Heads up: Some or all of the identifications affected by this split may have been replaced with identifications of Pica. This happens when we can't automatically assign an identification to one of the output taxa. Review identifications of Pica pica 144106

Taxonomic Split 57605 (Committed on 17-01-2020)

This is a retroactive split - Pica mauritanica etc. were added when they should have been split off from Pica pica. We are now splitting Pica pica after the fact to sort out existing identifications of Pica pica that now apply to Pica mauritanica etc.

Clements Checklist v2018 (Citation)
Added by rjq on 18 May, 2019 11:18 | Committed by loarie on 17 January, 2020
split into

Comments

@loarie @maxkirsch can this be committed? Issues are coming up with both P. pica sensu lato active, but also the elevated subspecies, e.g. https://www.inaturalist.org/observations/9190466

Posted by jwidness almost 5 years ago

New taxa have been atlased. They don't all have range maps as IUCN doesn't accept all the splits, only those of mauritanica and asirensis

Posted by rjq almost 5 years ago

I can make maps for P. bottanensis, P. serica, and P. pica sensu stricto. If I follow this map: https://bioone.org/ContentImages/Journals/jzoo/34/3/zs160119/graphic/f01_185.jpg would that work? Per the paper it came from (https://bioone.org/journals/zoological-science/volume-34/issue-3/zs160119/Deep-Phylogeographic-Breaks-in-Magpie-Pica-pica-Across-the-Holarctic/10.2108/zs160119.full) I would put serica, jankowskii, and anderssoni in the serica map.

Posted by jwidness almost 5 years ago

Hm, well I added maps for bottanensis and serica, but the P. pica sensu stricto map that I uploaded isn't showing up. I checked it for geometry errors and didn't find any, maybe something to do with the large size?

Posted by jwidness almost 5 years ago

Thanks. Have tweaked the atlases to match. Strictly speaking it doesn't need a range map for the split to be implemented.

Posted by rjq almost 5 years ago

The Pica pica sensu stricto map just needed some time. I also padded the atlases a little more where pica meets serica since the source map has a bunch of question marks there. It will just force genus IDs and anyone who feels strongly can move back to species level. Cool?

Posted by jwidness almost 5 years ago

That's great. Having some overlap is good, and I doubt there are many records from the overlap zone.

Posted by rjq almost 5 years ago

Ok, I think what's left to be done here is:

move all the remaining subspecies from P. pica sensu lato to P. pica sensu stricto (all of them belong in the strict definition)
commit this taxon change
update the frameworks for P. pica sensu lato and P. pica sensu stricto

@loarie @maxkirsch I think only you guys can finish this off

Posted by jwidness almost 5 years ago

@loarie is there some lag of old IDs being updated acc. to split? Loads of observations have the inactive ID, some are updated, and some have both updated and inactive IDs (eg www.inaturalist.org/observations/390279)

Posted by jakob over 4 years ago

Does seem to be going very slowly. The other taxon changes have gone through.

Posted by rjq over 4 years ago

Just for my curiosity/tracking purposes, there are currently:
8744 P. pica sensu lato obs
31 P. mauritanica
0 P. asirensis
16 P. bottanensis
1054 P. serica
2600 P. pica sensu stricto

There are 27 observations at Pica (genus) that have at least one active ID of the inactive P. pica sensu lato.

Posted by bouteloua over 4 years ago

Have also been keeping an eye on these, the numbers of P. pica sl have gone down by about 400 in 10 hours. This may be people changing their IDs, with no automatic changes being made.

Posted by rjq over 4 years ago

yeah definitely taking a long time - well we knew that changing this taxon with > 10k obs was going to test the system. I guess its good to find those limits though so we can try to figure out how to make them scale better (glad we didn't change mallards ;)

Posted by loarie over 4 years ago

@loarie only 220 changes have gone through in 19 hours - does this mean that the taxon change isn’t actually being processed by the system, and the only changes are people actively updating their records? The other changes were much quicker, including the Ruddy Duck split with 9000+ obs.

Posted by rjq over 4 years ago

we are now caught up on delayed jobs so anything still not sorted is an error. Looks like there was an error associated with this "undefined method `observations_places' for nil:NilClass" We'll look into it - I suspect it has to do with some bug with ID sorting by atlas that this split uncovered - once we identify and fix the bug, we can re-split

Please let us know if you see any other changes that looked like they failed (ie are still not sorted now that everything has finished processing)

Posted by loarie over 4 years ago

thanks for the update

Posted by bouteloua over 4 years ago

ok this was rerun (with the problematic ID without an observation that crashed it removed) and it looks good now. The remaining content on Pica pica 144106 appears to be from people who opted out of the CID and opted out of taxon changes creating new IDs.

So it looks like 2 bugs so far:
(1) the updated obs.taxon bug: https://github.com/inaturalist/inaturalist/issues/2559
and
(2) this commit_records bug when identification.observation doesn't exist https://github.com/inaturalist/inaturalist/issues/2562

there's also lots of ES indexing delays, e.g. the 16k obs that still get returned by the API under taxon_id 15984 even though they are now under taxon_id 980017 https://www.inaturalist.org/observations?verifiable=true&taxon_id=15984&place_id=&preferred_place_id=1&locale=en from this splithttps://www.inaturalist.org/taxon_splits/67564
we're looking into whether this is a 3rd bug or just slow ElasticSearch indexing

Posted by loarie over 4 years ago

Add a Comment

Sign In or Sign Up to add comments