With rapid socio-economic development in the Taihu Lake Basin, pollution of Taihu Lake water is increasingly getting more serious. Thus government at all levels has allocated considerable resources (money and manpower) to this issue in the Taihu Lake Basin. As a result of their efforts, point source pollution has been better controlled around the basin. But non-point source pollution is still serious due to its dispersion and concealment, and its random nature, and consequent difficult monitoring. So it is still the key to resolving pollution in the basin and is a difficult point to which attention needs to be paid both currently and in the future. The non-point source pollution is the result of multiple factors such as soil, topography, hydrology, socio-economic development, management methods and land use, and it is closely related to the landscape distribution pattern. Reasonable landscape distribution can improve water quality by reducing the output of non-point source pollution, and hindering the transfer of watershed pollutants to receiving waters. Landscape pattern index can also be a good description of landscape heterogeneity, and the location-weighted landscape contrast index (LCI) can quantify non-point source pollution. However, little work has been done on the concrete relationship between the water quality index and the landscape index.
In order to investigate responses of water quality to landscape pattern in the Taihu Basin watershed, 3 typical streams connected to Taihu Lake in Yixing City were chosen as test plots. Indicators which reflect landscape characteristics such as source-sink landscape location-weighted Landscape Contrast Index (LCI), Number of Patches (NP), Edge Density (ED), Contagion Index (CONT), Shannon's Diversity Index (SHDI) and Aggregation Index (AI) were analyzed. Some parameters which reflect water quality such as Total Nitrogen (TN), Total Phosphorus (TP), and Chemical Oxygen Demand (COD_Mn) were also measured. Correlation analysis and path analysis between the landscape pattern and stream water quality were used to show significant positive correlations between LCI and TN, TP, and COD_Mn with direct action coefficients of 0.266, 1.512 and 0.979, respectively. Though LCI and TN were significantly negatively correlated, there was a less direct effect on TN. Rather, LCI has an indirect effect on TN concentrations via other landscape pattern indices. ED had a significant effect on TN, COD_Mn and NH₄⁺-N, with direct action coefficients of 0.740, -0.189 and 0.852, respectively. CONT showed significant positive correlations with TN and NH₄⁺-N, but SHDI was negatively correlated with TN and NH₄⁺-N. The higher the values of CONT and SHDI, the better adhesion between landscape patches. The non-point pollution load assessment will be reduced, retained and transformed, and the water quality will be relatively good if better balanced allocation of landscape elements is adopted.
DingKua is the only port for which the LCI is less than 0. It has a proportion of 'sink’ landscape which is the largest of the three ports. The woodland area is 59.46% and the open forest area is 2.64% of the DingKua 5km buffer zone. Forests have the effect of ecological restoration and pollutant interception. So the water quality at DingKua is the best of the three ports.
Among the other indexes the relationship was not significant. It was concluded that landscape pattern had a certain influence on stream water quality. Reasonable allocation of the landscape pattern is able to effectively combat non-point source pollution, and improve water quality in the watershed.