Characterisation of soil potassium supply as derived from sorption-desorption experiments

Plant K uptake depends on soil K supply and this can be modelled using the K concentration in the soil solution (CK), the soil K buffer power (SK) and the effective K diffusion coefficient. With the appropriate sorption-desorption curve, the parameters CK and SK can be estimated from the equilibrium K concentration (CK,0) and from the slope of the curve at CK,0 (SK,0). However, buffer power is frequently estimated by the ratio of the soil exchangeable K content (EK) to CK. Up to now, SK,0 had not been compared with this ratio, nor had CK,0 been compared with any CK estimate from soil solution extract measurements. To address this question, we collected 45 soil samples from 15 K fertilisation trials in France. The soils differed widely in their physicochemical characteristics, soil solution K concentration and buffer power. For each soil sample, a sorption-desorption curve was established from 16-h experiments performed at the estimated Ca concentration of the soil solution. The CK,0 estimates were compared with the K concentrations measured in the soil solution obtained either by direct centrifugation (CCK) or by centrifugation with 1,1,2-trichlorotrifluoroethane (TFECK). On average, CCK values were 16% higher than TFECK values, whereas CK,0 values were intermediate between the two. The K buffer power increased when CEC increased and when CK decreased. Multiple linear regressions using either CEC at the soil pH and TFECK-1 or EK and TFECK-1 as independent variables accounted for more than 98 or 95% of the variability of SK,0. The buffer power estimated by the ratio of EK to TFECK overestimated by 100% the SK,0 value obtained from the sorption-desorption curves.