The present study examined a preliminary analysis for improving model structure of fuzzy habitat preference model for Japanese medaka (Oryzias latipes) dwelling in agricultural canals in Japan. The present model employed a simplified fuzzy reasoning method for evaluating habitat preference of the fish based on the relationship with physical habitat characteristics observed in the field survey. The model parameter was optimized by using a simple genetic algorithm, in which number of fuzzy

more »

... hip function was fixed. In the present analysis, number of fuzzy membership function was changed while the other methods were fixed as the original model. The model performance was evaluated based on mean square error between observed and predicted fish population density, and by using two different data sets. As a result, there was no clear tradeoff between number of fuzzy membership functions and prediction accuracy. By contrast, calibration and validation results showed a slight tendency of tradeoff. Further studies on clarifying the tradeoffs would be necessary for improving the model structure in an effective way.