Density estimation of the giant jellyfish Nemopilema nomurai around Japan using an alternative modified detection function for left truncation in a line transect survey

Abstract

Information on the distribution and bloom scale of the giant jellyfish Nemopilema nomurai is necessary to understand the impact on fisheries. In October 2009, we had an opportunity for visual counting of N. nomurai around Japan from a research/training vessel. To take into account possibly severe missed detection of N. nomurai in the close vicinity of the vessel, detection functions were modified by introducing an additional parameter δ expressing the proportion of detectable N. nomurai within a threshold perpendicular distance. Some covariate effects in the detection process were also investigated. The δ-modified hazard-rate detection function with δ variation parameter of jellyfish body size was chosen as the best-fit one by AIC model selection. The results suggested that a higher proportion (>30 % or more) of N. nomurai were undetected in the vicinity of vessel. The δ-modified model had an advantage of no loss of left-truncated data and provided better estimation of density with the smaller coefficient of variation than the conventional left truncation approach. It is applicable to other line transect surveys, e.g. aerial surveys. High densities of N. nomurai were found in the central part of the Sea of Japan and off Iwate Prefecture.