The nutrient effluent load to Tokyo Bay has been decreasing in recent years due to the enactment of the Water Regulations. However, it is difficult to establish whether there is any direct effect of the reduction in nutrient loading on red tide frequency. In the present study, I constructed the model using parameters of the red tide criteria (red tide index) and assessed the suitability of this model to predict future red tide occurrences in the Tokyo Bay based on changes in the coefficient of determination (R
2) of the model. Using the stepwise method, I developed a stable multiple regression model with a high R
2 for data from 1986 to 1998 from a computerized database. Particulate organic nitrogen (PON) , an index of red tide, was used as the target variable. The adjusted R
2 (R
2adj) of the model based on data from 1986 to 1998 was 0.772. Application of this model to new annual data from 1999 to 2003 could be well explained by the model based on previous data. However, from 2004 when the regulations were implemented, there was a substantial decrease in R
2, suggesting that environmental changes arising from the introduction of total loading control may affect the model developed in this study.
View full abstract