ArticlesIDENTIFYING ASSOCIATIONS AMONG SOIL AND SITE VARIABLES USING CANONICAL CORRELATION ANALYSISTan, Z. X.1; Lal, R.1; Smeck, N. E.1; Calhoun, F. G.1; Gehring, R. M.2; Parkinson, B.2 Author Information 1School of Natural Resources, The Ohio State University, 210 Kottman Hall, 2021 Coffey Rd., Columbus OH 43210. 2USDA-NRCS, 200 N High St., Columbus, OH 43215. Dr. Tan is. corresponding author. E-mail: [email protected] Received Aug. 16, 2002; accepted Jan. 9, 2003. Soil Science: May 2003 - Volume 168 - Issue 5 - p 376-382 doi: 10.1097/01.ss.0000070912.55992.d5 Buy Metrics Abstract Alfisols are the predominant soil resource in Ohio. An understanding of the relationships among their soil and site variables will help in planning judicious future land use to maintain soil quality and sustain agriculture. The objective of this study was to identify associations and their significance between dependent variables of surface horizons of Alfisols (consisting of bulk density, base saturation, CEC, pH, Munsell value and chroma) and descriptive variables (including drainage class, elevation, sand, slope gradient, soil organic carbon) using canonical correlation analysis (CCA). The association of SOC concentration with bulk density was found to be the most important relationship among the selected variables, followed by the association of soil clay content with CEC. The association of drainage class and Munsell chroma was identified as being dominant and significant from the third pair of canonical variates. Despite pronounced variations in the means of all selected variables among land uses, these associations and their significance were consistently confirmed, regardless of land use, and were responsible for about 60% of the variance within each set of variables. Additional associations could be identified from other canonical variates but were less explicit and varied widely among land uses. As a means of identifying associations and the order of their importance among multivariates, the CCA has become a promising technique for simplifying, and simultaneously analyzing, two sets of complex data consisting of biotic and abiotic variables. © 2003 Lippincott Williams & Wilkins, Inc.