Short-term release and fixation of K in calcareous clay soils. Consequence for K buffer power prediction

Non-exchangeable K always contributes to some degree to plant nutrition. To understand this contribution and to deduce a predictive buffer power model the release and fixation of K was studied on a range of 44 calcareous clay soils. Short-term K sorption and desorption experiments (16 h) were followed by the measurement of soil exchangeable K (ammonium acetate extraction). Soil K-Ca exchange properties and the contributions of exchangeable K and non-exchangeable K to K dynamics of the soil-solution system were estimated. The change in the amount of non-exchangeable K during the experiment was generally proportional to the initial constraint imposed to the soil-solution system (phi), i.e. the solution: soil ratio multiplied by the difference between the solution K concentration imposed at the beginning of the experiment and the solution K concentration for which neither sorption nor desorption of K would occur. The proportionality coefficient (beta) called the 'soil ability for K release and fixation' was identical for release and fixation for 36 soils, whereas eight soils showed some difference between their ability for fixation and their ability for release of K. When beta was considered identical in the release range and in the fixation range for all soils it was inversely proportional to the initial K saturation ratio of the CEC corrected for the amount of ammonium extractable K which was not in exchange equilibrium with Ca. This quantity of ammonium extractable K which is not in equilibrium with Ca probably contributes only partially to plant nutrition. For the studied soils this quantity contributed a large proportion of exchangeable K (26-65%) especially in soils with a small K content. Due to the contribution of non-exchangeable K to soil-solution K dynamics the buffer power of the system does not only depend on exchange properties but also on soil release and fixation properties. When beta was taken into account buffer power was better estimated than when it was deduced from ion exchange alone.