Palaeobotanical experiences of plant diversity in deep time. II: How to measure and analyse past plant biodiversity

Determining the diversity of past floras helps with interpreting both the history and predicting the future of vegetation change. For global-scale and regional-scale diversity studies especially, secondary data are often used but local-scale studies tend to be based on survey data that require rigorous sampling. The correct sampling strategies depend on the types of fossils being investigated, including their physical size, and whether the aim is to determine taxonomic richness or relative abundance. Describing and comparing diversities can use a range of different metrics, depending on whether binary presence / absence or abundance data are available. Each metric provides a different insight into the diversities and the choice of which to use depends on the research question being investigated. Various numerical approaches are available for identifying spatial and stratigraphical diversity patterns, mainly classificatory techniques (e.g., cluster and parsimony analyses) and ordination (e.g., Detrended Correspondence Analysis, Nonmetric Dimensional Scaling). The choice of technique again depends on the research question, but often it has proved useful to run both types of analysis in tandem. This article is illustrated by past biodiversity case studies from throughout the fossil record, dealing with floras ranging in age from the Devonian to the last few centuries.