The derivation of log‐transformed abundance data for the quantitative analysis of macroinvertebrate traits – an addendum to ‘A macroecological perspective of trait patterns in stream communities’ by Heino et al. ()

Abstract

Summary Heino et al. () identified the weak explanatory power of abundance data as an important limitation of macroinvertebrate trait analysis. This limitation may be an artefact of analytical design. The widespread practice of combining logarithmically transformed abundance data with trait frequencies, log(x + 1)*(trait frequency), represents a nonlinear abundance weighting of trait frequencies, as opposed to an expression of trait abundances per se. Because the addition of logarithmic data is equivalent to multiplication on an arithmetic scale, summing these abundance‐weighted frequencies provides an inconsistent scaling of trait abundance that may confound quantitative comparison. We provide examples of the options for estimating trait abundance from data on macroinvertebrate abundance and trait frequencies and discuss the meaning of numerical data in the context of analytical objectives. In the light of the contrasting methods that have been employed to analyse trait data for benthic macroinvertebrates, the conclusion that the explanatory power of trait abundance is inferior to that of taxonomic abundance may be premature.