Figure 4: Spatial assessment of the honeybee experimental homing failure, as delineated by the ED₂₀ (effective dose that leads to a 0.20 mortality rate due to homing failure).

Mortality due to homing failure, m_HF, is given by the relative difference of control versus treated homing probabilities. Conditional predictions show the sharp transition from (a) low-effect conditions at 28 °C to (b) high-effect conditions at 20 °C. Thick lines show the 1-ng ED₂₀ limits, delineating the landscape contexts where honeybees exposed to a 1-ng dose will be subject to a mortality m_HF >0.20. This area covers 34–75% of the whole landscape depending on the temperature (c), but may be much reduced if one focuses on lower exposure scenarios. Eventually, for doses <0.320 ng per bee at 20 °C, the landscape is expected to be exempt of critical homing failure risks (ED₂₀ landscape coverage=0%). For the sake of comparison, horizontal lines indicate worst-case scenarios of thiamethoxam intake by honeybees for 1 h flying, or to forage a day, based on 20–40% sugar nectars²⁶. In particular, the per-hour exposure scenarios (for example, 0.184–0.276 ng per bee for a 20% oilseed rape nectar) fall in the 0% range for ED₂₀ landscape coverage. (d) ED₂₀ conditional predictions use the local geographic information on the network of hedgerows and forest edges.