Columbidae. This
happens when we can't automatically assign an identification to one of the
output taxa.
Review identifications of Streptopelia 2926
Taxonomic Split 121567
(Tehty 28-03-2023)
|
|
jakaa |
|
Kommentit
Should Nesoenas also be included as an output, given the transfer of Malagasy turtle dove to that genus?
Good point, will add this. There may be the odd genus ID on Comoros etc that could refer to Nesoenas
Given that the input and output both have 2926-Streptopelia - how will we know when this swap is completed?
I dont find the link at Review identifications of Streptopelia 2926 helpful. It is too cluttered and overly split into 5,976 pages - are we seriously expected to help?
Why not use as the link the identify tool for Columbidae above generic rank?
https://www.inaturalist.org/observations/identify?quality_grade=any&taxon_id=2715&lrank=subtribe
helpers can then filter by their region and use their local knowledge to assist.
(would there be a need for review at generic level - will all IDs of genus Streptopelia become Columbidae, or would any stay as Streptopelia?)
Any idea on the number of Research Grade observations that have become Needs ID?
I see quite a lot of observations initially to Streptopelia, but with subsequent refinements, to species that were Research Grade, which are now Needs ID because the 1,2,3 or 4 IDs to Streptopelia are clashing with the species level IDs in the new genus.
Or are these observations still in the update pipeline where individual Streptopelia IDs will become Columbidae??
@tonyrebelo this split needs a tweak, so there may be some clashes showing. Will re-run it
@rjq - Strangely, my Streptopelia ID for a Streptopelia decaocto in Miami got reassigned to Spilopelia, which is an Old World genus. Not sure what went wrong. https://www.inaturalist.org/observations/68125068.
@rjq - So there are a tiny number of Spilopelia chinensis introduced in California (<100 observations), but there's a huge number of Streptopelia decaocto in the U.S. (>21,000 observations), so I don't think all Streptopelia IDs in the U.S. should get mapped to Spilopelia. If you have to assign the U.S. to one genus or the other, it should definitely be assigned to Streptopelia, not Spilopelia.
@zygy it appears the atlases were incomplete/incorrect at the time the split was committed: checking just now, the US was left off entirely from the atlas for Streptopelia (orders of magnitude more widespread and abundant than Spilopelia in the US), but the entire country was included on the atlas for Spilopelia (represented in the US by S. chinensis exclusively in Hawaii and southern California, the latter population very small and declining), incorrectly swapping genus-level Streptopelia IDs across the US to Spilopelia when essentially all should've remained Streptopelia. (There were several other major regions that should've been included in one or both atlases that were missing as well.) I've fixed the atlases, but unfortunately I'm not sure anything can be done about the large number of incorrectly-swapped IDs at this point (besides maybe another split sort of in reverse, splitting Spilopelia into Spilopelia and Streptopelia?).
Update: I'm pretty sure the Streptopelia atlas was correct when I looked yesterday, but a bunch of places with tons of Streptopelia observations (including the US and several countries in Asia) were somehow automatically removed? I just added a bunch of those missing places, only to reload the atlas and find that those and a bunch more were now missing (which I then re-added - hopefully they'll stick). Not quite sure what's going on, but I don't think it was intentional.
And now every additional US state has been added somehow to the Spilopelia atlas, which I'm removing as fast as I can but in the meantime may have resulted in a few IDs being swapped up to Columbidae when they should've stayed Streptopelia...
@loarie is there any way to stop and revert this swap to fix the large number of Streptopelia IDs erroneously switched to Spilopelia?
https://www.inaturalist.org/observations?locale=en&lrank=genus&place_id=1&preferred_place_id=7278&subview=map&taxon_id=1425710&verifiable=any so far there are 107 genus-level Spilopelia observations in the US alone, and another 52 in India (can't remember what other places were missing from the Streptopelia atlas when I first checked, but it was a few), that should not have been swapped to Spilopelia (mainland US ones should've stayed Streptopelia, except in Los Angeles County where they should've been bumped up to Columbidae due to the presence of S. chinensis, and ones in India should've also been changed to Columbidae since both genera are widespread and common there), and that's not counting however many observations exist that had been research grade with a genus-level Streptopelia and 2 species-level S. decaocto IDs, for example, that now have community IDs of Columbidae because those genus-level IDs were erroneously swapped to Spilopelia
I’m not sure what has happened here. The split was committed initially and then reversed, and then recommitted. Between the first and the second time the split was committed, I added an atlas for Nesoenas and made a few changes to the Spilopelia atlas (not to the US though), but I didn’t make any changes to the Streptopelia atlas.
Yeah, I could've sworn the atlases looked fine yesterday (before the reversal and recommitting of the swap) - none of the countries missing when I checked an hour ago were missing then. Given that a bunch of places were missing from the Streptopelia atlas when I checked around an hour ago (including the US and much of Asia), then half an hour ago [after adding those places, closing the atlas, and reopening it] even more countries were missing (including most/all of the previously missing countries, but also pretty much every country in Africa, all of which were there an hour ago if I remember correctly), without any sign of those countries having been removed in the intervening time, I'm pretty sure there was some bug affecting the atlases (rather than someone manually removing places)
Thankfully it seems the atlases are stable now (no places seem to have been unlisted without a trace in the last half hour or so), but there are still the already-swapped IDs to deal with
The African countries were definitely in the Streptopelia atlas before and after the first split, as I checked this atlas while making the Nesoenas one.
I guess the split can be reversed again and recommitted again. Could making atlas changes between committing and reversing a split be an issue?
we need to revert this for a second time? I'll do that (once you confirm) but then wait for I comment that the reversion is complete, but then I'll leave checking and committing the altered version to you. sound good?
That sounds good to me. If you revert and then let us know when complete, @maxkirsch and I can recheck the atlases and recommit. Not sure what went wrong with the atlases but seems to have been a bug.
ok reverted - I won't touch it anymore
@loarie thanks. Reverting the split seems to have again done strange things to the atlases. The atlas for Nesoenas has disappeared, and the atlas for Spilopelia has lost nearly all its content - only two countries are left in the atlas. The atlas for Streptopelia looks unchanged though.
Angola and Saudi Arabia were the two countries left on the Spilopelia atlas when I checked, in case that's significant; I just re-added everything missing (I think - feel free to modify it)
Maybe we should wait a bit to make sure the atlases don't change again before committing? Unless it's the act of committing that was changing the atlases somehow?
Just in case something's somehow affecting these atlases specifically, should we delete the Streptopelia and Spilopelia atlases and recreate them?
recreating the atlases won't help so please don't do that. I suspect whats happening is that the observations (which are moving around alot with all these committed and reverting changes) are creating/destroying listed taxa via some bug. Maybe we should wait until all these recent changes churn through the system
Thanks - won't delete the atlases then, and won't commit this for the time being.
Just noticed that the Phalacrocorax genus atlas was affected by the same bug, but it seems it didn't cause any problems so no follow-up needed there (maybe the atlas only changed toward the end of or after the split?) - a quick glance at some of the areas where Phalacrocorax overlaps with one other cormorant genus showed a bunch of IDs switched up to family-level as they should've been.
(I quickly re-added everything missing from Phalacrocorax in case the split was still committing, but if I come across another messed-up atlas in a not-yet-committed or already committed split, I'll leave it as is, in case the missing or still-included regions can potentially help diagnose the bug)
Have added an atlas for Nesoenas, and made a few minor changes to Streptopelia (adding a few missed off tiny countries). Spilopelia looks fine. @loarie @maxkirsch is it time to try again?
ok committing
@loarie the taxon changes for the new genus Spilopelia have been drafted (change group 'Spilopelia genus change'). The situation has got a bit confused as someone has already grafted Streptopelia chinensis to Spilopelia (hence the red observations on this map) - though this could be reversed