A deeper understanding of system interactions can explain contradictory field results on pesticide impact on honey bees

Abstract

While there is widespread concern regarding the impact of pesticides on honey bees, well-replicated field experiments, to date, have failed to provide clear insights on pesticide effects. Here, we adopt a systems biology approach to gain insights into the web of interactions amongst the factors influencing honey bee health. We put the focus on the properties of the system that depend upon its architecture and not on the strength, often unknown, of each single interaction. Then we test in vivo, on caged honey bees, the predictions derived from this modelling analysis. We show that the impact of toxic compounds on honey bee health can be shaped by the concurrent stressors affecting bees. We demonstrate that the immune-suppressive capacity of the widespread pathogen of bees, deformed wing virus, can introduce a critical positive feed-back loop in the system causing bistability, i.e., two stable equilibria. Therefore, honey bees under similar initial conditions can experience different consequences when exposed to the same stressor, including prolonged survival or premature death. The latter can generate an increased vulnerability of the hive to dwindling and collapse. Our conclusions reconcile contrasting field-testing outcomes and have important implications for the application of field studies to complex systems.

Fig. 1. The health of honey bees as influenced by multiple factors and their effects.

In the conceptual model of bee health bar-headed lines denote negative effects while arrow-headed lines indicate positive ones. See text and Supplementary Table 2 for explanation of lettered effects.

Fig. 2. The equilibria and some orbits of the full system in the projected phase plane of honey bee health (x_HB) and level of viral infection (x_VI).

Equilibria represent the values of the state variables where they do not change and are indicated with dots, while the orbits are the values that the state variables can assume while approaching the equilibria and are represented with lines. a Orbits and the unique equilibrium without immune-suppression, in presence of a low level of parasites. b Orbits and the unique equilibrium without immune suppression, in case of a high level of parasites. c Orbits and the three equilibria with immune-suppression; two orbits exiting from close initial conditions are marked with thick lines. d Equilibria of the subsystem of bee and virus for increasing immune-suppression. p is a function of the level of viral infection v that vanishes at equilibria; top curve: at low immune-suppression there is one equilibrium at high bee health; bottom curve: at high immune-suppression there is one equilibrium at low bee health; intermediate values of immune-suppression can cause three equilibria.

Fig. 3. Distribution of individual lifespans of honey bees under different conditions.

a Early in the season when the prevalence of an immune-suppressing virus is low (white bars) and later when all bees are virus infected (gray bars). b Treated or not (gray and white bars, respectively) with a virus administered to mature larvae through the diet. c When exposed to a toxic compound, when the prevalence of an immune-suppressing virus is low (white bars with diagonal pattern) or when the virus is widespread (gray bars with diagonal pattern); the corresponding distribution of honey bees sampled early or late in the season and not exposed to the toxic compound as a control (white and gray bars, respectively). d As (c) but exposed to a sub-optimal temperature in place of a toxin. Source data are provided as a Source data file.

Fig. 4. Dependence of the colony population at equilibrium on the death rate m of forager bees for varying death rate n of juveniles hive bees.

For a forager death rate m exceeding a critical value (black dot) the only stable equilibrium is zero, corresponding to colony failure. The premature death of hive bees (denoted by increasing values of n, represented by the blue curves) moves that critical value left, meaning that colony failure can occur for lower foragers’ death rates. Black line: n = 0, blue: n $\in$ (0, 1), dots: m(n). The parameter values are L = 2000, w = 27,000, α = 0.25, and σ = 0.75 as in a previously published report.