The most used non-selective herbicide in the world is glyphosate, the active ingredient of popular formulations such as Monsanto's Roundup. The global use of glyphosate is still in 2017 on the rise. There are concerns for the environmental safety of both glyphosate itself, its degradation product AMPA, the cation it is distributed by, surfactants and other adjuvants. Glyphosate is generally expected to degrade before leaching to groundwater resources can occur, but under the Danish Pesticide Leaching Assessment Programme glyphosate has been found in groundwater above the limit value.

Glyphosate is water soluble, and binds to soil surfaces in complex ways still not fully understood. This study focuses on the strength of the bonds made between glyphosate and soil, and the soil parameters controlling the strength. Models for predicting the linear glyphosate sorption coefficient at the glyphosate concentration 0.23 mg\L are discussed and improved.

A sample set is carefully selected. Danish fields with pronounced gradients in clay content, pH and content of organic carbon are included. In junction with Greenlandic fields a gradient in soil age is obtained, ranging from areas recently introduced to farming, areas that have been farmed for only a few generations and the centuries old farmlands of Denmark. Soils of different origin are included, ranging from the glacial deposits of Denmark to the crystalline rocks of Greenland. Inability to obtain data rendered an attempt at including soils of volcanic origin unsuccessful for the most part.

Soil parameters are obtained from the knowledge base of Aarhus University, Foulum, or measured in the laboratory. Glyphosate sorption coefficients are measured by batch experiments on C¹⁴-marked glyphosate and liquid scintillation counting. Linear regression analyses are performed in Matlab.

The soil parameters of pH, content of clay, content of fine silt, content of oxalate extractable phosphorus and most of all content of oxalate extractable iron prove important in predicting glyphosate sorption coefficients. Parameters of electrical conductivity, content of organic carbon and oxalate extractable aluminium are found to have only a weak or no influence. The historical management and origin of the soils are found to influence the relationships between soil parameters and glyphosate sorption coefficient.

Pedotransfer functions for predicting a glyphosate sorption coefficient based on oxalate extractable iron and either pH or oxalate extractable phosphorus are found to have coefficients of determination above 0.6 at regional level or higher.