Some considerations were made about the minimal number of samples for obtaining population fluctuation curves of mosquitoes on the basis of light trap collections in a given area and in a week. (1) In dealing with these subjects, the mosquito counts were transformed into y=log(x+1) values to normalize the data, and seasonal fluctuation curves were expressed by the Williams' means (Mω=antilog y^^^--1) obtained from the data thus transformed. Two kinds of the fluctuation curves were obtained from the periodical sampling data : one kind of the curves was derived from each of the mean values Xi (i=1,2,…, 15,16) of log-transformed mosquito counts in the randomly and successively accumulated samples, and the other from the mean values Y (=X_<17>) calculated using the total sample. The similarity between the two kinds of fluctuation curves was compared by γ^2 values (coefficients of determination) between the two means, Xi and Y. The degree of agreement between Xi and Y was expressed by a and b values of the regression line, Y=a+bXi; and the precision of Y estimation by Xi was evaluated in terms of the ratio, σY・X_i/Y^^^- (%) (σY・X_i=standard error in Y estimation by Xi) (Figs. 1,2). As a result, about 8 sampling sites among the total of 17 were sufficient to obtain about 90% closeness of the two kinds of the fluctuation curves expressed by the Williams' means (Fig. 2). (2) The same procedure was applied to obtain the minimal number of samples in a week in a given sampling site. As a result, the light trap collections in three nights a week gave 90% closeness to the every-day sampling in the seasonal fluctuation pattern, when the trap was operated at about the same intervals (Figs. 3,4) (3) Discussions were given of the differences between the results in this paper and those reported so far with respect to the operation schedules of light trap in a week, from the view points of variability and convergency of each accumulated samples to the total ones.

抄録全体を表示