Data Package Metadata   View Summary

We compiled a dataset consisting of previously published and unpublished data on plant community composition that spanned eight precipitation/aridity gradients in grassy biome regions on all continents, excluding Europe and Antarctica. These data were selected because 1) they encompassed relatively broad gradients of precipitation and temperature, and 2) plant composition was sampled using consistent methodology across all sites within each regional biome. We were able to find eight datasets (referred to here as “biome”) that met these criteria; however, it was difficult to control for fire and grazing history. That said, even though fire and grazing are important factors affecting richness and evenness in grasslands, we believe that at the scales of our analysis (regional, global) that precipitation is likely an overriding factor influencing water availability-diversity patterns. Each dataset consisted of six or more sites located along a precipitation gradient. For each biome, plant species composition was determined using the same sampling effort per site, using either percent cover or Braun-Blanquet or decimal cover classes. We obtained datasets directly from investigators rather than from published papers, allowing us to conduct a meta-level analysis.

For each biome, we calculated the average relative cover or abundance value of each species across all samples (e.g., plots or transects) at each site to create a single representation of the community at a site. We then calculated richness (the number of species at a site) and evenness for each site. We chose Evar as our evenness metric because it provides an intuitive measure of evenness by calculating the variance of species abundances and converting this variance measure to a bounded 0-1 scale, in which 0 is an uneven community and 1 is a perfectly even community. We dropped sites with four or fewer species (3 sites out of 115), because evenness metrics are unreliable at low species richness.

All code can be found at github.com/SallyKoerner/DomDiv