Examination of factors affecting seabird bycatch occurrence rate in southern hemisphere in Japanese longline fishery with using random forest

We analyzed the factor affecting bycatch occurrence rate. Random forest was applied to analyze.We constructed four models examining effect of species group, season, year, environmental factors, distance from the colonies, a lunar phase, and catch of fish. Our model was likely to be a statistically appropriate model because out of bags is an acceptable range though a little high. Dominant variables in common with analyzed four models were latitude, longitude, elapsed days from the first day of the year, number of observed hooks, species group, sea surface temperature in this study. Also year, cruise ID and lunar phase were dominant variable sin common with two to three models. Those variables would have the large impact on bycatch occurrence rate. Thus, it was suggested that those variables should be considered in the comparison between CPCs and in the collaboration work.