Fig 1.
a) Stylized landscape (green patch seminatural grassland and blue patches oil seed rape). The total area of the patches is 79 hectares, the beehive is located in the centre of the landscape. b) This stylized landscape would translate into a short period of mass flowering of the oilseed rape and an assumed background resource available around the year from the seminatural grass land (black line) or during a reduced duration (Environmental stress scenario, red line). The amount of pollen which is not shown here shows the same temporal dynamics.
Fig 2.
Persistence P measured as the percentage of honey bee colonies (n = 100) that had more than 5000 adult bees at the end of a simulated years for three scenarios.
Fig 3.
Correlation between outputs of the simulation model for a) the Reference scenario and b) the difference of correlations between the Reference and the Environmental stress scenario. A positive value in the correlation matrix means that the correlation was stronger in the Reference scenario than in the Environmental stress scenario.
Table 1.
Overview if and when the averages of 100 simulation runs could be distinguished between the reference and the Environmental stress scenario.
Stress was detected earliest for the number of pupae (Tdisc = 86 d) and even earlier for the brood-bee ratio and could not be detected at all for the number of flights and number of mites. We also reported the day during the first year of Environmental stress when the maximum difference between average values occurred Tmax and the maximum normalized reduction Rmax at this day (see text for details).
Fig 4.
Temporal dynamics of five response variables and the brood-bee ratio for the Reference scenario (black line represents the mean, grey area represents the 90% confidence interval), the Biotic stress scenario (blue line, 90% CI is not shown to increase readability) and the Environmental stress scenario (red line represents the mean and the orange area represents the 90% confidence interval).
N = 100 runs for all three scenarios. Shown is the first year in which the stress was introduced (after a burn in phase of two years) to assess whether stress can in principle be detected at an early stage.
Fig 5.
Number of pupae over time and as a function of cumulative foraging hours.
Fig 6.
To check the robustness of the results the Environmental stress was applied at random days, i.e. days where the background resource from the grasslands was not available is selected randomly from all 365 days.
We varied the number of stress days from 5 to 365 and repeated the simulation 100 times for all numbers. With number of stress days, the maximum reduction of pupae was steadily increasing. The day when the reference scenario could be discriminated from the Environmental stress scenario was declining with the number of stress days.