I checked the Ecolog archive and couldn't find an answer to this question (see subject line). In a factorial experiment on forest tree regeneration, I have a number of plots where abundance, and therefore species richness is zero. I want to use ANOVA to test for treatment effects on diversity, but the normal calculations result in a missing value for Shannon diversity, being that ln(0) is undefined. Simply replacing missing values with 0 is misleading, because that is the value of H' when there is one species. Throwing these plots out is also undesirable, as having 0 species is useful information. My thought is to replace the missing value of H' with -3 or some other negative number, which would "back-transform" with exp(H') to a number close to zero. This "corrected" H' would only be used in reporting means and testing for effects, not for describing any particular plot.
Does anyone have any comments on this approach, or a suggestion as to a better approach? Please send answers to me: [EMAIL PROTECTED] Thanks, Tim -- --------------------------------- Tim Nuttle, PhD Research Assistant Professor Department of Biological Sciences University of Pittsburgh A234 Langley Hall 4249 Fifth Avenue Pittsburgh, PA 15260 USA ---------------------------------
