Search for a command to run...
Abstract Not only are soil erosion and overall loss of soil fertility serious issues for loess‐derived soils of developing nations, such as Iran, but they are also global problems. This research investigated the role of land use change and its effect on soil degradation in cultivated, pasture, and urban lands, when compared with native forest in terms of declining of soil quality and fertility. Multivariate statistical methods including principal component analysis and cluster analysis were employed to determine the relative magnitude of anthropogenic and natural influences on soil quality. Partial least squares (PLS), principal component regression, and ordinary least squares regression were used to predict soil cation exchange capacity using soil characteristics. Principal component analysis identified five primary components of soil quality. PC1 explained 41·01% of the total variance, PC2 accounted for 15·29%, and PC3 accounted for 11·3%. PC4 and PC5 accounted for 7·37% and 4·19% of the total variance, respectively. Cluster analysis revealed that the lowest soil quality occurred in urban soils. An exponential semivariogram using PLS predictions revealed stronger spatial dependence among cation exchange capacity [ r 2 = 0·80, root mean square error (RMSE) = 1·99] than the other methods, principal component regression [ r 2 = 0·84, RMSE = 2·45] and ordinary least squares [ r 2 = 0·84, RMSE = 2·45]. Therefore, the PLS method provided the best model for the data. In stepwise regression analysis, mean weight diameter and labile carbon were selected as influential variables in all soils. This study quantified reductions in numerous soil quality parameters resulting from extensive land‐use changes and urbanization in the Ziarat watershed in northern Iran. Copyright © 2016 John Wiley & Sons, Ltd.