Non-publicly available data sources

[bmedlyn]

Hi guys, what is the policy on including data from sources that are not publicly available? I'm looking for waterlogging tolerance data and they all come from this source

austraits$sources[[488]]
[1] M. White, S. Sinclair, and D. Frood. Victorian Vital Attributes
Database. database. Department of Environment, Land, Water and Planning, Victoria, 2020.

I can't find this online, meaning I can't check how the values are derived. The values are a bit useless as a result. Could you perhaps introduce a policy that trait values are only included from sources that are publicly accessible - and that citations need to include a doi / web location?

[ehwenk]

@bmedlyn Thanks for this comment. I've moved this issue to austraits.build and will open a related comment on traits.build shortly.

This is a tricky issue, since a large proportion of AusTraits (https://github.com/traitecoevo/austraits.build) data come from unpublished data sources or from government databases/reports that we have deemed reliable efforts, but which indeed are not available online. For instance, the dataset mentioned is a substantial effort from leading research organisation and we trust the integrity/accuracy of the "experts", but they have not provided an explicit description of the methods used, leading us to enter "compilation from diverse sources and expert opinion" in the methods field. Indeed, methods are usually much more generic for categorical traits, where one often has to assume the person creating the dataset did make explicit, informed decisions based on their expert knowledge.

However, this issue clearly indicates the need for an additional column in the methods table that includes citation type (bibtype) and that will be included in the next traits.build release.