Effects of “source-sink” landscape on spatiotemporal differentiation of phosphorus output in multiple watersheds in mountainous areas based on the modified location-weighted landscape index

Abstract

Optimizing the landscape structure and spatial configuration of the basin is one of the most important ways to reduce the harm of non-point source pollution (NPSP) to water bodies. The coupling relationship between the “source-sink” landscape pattern and NPSP process had long been a research hotspot of “source-sink” landscape theory. However, the “source-sink” landscape index represented by the classic location-weighted landscape index (LWLI) is suitable only for watersheds with similar environmental backgrounds, and cannot reflect the spatiotemporal heterogeneity of phosphorus output between watersheds with large differences in environmental backgrounds. In this study, 15 tributary watersheds in the Poyang Lake Basin were selected as the case study. The characteristics of “source-sink” landscape structure in 2020 were described by using landscape area structure indices. Moreover, the classical LWLI was modified by adding the topographic wetness index, soil erodibility index and vegetation interception index, and by evaluating the spatiotemporal differences in the effect of “source-sink” landscape patterns on phosphorus export in three hydrological periods (normal, wet and dry) by pearson correlation analysis and redundancy analysis. We found that (1) the proportion of “sink” landscape (greater than 58%) was greater than that of “source” landscape (less than 40%) in all sub-watersheds. In addition, the results of Lorentz curves showed that when the elevation and slope increased, the proportion of “source” landscape decreased and the proportion of “sink” landscape increased. (2) the classical and modified LWLI values were less than 0.5, which indicated that the potential contribution of the “sink” landscape was greater than that of the “source” landscape. (3) The correlation between the classical LWLI and total phosphorus (TP) in the wet period and normal period was stronger than the correlation between all “source-sink” landscape area structure indices and TP, indicating that the LWLI was more effective in describing TP loss than were the “source-sink” landscape area structure indices. (4) the correlation of the modified LWLI with TP in the three periods were stronger than those of the classical LWLI with TP, which indicated that the modified LWLI could better indicate the coupling relationship between the “source-sink” landscape pattern and the phosphorus output by introducing the factors of terrain runoff, soil erodibility and vegetation interception. This study can help with the evaluation of the spatial and temporal heterogeneity of nutrient loss in different watersheds with various landscape patterns, and provide a basis for scientific prevention and control of NPSP.