PREDICTION OF EFFECTIVE CATION EXCHANGE CAPACITY IN LOW pH SOILS.

In soil survey, there is a need to estimate effective cation exchange capacity (ECEC) for low pH soil horizons ((less than)5.5) where measured data are not available. The objectives were to determine if ECEC prediction equations could be developed from readily available soil properties and to test different data stratification approaches to improve predictability. Measured data in the National Soil Survey Characterization database were used for model development. Data stratification by Andisol and non-Andisol soil orders provided the best-fit models for the high organic C ((greater than)8%) soil layers. Total C and pH (CaCl2) explained 44% and 48% of the variation in ECEC for the Andisol and non-Andisol data, respectively. For soil layers with (less than or equal)8% organic C, stratifying the data by taxonomic order produced overall better-fitting models than by stratifying by taxonomic mineralogy. The eight taxonomic order equations developed (Andisols, Alfisols, Entisols, Inceptisols, Mollisols, Oxisols, Spodosols, and Ultisols) explained between 22% and 75% of the variation in ECEC, with the Oxisol and Andisol orders having the lowest predictability. Of the mineralogy class equations developed, only three (parasesquic, smectitic, and isotic) produced an R2 (greater than) 0.5. The four CEC-activity class strata, as a group, produced equations with the overall highest R2 and lowest root-mean-square errors. The CEC-activity classes explained between 62% and 88% of the variation in ECEC. Validation of the best-fitting models with an independent data set showed significant differences from unit one slope and zero intercept, except for the smectitic model. Predicting ECEC from readily available soil properties resulted in mostly moderate to a few strong ECEC prediction equations; weak predictions resulted for the Oxisol and Andisol strata.