ABSTRACT
Within a multi-species study, species do not necessarily represent independent data points. The data set used by Foster and Janson (1985) to look at the relationship between seed size and establishment conditions for naturally occurring tropical trees was re-analyzed, to take into account the effect of relatedness among species. The re-analysis showed that 1) this data set does not support the hypothesis that large-seeded species are more likely to establish in small gaps or shade than are small-seeded species, and 2) more than 1/3 (16 of 39) of the data points were extraneous to the test of the hypothesis. It is recommended that all ecologists, and not just those interested in evolutionary questions, consider species relatedness prior to inception of any multi-species study.