Taxonomic bias in occurrence information of angiosperm species in China.

Abstract

Taxonomic bias is a well-known shortcoming of species occurrence databases. Understanding the causes of taxonomic bias facilitates future biological surveys and addresses current knowledge gaps. Here, we investigate the main drivers of taxonomic bias in occurrence data of angiosperm species in China. We used a database including 5,936,768 records for 28,968 angiosperm species derived from herbarium specimens and literature sources. Generalized additive models (GAMs) were applied to investigate explanatory powers of 17 variables on the variation in record numbers of species. Five explanatory variables were selected for a multi-predictor GAM that explained 69% of the variation in record numbers: plant height, range size, elevational range, numbers of scientific publications and web pages. Range size was the most important predictor in the model and positively correlated with number of records. Morphological and phenological traits and social-economic factors including economic values and conservation status had weak explanatory powers on record numbers of plant species, which differs from the findings in animals, suggesting that causes of taxonomic bias in occurrence databases may vary between taxonomic groups. Our results suggest that future floristic surveys in China should more focus on range-restricted and socially or scientifically less "interesting" species.