An inventory of biodiversity data sources for conservation monitoring

Biodiversity Information Facility, GBIF, and the Ocean Biogeographic Information System, OBIS, and life-history parameters from FishBase); and complex elements that rely on several other elements (e.g. the EMODnet Portal, which can combine multiple datasets in interactive maps, drawing on data sets from, for example, the European node of OBIS, the CITES Trade database, and the IUCN Red List of Threatened Species). This framework could be integrated with the approach taken by Konig et al. [36] who conceptualised widely-used biodiversity data types according to their domain (the aspect of biodiversity that is described) and informational resolution (how specific the description is). Such an ontology would help in refining the inven- tory of data sources and help clarify the numbers and types of sources available. For example, an assessment of data sets looking at threats ([24]; S1 Table) considered multiple data sets looking at the same issue (e.g. forest cover) or multiple versions of the same data set (e.g. Glob- Cover 2.2 and GlobCover 2.3) or disaggregated data from one agency (e.g. FAO data on differ- ent types of livestock–buffalo, cattle, goats, sheep, pigs, poultry, etc.). The current study lumped data sources for efficiency of use and analysis. An ontology would therefore help in comparing different data source inventories. As well as developing an ontology of data source types, existing data sources should strive to make it clearer (preferably on the homepage of their websites) which data they are sourcing

and from where, and if it is instantly, openly and freely available. Address data gaps and biases. Conservation practitioners and scientists, as well as other biodiversity data users in governments., NGOs and businesses, need to identify and systemati- cally fill data gaps. All stakeholders should, at the very least, be collecting primary data to mon- itor their biodiversity priorities and projects. In addition, efforts need to be made to focus on data collection in the tropical regions housing most of the species and on those taxa that are most under-represented in data sets, such as marine and aquatic species, small mammals, invertebrates, fungi, plants and wide-ranging species [4, 13, 29, 32, 34, 37, 38]. More data from tropical countries are also needed if we are to be able to predict the impacts of climate change on tropical species [39] and assess projected range to set conservation priorities [40]. Even the larger and more widely known and used global data sets, such as GBIF, the Living Planet Index and the IUCN Red List of Threatened Species, tend to have data skewed taxonomically and geographically [29, 32, 37], and more detailed breakdowns of gaps would help focus data collection efforts [41]. As well as data on the state of biodiversity and conservation responses, threat monitoring needs to become more mainstreamed into conservation projects [42]. Data need to be collected regularly and efforts made (at least in some cases) to come as close to possible to real-time monitoring. Satellite-based remote sensing data is becoming

cheaper and more easily available [43], suggesting real-time monitoring of biodiversity from space (especially the state of, and loss of, key habitats) will continue to expand. Technological advances have ensured that satellite-based remote sensing is increasingly complemented by the newest generation of Earth-based sensors [13, 44], including camera traps [45], acoustic recording devices [46] and unmanned aerial vehicles or drones [47], as well as by environmen- tal DNA monitoring [48]. Some projects are also underway to provide more integrated moni- toring systems. For example, the conservation agency Resolve, with its NGO partners like the National Geographic Society and the Zoological Society of London, and its technical partners like Google and Microsoft, are working to try to establish a network of space-based and Earth- based sensors in key conservation areas that will lead to real-time monitoring of species, habi- tats and illegal activities [49]. Therefore, we may not be far away from having live-streamed updates from key conservation areas. Although remote sensing offers many opportunities for enhancing biodiversity monitoring [44], current technology tends to bias data collection towards trees, large mammals and birds. While, in many instances these may be good indicators of broader ecosystem health, a more PLOS ONE Biodiversity data sources PLOS ONE | https://doi.org/10.1371/journal.pone.0242923 December 2, 2020 7 / 14