Texture specific regression models for predicting soil EC_e values from EC_1:2.5 for effective soil salinity assessment in Tanzania

Abstract

Electrical conductivity of saturated soil paste extract (EC_e) is a standard laboratory soil salinity measurement. However, due to difficulty of ECe measurement, electrical conductivity of soil to water suspensions (EC_soil:water) such as EC_1:2.5 are used and its values converted to EC_e for salinity interpretation in crop production. This study was conducted to develop texture specific regression models for predicting EC_e values from EC_1:2.5 for Tanzanian soils. A total of 198 composite soil samples at 0 – 30 cm depth were collected from Kiwere, Dakawa, Sakalilo and Mwamapuli irrigation schemes in Iringa, Morogoro, Rukwa and Katavi Regions respectively and analyzed for soil texture, EC_1:2.5 and EC_e using standard laboratory methods. The dominant soil textural classes were clay, sandy clay loam, sandy clay, and clay loam. There were significant differences (P<0.05) between mean values of EC_1:2.5 and EC_e (dS m^-1) in all textural classes. The regression models indicated significantly strong linear relationships between values of EC_1:2.5 and EC_e for all textural classes with R2>0.90 and P<0.001 for both regression models with and without intercept. The regression models without intercept performed better in predicting soil EC_e from EC_1:2.5 than regression models with intercept by having higher P-values, slope value closer to 1.0 and lower RMSE values between measured and predicted EC_e. The study recommends regression models expressed as EC_e = 2.0963 EC_1:2.5 for clay; EC_e = 2.7714 EC_1:2.5 for sandy clay loam; EC_e = 2.3519 EC_1:2.5 for sandy clay and EC_e = 2.0811 EC_1:2.5 for clay loam soils for predicting soil EC_e from EC_1:2.5 in Tanzania.