Prediction of field efficacy from greenhouse data for four auxenic herbicides

摘要

The proposed paper will compare greenhouse and field efficacy data in light of concerns for offsite movement of herbicides. A central issue of environmental concern is how well greenhouse or laboratory data collected on a few species can predict injury to a larger, more diverse set of species in the field. A retrospective analysis of four auxenic herbicides shows that most efficacy data was generated to predict, with a high degree of certainty, the application rates required to cause 90% injury. There was little rate response data generated on the same species in both the greenhouse and field sufficient to estimate the 25% or 50% injury that is the environmental endpoint for most regulatory concerns. For those species where direct comparisons coufdbe made, the greenhouse to field injury varied from approximately equal to as much as 20X with large variations between species. For the species with the lowest EC25 values, the greenhouse data over predicted the field injury. Alternatively, a species sensitivity distribution uses all available data, and is predictive of injury to plant populations. Initial results suggest that the field and greenhouse data can be adequately modeled as log-normal distributions, were non-parallel, and can be used to predict the maximum application rates that are protective of 95% of species.