Temporal and spatial variation in biological sampling may generate bias leading to shortfalls of biological knowledge, including undescribed species, poorly described species distribution, and a lack of information on tolerances and responses of species to e.g. climate changes.
In this study, researchers compiled a database of arthropod, vertebrate and angiosperm records in Brazil using data mainly from the GBIF network to test for sampling bias. Comparing taxonomic groups, they find significant differences in sampling effort with angiosperms having larger areas of high density than arthropods and vertebrates. Across all groups and biomes, however, the highest density of records falls within one kilometre of access routes, and this density increases with number of routes.
The study notes the importance of considering such collection bias in biogeographic studies, and suggests focusing on poorly sampled locations in future biodiversity surveys.