An inventory of biodiversity data sources for conservation monitoring

updated, but it was not always easy to judge how these criteria were met from the information available. It was also not always easy to assess how accessible data were and if they were free- of-charge. Indeed, even if data are freely accessible they are not always easy to use [24]. There- fore, we note that some data sources in the inventory may still prove challenging for monitor- ing purposes when investigated in more detail. Multiple, linked data sets managed by a single agency and consolidated in a single database were considered as one source. For example, the FAO Fisheries and Aquaculture database has numerous data sets (e.g. Global Production, Global Capture Production, Global Tuna Catches by Stock, Global Aquaculture Production, Atlas of Tuna and Billfish Catches, Global Number of Fishers, Consumption of Fish and Fishery Products, etc.) but was counted as one fisheries data source. Similarly, habitat and land cover data were not disaggregated by remote sensing data set but, instead, we identified data sources that consolidated from different sets. For exam- ple, forest cover data sources are numerous (e.g. Global Forest Change, Global Forest Watch) but all in turn source their data from a variety of remote sensing programmes, such as ESA’s GlobCover and NASA’s Landsat. The intent was to identify wherever possible pre-prepared data of use to managers, not raw data files for GIS analysis. Data sources were clustered based on their uses for monitoring biodiversity state, biodi- versity threats and pressures, and conservation responses, as per the Pressure-State-

Response indicator model widely used in biodiversity conservation monitoring [4, 7, 25]. The data sources that could potentially be used to monitor pressures and responses were clustered under the threats and conservation actions categories developed by IUCN and the Conservation Measures Partnership, CMP [26]. For this exercise, the newer, revised ver- sions of both classifications were used: the CMP Direct Threats Classification (v 2.0 of 2016) and the CMP Conservation Actions Classification (v 2.0 of 2016) [27]. In each case, we developed examples of related indicators and indicator categories associated with each action and each threat and then listed examples of global databases and platforms of poten- tial use in monitoring them. Note that, in some cases, a database could be used for more than one indicator. For example, forest cover data can show the state of the habitat, but a trend in forest cover can also monitor a trend in forest loss, which is a threat. Note too that state metrics–such as species populations trends and habitat cover–can also be used to mon- itor the impact of various threats (e.g. deforestation rates in relation to infrastructure devel- opment or mining). Results, conclusions and discussion A total of 145 global data sources were identified of potential value in conservation monitoring (Fig 1): 40 data sources for biodiversity state, with at least 33 open access (S1 Table); 44 data sources for biodiversity pressures and threats, with at least 38 open access (S2 Table); 33 data

sources for conservation responses, with at least 28 open access (S3 Table); and 28 data sources that cut across multiple indicator types, with at least 27 open access (S4 Table). Twenty-six types of report were identified (S5 Table) that regularly synthesise biodiversity data from diverse sources. These helped identify some of the data sources. The shortlist of data sources that, from this preliminary analysis, appear to be the most useful in accessing data for monitor- ing biodiversity were compiled (Table 1). While 86.9 per cent of the data sources allowed instant, open and free access to at least some of their data, the usefulness of that data and the relevance of the formats was not ascer- tained. The degree to which it was possible to assess what data are available and accessible, the origins of the data, the precise indicators the data measure, and the length of any time series available was very variable and the information was often complicated to find. In some cases, PLOS ONE Biodiversity data sources PLOS ONE | https://doi.org/10.1371/journal.pone.0242923 December 2, 2020 3 / 14