Figure 1: Generic dose–response function of the risk of homing failure in exposed honeybees.
From: Pesticide risk assessment in free-ranging bees is weather and landscape dependent
(a) The predicted dose–response function (continuous line±estimate s.e. as dashed lines) derived from a binomial logit LME model, specifying random intercepts for colony and release point effects, with n=978 individuals split into 60 homing trials from 22 release points around nine colonies. Quartile boxes and dots overlaying predictions display observed homing failure proportions for trials with >15 individuals. (b) A posteriori statistical power analysis, appraising the ability of the LME to detect a significant homing failure dose–response with the current sample size. Statistical power was measured as the proportion of significant effects out of 1,000 simulated homing failure data sets randomly recomputed with a constrained effect size (percentage increase at 1 ng). The observed effect size (46% increase at 1 ng, filled dot) had a 99% chance of being detected given our sample size.
