Analysing Spider Web-building Behaviour with Rule-based Simulations and Genetic Algorithms

We analysed the web construction-behaviour of an ideal orb web-building spider using a computer model that constructed artificial webs with a Rule Based Simulation and optimised them with a Genetic Algorithm (GA). The GA simulated adaptation towards an adjustable ecological niche in a population of artificial (cyber) spiders. Web construction behaviour in each cyber spider was controlled by a set of rule parameters (encoded in artificial genes) resulting in different web shapes. We then statistically compared the best cyber webs from our optimisation runs with real webs ofAraneus diadematusbuilt under controlled, experimental conditions in the laboratory. Web characteristics like spiral distances, eccentricities and vertical hub location could to a large degree be accurately simulated with the model. In addition, falsification of working hypotheses using the model pointed to incomplete or wrong assumptions about the behaviour under study which became apparent when web geometry adapted poorly.