An Adaptive Modeling Technique for Instream Fish Habitat Preference of Japanese Medaka (Oryzias Latipes)

Abstract

It is widely known that habitat selections of riverine fish differ wlthin and between rivers. In our past study, the preference intensity of Japanese Medaka (Or~/zias latipes) to three environmental factors of water depth, current velocity and cover ratio was quantified on laboratory open-channel experiments for developing a general habitat preference model. A simplified fuzzy reasoning method was introduced in consideration of essential vagueness of fish behaviors. The fuzzy preference intensity model was then optimally searched with a simple genetic algorithm and was successfully verified by both laboratory water tank experiments and on-the-spot examination. The results indicated that this general model showed agreement between predicted and observed spatial distribution of target fish, but the habitat preference models are still desired to be developed through field studies. In this study, we propose an adaptive modeling technique for instream fish habitat preference by conjugating the fish preference intensity model developed in laboratory experiments. The adaptive prediction model was also deterrnined by simple genetic algorithm, which enabled us to model the habitat preference of instream resident fish even with insufficient data.