PubAg

Main content area

Landscape patterns regulate non-point source nutrient pollution in an agricultural watershed

Author:
Wu, Jianhong, Lu, Jun
Source:
The Science of the total environment 2019 v.669 pp. 377-388
ISSN:
0048-9697
Subject:
agricultural land, agricultural watersheds, dry season, forest land, intensive farming, landscapes, nutrients, pollution, probability, river water, statistical analysis, subwatersheds, variance, water quality, wet season, China
Abstract:
Landscape pattern critically affects hydrological cycling and the processes of non-point source nutrients pollution. However, little is known about the quantitative relationship between landscape characteristics and the river water quality, and very few studies have addressed the abrupt changes in river water quality with the gradient of landscape metrics. The present study was conducted in a typically intensive agriculture watershed of eastern China including 13 sub-watersheds with different landscape pattern metrics. We adopted redundancy analysis, nonparametric deviance reduction approach, bootstrap sampling and other statistical methods to reveal the quantitative relationship between landscape pattern metrics and water quality variables; then, the phenomenon of an abrupt change in river water quality was explored with different landscape pattern gradients. The results show that landscape pattern significantly affects river water quality, and this effect was quite different in dry and rainy seasons. In the studied watershed, landscape pattern metrics could respectively explain 71.1% and 55.3% of the total variance in the river water quality in dry and rainy seasons. The configuration metrics of landscape pattern had a stronger ability than their composition metrics to explain the variance in water quality. In the dry season, largest patch index of forestland (LPIfor), the most important landscape index, explained 37.9% of the total variance in water quality. While, in the rainy season, the most important landscape index was the largest patch index of farmland (LPIfar), and it could explain 32.4% of that variance. In the studied watershed, when the LPIfor was <35% or LPIfar was over than 50%, water quality would typically change abruptly, at which the probability of a change in river water would suddenly rise substantially.
Agid:
6332975