Occurrences seem to have been removed in some GBIF datasets - was it intentional?

[ManonGros]

Hi PLAZI,

My colleagues and I recently noticed that for at least two datasets, the occurrences seem to have disappeared.

For example, this dataset https://www.gbif.org/dataset/e542d309-1af6-465f-816b-3ef3e62a526e has occurrences indexed on GBIF but there aren't any in the archive: https://tb.plazi.org/GgServer/dwca/C56BA308C17E0264FF88FFD6FFC6C32D.zip
GBIF has a safeguard to prevent accidental deletion of occurrences so if all the occurrences disappear from a given dataset, they are still in the index. This is why the occurrences are still on GBIF. We can removed those from the index manually but I want to make sure this wasn't an accidental deletion.
The other dataset concerned is this one: https://www.gbif.org/dataset/91a51391-41b1-4184-bb62-6cd3bbbfdd74.

Both datasets contain material from specimens in institution that aren't well represented on GBIF, the occurrences are valuable.
Were they deleted on purpose? Let me know, thanks!

[flsimoes]

Hi @ManonGros I honestly don't know, it seems to be something on the GBIF side of the operation.

On our side it seems normal
See:

• Article link on TreatmentBank -> https://tb.plazi.org/GgServer/summary/C56BA308C17E0264FF88FFD6FFC6C32D
• GBIF links to materials -> https://tb.plazi.org/GgServer/srsStats/stats?outputFields=doc.articleUuid+tax.name+matCit.id+matCit.gbifOccurrenceId&groupingFields=doc.articleUuid+tax.name+matCit.id+matCit.gbifOccurrenceId&FP-doc.articleUuid=C56BA308C17E0264FF88FFD6FFC6C32D&format=HTML

I'll have a further look and get back to you

[Carol-Sokolowicz]

I did the quality check on GGI.
If the problem persist let us know

[ManonGros]

Thank you Carol! The data looks good for the first dataset! Would it be possible to do the same for this other one: https://www.gbif.org/dataset/91a51391-41b1-4184-bb62-6cd3bbbfdd74? Thanks!

[flsimoes]

Thank you Carol! The data looks good for the first dataset! Would it be possible to do the same for this other one: https://www.gbif.org/dataset/91a51391-41b1-4184-bb62-6cd3bbbfdd74? Thanks!

Should be updated in the next couple of hours

[spalp]

Thank you both.
I found yet another dataset, which previously had occurrences, but now returns 0 occurrences: https://www.gbif.org/occurrence/search?dataset_key=50e48d61-aaca-4af9-9ef3-d2104aba3b8b

[Carol-Sokolowicz]

This has been updated, if there's something elso wrong let me know
https://www.gbif.org/occurrence/search?dataset_key=50e48d61-aaca-4af9-9ef3-d2104aba3b8b

[myrmoteras]

@ManonGros is there a way on GBIF side to check, whether there are more datasets that have been disappearing? Just to understand the issue on our end.

[muttcg]

@myrmoteras More than 2600+ Plazi checklist datasets were affected.
List of GBIF Registry dataset keys: plazi.csv

[myrmoteras]

@muttcg thanks. This might have to do with a reaanalysis of the articles using more advanced tools, especially older articles/datasets.
We look into this asap

[gsautter]

@myrmoteras your assumption seems to be pretty much spot-on ... since we retro-applied quality control to our older IMFs, occurrences might well have been filtered from several DwCAs ... as we catch up on inspecting these IMFs, the occurrences should reappear, though, at least the correct ones.