Self-Organization of working bees on a honeybee comb

The focus of our project is to characterize the self-organizational components of honey bee colonies, and to describe how they adjust their resource management and nursing efforts precisely and quickly to changes in the supply and demand status of the whole hive. This ability is impressive, because it results (collectively) from behavioral changes by thousands of individual actors. Neither do workers inspect the whole hive to inform themselves about the current colony status, nor are there any special-ized organizer-bees doing that. Our recent behavioral studies have shown that nurse bees quickly adapt their nursing-strategies if the pollen influx decreases (e.g., in bad weather). But the question of how this process works remains unanswered. We assume that a set of simple behavioral rules, responding to the flow of information via interadult-feedings and nurse-to-brood feedings, searching-time and meeting probabilities, function to self-organize the nurses. We assume that also the comb, acting as a communication platform, message pad and memory bank, plays an important role. Our project will include a set of laboratory experiments with different cohort sizes and age distributions of bees and brood in combination with different levels of resource influx. We will separate a set of nurses of larvae from the resource areas to make them solely dependent on the flow of information through interadult contacts. The behavioral patterns of the nurse bees as well as their nursing acts will be automatically processed by advanced video processing system and by manual (human interpreted) video observations. These data will be complemented by physiological analyses (weight, protein content) of all actors (adults and larvae) at the end of the experiments. Observations of nurses in the full-hive will collect additional data in the full-society context. Using this information, we will implement a multi-agent model of a brood comb. It will model a set of nurse bees, resource income and deposition, resource consumption, interadult-feedings and nursing of a cohort of brood. This model will be used to perform simulation experiments to predict brood survival and the self-organization of nurses under different conditions. We will implement the multi-agent model based on a framework (also part of the project), that explicitly allows and supports the building of comb-based multi-agent models. It will take into account the special "infrastructure" and "services" the comb offers as an environment, as a place of storage and brood-growing, as a collective memory and as a platform for indirect and direct communication. In recent years, studies of self-organizational processes in ant colonies have led to the invention of the so called ANT-algorithms and Ant-Colony-Optimization algorithms that are now used widely in regulating networks such as telecommunication or postal services. In addition to contributing to the fields of social insect and bio-modeling research, we hope that our mathematical characterization of the ability of bees to collect, process, store and distribute a variety of resources in their highly regulated society might also lead to new inspirations in industrial, technical and commercial fields.

The goal of our project was to investigate self-organisation of worker bees an a comb. This topic was addressed by a series of experimental approaches, by a series of computer modelling and by simulation. In a first step we explored how much a honeybee colony can be reduced in terms of its complexity without loosing the ability to rear brood. After a series of experiments, we succeeded in creating extremely small honeybee colonies consisting of 120-180 bees, kept under total video observation. Using this "dwarf-hive", we successfully observed brood rearing in colonies that were under full video control. In this experiments we tested weather brood that was deprived from food is preferentially fed and if pollen deprived nurse bees feed brood less than saturated nurse bees. The results were implemented into a computer simulation of self-organised brood rearing in honeybees, which is the first of its kind so far in the world. In addition to the experiments with "dwarf hives", we conducted successful experiments in the context of a full- sized Observation hive (several thousands of bees). We investigated how reduction of workforce (removing half of the nurse bees) and reduction of workload (removing half of the brood) affects the nursing behaviour of nurse bees and`the allocated nursing time received by larvae. A spin-Off of these experiments was the development of a method to determine the age/size of larvae from photographs. In an addition experiment, we investigated how deprivat on of pollen affects trophallactic contacts among adelt bees. This is important to understand the rote of trophallaxis - the mouth-to-mouth transfer of food between adult bees - in the process of the regulation of pollen stores in a honeybee colony. We successfully implemented computer simulations of self-organised process that work within the honeybee colony, with focus an communication about food sources, an trophallactic interactions among adult bees, an division-of labour that is regulated by several sets of intra-colonial stimuli and an the self- organised regulation of brood nursing. These simulation experiments led to several publications and were also presented at international conferences. As a first derived technical applicationof the project`s outcome, a`honeybeeinspired control-algorithm for robot swarms was developed which is based an inter-adelt feedings of honeybees. A second derivation from that project is a honeybee-inspired optimisation-algorithm that is intended enriches the field of bio-inspired optimisation (Genetic Aigorithms, Evolutionary Strategies, Ant-Golony- Optimisation, ANT-Aigorithm, ParticleSwarm-0ptimisation).