ABSTRACT
Transfer parameters for contaminants in the environment are notoriously variable. It is not unusual to measure a geometric standard deviation (GSD) of tenfold, so that the 5th and 95th percentiles are 10,000-fold apart. Despite many years of research, there remain large uncertainties, and often risk assessors will use on-site data for the transfer parameters. This Perspective assesses, using radionuclides as an example, whether a few site-specific measurements are more reliable than the more generic data compiled by researchers around the world. Using the example of plant/soil concentration ratios for several radionuclides, if one knew nothing about the soils and plants to be assessed, the GSD would be about 5.7-fold. If one had extensive knowledge about the soils and plants, the GSD would only diminish to about 3.2-fold. If a GSD of 3.2 is the level of residual natural variation to expect for plant/soil concentration ratios (the error term in a formal analysis of variance), then any on-site data that are within an order of magnitude of generic data could not be considered significantly different from the generic data. An appropriate conclusion in this case would be that unless on-site data are significantly different from the generic data, then the on-site data should not be used exclusively and the generic data should be considered suitable. The inherent variability of transfer parameters is so large that generic data may often be the best choice.
ACKNOWLEDGMENT
ECOMatters Inc. is acknowledged for support in preparation of the manuscript. This article was presented as a keynote address to an international conference in Rokkasho, Japan, and the interest of the organizers in this perspective is appreciated. I thank two anonymous reviewers for their constructive comments.
Notes
a The average GSD for 16 to 19 elements shown, with the range in brackets.
b Detailed models were developed for only four elements; this is an estimate for a more complete suite of elements.