Making informed decisions in conservation planning requires good knowledge about species distributions. Gaps and biases in spatial data, however, may skew results and affect our ability to accurately predict the distribution of species.
This study explores the use of ‘ignorance scores’ to evaluate sampling effort and bias—in a case study of species present in the Caatinga semi-arid tropical forest in Brazil. Downloading all GBIF-mediated occurrences in the region, researchers organized the data into taxonomic reference groups of similar collection methods and calculated scores for each 10x10 km cell.
Their results showed a staggering taxonomic bias with high ignorance scores across the vast majority of cells for all groups, except plants—suggesting a preference towards plant recording in the region. For amphibians, the percentage of grid cells with no records was more than 99 per cent.
Exploring reasons for recording biases, the authors point to road density as a main factor, while population density and distance to nearest university are secondary predictors.