Flagger | Content Author | Content | Reason | Flag Created | Resolved by | Resolution |
---|---|---|---|---|---|---|
kueda | Blue Dicks (Dichelostemma capitatum) |
it might need to move to Dipterostemon |
Mar. 13, 2020 17:34:14 +0000 | rupertclayton |
Taxon move complete |
The paper seems well enough reasoned, so I could support the deviation (though POWO might well adopt it too). For now I have added the Dipterostemon names as synonyms to the existing Dichelostemma taxa for cross-reference purposes.
Since it will affect at least 9385 observations, probably need to check for fairly broad buy-in first.
@rupertclayton @jrebman @diego4nature @graysquirrel @grnleaf @sganley @stevejones @serpophaga @charlie @catchang @flower_prof @jlmartin @matt_g @oscargsol @sedgequeen @paloma @glmory @silversea_starsong @terrydad2 @efmer
...and anyone feel free to tag in others.
We love following the Jepson for regional plants. Makes more sense than following POWO.
We not really agree with Dipterostemon capitatus subsp. pauciflorus with the Jepson Key.
The long pedicels are found in many Dipterostemon capitatus subsp. capitatus.
We tried to find something to separate the plants with few flowers and long pedicels.
So far the Jepson key is not working in the field here in the Anza-Borrego Desert
I know @tchester has his doubts.
@efmer Changing the genus name, which is the topic of this discussion, won't change whether the subspecies are accepted or not, and they are not discussed in the paper making this change, other than to change their genus names.
For the record, I definitely agree with you that we do not have ssp. pauciflorus in the Anza-Borrego Desert.
However, it might well be a different story east of California, which is where the type specimen is found. If so, I also agree that the Jepson Manual key is not sufficient to separate the subspecies, leading to confusion as to the presence of ssp. pauciflorus in the Anza-Borrego Desert.
@rupertclayton and I have been discussing similar confusion with the plants just east of the Sierra Nevada, which seem to have the light bracts ascribed to pauciflorus, but the pedicels and flowers ascribed to subsp. capitatus. (https://www.inaturalist.org/observations/34528680) The plants near the west edges of the deserts may be transitional between the two subspecies.
But yes, this doesn't really pertain to whether we adopt the new genus or not.
I agree that the move to Dipterostemon capitatus is well supported having been indicated by phylogenetic studies from Pires and Sytsma in 2002 and formally proposed by Preston in 2017. Preston and Pires updated the Jepson eFlora with this change in December 2019.
I can see some value in having iNat adopt this change ahead of FONA/POWO. If it's going to happen eventually, we might as well start now. It could also help people think about distinguishing these genera in the field or when they ID. Conversely, there's some value in iNat having a consistent process that requires fewer taxon-specific decisions (and therefore to stick with whatever POWO decides). Then again, maybe iNat is becoming more of a place where these taxon-specific conversations should happen. So, I'm easy either way.
As to the subspecies. Preston has a 2014 paper that makes a clear case for the distinct identity of ssp. lacuna-vernalis. He also identifies (but doesn't formally describe) another likely Central Valley subspecies with characters midway between ssp. lacuna-vernalis and "classic" ssp. capitatum.
But there seems to be broad agreement that other populations of Blue Dicks - from Oregon to Baja California and across the deserts to Utah and Sonora - are poorly characterized by lumping them into just ssp. capitatus/m and ssp. pauciflorus/m. These await someone with the time and skills to perform a phylogenetic analysis and figure out the best-supported subspecies groupings. Meantime, we'll have to make our best judgments about which of these buckets a particular observation falls into.
Oops! Did I kill the discussion?
As the flag was originally raised by @kueda, maybe I should assume that iNat staff are OK with the idea of making well-supported taxon changes ahead of POWO. In that case, I'd vote yes for the move to Dipterostemon. I see several other votes for approval here (some conditional on broad support) and a couple people who don't feel well-enough informed. Should we leave this open a while longer to gather other views?
Robert Preston also put together a poster explaining the evidence and rationale for the name change: http://www.norcalbotanists.org/files/NCB_2017Poster_29_Preston.pdf
Hello all. This discussion seems to have been dormant for a while, but it seems that everyone with an opinion voiced support for the move proposed by @kueda. Meanwhile we're starting to see the first Blue Dicks observations of 2021, and it would probably be best to make this happen before the trickle becomes a deluge.
On that basis, I'll create the new genus and stage the taxon move. Feel free to jump in with comments.
Well, here's the draft batch of four taxon swaps. I created inactive taxon entries for the new genus Dipterostemon, the renamed species and its three subspecies. And I drafted taxon swaps for the species and the three subspecies.
Having chosen a species with 12,000+ observations for my first taxon change I would appreciate it if someone could check it over for any mistakes before I commit the change.
The only other thing I can think of is whether Dichelostemma needs to be split into Dichelostemma (sensu stricto) and Dipterostemon? i.e. if there are existing IDs of Dichelostemma that could refer to Dipterostemon (IDs, not observations), then we'd want to do a taxon split for the genus so that those IDs can either be transferred to Dipterostemon based on range or bumped to the common ancestor
(and after the changes are committed, to wire up taxon framework relationship deviations, which I can help with)
Right now there are about 900 Dichelostemma observations identified only to genus. And if I remember right, the range of Dipterostemon overlaps with most of Dichelostemma sensu stricto. So we may be talking about a lot of legitimate (s.s.) Dichelostemma IDs that get bumped back to Subfamily Brodiaeoideae IDs. I wonder if it would be less disruptive to just go through and manually identify the Dipterostemon observations in that group? What do you think @rupertclayton?
Good point @jdmore. Also, I probably shouldn't have started this discussion in two places at once. I'm closing out the discussion in this forum thread and we'll continue it here on the flag.
Apologies for "thinking out loud", but here goes...
First of all, I thought that there might be an issue with observations that are IDed at genus level due to conflicting species IDs, but actually, these will be fine whether or not I explicitly split the genus. For example this D. capitatus that has a D. multiflorum ID will get bumped up to Brodiaeoideae. And this D. multiflorum that has a D. congestum ID will stay in Dichelostemma as all the current IDs are in that genus.
So we really just need to consider those observations that received explicit Dichelostemma genus-level IDs. We could use atlases to retain some of these as Dichelostemma sensu stricto (e.g. WA observations), move others to Dipterostemon (e.g. observations in Mexico, AZ, NM, UT and NV), and bump many of them up to Brodiaeoideae (most of the California observations, where D. capitatum is sympatric with one or more Dichelostemma species).
In theory, we might decide that it would be best to try to add IDs for plants with only explicit Dichelostemma genus-level IDs in advance of the split, but I'm not sure that's really worthwhile. I am working through adding IDs to all Brodiaeoideae observations, but it's going to take a while longer. Meantime, if the taxon change causes some iNat observers and identifiers to revisit IDs for these observations, that's probably not a bad thing, and I don't see a scenario where making the taxon change (including the genus-level split) makes an ID actually incorrect (although some will be less precise).
So it seems my main remaining tasks are:
Ensure there are correct atlases for both genera
Create the taxon split from Dichelostemma into Dichelostemma + Dipterostemon
Check with an admin about the workload this will cause.
Commit the changes.
I don't see a scenario where making the taxon change (including the genus-level split) makes an ID actually incorrect (although some will be less precise).
One added workload that a genus split can cause, and to be aware of, is with explicit disagreements, e.g.
ID 1: Dichelostemma multiflorum
ID2: Dichelostemma (explicit disagreement)
Observation taxon = genus Dichelostemma
The taxon change could turn ID 2 into Brodiaeoideae, and it won't transfer that disagreement. So the observation will now be labeled as D. multiflorum again instead of genus genus Dichelostemma. Which may be the incorrect label...if IDer 2 is correct and IDer 1 is not. :)
Thanks both.
@bouteloua: Is there a strategy to avoid or mitigate the explicit disagreement issue you mention? Or is it just something I need to be aware of when reviewing the results?
@jdmore: Right now the destination taxa are all inactive. If I commit the changes from the bottom up I assume that means I would start by activating the new genus, species and all subspecies; then do the subspecies migrations and check the results; and then move on to the species and last the genus split. Are there any issues if the target taxa are active at the same time as the source ones?
@rupertclayton - nothing I can think of to mitigate that issue, just something to add to a like, "post-taxon change clean-up effort"
Hi again @jdmore. I just thought of a potential problem with committing the changes piecemeal. Let's say I have an observation with a couple species-level IDs as Dichelostemma capitatum and a couple more IDs as ssp. capitatum. Right now, this will have a community ID as Dichelostemma capitatum ssp. capitatum. If I commit all the changes together, it will get a new community ID as Dipterostemon capitatus ssp. capitatus. But if I only commit the subspecies IDs, this will get a community ID of Brodiaeoideae (at least for a while), because the species and subspecies IDs will conflict.
I'm thinking that the disruption caused by several thousand observers having their observations bumped up to Brodiaeoideae (even if only for a short while) probably outweighs the benefit of being able to separately review the results of each taxon change.
Yep, good point. I'll defer to @bouteloua on whether sequence makes any difference in cases like this. (Maybe only when using "Move children to output?" which it looks like you are wisely not using?) Even if the changes are committed as a group, things aren't likely to be perfectly simultaneous, so the scenario you outline above could still briefly happen.
Thanks @jdmore. I decided against "Move children to output?" because the species and subspecies names are changing Latin gender (i.e. spelling).
I need to work on atlases for all the target taxa. Presumably I can ignore those for the source taxa?
Thanks @kueda. Just realized that I need to create an atlas for the Dichelostemma genus because that's one of the outputs from the split, and to do that I really need to create atlases for all the Dichelostemma species, so I guess there's no shortcut.
you can make a genus atlas here: https://www.inaturalist.org/atlases/new?taxon_id=50653
Thanks @bouteloua. May take a couple days before I can get the time to work on the atlases.
Hello again. I have created atlases for all the target taxa. It looks like I'm ready to move ahead with committing the four taxon swaps and one taxon split in this change group.
As this will touch around 12,000 records, I would be grateful if an admin (@tiwane, @loarie, @kueda ?) could tell me if there any concerns about when and how I trigger the commit.
Many thanks, Rupert
Dang, complex. I think this looks ok. @loarie, what's the best order to commit things in when all the taxa have their own swaps like this? Starting with the subspecies?
This looks good to me except that order of operations will be important because active taxa can't have inactive parents and vice versa. (I recommend leaving "Move children to output?" on Taxonomic Swap 89050 unchecked as you have it because that won't deal with the um to us ending change properly - it should work fine manually as long as you follow the order of operations proposed here). Generally you want to first activate output taxa from coarse ranks to fine ranks and then commit taxon changes from fine ranks to coarse ranks. Here's the 6 specific steps I'd recommend:
step 1. activate output genus:
Dipterostemon 1196714
step 2. activate output species:
Dipterostemon capitatus 1196784
step 3. activate output ssp:
Dipterostemon capitatus capitatus 1196824
Dipterostemon capitatus pauciflorus 1196829
Dipterostemon capitatus lacuna-vernalis 1196831
step 4. commit ssp Taxonomic Swaps:
Taxonomic Swap 89052: Dichelostemma capitatum capitatum 59083 -> Dipterostemon capitatus capitatus 1196824
Taxonomic Swap 89054: Dichelostemma capitatum pauciflorum 79801 -> Dipterostemon capitatus pauciflorus 1196829
Taxonomic Swap 89055: Dichelostemma capitatum lacuna-vernalis 863851 -> Dipterostemon capitatus lacuna-vernalis 1196831
step 5. commit sp Taxonomic Swap:
Taxonomic Swap 89050: Dichelostemma capitatum 50654 -> Dipterostemon capitatus 1196784
step 6. commit genus Taxonomic split:
Taxonomic Split 89064: Dichelostemma 50653 -> Dichelostemma 50653, Dipterostemon 1196714
@loarie this is great sequencing information to know about for future splits like this too! Maybe we could find a place in the Curator Guide to note that sequencing can be critical, and include a link back to your comment here?
Another question I would have, when large numbers of observations are involved like this: is there any need to space out the steps in the sequence to allow indexing operations to catch up?
jdmore - yes we're working on better documentation for taxon changes. I had written this in the past which includes text on order of operations but, in addition to integrating it better into iNat documentation and making it more readable and concise I'd like to update some things (e.g. include using a split input as its output) and revisit some of the norms laid out in the doc (e.g. is a taxon change always necessary when the name change is trivial: Species oneum -> Species oneus ? Likewise, is it really necessary to split all they way up the tree when a species is reshuffled: Species A moved from Order A to Order B - do we really have to split the Genus, Family and Order too or can we incur some misinterpretation of existing IDs?) hope that tangent didn't derail too much
Re: server-side impact. The load of committing these shouldn't be an issue. We have noticed some weird bugs with some big taxon changes such as the community ID not getting updated etc. but have had trouble reproducing them and thus confirming/fixing them. It would be great to use this Blue Dicks example as an opportunity to help with that. If anyone sees anything weird please note it hear and we can try to replicate it while its fresh.
Thanks again for all your help on this.
Thanks @loarie, @jdmore and @kueda for the input. Very helpful to have that sequence of steps. As everything now appears to be in place, I'm planning to start the process this morning. Not knowing much about what's involved on the backend, my plan is leave an hour between each of the three activation steps, and then do a review of relevant observations after each phase of taxon change commits.
OK. All target taxa have been activated and I have committed all the taxon changes in ascending rank order. Based on the assurances from @bouteloua and @loarie I didn't wait long between these steps, figuring that would minimize the amount total amount of reindexing and the chance that observations would get misleading intermediate IDs because they have some IDs in Dichelostemma and other Dipterostemon.
I'll start looking at the results. (At this point only Dichelostemma capitatum ssp. lacuna-vernalis observations appear to have moved.)
Interestingly, Plants of the World Online now appears to recognize the resurrection of Dipterostemon as a monotypic genus for Blue Dicks. I can't really tell when this change was made at POWO, but I guess iNat is now back in sync and not deviating for these taxa.
And I see that you added all the taxon framework changes. Thanks @loarie.
Is it normal for an inactivated taxon's ID's to hang around for so long? I'm seeing the subspecies IDs automatically update but "Dichelostemma capitatum" IDs are still hanging around.
This page shows 822 observations still under the old name:
https://www.inaturalist.org/observations?locale=en-US&place_id=any&preferred_place_id=1&taxon_id=50654
And there are many other observations like this that should be research grade under the new name but are not because the "Dichelostemma capitatum" IDs haven't swapped and aren't being counted:
https://www.inaturalist.org/observations/69996209
Hi @paulexcoff. This is the first taxon swap I've done, so I don't know what's normal. The 822 Dichelostemma observations you mentioned are now down to a single one where the observer chose to opt-out of community ID. And the Needs ID observation you linked to is now back at Research Grade.
I've been monitoring the observation counts listed in the change group. It took all of yesterday for the update process to work through the three subspecies and the core Dichelostemma capitatum species-level IDs. The final piece was to work through the genus-level Dichelostemma IDs, which I think may now be complete. I'm not sure what performance would usually be expected, but there were more than 12,000 observations to update and various overlapping atlases to evaluate.
It did make me wonder whether the update process is multi-threaded. Usually, that would be a great way to improve performance. But in this case it could have been problematic because it was important to process these changes starting with subspecies-level changes, and moving on to species- and genus-level changes.
I'm noticing that one welcome side effect appears to be a big increase in the number of observations ID'ed to subspecies level. Previously, I had added a subspecies ID to roughly half of the Blue Dicks observations, but many of these still had community IDs at species or genus level due to the way iNat doesn't allow an infrataxa ID to "lead" the community ID (unless it's the first/only ID).
It looks like two things are happening:
First, a lot of observers and other identifiers are getting notifications about updated Blue Dicks IDs (from the taxon change) and this leads them to revisit the observation and maybe agree to the subspecies ID.
Second, because the subspecies taxon changes were processed first, the coarser IDs are now treated as having been made after the subspecies IDs, and so iNat's Community ID algorithm is willing to accept these IDs at subspecies level. Here are a couple of examples: https://www.inaturalist.org/observations/42967800 and https://www.inaturalist.org/observations/42557241
@paulexcoff can you share a URL to a specific observation? I suspect its working as intended and is just confusing due to the many nuances for how the community ID works. But it could be a bug. If you share the URL I'll take a look
Oh sorry that was unclear. Everything I was talking about last night has been fixed. I was discussing what was going on in the examples @rupertclayton shared which aren't necessarily problematic.
got it - yes. Observations with multiple IDs have an observation taxon
and a community taxon
. In most cases they are the same, but because people preferred observations with 1 species level ID followed by 1 subspecies level IDs becoming research grade with the observation taxon
at the species level rather than rolling forward and becoming needs_id with an observation taxon
at the subspecies level, there's a caveat where a single subspecies level ID's doesn't roll the observation taxon
forward. So as you say the order of IDs does impact that caveat. But the community taxon
is the same either way
Congrats, @rupertclayton ! The end of a long journey.
This would represent a deviation from POWO (http://plantsoftheworldonline.org/taxon/urn:lsid:ipni.org:names:82063-2), but Jepson has adopted this change: https://ucjeps.berkeley.edu/eflora/eflora_display.php?tid=108599. The paper underlying this change is
Preston, R.E. 2017. New nomenclatural combinations for blue dicks (Dipterostemon capitatus; Asparagaceae: Brodiaeoideae).
Phytoneuron 2017-15: 1–11. Published 22 February 2017. ISSN 2153 733X. http://www.phytoneuron.net/2017Phytoneuron/15PhytoN-Dipterostemon.pdf