- Group Name: Honey bee
- Group participants names: Gugler Stefan, Huwyler Elias, Tschopp Fabian
- Project Title: Hive Simulation
Honey bees (Apis sp.) are essential insects for pollination and are therefore associated with our food supply. The pollination rate again is connected to the health and size of a hive. It is important to know under what conditions a hive can grow and how much it takes for a colony to become extinct. Most research done so far has been experimental and there are numerous parameters to take into account.
Bees have an utterly complex social system. In any case, it is not possible to take all aspects into account. Some of them won't be used at all, while others are simplified with probability variables from real world measurements [1]. Our first approach is largely based on an existing simulation which works with a dynamic system and continuous update inside the hive [2]. A reasonable update rate for this is one up to a few hours. Simulation length is from a few months to a few years (simulated time). Parameters and variables taken from the existing simulation [2]:
- Brood rate
- Adult bees emerging
- Rate of change of foragers
- Recruitment of foragers
- Rate of change in food stores
- Total hive weight (depending on 5.)
- Indicators for hive healthiness (depending on 1. and 5.)
- Forager mortality
- Sudden unexpected events (increased death rate, infections)
The second approach adds a (graphically implemented) environment simulation to refine the processes. This will mainly affect parameters 5. and 8. (first approach) by substituting the fixed rates with values obtained by fine-grained environment simulation (time resolution of approximately one second per iteration step and one day per trigger). Environment simulation depends on parameters 1. to 9. of the first model, which can trigger a significant re-evaluation of the environment to redefine parameters 8. and 5.
This hybrid approach seems to be a good model because fixed rates [2] are too abstract for such a complex model [1], but evaluating all parameters and aspects becomes too complicated for meaningful statistical analysis.
Basic questions:
- Statistical analysis of the existing model assuming fixed rates
- Hive survival probability over years under different initial conditions
- Hive growth rate and healthiness
Advanced questions:
- The effect of smog on food supply/pollination
- The effect of smog on forager bees (higher death rate)
- Hive growth and relation to food availability (seasonal changes, different environment settings)
- Differences between a hive-only simulation and an environmentally interacting model (population dynamics depends on environment)
It is expected that our simulation will at least reach similar results as the dataset from the reference paper used [2]. This answers our basic questions. Furthermore, we expected using a dynamic environment will give similar results for the total weight of the hive as found in the reference book [1].
[1] Seeley, T. 1995. The wisdom of the hive - the social physiology of honey bee colonies
[2] Khoury et al. 2013. Modelling Food and Population Dynamics in Honey Bee Colonies
[3] Karaboga, D. 2005. An idea based on honey bee swarm for numerical optimization
[4] Girling, R. 2013. Diesel exhaust rapidly degrades floral odours used by honeybees
[5] Khoury et al. 2011. A Quantitative Model of Honey Bee Colony Population Dynamics
The bees in the environment outside the hive are simulated with an agent-based model (continuously updated). The environment itself (outside the hive), in which the bees will navigate, is based on a 2D-cellular automaton which is updated sequentially with fixed borders (scent and smog diffusion, flower growth). Inside the hive, a simplified model is used, assuming all parameters at first and including the parameters of the environment simulation later. Dynamical systems (SIR Model, ageing and reproduction rate, Lotka-Volterra equations) are used inside the hive.
Datasets for comparison and model refinement:
- Experimental data from measurements [1]
- Simulation data from already existing simulation [2]
- Advanced parameter datasets for scent and smog [4]