Nonparametric predictive inference for exposure assessment.

Exposure assessment for food and drink consumption requires the combining of information about people's consumption of products with concentration data sets to provide predictions for chemical intake by humans. In this article, we present a method called nonparametric predictive inference (NPI) for exposure assessment. NPI is a distribution-free method relying only on Hill's assumption A(n). Effectively, A(n) is a postdata exchangeability assumption, which is a natural starting point for nonparametric statistics. For further discussion we refer to works by Hill and Coolen. We illustrate how NPI can be implemented to produce predictions for an individual's exposure based on consumption, body weight, and concentration data. NPI has the advantage that we do not have to assume a distribution to implement it. There may, however, be information available to suggest a distribution for a random quantity. Therefore, we present an NPI-Bayes hybrid method where this information can be taken into account by using Bayesian methods while using NPI for the other random quantities in the model.

Estimating the rate of poisoning by insecticide-treated seeds in a bird population.

Poisonings of granivorous birds by cereal seed treatments have been reported in the UK, but the true frequency of poisonings is unknown. We aimed to measure the rate of mortality due to poisoning by seed treatments in woodpigeons (Columba palumbus) in an area of East Anglian farmland where the risk from these compounds appeared high. Woodpigeons were fitted with temperature-sensing radio-tags and monitored daily during the winters of 1994/95 and 1995/96. Birds' deaths were detected and attempts made to retrieve carcasses for post-mortem examination including, where possible, analysis for pesticide residues. Ninety-one woodpigeons were monitored. Eleven birds died, but the causes of their deaths were uncertain: one contained a low residue of insecticide and in the other ten cases, no carcass was recovered, so no analysis was possible. Therefore, the number poisoned by pesticides could lie anywhere between zero and eleven. During 1994/95, estimated mortality ranged from 0% to 52%, depending on how many (if any) of the 11 casualties were poisoned. During 1995/96 there were no casualties. Using conservative diagnostic rules for classifying birds as poisoned by OP seed treatments, no link was found between the availability of treated fields in the study area and the rate of poisonings, and there were no significant differences between the two study years. For reasons discussed in the paper, true mortality resulting from exposure to insecticide seed treatments was considered likely to lie in the range 0-5%.

Case Study Part 2: Probabilistic modelling of long-term effects of pesticides on individual breeding success in birds and mammals.

Long term exposure of skylarks to a fictitious insecticide and of wood mice to a fictitious fungicide were modelled probabilistically in a Monte Carlo simulation. Within the same simulation the consequences of exposure to pesticides on reproductive success were modelled using the toxicity-exposure-linking rules developed by R.S. Bennet et al. (2005) and the interspecies extrapolation factors suggested by R. Luttik et al. (2005). We built models to reflect a range of scenarios and as a result were able to show how exposure to pesticide might alter the number of individuals engaged in any given phase of the breeding cycle at any given time and predict the numbers of new adults at the season's end.

Case Study Part 1: How to calculate appropriate deterministic long-term toxicity to exposure ratios (TERs) for birds and mammals.

In the European Union, first-tier assessment of the long-term risk to birds and mammals from pesticides is based on calculation of a deterministic long-term toxicity/exposure ratio (TER(lt)). The ratio is developed from generic herbivores and insectivores and applied to all species. This paper describes two case studies that implement proposed improvements to the way long-term risk is assessed. These refined methods require calculation of a TER for each of five identified phases of reproduction (phase-specific TERs) and use of adjusted No Observed Effect Levels (NOELs) to incorporate variation in species sensitivity to pesticides. They also involve progressive refinement of the exposure estimate so that it applies to particular species, rather than generic indicators, and relates spraying date to onset of reproduction. The effect of using these new methods on the assessment of risk is described. Each refinement did not necessarily alter the calculated TER value in a way that was either predictable or consistent across both case studies. However, use of adjusted NOELs always reduced TERs, and relating spraying date to onset of reproduction increased most phase-specific TERs. The case studies suggested that the current first-tier TER(lt )assessment may underestimate risk in some circumstances and that phase-specific assessments can help identify appropriate risk-reduction measures. The way in which deterministic phase-specific assessments can currently be implemented to enhance first-tier assessment is outlined.

A new interpretation of avian and Mammalian reproduction toxicity test data in ecological risk assessment.

The long-term risks of pesticides to wildlife in the EU currently are assessed by comparing the lowest no-observed-effect concentration (NOEC) determined from the suite of endpoints measured in existing avian and mammalian laboratory reproduction tests with estimated exposure concentrations by calculating Toxicity to Exposure Ratios (TERs). Regulatory authorities experience difficulties when assessing long-term risks because of the lack of accepted methods to improve the ecological realism of exposure and toxicity estimates and understand risks at a population level. This paper describes an approach for interpreting existing avian and mammalian toxicity test data that divides breeding cycles into several discrete phases and identifies specific test endpoints as indicators of direct pesticide effects possible at each phase. Based on the distribution of breeding initiation dates for a species of concern and the dates of pesticide applications, this approach compares the phase-specific toxicity endpoint with the expected pesticide exposure levels during each of the breeding phases. The fate of each breeding attempt is determined through a series of decision points. The cumulative reproductive response of individuals in a breeding population based on this decision framework provides a means of examining the estimated risks over the course of the breeding season and deriving an overall metric of the impact of the pesticide on reproduction. Research needed to further improve the approach is discussed.

Assessing potential exposure of birds to pesticide-treated seeds.

Seed treatments are widely used for crop protection and present a particular risk to granivorous birds. UK risk assessment for seed treatments has tended to focus on highly granivorous species; however, under some conditions, non-granivorous birds will take seeds. Better data is needed on which species eat seeds for which pesticide treatments are used. To identify which species will take and eat a range of crop seeds in common usage in the UK, birds visiting bait stations at which untreated seed was presented were video recorded. Information was also obtained on how much seed is taken by individual birds. The seeds tested were wheat, barley, maize, oilseed rape, grass, peas and pelleted sugar beet. For many of the species observed at the bait stations, the amounts of seed consumed during single visits were sufficient to pose a potential risk (toxicity-exposure ratio < 10) if the seed had been treated with one of the more acutely toxic seed treatments. Previous studies have shown that de-husking of seeds can substantially reduce birds' exposure. This paper provides information on which of the species recorded de-husked which seeds, in field conditions. The use of these data in pesticide risk assessment is considered.

Intake estimation of polychlorinated dibenzo-p-dioxins, dibenzofurans (PCDD/Fs) and polychlorinated biphenyls (PCBs) in salmon: the inclusion of uncertainty.

Dioxins and dioxin-like PCBs are given toxic equivalency factors (TEFs) in order to calculate the combined toxic equivalence (TEQ) of these contaminants in a sample of food. This study calculates the probability of an average consumer exceeding the recommended tolerable daily intake of 1-4 pg WHO-TEQ kg(-1) bw day(-1) as the amount of salmon in the diet is increased. Probabilistic risk analysis is used to account for the known uncertainties in this calculation. When the TEF uncertainty was excluded with no salmon consumption, the background dietary intake ranged from 1.36 to 1.78 pg TEQ kg(-1) bw day(-1). A weekly consumption of three standard salmon portions resulted in a 36% chance of exceeding the upper limit of the TDI. Inclusion of the TEF uncertainty increased the background dietary intake range from 2.1 to 4.4 pg TEQ kg(-1) bw day(-1), and the weekly consumption of one salmon portion resulted in a 79% chance of the average consumer exceeding the upper TDI. The most important factors contributing to the uncertainty in these results were, in order of magnitude, the TEF for PCB 126 and the sampling uncertainty (sample size) followed by the measurement uncertainty of PCB 126. We recommend that it is more important to increase sample size and produce more precise estimates in the TEF than to improve analytical accuracy.