Effects of land use on water quality at different spatial scales in the middle reaches of Huaihe River

Abstract The impact of land use on water quality is spatially scale-dependent. Exploring the relationship between land use pattern and river water quality provides an important basis for water quality safety and land planning. This study explores the effect of land use patterns on surface water quality using land use data and water quality data of 16 sampling points in the middle reach of the Huai River. It investigates the water quality status of the middle reach of the Huai River and the relationship between land use patterns and water quality indices at 100, 500, and 1000 m and watershed scales utilising Spearman correlation, redundancy analysis (RDA), and multiple linear regression analysis. Multiple linear regression analysis showed that the water pollutants in the middle reach of the Huai River are mainly total nitrogen (TN) and total phosphorus (TP). Redundancy analysis (RDA) indicated that the impact of land use patterns on water quality is significant on watershed scales. Spearman correlation indicated that water quality and land use showed significant correlation differences, the building land was positively correlated with TP, TN and NH3-N, the cultivated land was positively correlated with TN, and the grassland was negatively correlated with TP and NH3-N. This studies have revealed that building land acts as a 'source’ of pollutants, grassland purifies the load of TP and NH3-N, and forest land has the effect of interception and deposition of pollutants. The results can provide important information for land use planning and water quality protection measures at multiple spatial scales.


Introduction
River water quality is a critical factor that cannot be ignored in comprehensive regional development, and water quality safety is the fundamental guarantee for human health and life safety (Layhee et al. 2015;Khan et al. 2021;Scanlon et al. 2022).With the rapid and healthy growth of China's economy, the population density rises, and the water quality of rivers is significantly affected by human activities, increasing various pollutants (Schmidt et al. 2019).As the main carrier of pollutants, land use forms non-point source pollution by changing the hydrological process and nutrient migration of the basin, which directly leads to the degradation of river water quality.Therefore, it is of great practical significance to analyze the relationship between land use and water quality for water quality management, land use planning and ecological environment restoration (Yang et al. 2022).
Land use is a comprehensive reflection of human activities, and its pattern is a crucial aspect affecting surface water quality (Chen et al. 2019).The relationship between regional land use patterns and water quality has been a hot topic in hydrological research.However, the analysis of land use is usually based on a single scale.In recent years, domestic and foreign scholars have analyzed the relationship between river water quality and land use based on different spatial scales.Shi et al. (2017) conducted a preliminary study on the relationship between land use and water quality.Ou et al. (2016) and Shen et al. (2020) reported that land use is the root cause of pollution sources.Studies have demonstrated that reducing pollution sources is the fundamental way of curbing water quality degradation (Ou et al. 2016).described that land use in the buffer zone has a higher degree of interpretation of water quality changes, while Zhang et al. (2019) believed that land use in the sub-basin scale has a greater ability to explain water quality.Due to regional specificity and human characteristics, the water quality response to land use patterns at various scales is quite different, and the optimal scale is not identical, so it is not easy to draw a unified conclusion (Tian et al. 2020).Therefore, it is necessary to compare different spatial scales, which is helpful to find out the response scale of the correlation between water quality and land use.
In recent years, the degradation of water quality in the Huaihe River Basin cannot be ignored.As one of the seven major basins in China, it is also an important agricultural area in South China, gathering a large number of agricultural population.Agricultural nutrition enrichment, domestic sewage and industrial wastewater discharge lead to river water degradation, especially non-point source pollution.At present, the study of water quality in Huaihe River Basin based on different spatial scales is extremely rare.Therefore, this paper takes the middle reaches of the Huaihe River as the research object, using land use data, combined with the measured data of water quality in the study area, using Spearman, RDA and multiple linear regression analysis to explore the impact of land use on water quality in the middle reaches of the Huaihe River.Our goals are: (1) Quantify the relationship between land use pattern and water quality, (2) Determine the multi-scale response of land use to water quality.Our original contributions to the literature include: (1) determining the impact of land use indicators on water quality in the middle reaches of the Huaihe River, (2) exploring the relationship between land use and water quality at multiple spatial scales.In order to provide scientific basis for water quality management and water pollution control in the middle reaches of Huaihe River.

Study area
The middle reach of the Huai River is located in Anhui Province, China, which is located at 32 35 0 $32 50 0 N, 116 45 0 $117 5 0 E, and has a temperate semi-humid monsoon climate with large temperature fluctuations.Plains dominate, hills and mountains coexist, the water system is advanced, and the river network is complex.May through October is the heaviest precipitation season.With the development of agriculture and industry come a variety of pollutant emissions and severe environmental contamination.The primary crops are rice, corn, and wheat, and the breeding industry is mainly cattle and sheep.

Dataset acquisition
In September 2021, 16 water samples were collected at a depth of 10 cm in the tributaries and mainstream of the study area, and the samples were placed in 5 L pickling plastic containers.The obtained water samples were put in a refrigerated incubator and then transferred to a laboratory refrigerator at 4 C for storage in the dark.The water quality indicators were determined, including TN, TP, NO 3 -N, and NH 3 -N.Table 1 indicates the determination method, repeat three times for each sample.
The remote sensing image and DEM data of Landsat 8 in 2021 were obtained from NASA.In ENVI 5.5 software through image mosaic, cutting, atmospheric correction, radiation correction pretreatment, image preprocessing using support vector machine algorithm (SVM) classification divides land categories into arable land, forest land, grassland  ( Zare et al. 2019), building land, water, and other lands (Figure 1).Finally, the total accuracy of land use classification verified by the confusion matrix is 86.5%.Sixteen watersheds were extracted using digital elevation model (DEM) data with an accuracy of 30 m and ArcGIS 10.6 Arcswat plug-in with 16 sampling points as the outlet.Watersheds are catchments controlled by sampling points.This paper selects four spatial scales of 100, 500, and 1000 m buffer zone and watershed and creates 100, 500, and 1000 m band buffer zones on the river bank to investigate the relationship between land use patterns and surface water in the middle reach of the Huai River.

Data analysis
Spearman correlation analysis between the proportion of land area and water quality indicators in four spatial scales was conducted using R4.1.2to perform.RDA analysis can interpret the relationship between environmental factors (explanatory variables) and species (response variables), and quantitatively analyze the impact of various environmental factors on different species.Based on the advantages of RDA method for quantitative analysis and intuitive interpretation of the relationship between multiple explanatory variables and multiple response variables, this method has been maturely applied to other research fields such as environment to identify the main influencing factors, analyze the relationship between multiple factors and judge the degree of influence.Therefore, the area of each land type was used as an explanatory variable.The water quality parameters were utilised as response factors for multiple linear regression analysis, and the coefficient of determination (R 2 ) value was employed to express the degree of land use pattern interpretation of water quality (Zong et al. 2020).

Water quality characteristics
Table 2 and Figure 2 display the statistical results and spatial distribution of water quality parameters in the middle reaches of the Huaihe River, respectively.It can be seen that the TN concentration in the middle reaches of the Huaihe River shows the distribution characteristics of high upstream and low downstream, with a significant difference between sampling points.Based on the 'Surface Water Environmental Quality Standard' (GB 3838-2002), one sampling point exceeds the 2.00 mg L À1 TN limit for Class V water.The concentration of total phosphorus (TP) decreases gradually from upstream to downstream, and two sampling points surpass the standard limit of TP 0.4 mg L À1 (Zhang et al. 2020).The concentration of NH 3 -N is low upstream and high downstream, and the difference in concentration between sampling points is minimal.The concentration of NO 3 -N is greater upstream and lower downstream.Therefore, the primary pollutants in the water body of the middle reach of the Huai River are TN and TP, with the highest pollutant concentrations primarily found in cultivated and building land.

Analysis of the relationship between land use and water quality indicators
The land area is an explanatory factor, and the water quality parameter is the response factor for multiple linear regression analysis.The R 2 value represents the degree of interpretation of the proportion of land area to water quality parameters.Table 3 indicates the land use pattern has the greatest explanation for TN, the land use pattern within 100 m has the highest explanation for NO 3 -N, and the land use patterns of 500 m and 1000 m have the largest explanation for TN.The land use pattern at the watershed scale has the greatest explanation for TP.Table 4 illustrates the Spearman correlation analysis between the proportion of land type area and water quality indices at four spatial scales.The results showed that TN is positively correlated with cultivated and other lands and negatively correlated with grassland within 1000 m, with a correlation coefficient of À0.566.Total phosphorus is positively correlated with the proportion of cultivated land and building land in the buffer  zone and watershed scale, and the correlation is the largest on the watershed scale.NH 3 -N is negatively correlated with grassland and is the most significant in the range of 1000 m, with a correlation coefficient of À0.377.NO 3 -N is positively correlated with cultivated land, building land, other land and water area and has a strong correlation with 1000 m cultivated land, and the correlation coefficient is 0.492.Redundancy analysis (Table 5) shows a significant correlation between cultivated land and water quality within 100 to 500 m, and there is a significant correlation between other land uses and water quality at all spatial scales.Building land indicates a more obvious correlation with water quality on the watershed scale.There is a significant correlation between forest land and water quality in the range of 100 m, and the correlation between water area and water quality is the strongest within 500 m.In general, the land use pattern at the watershed scale better explains the relationship with water quality, while at the buffer zone scale, the interpretation of land use pattern to water quality decreases with the increase of buffer zone.

Discussions
Overall, the water quality in the upper reaches of the middle reach of the Huai River is relatively worse, which can be attributed to more residential areas in the upper reaches, and more complex human activities, so the surrounding water pollution is more serious.Total phosphorus was negatively correlated with the proportion of grassland, forest land, and water area in the buffer zone scale and watershed scale, indicating that the increase of grassland and forest land area inhibited water pollution to a certain extent.Grassland and forest land affected water conservation and water purification.Hong et al. (2016) discovered that the water area can intercept, purify, and precipitate phosphorus, and the water area of the middle reach of the Huai River has a positive effect on TP.Total phosphorus is positively correlated with cultivated land in the buffer zone and watershed scale, and TN, NH 3 -N, and NO 3 -N are strongly negatively correlated with cultivated land in the buffer zone and watershed scale, which is closely related to the absorption capacity of cultivated land crop types and crop types to nitrogen and phosphorus in the middle reach of the Huai River (Liu et al. 2021).Nitrogen and phosphorus fertilisers must be applied to the growth of rice and other crops.
Studies have shown that rice can absorb most of the nitrogen in the water, and phosphorus absorption is relatively small.The strong absorption capacity of rice and the deposition of nitrogen in the soil make the nitrogen slow down in the surface runoff process (Xiao et al. 2016).The content of NH 3 -N and NO 3 -N is affected by the migration and transformation of nitrogen.Microorganisms in the cultivated land cause nitrification and denitrification between nitrogen.Therefore, there is a negative correlation between the nitrogen in the cultivated land and the water body.This is because animal husbandry in the building land is developed.The excretion of livestock such as cattle and sheep, the domestic wastewater generated by human activities, and animal residues increase nitrogen and phosphorus content, indicating that human activities in the building land have a 'sink' effect on TN and NH 3 -N content.However, TN and NO 3 -N are positively correlated with grassland and woodland in each buffer scale.Different from other research results, it indicates that woodland and grassland will become the source of nitrogen pollutants in various periods.This is a result of the animal manure produced by local herders and the soil erosion caused by forest and grass.The nutrients in the water flow into the river through surface runoff, resulting in deterioration of water quality.At the same time, according to the multiple regression analysis, the land use pattern at the buffer scale has a higher degree of explanation for TN and NO 3 -N.The land use pattern in the watershed has a greater explanation for TP and NH 3 -N since nitrogen and phosphorus are readily adsorbed by soil, and the nitrification and denitrification processes involved by microorganisms are finally converted into gases (Wang et al. 2014).
The results of redundancy analysis show that the impact of land use patterns on water quality is more interpretable at the watershed scale, which is similar to the results of Ren et al. (2022).The critical land types affecting water quality are different at various spatial scales.The correlation between cultivated land and water quality is the largest within the 100-500 m buffer zone, whereas the correlation between grassland and water quality is greater within the 1000 m buffer zone, and the correlation between water land and water quality is the highest within the watershed scale.It demonstrates that agricultural nonpoint source is the main cause of water pollution.Reasonable control of cultivated land and planning and management of grassland are helpful to maintain the sustainable development of surface water in the middle reach of the Huai River.

Conclusions
1.The water pollutants in the middle reach of the Huai River are mainly TN and TP, The average mass concentrations were 1.24 mg Á L À1 and 0.21 mg Á L À1 , which were in the Class IV standard of 'Surface Water Environmental Quality Standard' (GB 3838-2002).The mass concentrations of NH 3 -N and NO 3 -N are low, The average mass concentrations were 0.56 mg Á L À1 and 22.75 mg Á L À1 , showing prominent spatial differentiation characteristics.2. Land use at various spatial scales explains water quality variations, and land use patterns at the watershed scale have a higher degree of explanation for water quality changes.
3. The relationship between water quality and cultivated land and building land in the middle reach of the Huai River is the most significant.Building land is positively correlated with TN, TP, and NH 3 -N, indicating that building land has a 'source' effect on surface water quality.TP is positively correlated with cultivated land, demonstrating that cultivated land has a 'source' impact on TP.TP and NH 3 -N are negatively correlated with grassland, revealing that grassland had a purifying influence on TP and NH 3 -N.TN, TP, and NO 3 -N are positively correlated with forest land, indicating that forest land had a 'sink' effect on surface water quality.In general, the water quality of the middle reach of the Huai River is mainly affected by agricultural non-point sources and urbanisation development.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Figure 1 .
Figure 1.Land use and watershed map of study area.

Figure 2
Figure 2 depicts the land use proportion of each sampling point at different spatial scales.It revealed that the land use in the middle reach of the Huai River is dominated by cultivated land, and the proportion of grassland and forest land in the study area is relatively small.The land use types near the sampling sites are mainly water areas and cultivated land.The proportion of grassland and forest land increases with spatial scale, while the water area decreases (Figure3).

Figure 2 .
Figure 2. Water quality at sampling points.

Figure 3 .
Figure 3. Area ratio of land use types.

Table 1 .
Determination method of water quality index.

Table 2 .
Descriptive statistics of water quality parameters.

Table 3 .
Explanation of water environment parameters utilising multiple linear regression models of land use patterns at various buffer scales.

Table 4 .
Spearman correlation coefficient between water quality parameters and land use types.

Table 5 .
Interpretation of water quality index by land use based on redundancy analysis.