Quadrinomials are excluded from the classification

[GoogleCodeExporter]

What steps will reproduce the problem?
In VASCAN we have 44 accepted taxa (all varieties) with a quadrinomial name, 
for example:
Achillea millefolium subsp. lanulosa var. lanulosa (Nuttall) Piper
http://data.canadensys.net/vascan/taxon/2765

This name also appears in the GBIF Checklist Bank:
http://ecat-dev.gbif.org/usage/100018762

But for some reason, all links to the higher classification are lost (see the 
Classification box on the right of the page). This happens for all 44 
quadrinomials in our database. They all appear as root taxa (see here: 
http://ecat-dev.gbif.org/usage/100018762). This incorrect information is copied 
as such on the GBIF test data portal: 
http://testportal.gbif.org/species/browse/resource/13038

What is the expected output? What do you see instead?
In our Darwin Core Archive, all 44 quadrinomial varieties are correctly linked 
to their subspecies parent (via parentNameUsageID). On the Checklist bank (and 
GBIF data portal), they should:
1) Be linked to their subspecies parents as well
OR
2) Be linked to their species parents if for technical reasons the Checklist 
bank doesn't allow links between infraspecific taxa.
The final result should be that none of the quadrinomial taxa appear as root 
taxa.

Original issue reported on code.google.com by peter.de...@gmail.com on 31 Aug 2011 at 3:30

[GoogleCodeExporter]

Interesting finding, Peter.
I dont know the exact problem yet, but Im diving into this. 

Checklist bank should work with quadrinomials when representing the original 
data. It will convert it to a trinomial though when integrating it into our 
GBIF Nub taxonomy. Investigating...

Original comment by wixner@gmail.com on 5 Sep 2011 at 11:57

• Changed state: Accepted

[GoogleCodeExporter]

Im not sure what had happened before, but I forced a reindexing of VASCAN and 
the problem is gone!
Achillea millefolium subsp. lanulosa var. lanulosa (Nuttall) Piper 
is linked to Achillea millefolium subsp. lanulosa now as it should be:
http://ecat-dev.gbif.org/usage/100018762

There are still 2 root taxa though, see issue64, but no more quadrinomials:
http://ecat-dev.gbif.org/checklist/1016

Because automatic indexing in CLB now requires archives to pass the 
http://tools.gbif.org/dwca-validator/ VASCAN hasnt been reindexed for a while, 
as the validator was not aware of pro parte synonyms given as concatenated ids. 
This is fixed as of today and that 'might' have been the source of the problem. 

Btw, there is still a minor validation issue with VASCAN having a verbatim 
"null" value in the parentNameUsageID column: 
http://tools.gbif.org/dwca-reports/3f8a1297-3259-4700-91fc-acc4170b27ce.html

Original comment by wixner@gmail.com on 6 Sep 2011 at 9:17

• Changed state: Invalid