Variation in transfer factors for stochastic models: soil-to-plant transfer

The scientific literature is a good source of information about soil-to-plant transfer, usually documented as mean concentration ratios. It does not provide, however, good information about variation in concentration ratio values, because most studies are done with limited numbers of crops and soils. Fortunately, a suitably large and diverse database is available from the International Union of Radioecologists that can be used to investigate variation. It has a total of over 7,000 concentration ratio values for over 22 elements. This paper reports on analysis of the geometric standard deviations obtained from the database. We assert that geometric standard deviation values are comparable and transferable among elements, so that geometric standard deviations calculated from large databases can be applied to other elements when there is a paucity of data. For a fully generic situation, where neither crops nor soil types are known, the geometric standard deviations is typically about 6. This implies a 1,300-fold range to encompass 95% of the data. For concentration ratios obtained for crops of interest on a specific site, the geometric standard deviation values are much lower, about 1.5. This still implies a fivefold range to encompass 95% of the data. Other values of geometric standard deviations for intermediate levels of information are provided. These estimates of geometric standard deviation are important to and appropriate for stochastic simulations of the impacts of soil contamination on crops and subsequent food chains. They are also important to the inclusivity of concentration ratios value distributions for unusual but perhaps important crops and settings.