uriarte M, Yackulic CB, Lim Y, Arce-Nazario JA (2011) Influence of land use on water quality in a tropical landscape: a multi-scale analysis. Landsc Ecol 26:1151–1164 US National Ocean Council (2013) National ocean policy implementation plan https://obamawhitehouse.archives.gov/sites/default/files/ national_ocean_policy_implementation_plan.pdf. Accessed 16 Apr 2013 Wilkes G, Edge T, Gannon V, Jokinen C, Lyautey E, Medeiros D, Neumann N, Ruecker N, Topp E, Lapen DR (2009) Seasonal relationships among indicator bacteria, pathogenic bacteria, cryptosporidium oocysts, giardia cysts, and hydrological indices for surface waters within an agricultural landscape. Water Res 43:2209–2223 Ye Y, He X, Chen W, Yao J, Yu S, Jia L (2014) Seasonal water quality upstream of Dahuofang reservoir, China – the effects of land use type at various spatial scales. CLEAN – Soil Air Water 42:1423–1432
Yu S, Xu Z, Wu W, Zuo D (2016) Effect of land use types on stream water quality under seasonal variation and topographic characteristics in the Wei River basin, China. Ecol Indic 60:202–212 Zhang L, Song X, Xia J, Yuan R, Zhang Y, Liu X, Han D (2011) Major element chemistry of the Huai River basin, China. Appl Geochem 26:293–300 Zhang Y, Xia J, Liang T, Shao Q (2010) Impact of water projects on river flow regimes and water quality in Huai River basin. Water Resour Manag 24:889–908 Zhang Y, She D, Xia J, Zhai X, Zuo Q (2015) Causal analysis on the specified paroxysmal water pollution incidents in the Huai River basin, China. Environ Eng Manag J 14:139–151 Zhou P, Huang J, Jr PR, Hong H (2015) New insight into the correlations between land use and water quality in a coastal watershed of China: does point source pollution weaken it? Sci Total Environ 543:591– 600 Zhou T, Wu J, Peng S (2012) Assessing the effects of landscape pattern on river water quality at multiple scales: a case study of the Dongjiang River watershed, China. Ecol Indic 23:166–175 Zuur B, Ieno E, NeilWalker SA, Smith G (2009) Mixed effects models and extensions in ecology with R. Springer, New York