Abstract

Modern biodiversity hotspots are characterized by high species diversity and by biotas facing a substantial threat of extinction, largely due to a high proportion of endemic taxa living in these regions. Theoretically, hotspots of biodiversity are areas of particular interest in the fossil record because of the relatively high quality and quantity of data that they may contribute to a global understanding of vegetational response to changes in climate, tectonic uplift, and ecological disturbance. Current models for climatic reconstruction that depend on leaf physiognomy are based on data sets in which species-rich tropical floras are less well represented, relative to temperate floras. Eight modern Neotropical floras from a range of precipitation regimes were evaluated to determine the influence that high source floral diversity has on reconstruction of mean annual temperature (MAT) and mean annual precipitation (MAP). Floras are drawn from sites in Costa Rica to southern Peru, having species richness from 55 to >400 species per plot. MAT of the sites spans a range of 24 to 28°C, and MAP ranges from ∼1600 mm to 3000 mm. By subsampling the modern floras in rank order of dominance (basal area), the importance of collecting intensity and completeness on subsequent assessments of MAT and MAP is evaluated. Biodiverse floras are good at reconstructing MAT if at least 50% of the species are included. When only 25 species are used for temperature calculations, the accuracy of the parameter is compromised, but a ±3°C error encompasses the majority of the deviation. Application to the early Paleocene Castle Rock fossil flora of Colorado confirms the validity of subsampling in high-diversity fossil applications. However, reconstruction of MAP is fraught with problems that do not appear to be related to biodiversity of the floras. Errors on estimates of MAP currently are so large as to make the values too vague to be useful in most applications. This study has accepted a 20% error as necessary, but the applicability of data with errors > 20% is questionable in situations where rainfall is >1500 mm per year. MAP estimates using leaf area are almost universally underestimates of actual MAP, and frequently are >400 mm in error. Exploration of these data indicates that effort would be well placed in investigating the relative importance of precipitation parameters in altering leaf morphology before choosing one to reconstruct climates of the past.

You do not currently have access to this article.