1996 年 46 巻 3 号 p. 269-289
The influences of sample size on the index values of species diversity were examined for various indices which have been used hitherto in community studies, together with several indices newly proposed in this paper. Samples of various sizes ranging from 50 to 4,000 individuals were taken randomly from each of nine types of artificial communities which were set up using paper tips, each representing an individual, and the calculated index values of these samples were compared with each other for each community. The paper tips were made by cutting a card board in about 1×1 cm size. The indices which were least affected by sample size were divisible into three groups. The first group included the β index and allied ones. The indices of the second and third groups had values corresponding to the square root and the logarithm of the respective index values in the first group. The first and the second group satisfied the following quantitative relationship, which has preferable characteristics of diversity index : diversity=richness×evenness. A new method was proposed for estimating the total number of species in the mother community from a sample. The results of comparison between the estimated and actual numbers of species in artificial communities showed that the method might be effective for practical use. The samples of artificial communities were compared with the samples of natural communities, and a number of examples which showed fairly good similarity of structure were found in both communities. It is suggested that not only in artificial communities but also in natural ones, the number of species found in a sample would not reach half the number of species in the mother community when the sample size is smaller than 100 and the value of the β index is larger than 10.