Variance Tests for Species Association

Abstract

Schulter recently discussed species association tests for handling situations where the data consist of a two—way table (species by samples) of either presence/absence or density data. In this paper, connections between those association tests and standard statistical tests are established and used as a guide to proper interpretation of the association tests. Association tests are sensitive to purely sample—to—sample differences and in some cases will not reflect species interactions at all. However, in cases where the association tests are appropriate, the connections furnish information on the choice of accurate critical values.