Over-Dispersion in Grassland Communities and the Use of Statistical Methods in Plant Ecology

Abstract

The data presented by T. L. Steiger in "The Structure of Prairie Vegetation" [see B. A. 6(2): entry 3393] are studied statistically to determine the mode of distribution of the recorded spp. Of 44 spp. tabulated, the relative variance exceeds 1 in 42 cases, and is less in but 2 cases. In random dispersion the relative variance equals 1 and the excess over 1 indicates that most of the spp. were less uniformly dispersed than random and tend to be patchy. The relative variance tends to increase with increase of the mean density of each sp. When the quadrats were divided into classes having 0, 1, 2, 3, etc., individuals of a sp. per quadrat, there was also a deviation in the size of the classes from what would be expected from random dispersion. And the deviations increased with mean density. The author concludes that in the prairies of E. Nebraska the individuals of most of the spp. show strong over-dispersion. He suggests this may be due to the methods of reproduction of the spp., as vegetative propagation tends to form denser patches around the older plants which have a greater than mean density. He also suggests that over-dispersion may greatly increase the difficulty of determining the mean density of the different spp. from a set of random quadrats.