How to fill the biodiversity data gap: Is it better to invest in fieldwork or curation?

Data gaps and biases are two important issues that affect the quality of biodiversity information and downstream results. Understanding how best to fill existing gaps and account for biases is necessary to improve our current information most effectively. Two current main approaches for obtaining and improving data include (1) curation of biological collections, and (2) fieldwork. However, the comparative effectiveness of these approaches in improving biodiversity data remains little explored. We used the Flora de Bogotá project to study the magnitude of change in species richness, spatial coverage, and sample coverage of plant records based on curation versus fieldwork. The process of curation resulted in a decrease in species richness (synonym and error removal), but it significantly increased the number of records per species. Fieldwork contributed to a slight increase in species richness, via accumulation of new records. Additionally, curation led to increases in spatial coverage, species observed by locality, the number of plant records by species, and localities by species compared to fieldwork. Overall, curation was more efficient in producing new information compared to fieldwork, mainly because of the large number of records available in herbaria. We recommend intensive curatorial work as the first step in increasing biodiversity data quality and quantity, to identify bias and gaps at the regional scale that can then be targeted with fieldwork. The stepwise strategy would enable fieldwork to be planned more cost-effectively given the limited resources for biodiversity exploration and characterization. •Curatorial work from collections makes to recover key geographical and taxonomical species information at a regional scale.•Curation of biological collections is more efficient in producing new information compared to fieldwork.•Given the limited resources for biodiversity study, curatorial work should precede fieldwork as it is more cost-effective.