Citizen science yield different results compared to traditional databases regarding species distribution analysis

Abstract

Citizen Science (CS) databases such as GBIF and iNaturalist (iNat) are increasingly used in biodiversity inventories andfor distributional analysis. However, “traditional” databases might also be derived from voucher specimens and bibliographicreferences. Many studies recognized the biases in databases, and there are unevaluated differences between themin distribution analysis. I performed a comparison between databases of distributional data for 12 species of mustelids,inhabiting mainly South America. The number of records, temporal and geographical provenance and levels of uncertaintywere compared. Geographic Information Systems and ellipse niche models were used to compare the geographical andthe environmental spaces, following Hutchinson’s niche duality theory. The traditional database presented the highestnumber of records, followed by GBIF and iNat. Differences regarding records per country were found between databases.The three databases presented the highest number of records in the last 30 years and levels of uncertainty weredifferent between databases. The estimated geographical areas for the 12 species were different between databases, withthe traditional one presenting ca. 20% larger areas. The estimated environmental niches showed differences between themajority of species (i.e., no overlap) comparing databases. The use of different databases yields different geographical andenvironmental areas in distribution analysis.