Soil CEC Predicting Model of Tobacco-Planting Fields in Chenzhou, South Hunan Province

Abstract

Background: Cation exchange capacity (CEC) is a basic but important soil property of soil fertility or quality, CEC predicting model is often derived from other soil properties measured more easily because the traditional method determining CEC is time-consuming and laborious. It is necessary to establish a new CEC prediction model for a new region because CEC predicting model usually is dependent on the study region. Objective: Chenzhou City is the most important and typical tobacco-planting region with tobacco-rice rotation in Hunan province and China, this study was conducted to establish CEC predicting model for the tobacco-planting fields in Chenzhou because so far no CEC predicting model is available for tobacco-planting fields in Chenzhou and in China. Method: In total 1055 topsoil samples (0∼20 cm) were collected in 2015 from the tobacco-planting fields in Chenzhou, soil properties included the particle size composition, pH, soil organic matter and various nutrients were determined, the status of CEC were assessed, and then CEC predicting models were setup in different regions in Chenzhou. Result: The results showed that CEC in Chenzhou was ranged from 3.50 to 48.50 cmol (+) kg-1 with a mean of 22.05 cmol (+) kg-1, averagely belonged to the very high grade (>20 cmol(+) kg-1). There were significant differences in CECs in different regions in Chenzhou, which was the highest in Jiahe (23.83 cmol(+) kg-1) but the lowest in Anren (15.78 cmol(+) kg-1). CEC was significantly correlated with different soil properties in different regions, which was significantly correlated with coarse sands, fine sands, clays, pH and total P in Chenzhou (R= 0.312**∼0.445**), significantly correlated with coarse sands, silts, fine sands, clays, pH, total P, exchangeable Ca2+, Mg2+ and available Zn in Suxian (R= 0.430**∼0.684**), significantly correlated with coarse sands, fine sands, silts, clays, pH, total P, available B and Cu in Yongxing (R=0.321**∼0.605**), significantly correlated with coarse sands, fine sands and clays and total P in Guiyang (R=0.330**∼0.477**), significantly correlated with coarse sands, silts and total K in Yizhang (R=0.326**∼0.466**), and only significantly correlated with fine sands in Jiahe (R=0.350**). The accuracy of CEC predicting model usually was lower when less properties involved. Based on the comparison of the R2 and RMSE of the established CEC predicting models, it is recommended that the total model for Chenzhou could be used for Guiyang, Jiahe and Yizhang, while the regional models should be selected for Yongxing, Anren and Suxian. Conclusion: This study proves further that different soil properties were most important for CEC predicting models in different regions, new CEC predicting models must be setup for a new study region, and soil organic matter is not a variable in soil CEC predicting models for tobacco-planting fields in Chenzhou, which are different from some previous studies.