Environmental Earth Sciences   2015 年 74 卷 5 期

2015.10.01

Establishing nutrient criteria for regional lakes is necessary to assess human impact on lake aquatic ecosystems and protect water quality and biotic integrity. Multiple linear regression models, in which the ratios of land use to mean lake depth (DEP) are the predictor variables, and the logarithms of physico-chemical variables and Chl a concentrations are the dependent variables, were developed to predict physico-chemical variables and chlorophyll a criteria for Yungui Plateau Ecoregion lakes. The contemporary land use data of 22 lake watersheds were analyzed and employed to develop the spatial relationship with the regression models. The data of five lake watersheds in four periods were used to verify the accuracy of the regression models, and to test their applicability in time scale. The intercept of these models (i.e., expected physico-chemical variables and Chl a concentrations in the absence of human activities) represents the criterion concentrations. Results suggested that the percentages of other construction land/DEP had strong positive influences on the concentrations of all variables (except electrical conductivity). The multiple linear regression models offered a potential method for regions with heavy anthropogenic disturbances to develop the physico-chemical variables and Chl a criteria.