Assessment of the quality of Rosa (Rosaceae) observations on Inaturalist and practical recommendations for observers. A case study for selected regions of Russia

. The quality of biodiversity data plays a big role in understanding the actual distribution of organisms. This is especially important for taxonomic groups that are difficult for species identification. In this study, we assessed the identification accuracy of Rosa (Rosaceae) observations uploaded on the iNaturalist online platform. For this purpose, all Rosa observations, made in July and August 2021-2022 and uploaded to the iNaturalist platform until 13.12.2022 from seven regions of the center of European Russia (Kostroma Region, Nizhniy Novgorod Region, Tambov Region, Tula Region, Republic of Mordovia, Republic of Tatarstan, Chuvash Republic) and from two regions of the Trans-Urals Region (Tyumen Region and Kurgan Region). Among all considered taxa, five Rosa species (R. acicularis, R. glauca, R. rugosa, R. spinosissima, R. villosa) were marked as well-identified ones. As a rule, identifications of these species are not problematic for either iNaturalist-platform experts or machine generated identification mechanism acting on this resource. These species were excluded from the final evaluation. The obtained results showed that in eight out of nine Russian regions of Russia, 62.7±11.3% of Rosa observations cannot be identified to the species level due to the lack of morphological traits presented in photographs of these observations. Only in the Tyumen Region, we found a relatively high proportion of Rosa observations (69.4%), which can be identified to the species level. This indicates a high awareness degree of iNaturalit-observers on the species composition of wild roses in the Tyumen Region and/or about morphological traits, which are important to show on photographs to identify these Rosa observations to the species level. To reduce a proportion of unidentifiable Rosa observations uploaded on the iNaturalist platform, we proposed practical recommendations for making photographic observations of wild roses, and we pointed out the key features necessary to identify them. The iNaturalist platform is a powerful tool for collecting biodiversity data. However, this crowdsourcing data needs to be as accurate as possible to be useful in scientific research. The recommendations we have compiled should help improve the accuracy of data on the genus Rosa uploaded on the iNauralist platform, and they may be also applicable to other plant groups.