An Artificial Neural Network for simulating Predator-Prey Relationships in an environment based on the concept of camoflage. This work, entitled "EVOLUTION OF PREY POLYMORPHISM INDUCED BY LEARNING PREDATORS", was published in Journal of Biological Systems and can be found at (https://doi.org/10.1142/S0218339011003944).
A prey species using crypsis to avoid predators has the opportunity to evolve polymorphic crypsis when it is being exposed to two (or more) habitats with different backgrounds. Here, we investigate when this phenomenon can occur, in a simulation study with a sexually reproducing prey and a predator that can learn to find hiding prey, represented by an artificial neural network. Initially, the prey is well adapted to one habitat, but tries to expand its range by invading another, different, habitat. This can cause the prey to evolve toward an intermediate phenotype, equally cryptic in both habitats. The prey can also fail in adapting to its new environment, and stay the same. Alternatively, it can evolve polymorphic crypsis. We find that the evolutionary outcome depends on the amount of dispersal between the habitats, with polymorphic crypsis evolving for low dispersal rates, an intermediate phenotype will evolve for intermediate dispersal rates and no adaptation to the new habitat will occur for high dispersal rates. The distribution of phenotypes of the prey will also vary for different dispersal rates, with narrow distributions for low and high dispersal rate and a wide distribution for intermediate dispersal rates.
Keywords: Crypsis; Artificial Neural Network; Heterogeneous Environment; Dispersal; Local Adaptation