Size bias in traditional analyses of substitutive competition experiments.

The objective of this study was to examine the claim that traditional measures of competitive performance in substitutive experiments are biased towards larger plants. Results from a three-year diallele experiment of 6 marsh plant species were analyzed using both Relative Yields (a traditional analysis) and the Relative Efficiency Index (a recently proposed analysis presumed to be size-independent). In adddition, a mechanistic model of competition was used to explore the behavior of both methods of estimating competitive performance.Results from the three-year experiment showed that Relative Yields (RYs) were correlated with the initial sizes of plants for the first two years but not the third. By the third year, RYs were highly correlated with Relative Efficiency Index values (REIs) suggesting that the effects of initial size were eventually overcome. Model results showed that RYs are inherently biased in favor of larger plants during the early phases of competition while REIs are not. Further, model analysis confirmed that the size bias associated with RYs declines with increasing duration of the experiment. It is concluded that current generalizations about the relationship between plant size and competitive ability may be biased by the procedures used to analyze competition experiments.