Abstract
Some considerations were made on the conventional interpretation of results of mosquito sampling with light traps, on the basis of the statistical analysis of the distributional patterns of adult populations. 1) The statistical analyses were made on the relationship between means and variances calculated from field data of three dominant species sampled at 17 sites in Nagoya City (Fig. 1) during the period from June to September 1972. It was revealed that spatial distribution of individuals of each species seemed to be an aggregated type and to fit well to the negative binomial distributions with a common parameter k (Fig. 2 and Table 1). 2) Comparison was made among three methods of transformation, log(x+1), log(x+k/2) and √<k>sin h^<-1> √<kx>, in their efficiencies of stabilizing variance and normalizing variate distribution (Fig. 3). The three transformation methods were compared also in their practical usefulness. As a result, the log(x+1) transformation was adopted because of its facility in practical use and its adequate efficacy in normalizing data. 3) The log(x+1) transformation was applied to the field collection data of Culex pipiens pallens Coquillett, and several advantages were discussed from the viewpoints of normality of observed values and effectiveness in the significance test of means. Williams' mean (Williams, 1937; Bidlingmayer, 1969) was proved to be more suitable than arithmetic mean in interpreting fluctuation of population size in an area as a whole and in comparing abundance of mosquitoes between sampling sites (Figs. 4,5,6 and 7). 4) The relationship was analyzed between seasonal fluctuation pattern of C. p. pallens in all the sites and that in each of these sampling sites (Fig. 8). As a result, it seemed very difficult to estimate fluctuation pattern in the whole area from a result in a certain sampling site; so that it will be necessary to take sufficient samples for the consistent estimation of population fluctuation in the whole area.
