1997 年 63 巻 5 号 p. 708-714
Based on the separation of different cohorts from compound multiple normal distributions, a method is proposed which can simultaneously analyze multiple length frequency data sets even when there is fluctuation in the interannual growth rate. Each parameter can be estimated independently by data set or as common for some or all data sets. Parameters can also be selectedas unknown or can be fixed. Through simultaneous analysis of multiple data sets, accurate estimationscan be carried out allowing utilization of all information inherent in the data. Carapace-length frequency data sets of male Japanese spiny lobster Panulirus japonicus sampled at Wagu, Mie Prefecture, Japan are analyzed as an example. Data for 38, 486 individuals sampled over five fishing seasons (October to April) from the 1990-1991 to the 1994-1995 season are analyzed. Judging from the comparison of AIC values, optimality of the model increased through the introduction of variations to allow for interannual and seasonal growth fluctuations, shifts in the standard deviation with growth, and the totalmortality coefficient for older age groups.