BAYESIAN HIERARCHICAL ANOVA MODEL OF STOCHASTIC SEASONALITY FOR Diodon holocanthus IN NORTHERN TAIWAN

Detecting seasonal variation in ecological phenomena is not always possible by use of traditional methods. The pattern of seasonality fluctuates because of the stochasticity of environmental factors, correlations among organisms, and human activity. Here, we propose a Bayesian hierarchical analysis of variance (ANOVA) model of stochastic seasonality for detecting the annual fluctuation of seasonality. Using 11-year data recorded monthly at 2 nuclear power plants in northern Taiwan, we examined monthly fluctuations in fish species. To illustrate the performance of the proposed model, we conducted a simulation that reflects the varying seasonality of species. The most dominant species, Diodon holocanthus, exhibited a shift in peak abundance from March to July that the traditional ANOVA model failed to detect. Our results suggest that impingement data obtained from the intakes of power plants are useful in studying the long-term temporal variation of fish species. The proposed model provides an in-depth examination of temporal variation for ecological analysis.