A case of benign neglect

Global environmental and socioeconomic databases and websites

Observations

The databases andwebsites category includes various sources. Within the 33 sources that provided hits for keywords related to pastoralism and/or rangelands, 21 were categorized as data sets and statistics, 11 as knowledge repositories and 1 as a GIS portal. The knowledge repositories contained collections of published and unpublished literature, news and media reports, manuals and guidelines, case studies and other documents. As was the case with the Google word cloud (Figure 4), searching the databases for pastoralist (or pastoral) and rangeland returned less information from the 92 databases than their metonyms. However, this increased uncertainty in the availability

of information, since many of the metonyms do not necessarily relate to pastoralism or rangelands. For example, the term ‘pasture’ could relate to either natural (rangelands) or cultivated pastures (not rangelands), which are not distinguished in the data. In Eurostat, ‘nomad’ appeared with the terms ‘vagrants’ and ‘homelessness’ and was not related to pastoralism at all. In the World Database of the Key Biodiversity Areas some sites were labelled as ‘ranch’, though information was lacking for these and only concerned biodiversity. The Statistical Data Warehouse of the FAO is a repository holding all of the organization’s digitized maps and statistics in one place. However, its search engine was not sensitive to first-tier keywords and returned no directly relevant

The 100 identified and publicly accessible databases and websites screened for the study were rated according to disaggregation, accessibility and confidence level. A high rating meant that data were available specifically for both pastoralists and rangelands (or their metonyms), that they could be easily searched and accessed, and that they were verified by a third party or a scientific method. A low rating meant that data were not disaggregated and only provided general information on populations or ecosystems, that they were difficult to access, and that there was no visible attempt of verification. In eight cases labelled “no access”, it was not possible to rate one or more of these indicators due to insufficient information, password protection or the website being under construction. In total, 92 databases were screened, 59 of which had no hits for any metonym keywords and were thus labelled “no information” (see Figure 8). Most of the databases reviewedwere easily accessible, though the ability to search these varied. Boolean searches were not possible for search engines where search paths were limited to single words. Some data sets could be searched only through a specific geographical site, a specific species or according to pre-set subjects. Some databases had sections that were passwordprotected andothers required training to understand how to navigate the site. Screening databases and websites for relevant information was therefore a very time-consuming process. Most of the databases were rated as having only a medium confidence level, as their data were not subject to rigorous verification, with the majority based on some formof self-reporting. These databases may therefore be biased in what they report.

High Medium Low No information No access

70

60

50

40

30

20

10

Accessibility

Availability

Con dence level

Figure 8: Data sources rated according to accessibility, availability and confidence level Note: Databases that were under construction or could not be accessed for other reasons are labelled “no access”.

37