search_species returning outdated species_ids

[cjardine-bsc]

Reported by David Iles.

When using search_species, the species_code being returned for Northern Flicker has species_status = 0.

NOFL <- naturecounts::search_species(name = "Northern Flicker")

A tibble: 7 × 5

species_id scientific_name english_name french_name taxon_group

1 10460 Colaptes auratus Northern Flicker Pic flamboyant BIRDS
2 10470 Colaptes auratus auratus/luteus Northern Flicker (Yellow-shafted) Pic flamboyant (auratus/luteus) BIRDS
3 10480 Colaptes auratus [cafer Group] Northern Flicker (Red-shafted) Pic flamboyant (groupe cafer) BIRDS
4 10485 Colaptes auratus luteus x cafer Northern Flicker (Yellow-shafted x Red-shafted) Pic flamboyant (hybride forme dorée x rosée) BIRDS
5 40560 Colaptes auratus mexicanoides Northern Flicker (Guatemalan) Pic flamboyant (mexicanoides) BIRDS
6 41776 Colaptes auratus chrysocaulosus Northern Flicker (Cuban) Pic flamboyant (chrysocaulosus) BIRDS
7 41777 Colaptes auratus gundlachi Northern Flicker (Grand Cayman I.) Pic flamboyant (gundlachi) BIRDS

Guatemalan Flicker was split from Northern Flicker and 10460 refers to the species complex for both. The correct species_ids that should be returned for the “true” Northern Flicker (sans Guatemalan) is now species_id 48798.

Is search_species maybe using a cached table?

[steffilazerte]

I just updated the package which will use the newer metadata by default, however there is also nc_metadata() function which users can use to update their local metadata as needed (without waiting for me to update the package).

So I would recommend updating the package. If that doesn't work, use nc_metadata() to update the species tables (although this should only be necessary if the species change was very recent). If that doesn't work, I'll check, but it would most likely be a problem with the data on the server (but I'm assuming you've already checked that).

[cjardine-bsc]

Thank you!

Sounds like updating the R package needs to be part of the annual taxonomic package. @denislepage do you have a place you've documented the taxonomy steps that we can add this to?

[denislepage]

Noted. Thanks Steffi.

It'd be good eventually to have a way to initiate the update without a R package change. Taxonomy is an annual event. Perhaps an API query that can be done from the R package side that returns the date when some of the lookup tables were updated.

We'd also have to check whether the package includes deprecated species_id, for which people may still have data associated. Otherwise, updating the metadata will potentially lead to unrecognized species_id's. Another approach is make a copy on the local instance when an update happens, so people can resolve the older codes, etc.

[steffilazerte]

A couple of things to bear in mind

• Users can update their local metadata with nc_metadata() and see when it was updated with nc_metadata_version() but this isn't a true 'version'
• We have an open issue to check metadata case-by-case for out of date versions (Case by case metadata check #37) with the idea that we should alert users when the metadata is out of date but we need the API to send this information somehow (it doesn't necessarily have to be case by case, but would be most transparent if it were).
• As things work right now, updating the metadata locally won't change any pre-existing database tables, but could result in adding species with different ideas to the current database
• However there is a check comparing the current version of the package to that version when the database was created. If that fails, the users get a warning that they should create a new database and redownload the data. This could be expanded to check against metadata versions and promt the user to start with a new data base if the metadata versions aren't the same.