Overview of the remaining GLiM models following the stepwise backward elimination process of nonsignificant terms (α = 0.05) testing for the effect of carbaryl, population, selection history and their interactions on mortality of D. magna in standardized acute toxicity experiments. (A) Results of general model analysing the data of all population × selection treatments; (B) Results of targeted comparisons between selection histories: control versus original, carbaryl-selected versus original, and carbaryl-selected versus control. The two populations obtained in the selection experiment (control and carbaryl-selected population) were compared with each other, and each of them was also compared with the original population in a paired analysis. ‘Carbaryl’ refers to different carbaryl concentration levels in the acute toxicity tests, ‘population’ refers to the pond of origin of the populations, and ‘selection history’ refers to the selection history of the populations (original = prior to selection experiment; control = experimental selection in the absence of carbaryl; carbaryl-selected = experimental selection in the presence of carbaryl)