3.1. Characteristics and Variations of Water Physic-Chemical Parameters in Different Topographic Relief Areas
According to the limit value standard of water quality indicators detected in comparison with the Environmental Quality Standard for Surface Water (GB3838-2002) based on the sampling results of water quality in the two stages of the Chishui River Basin, the statistical values of water physical and chemical parameters in STR and LTR in the two stages are listed, including the maximum, minimum, average and standard deviation of each water quality indicator (
Table 5).
In general, the pH value of river water during these two periods ranges from 7.4 to 8.5, which is slightly alkaline in general. The average water temperature in the wet season is 21 °C higher than 10 °C in the dry season. The TN content in the watershed is far higher than the Class III limit, and the nitrogen pollution is serious; TP content is lower than Class III limit, and phosphorus pollution is small. There are seasonal differences in the physical and chemical parameters of watershed water. TP, TN and EC are lower in the rainy season than in the dry season. The Nemerow pollution index is also smaller in the rainy season, and the water quality is better.
On this basis, through Kruskal-Wallis test, this study found that TP, TN and EC also have significant differences in LTR and STR. TP, TN and EC of STR in Chishui River Basin are higher than LTR, and Nemerow pollution index is smaller in LTR area.
The ion absorption capacity of the thin soil layer in the Chishui River Basin is weak, which easily leads to the loss of ions in the soil to the river. Compared with the three kinds of water quality standards, the exceeding standard rate of total nitrogen is much higher than that of total phosphorus, and the pollution of total nitrogen to water quality is more serious than that of total phosphorus; According to the calculated average Nemerow pollution index, the water quality of the two regions in the two periods from good to bad is LTR in wet season, STR in dry season, LTR in wet season and STR in dry season.
3.2. Relationship between Landscape and Water Quality in Different Topographic Relief Areas
The landscape types are divided into 18 categories according to slope and land use type (
Figure 3). In different terrain undulating areas, the landscape composition is significantly different. First, from STR to LTR, the area of dry land, shrubbery and construction land decreased, and the area of forest land increased significantly. Second, in STR, the proportion of dry land and shrubbery is the largest, with the sum of the two accounting for 63.5%. The construction land is concentrated in STR. In LTR, forest land is the main landscape, accounting for 42.2%. Forest land is the only landscape type that increases with the increase of fluctuation range, and it is mainly gentle slope forest land and steep slope forest land. STR is an area with frequent human activities, which is suitable for farming and construction and shows the deepest human conflict. In LTR, the impact of human activities is small, and the ecological environment is good. Pearson correlation analysis is conducted on the water quality indicators (EC, TP and TN) and landscape composition indicators of the Chishui River Basin (
Figure 5). The results of correlation analysis show that the proportion of landscape types and water quality indicators are different in different water periods and different topographic relief areas. Dry land, shrubbery and construction land are positively correlated with water quality indicators, and are the main “source” landscapes that pollute river water quality; There is a significant negative correlation between forest land and water quality indicators, which is a “sink” landscape. Grassland was positively correlated with TN and EC, and negatively correlated with TP.
The study area is located in karst mountain area, with poor soil and high erosion risk. There is a lot of dry land agricultural activities. Due to excessive fertilization, pollutants such as chemical fertilizers and pesticides are easy to enter rivers through surface runoff, soil flow, shallow groundwater and other channels during the flood season, resulting in eutrophication of river water body and deterioration of water quality. In recent years, the implementation of the policy of “returning farmland to forests and grasslands” has increased the proportion of shrubbery and grassland. Due to the nitrogen and phosphorus residues caused by previous farming methods, the ability of shrubbery and grassland roots to intercept pollutants is weak, and the slope is inclined. With the help of gravity and terrain, the erosion ability of runoff is enhanced, making it easier for surface pollutants to migrate to rivers. The construction land is a densely populated area where people live and work. There are many ways of industrial point source pollution and rural urban non-point source pollution. The laying of impervious surface and the design of pipeline drainage led to the accumulation and discharge of pollutants, making the construction land become the main source of river water pollution. Forest land can intercept surface runoff, capture sediment and absorb pollutants, thus significantly reducing the pollutant load in runoff.
In general, the correlation between water quality indicators and landscape composition indicators in STR is small, and the correlation is more significant in LTR. It is specifically shown in the gentle slope dry land, slope dry land, gentle slope forest land, steep slope forest land, shrubbery and grassland. In comparison of water quality and landscape composition in rainy season and dry season, different topographic relief areas are different. The water quality indicators of flat dry land and gentle slope dry land are more significant in rainy season, while those of construction land and dry season are more significant. It is concluded that the landscape of different topographic relief areas has different effects on water quality.
3.3. Relationship between Landscape Pattern and Water Quality in Different Topographic Relief Areas
Six landscape pattern indexes, namely, percentage of land use (PL), patch density (PD), largest patch index (LPI), edge density (ED), landscape shape index (LSI), and mean patch size (MPS), and 108 slope landscape pattern indexes were obtained based on the six landscapes of paddy field, dry land, forested land, shrubbery, grassland, and construction land with three slope grades.
The spatial patterns of landscape also played a key role in mediating ecological processes (i.e., energy flows, hydrological process, and nutrient cycles). PD and ED indicators are often used to characterize the fragmentation degree and edge density of the landscape. LSI indicators are perimeter area ratio, which increases with the irregular shape. High PD, ED and LSI means that human activities are strong in the landscape in the region, and the spatial distribution of patches tends to disperse, which may increase surface runoff and soil erosion, making the risk of water quality deterioration in the basin high [
34]. LPI and MPS indexes were used to characterize the maximum plaque proportion and average plaque area. Higher LPI and MPS means that the landscape patches in the region are relatively complete and fragmented. However, its impact on water quality depends on whether the patch belongs to “source” landscape or “sink” landscape.
According to the statistics of “source” and “sink” landscape indicators of STR and LTR (
Table 6), it is found that they are significantly different in different topographic relief areas.
The “source” landscape includes dry land, bushes and construction land; The “sink” landscape is forest land.
In STR, the PD, ED and LSI index of various slope “source” landscapes are greater than those in LTR, especially the gentle slope dry land and gentle slope irrigation area with small relief accounting for 21.9% and 24.3% of the total. It shows that the landscape fragmentation is more serious in STR, and the risk of water quality deterioration in the watershed is greater than that in LTR. And the results of single factor analysis experiment of landscape pattern index of geographical detector on water quality can be obtained (
Figure 6). The gentle slope shrubbery PD (0.49), gentle slope shrubbery ED (0.45), steep slope dry land LSI (0.57) and gentle slope construction land LSI (0.44) have significant negative effects on water quality in the STR in the wet season, while the STR in the dry season is the gentle slope dry land PD (0.46), steep slope dry land ED (0.50) Mild slope shrubbery ED (0.53) and mild slope dry land LSI (0.51). In STR, the higher PD, Ed and LSI of woodland indicated that the degree of fragmentation of woodland landscape was high, and the proportion of woodland area was only 15.7%, which could not effectively alleviate the deterioration of water quality. The degree of landscape fragmentation is high, the “sink” landscape such as forest land is cut into smaller patches, the interception capacity of pollutants output from adjacent “source” patches is reduced, and the risk of water quality deterioration is increased. Therefore, forest land plays an important role in maintaining water quality and providing positive ecological feedback. Improving forest land and increasing its connectivity is an effective means to improve water quality.
In LTR, the LPI and MPS of gentle slope forest land (21.6%) and steep slope forest land (18.8%), which account for the largest proportion, are both high. A large number of connected forests are considered to be able to filter more pollutants to improve water quality [
35]. And the results of single factor analysis can be obtained that LTR in the wet season, LPI (0.61) and MPS (0.72) of forest land on the steep slope have a greater impact on water quality. LTR in the dry season, LPI (0.43) of forest land on the gentle slope and MPS (0.54) of forest land on the steep slope. These landscape pattern indicators mean that forest land in LTR can effectively purify water quality.
The factor detection of geographical detectors was used to analyze the explanatory power of slope landscape pattern index to water quality in wet and dry seasons in STR and LTR. As shown in
Figure 6, the height of the histogram represents the magnitude of Q explanatory power, and the 108 landscape pattern indexes 360 ° counterclockwise from the positive x-axis successively represent the PL, PD, LPI, ED, LSI and MPS of each slope landscape (A1, B1, C1, A2, B2, C2, A3, B3, C3, A4, B4, C4, A5, B5, C5, A6, B6 and C6).
The difference of the impact of slope landscape pattern index on water quality is not only reflected in the wet season and dry season, but also in STR and LTR. In general, the average explanatory power of slope landscape pattern index to water quality in LTR (0.36 in wet season and 0.33 in dry season) is much higher than that in STR (0.60 in wet season and 0.52 in dry season).
The landscape pattern index with the largest or smallest explanatory power to water quality is different in different topographic relief areas. In wet season of STR, the MPS (0.10) of flat construction land has the least explanatory power to water quality, and the first three landscape pattern indexes are flat paddy field LSI (0.62) > steep slope paddy field LPI (0.61) > gentle slope grassland PL (0.60); In dry season of STR, the gentle slope paddy field ED (0.09) has the least explanatory power for water quality, and the first three are gentle slope dry land LPI (0.68) > flat grassland PL (0.56) > flat grassland MPS (0.55). In wet season of LTR, MPS (0.13) of steep slope construction land has the least explanatory power for water quality, and the first three landscape pattern indexes are steep slope dry land PL (0.84) > flat paddy field LSI (0.84) > steep slope dry land PD (0.83); In dry season of LTR, MPS (0.17) of steep slope construction land also has the least explanatory power for water quality, and the first three are flat land construction land PL (0.74) > steep slope dry land PL (0.73) > steep slope forest land PL (0.72).
In STR, only a few slope land types have landscape pattern indexes with strong water quality explanatory power, such as A1LSI, C1LPI and B5PL in the wet season, and B2LPI, A5PL and A5MPS in the dry season. In LTR, a certain landscape pattern index has a great explanatory power on water quality in all slope fields, such as PL, PD, ED and LSI in both wet and dry seasons.
3.4. Quantitative Analysis of Water Quality Influencing Factors in Different Topographic Relief Areas
This study uses factor detection and interactive detection methods of geographic detectors to distinguish the impact of spatial difference drivers, and further discusses the impact of different factors in STR and LTR on water quality.
The selected factors include the landscape pattern index that has a significant impact on water quality and the factors that have a significant impact on water quality in previous studies, such as temperature, precipitation, population density and GDP per capita.
The quantitative analysis results of the explanatory power of individual and interaction of significant factors on water quality are shown in
Figure 7. The explanatory power of individual factors on water quality is located at the hypotenuse of the triangle, and the explanatory power of interaction between factors on water quality is located at the intersection of the two in the triangle.
The result of factor detector analysis shows that the explanatory power of significant factors to water quality varies in different regions and seasons. The selected landscape pattern index has higher explanatory power to water quality than temperature, precipitation, population density and GDP per capita. The landscape pattern index factor in LTR has greater explanatory power to water quality than that in STR. In the area of small fluctuation in high water season, the LSI of flat paddy field, LPI of steep slope paddy field and PL of gentle slope grassland have the greatest explanatory power to water quality; In dry season of STR, LPI of gentle slope dry land, PL of flat grassland and MPS of flat grassland have the greatest explanatory power to water quality. In wet season of LTR, the most powerful explanatory factors for water quality are the steep slope dry land PL, the flat paddy field LSI and the steep slope dry land PD; In the area with great fluctuation in dry season, the flat construction land PL, the steep slope dry land PL and the steep slope forest land PL have the greatest impact on water quality.
The explanatory power of air temperature to water quality is greatly affected by the terrain. The explanatory power of air temperature to water quality in LTR (0.61 and 0.55) is higher than that in STR (0.47 and 0.49). However, the explanatory power of precipitation on water quality is more affected by seasons, which is significantly higher in the wet season (0.52 and 0.55) than in the dry season (0.19 and 0.40). This is because the Chishui River Basin is located in the transitional zone between the Yunnan Guizhou Plateau and Sichuan Basin. It is less affected by the ocean monsoon, and generally belongs to the temperate continental climate. The temperature is high in summer, and the rainfall is large. The winter is dry and cold, and the rainfall is small.
The results of the interaction detector experiment show that the interaction between the factors has greater influence on the water quality than that of each factor alone. The interaction results are binary enhancement.