Effects of publication bias on conservation planning

Conservation planning needs reliable information on spatial patterns of biodiversity. However, existing data sets are skewed: some habitats, taxa, and locations are under-represented. Here, we map geographic publication density at the sub-national scale of individual 'provinces'. We query the Web of Science catalogues SCI and SSCI for biodiversity-related publications including country and province names (for the period 1993-2016). We combine these data with other provincial-scale factors hypothesised to affect research (i.e. economic development, human presence, infrastructure and remoteness). We show that sites that appear to be understudied, compared with the biodiversity expected from their bioclimatic conditions, are likely to have been inaccessible to researchers for a diversity of reasons amongst which current or recent armed conflicts are notable. Finally, we create a priority list of provinces where geographic publication bias is of most concern, and discuss how our provincial-scale model can assist in adjusting for publication biases in conservation planning.
Comment: 10 pages; 3 figures; 1 table;R code on https://github.com/raffael-hickisch; data at https://zenodo.org/record/998889; interactive at http://bit.ly/publication_density_map