PREDICTING PHOSPHATE ADSORPTION-DESORPTION IN A SOILSCAPE

Modeling the P availability in a soilscape requires knowledge about th e three-dimensional variability of the phosphate adsorption-desorption (AD) curves. The high variabilities of soil parent material and of ag ricultural P input in a 1.5-km(2) study area were taken into account b y taking 2555 samples at the intersections of a 50-m grid. To reduce t he necessary laboratory work, pedotransfer functions (PTFs) were devel oped that could predict the AD curves from more easily obtainable soil characteristics. Therefore, AD curves were measured for only 40 repre sentative samples. The AD curves were fitted with an extended and rear ranged Freundlich equation with a constant adsorption energy parameter k = 0.29. The adsorption capacity parameter A of this equation was pr edicted from clay, dithionite-extractable Fe (Fe-d), and pH using a no nlinear regression method. With increasing pH, A decreased for 4 < pH < 6.5 but increased for 6.5 < pH < 7.5. The Fe-d could be replaced by the soil color yellowness component b, which was easily determined wi th a chroma meter. The equilibrium P concentration of the soil solutio n, c(eq) was closely related to a one-point measurement of the AD curv es, c(1.2) (the final concentration after adding 1.2 mg L(-1) P). Thus , only the four soil properties clay, Fe-d (or b), pH, and c(1.2) wer e necessary to predict complete AD curves using the developed equation s. The equations proved to be reliable when tested by an independently selected subset of the 2555 samples and by a functional validation. T he developed equations can therefore be used as PTFs.