Spatial scale effects of riparian zone landscape patterns on river water quality in the Dongting Lake Basin of China

Riparian zones can serve as a critical interface to regulate the river water quality. However, it is still lacking in the influence of the riparian zone landscape patterns on the river water quality. Targeted strategies of water pollution control have been hindered in optimizing the riparian zone landscape. This study aims to quantify the spatial scale effects of the riparian zone landscape patterns on the river water quality. Key indicators of the landscape patterns were also identified to obtain the optimal sizes of the riparian buffer. Accordingly, the Dongting Lake basin was selected as the study area due to the intensive agriculture and rapid urbanization in recent years. The dataset was collected from the high-resolution land use and surface water quality in 2023. The recursive feature elimination-random forest model was integrated with the redundancy analysis. The key indicators of the riparian landscape were identified for the specific and composite water quality variables. The results showed that the total nitrogen (TN) was first identified as the predominant pollutant, with the elevated concentrations primarily distributed in the Dongting Lake area and Chang-Zhu-Tan urban agglomeration, which was driven by the combined agricultural non-point source pollution and urban emissions. Secondly, the most significant influence of the riparian landscape patterns on the dissolved oxygen (DO) occurred at a buffer scale of 25 m. In contrast, the optimal buffer scale was 200 m for the total nitrogen (TN), total phosphorus (TP), and the chemical oxygen demand by manganese (COD_Mn), while the most effective buffer scale was 100 m for the ammonia nitrogen (NH₃-N). Thirdly, the explanatory power of the landscape pattern of the riparian zones to the composite river water quality demonstrated a nonlinear trend of initially decreasing, then increasing, and finally decreasing. With the maximum explanatory power of 41.87% at the 200 m buffer zone, the landscape pattern of the riparian zone was thus identified as an effective predictor of the composite river water quality variations. Fourthly, the dominant influencing indicator of the landscape on the composite water quality also shifted at the different scales: Patch density of cultivated land at 25 m, contagion index at 100 m, patch density of construction land at 200 m, and Shannon's diversity index at 800 m. The 50 and 400 m buffers were both characterized by the edge density of the construction land. The landscape patterns of the cultivated and construction land were the key influencing factors on the composite river water quality. A great contribution was made to the landscape patterns' influence on the river water quality at the different spatial scales in riparian zones. The 200 m riparian zone was recommended to optimize the landscape strategies for the spatially precise regulation. A hierarchical regulatory framework was integrated with the "source reduction–process interception–end purification". The proportion of the cultivated and construction land was reduced to increase that of the forest land. The interface of the cultivated land, construction, and forest land was significantly expanded for the optimal landscape pattern of the riparian zone. The finding can also provide a strong reference to improve the river water quality.