Taxon-level assessment of the data collection quality in Atlas Florae Europaeae : insights from the case of Rosa (Rosaceae) in Eastern Europe

By the method of data re-collection and re-assessment, we here test the completeness of distribution areas of the species and species aggregates of Rosa in Eastern Europe as mapped in volume 13 of Atlas Florae Europaeae (AFE), and discuss insights into the issues connected with the data. We found many new occurrences which are additions to the published maps: 1068 records of species and 570 records of species aggregates. The new occurrences are listed with references to the sources, and the updated AFE maps are provided. The greatest increase by new native occurrences was revealed for the species that are widespread or taxonomically complicated, and by new alien occurrences for the species that currently expand their secondary distribution areas. The mapping work published in 2004 is considered good, with minor omissions caused by possible oversights and incomplete sampling. The majority of new additions originated in the period after the original data collection. Nearly the same amount of new data originated from larger and smaller herbarium collections, underlining the value of small collections for chorological studies. We found that only ca 20% of new records based on herbarium specimens have been published, thus highlighting the need for data papers for publication of distributional data. The greatest increase by new records based on herbarium specimens was found for insufficiently studied territories (Belarus, central, northern and eastern parts of Russia), whereas the same level of increase for the territories with reasonably good coverage (Latvia) was achieved by observations. We conclude that the overall sparsity of published records in Eastern Europe is caused by a lower level of data collection rather than by poor data availability, and that floristic surveys based on herbarium specimens cannot compete in speed and density of records with observation-based surveys, which may become the main source of distributional information in the future.Peer reviewe