Our study used the RFID-technology to analyze the impact of insecticide compounds on honeybee foraging behavior. Decourtye et al.
already showed that fipronil at doses of 0.3 ng/bee reduced the number of foraging flights to the feeder and prolonged the duration of the homing flight 
. These effects were observable for up to three days. Similar effects were found for the compounds used in this study. In contrast to Decourtye et al.
, who conducted their tests under semi-field conditions, the described trails were conducted under field conditions, allowing the test colonies to normally provide themselves with necessary additional resources including pollen, water, and propolis. Furthermore, since the test is designed to detect effects on individual bees, the tested bees were fed defined amounts of the pesticide individually with the differently spiked sugar solutions instead of bulk feeding them in a cage.
By obtaining data describing a minimum of six different parameters of foraging behavior, sub-lethal effects for both substances, imidacloprid and clothianidin, used in this study were detected. Impairments were detected at doses of 1.5 ng imidacloprid per bee, which would equal a concentration of around 115 ppb (parts per billion) in nectar. These results are in agreement with previous studies, which tested the effect of imidacloprid on homing and foraging behavior 
. Yang et al. 
found effects on foraging behavior at concentrations as low as 50 µg/L (40–50 ppb). These documented concentrations are still more than twenty-five to fifty times higher than the residues found in the nectar of sunflowers (Helianthus
, 1.9 ppb) 
. Treatment with the lowest dose of imidacloprid (0.15 ng; 11.5 ppb), which is about five-fold higher than any residues found in nectar, had no recognizable effect on foraging behavior. Nevertheless, bees may be exposed to almost 100-fold higher doses than tested in our trials, as shown in a study concerning the insecticide contamination of guttation drops, xylem fluids that are excreted at the leaf margins 
in seed dressed crops. It remains unclear, though, if water foragers collect these fluids in the field.
This is the first study on foraging behavior of honeybees that presents sub-lethal effects after acute oral treatment with clothianidin. Dosages of 0.5 ng (38 ppb) negatively influence the foraging behavior and low dosages (0.05 ng; 3.8 ppb) can have effects on certain aspects of foraging behavior even if they did not have any significant effects on the number of feeder visits or on the total foraging time. Clothianidin elicited detrimental sub-lethal effects at somewhat lower doses (0.5 ng/bee) than imidacloprid (1.5 ng/bee). Bees disappeared at the level of 1 ng for clothianidin, while we could register the first bee losses for imidacloprid at doses exceeding 3 ng. This indicates a stronger impact of clothianidin compared to imidacloprid, which is in agreement with previous reports that both oral 
and contact toxicity 
levels are lower for clothianidin.
Both neonicotinoid insecticides are known to be partial agonists for different types of the insect nicotinic acetylcholine receptor (nAChR) 
. In vitro experiments that observed the action of imidacloprid and clothianidin on native nAChRs of cholinergic neurons from Drosophila
and nAChRs of the terminal abdominal ganglion neurons of the American cockroach 
showed greater agonist efficacy of clothianidin compared to imidacloprid. A similar effect on cholinergic neurons in adult honeybees could be an explanation for our findings that clothianidin elicits detrimental effects at lower doses than imidacloprid.
In general, both substances led to similar effects on the observed foraging parameters. The only difference was found when investigating the flight time to the feeder. Bees treated with imidacloprid needed more time to fly to the feeder while no effect on this phase was observed after administration of clothianidin. Relating this to the symptoms observed after administration of higher doses of both substances it could be another indication for differences in their efficacy. Higher doses of imidacloprid (≥3 ng) led to reduced movement, eventually leading to immobility and trembling movements, which is in accordance to previously reported doses of ≥2.5 ng by Lambin et al. 
. This might have occurred to some degree in the lower doses as well, but escaped visual observation. Higher doses of clothianidin (≥1 ng), in contrast resulted in an arched abdomen, which did not reduce the mobility of the bees like imidacloprid did. Girolami et al. 
reported that when their abdomens were arched, the bees still retained their flying capability. Taking this into consideration, it could explain why the detrimental effect on flight behavior by clothianidin was less pronounced compared to imidacloprid, though still significant compared to the controls. Furthermore, it might be hypothesized that imidacloprid and clothianidin aim at differing targets, i.e. different subtypes of receptors located on pre-motoneurons and motoneurons of honeybees, though their cell physiological actions are still unknown. This was already shown in a study of Thany 
for DUM-(dorsal unpaired median) neurons isolated from the cockroach Periplaneta americana
. Here clothianidin was found to act on imidacloprid-sensitive and -insensitive nAChR subtypes. DUM-neurons are known for their neuromodulatory role in altering the performance of motor patterns and are thought to be homologous to VUM (ventral unpaired median)-neurons 
of honeybees because of their similar morphology and physiology.
An increase in motor activity observed in a study by Lambin et al.
after topical application of 1.25 ng imidacloprid and subsequent introduction into an open-field-like apparatus 
. The orally administered dosages of imidacloprid in our study did not seem to increase motor activity at the feeder site.
Both substances led to a longer 1st and 2nd period of stay inside the hive before returning to the feeder. This is likely due to a prevailing toxic effect on the bees while they were inside the hive. They remained in the hive until the effect ceased and they were able to fly out again. This is consistent with the fact that bees that did not return to the hive after treatment were not registered again, but the majority of bees that made it back to the hive returned to the feeder later on.
In conclusion, we think that the method of RFID aided feeder tests could be of considerable help concerning risk assessment of insecticides. Subsequent to initial mortality tests to determine the LD50 of an insecticide, the RFID-method could be used in field-like tests to investigate possible sublethal effects of doses thought to be non-hazardous for honeybees. Furthermore, we think that the sensitivity would be increased considerably by higher distances between hives and feeder, which will be explored in further experiments.