Effects of the mixture of diquat and a nonylphenol polyethoxylate adjuvant on fecundity and progeny early performances of the pond snail Lymnaea stagnalis in laboratory bioassays and microcosms.

Effects of the bipyridylium herbicide diquat and tank-mix adjuvant Agral90 were investigated on various life history traits of the freshwater pulmonate snail Lymnaea stagnalis. Trait expression was measured in simple laboratory bioassays on small size groups of snails, and under more complex, indoor microcosm conditions, on larger groups of snails. Microcosms were provided with sediment, plants, and fish, thus allowing a more complex level of intra and inter-specific interactions to develop. Treatments were performed with substances alone or in mixture, at concentrations ranging from 4.4 to 222.2microgl(-1) for diquat, and from 10 to 500microgl(-1) for Agral 90, under a fixed ratio design. Adult growth was negatively affected by diquat and its mixture with Agral 90 both at the highest concentrations (222.2 and 500microgl(-1), respectively). Fecundity expressed differently in bioassays and microcosms, but no effect of the chemicals could be observed on this trait. Progeny development was impaired by 222.2microgl(-1) diquat and its mixture with 500microgl(-1) Agral 90, as reflected by longer development time and reduced hatching rate of clutches laid by the exposed animals, as compared to the controls. Hatching data suggested that diquat bioavailability was lower in microcosms than under bioassay conditions. Consistently, chemical analysis showed that diquat disappeared more rapidly from the water in microcosms than in bioassays. Moreover, the differential expression of several life history traits under bioassays and microcosms conditions was probably also influenced by the level of intraspecific interaction, which differed among the systems. When significant, the effect of diquat was attenuated by the presence of Agral 90, indicating antagonistic interaction between the two substances. Such a deviation from additivity was partly validated statistically.