Impact of an Oomen feeding with a neonicotinoid on daily activity and colony development of honeybees assessed with an AI based monitoring device

Abstract

Feeding experiments are standard tools in the pollinator risk assessment. The design (Oomen et al 1992) was developed to test insect growth regulators and herbicides. In recent years there was an update (Lückmann & Schmitzer 2015) on the outline in order to also focus on the advantage of different rates making a dose response design possible where exposure levels are known. Additionally, this design gives the possibility to test different rates for honey bee colonies foraging in the same landscape.The main objective of the experiment presented here was to determine the natural variability of foragers losses of hives fed with a sub-lethal neonicotinoid concentration compared to an untreated control. Other objectives were to see if the neurotoxic exposure results in any observable sub-lethal effects and to find out if losses can be correlated to hive development. This was assessed with traditional methods and a novel, visual monitoring device.

Current achievements and future developments of a novel AI based visual monitoring of beehives in ecotoxicology and for the monitoring of landscape structures

“…. 3 Activity at the hive entrance following contact with imidacloprid (Gonsior et al, 2019) The detection of whether individual animals were drones, worker bees or different genuses like wasps and whether they were dead or alive could be achieved with the use of a cloud based multi network. It was also possible to detect whether bees carried pollen or not with a certain probability.…”

“…Eight hives were monitored, of which four were fed with 500 g of a sugar solution, including a concentration of 200 µg imidacloprid/ kg of sugar solution, for ten consecutive days. The control group was fed the same amount of sugar solution during that exposure period (Gonsior et al, 2019). In a separate proof-of-concept study, the possibilities of machine learning algorithms were explored to perform localization, classification and pose estimation tasks on video recordings (Marstaller et.…”