Design and analysis of biodiversity field experiments

Abstract

The design and analysis of field experiments have figured prominently in the current debate about biodiversity and ecosystem function. These debates have identified important issues about species traits, functional groups, and community assembly, but a broader debate needs to include discussions about the scale (grain and extent) of experiments relative to natural spatial and temporal heterogeneity. In addition, alternative statistical analyses need to be explored that focus on comparison among several statistical models rather than simple hypothesis testing. Analyses of the first 2 years of data from a new biodiversity field experiment are used to illustrate these concepts. A traditional one‐way ANOVA demonstrates the expected increase in aboveground biomass with higher levels of vascular plant diversity. Further analysis demonstrates that this relationship is absent when the community contains either Arrhenatherum elatius or Holcus lanatus, two dominant species of grass. Variance in biomass is also a function of diversity and both spatial and temporal heterogeneity are significant factors in the analysis despite precautions taken to minimize them. These examples illustrate the degree to which the analysis of a field experiment influences the interpretation of the observed results. Ultimately, results from field experiments must be validated through continued comparisons among field experiments, mathematical models, laboratory trials, and mesocosm experiments.