Examining Collection Biases Across Different Taxonomic Groups: Understanding How Biases Can Compare Across Herbarium Datasets

Abstract

Specimen-based data are an invaluable resource for an increasing diversity of scientific fields, including global change biology, ecology, evolution, and genetics; however, certain analyses of these data may be limited by the non-random nature of collecting activity. Geographic, temporal, and trait-based collecting biases may consequently affect the understanding of species’ distributions, obviating the need to determine what biases exist and how they may impact further analyses. Trait-based biases were examined in herbarium specimen records of two abundant and diverse families (Asteraceae and Fabaceae) in a well-collected and digitized region (California) by comparing geographic-bias-adjusted simulations of random collections to actual collecting patterns. Collecting biases were fairly similar between families for a number of traits, such as a strong bias against collecting introduced species, while seasonal collecting biases showed a peak in activity in the Spring for both families. However, while there was only a dip in the fall for Asteraceae, Fabaceae were seriously under-collected for the majority of the year. These results demonstrate that significant collecting biases exist and may differ depending on the dataset, highlighting the importance of understanding the dataset and potentially accounting for its sampling limitations.