The effects of taxonomic rank on climatic calibrations: A test using extant floras of United States counties

Abstract

Fossil plants are often used to infer the paleoclimates in which they occurred by applying methods that match fossils to their nearest living relatives and, in turn, use the climatic preferences of the living species to infer paleoclimate. These methods of paleoclimatic reconstruction are termed taxon-based and are ideally accomplished using pairs of fossil and extant species. However, in practice, they are often performed using higher taxonomic ranks, such as genera, families, or even orders. The use of higher taxonomic ranks has largely unknown effects on the performance of the taxon-based methods. In this study, we tested the effects of using species, genera, families and orders on the accuracy and precision of climatic predictions from the taxon-based method known as taxonomic calibration. We performed taxonomic calibration using the extant floras of US counties and compared our predictions with the reported values for the counties. We made predictions for mean annual temperature, minimum temperature during the coldest month, average temperature during the coldest quarter, and mean annual precipitation for the US counties. We evaluated our results from taxonomic calibration for accuracy and precision using several measures of differences between reported and predicted values and r2, respectively. Our results showed that genera performed as well as or better than species for all climatic variables according to our measures of accuracy and precision, and families performed only slightly less well. Orders performed relatively poorly, but still showed an unexpectedly high percentage of accurate predictions for annual temperature and average temperature during the coldest month. Our results validate the use of taxonomic calibration for inferring climates using taxonomic data at the ranks of species, genus, and family and suggest that orders may be used cautiously in some cases. We discuss the applicability of our findings to studies of paleoclimates.