A statistical view of synthesizing patterns of species richness along productivity gradients: devils, forests, and trees

Robert Whittaker (2010) offers a critique of quanti tative research syntheses attempting to generalize species richness patterns along gradients of productivity. As he says, the results of such syntheses have been controver sial and disagreed in their conclusions. Beginning with a large research synthesis by Mittelbach et al. (2001), a number of authors (Gillman and Wright 2006, Partei et al. 2007, Laanisto et al. 2008) have attempted to classify patterns from individual studies by the shape of the responses; Whittaker and Heegaard (2003) criticized what they felt were methodological flaws in the Mittel bach et al. (2001) paper and the critique was rebutted by Mittelbach et al. (2003). Due to what he feels are per sistent methodological flaws in the papers attempting quantitative syntheses of this literature, and because he now believes that it is impossible to carry out meaning ful meta-analyses on this relationship, Whittaker (2010) recommends an end to meta-analyses on this topic, expresses concern over whether quantitative data syn thesis is a legitimate and repeatable approach to making sense of the data on this question, and suggests a profound change in the way meta-analyses are conduct ed and reviewed in ecology. We offer a short discussion of why we feel that Whittaker's dismissal of meta-analysis is inappropriate, add a brief critique of this literature of our own from a statistical perspective, and most importantly, point the way to improved statistical approaches. While the devil is certainly in the details, we also don't want to lose sight of the forest for the trees.