Premium
A Bayesian approach to assessing the uncertainty in estimating bioconcentration factors in earthworms—the example of quinoxyfen
Author(s) -
Fragoulis George,
Merli Annalisa,
Reeves Graham,
Meregalli Giovanna,
Stenberg Kristofer,
Tanaka Taku,
Capri Ettore
Publication year - 2011
Publication title -
pest management science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.296
H-Index - 125
eISSN - 1526-4998
pISSN - 1526-498X
DOI - 10.1002/ps.2105
Subject(s) - bioconcentration , bayes' theorem , probabilistic logic , environmental science , earthworm , statistics , mathematics , ecology , toxicology , biology , bayesian probability , bioaccumulation
BACKGROUND: Quinoxyfen is a fungicide of the phenoxyquinoline class used to control powdery mildew, Uncinula necator (Schw.) Burr. Owing to its high persistence and strong sorption in soil, it could represent a risk for soil organisms if they are exposed at ecologically relevant concentrations. The objective of this paper is to predict the bioconcentration factors (BCFs) of quinoxyfen in earthworms, selected as a representative soil organism, and to assess the uncertainty in the estimation of this parameter. Three fields in each of four vineyards in southern and northern Italy were sampled over two successive years. RESULTS: The measured BCFs varied over time, possibly owing to seasonal changes and the consequent changes in behaviour and ecology of earthworms. Quinoxyfen did not accumulate in soil, as the mean soil concentrations at the end of the 2 year monitoring period ranged from 9.16 to 16.0 µg kg −1 dw for the Verona province and from 23.9 to 37.5 µg kg −1 dw for the Taranto province, with up to eight applications per season. To assess the uncertainty of the BCF in earthworms, a probabilistic approach was used, firstly by building with weighted bootstrapping techniques a generic probabilistic density function (PDF) accounting for variability and incompleteness of knowledge. The generic PDF was then used to derive prior distribution functions, which, by application of Bayes' theorem, were updated with the new measurements and a posterior distribution was finally created. CONCLUSION: The study is a good example of probabilistic risk assessment. The means of mean and SD posterior estimates of log BCFworm (2.06, 0.91) are the ‘best estimate values’. Further risk assessment of quinoxyfen and other phenoxyquinoline fungicides and realistic representative scenarios for modelling exercises required for future authorization and post‐authorization requirements can now use this value as input. Copyright © 2011 Society of Chemical Industry