Syoiti Tanaka

Migration Model and Population Dynamics of Large-sized Yellowtails in the Pacific Ocean along the Japanese Coast Inferred from Tagging Experiments-I

A migration system model was constructed on the basis of information obtained from the tagging experiments. Two migration routes start from Sagami Bay and Kumano-nada area, central Japan. One is southward migration in spring in the direction of Kyushu, and the other is northward migration in summer to northern Japan. Recapture data within the year of release were applied to this system model and the fishing rate, transfer rate, and the rate of disappearance were estimated. In 1926-1938 period, about 80% of southward migrants disappeared from both Kumano-nada and Tosa Bay and discontinuity of the distribution of the population at these points was suggested. In 1963-1965 period the boundary at Kumano-nada disappeared and 85% of fish moved south from there toward Tosa Bay. More than a half of the fish which once disappeared from Sagami Bay and Kumano-nada in the spring season reappeared in northern Japan in summer. The fishing rate for southward migrants in the spring season was about 20% in 1926-1938 period, but in 1963-1965 the rate increased to more than 30% in Sagami Bay and Kumano-nada. The fishing rate was low in northern Japan being not more than 10% in both periods.

A Note on the Mathematical Model for the Estimation of Fishing Rate of Mojako by Means of Tagging Experiments on Drifting Seaweeds

The basic model for the number of tags N presented in the previous paper is modified, in order to take into account the fact that tags shed from seaweeds are not vulnerable to the fishery, and hence the coefficient of fishing κF can be applied only to retained tags NT, but not to the total tags N. The results from a modified model are compared with those from the previous one. It reveals that the results from the latter, which is very simple in expression, is a good approximation to that from the former and the error involved in the estimate of the coefficient of tag shedding M based on the previous model is thought to be negligible. This modification of the model does not affect the estimates of other parameters such as κF.

Estimation of the Coefficient of Tag Shedding from Drifting Seaweed

1) A trial for estimating the tag shedding rate from drifting seaweed was made. This was needed to evaluate the rate of exploitation of drifting seaweeds from the tagging experiments of them. 2) The maximum likelihood estimates were obtained for three groups in different years and/or areas of release by means of iteration. The precision of the estimates were also evaluated. 3) The values of the coefficient of tag shedding ranged 0.017-0.022 with the average of 0.019 which is close to 2% per day. The difference between years or areas of release was not demonstrated. The coefficients of variation of the estimates were roughly within the range of 20-30%.

Estimation of Fishiing Coefficient of Mojako by Tagging Experiments on Drifting Seaweeds-II

1) The fishing coefficient F, coefficient of drifting away D and rate of exploitation E of the drifting seaweeds are calculated for several experiments in each of which a substantial number of recoveries was obtained. 2) Most of the estimates of ακF are within the limits of 2 ?? 3% per day. Tag shedding rate M is estimated as 2% per day. If a is assumed to be one, many of the estimates of D fall within a range of 0.10 ?? 0.15, though extremely deviated values of less than 0.01 or more than 0.3 are also obtained depending upon the areas and periods of release. 3) The rate of exploitation is estimated at 7% at its minimum and 73% at its maximum. Fishing for mojako may have a substantial effect on the part of stock which enter into the fishing grounds. 4) Estimates of F obtained seem to be fairly reliable, But the effect of fishing on mojako stock would be more weak than that estimated here if mojako are also distributed outside the fishing grounds and some of them are migrating independently of seaweeds.

Estimation of the Catch of Autumn Chum Salmon in Otsuchi Bay in 1959

1) The catch of autumn chum salmon (Oncorhynchus keta) during 1959 season was estimated for 18 trap nets set in the Otsuchi Bay, Iwate Prefecture, northern Honshu (Fig. 1). Data were obtained from 10-day reports of daily catch submitted from each net. For the 10-day periods where reports were not available, statistics kept at the Otsuchi fish mar-ket were applied with correction by the mean percentage of the catch of each net sold at the market. 2) The total number of fish caught from early October to late December amounted to about 20, 000. The catch in Otsuchi and Kotsuchi Rivers (escapement) in the same period was about 5, 000, which is about 20% of the total return. (Table 1). 3) The percentage of the catch sold at the market varied from net to net, some having values of almost 100% and some almost zero. There was a trend for the percentage to decrease during the fishing season. (Fig. 2). 4) There was a resemblance in the pattern of the seasonal change of the catch between adjacent nets (Fig. 3). 5) Although in some cases the percentage sold at the market exceeded 100%, most of these ridiculous values occurred in the beginning or the end of the season and the catch in-volved formed only a negligible part of the total. And this inaccuracy would not negate the validity of the entire results. 6) As no direct evidence was given to verify the reliability of the fishermens reports, there may be a possibility of erroneous estimates. However, the above mentioned circum-stantial evidences suggested that the actual catches were represented fairly well in the re-ports.

Meaning of Mode in a Length Distribution of Fish-I

The purpose of this paper is to explain usefulness of the mode which appears in a length distribution of fish for separating a mixed group into homogeneous subgroups, e. g., by age. A mixture of two subgroups each of which distributes normaly with difference of means d, the ratio of standard deviations O and mixing ratio A are considered. When d becomes larger than a certain limit d0, two modes would appear very close to the respective means (Fig. 2). Ratio of the modal values are also near to A/O (Table 1). Sometimes, though d is less than d0, existence of heterogeneous group is indicated through a bump in the skirt of the distribution curve. These facts show that the mode can be made use of in isolating a mixed group into subgroups. An effort has been made to obtain the conditions of parameters which enable to separate the mode and have the bump appear (Fig. 3, Fig. 4). Furthermore, means and areas which were calculated for parts of the distribution curve separated at the point of minimum value, have been compared with the locations of modes and the ratio of modal values, respectively (Table 2, Table 3). As regard to the biases of estimations for the mean values of respective subgroups and the mixing ratio, the modes excel means and areas for the parts. However, it is necessary to discuss the varidity of the mode from the standpoint of sampling error, the problem which should be studied in future.