Flagger Content Author Content Reason Flag Created Resolved by Resolution
bouteloua northern seaside goldenrod (Solidago sempervirens)

needs to be split

Nov. 16, 2019 15:08:20 +0000 Not Resolved

Comments

From @astrobirder:

"There's a broad overlap zone between the two (MA to VA) but south of VA all Seaside Goldenrods should be S. mexicana and the ones north of MA (and inland in the Northeast) should be S. sempervirens. Is it possible to switch the ones south of the overlap zone to S. mexicana? The overlap zone is going to be a bit messy since the two species are pretty similar (separated by the number of flower rays and other flower details), but they're both in the subsection Maritimae, so the overlap observations could be set to that to make cleaning up the records a bit easier.

here's the eFloras key to the two species (they have them still at subspecies):
http://www.efloras.org/florataxon.aspx?flora_id=1&taxon_id=250067570"

Posted by bouteloua over 4 years ago

I have created the sectional and subsectional taxa for Solidago based on Semple's web page: https://uwaterloo.ca/astereae-lab/research/goldenrods/classification-and-illustrations

Maritimae is now fully populated (although we still have a few things at infraspecific rank that Semple separates as species), so the swap as proposed here can go forward by creating atlases for both taxa. I will work on placing other Solidago species into the correct subsection.

Posted by choess over 4 years ago

Can this be resolved?

Posted by rynxs about 1 year ago

have those atlases been made?

Posted by sbrobeson 6 months ago

I have set up the atlasing based on Semple's website, so it should be possible to go ahead with the split of S. sempervirens between S. sempervirens and S. mexicana. The introductions/waifs in Europe and outside the upper Midwest will get raised to section Maritimae, as will material around the Chesapeake and along the barrier islands down to Hatteras. South of there observations should shift smoothly to S. mexicana on split.

Posted by choess 6 months ago

@choess can the introduced populations be atlased as well? Why leave them out?

Posted by rynxs 6 months ago

The ones introduced in the Midwest (as shown on Semple's atlas) are atlased but I haven't checked the ones further afield, which in theory could be either taxon. It doesn't look like a huge number but should probably be tackled by a good goldenrod person.

Posted by choess 6 months ago

Ah, sorry. Misread your comment, and thought you were leaving out the upper Midwest observations from the atlases.

Posted by rynxs 6 months ago

split drafted, please edit as needed

Posted by sbrobeson 6 months ago

@ddennism @danielatha @jayhorn @dryopteris2 @sk321
if you have any input on range limits for S. sempervirens and S. mexicana north or south, in trying to facilitate this auto-reassignment of especially the S. mexicana name to records of S. sempervirens sensu lato.

Posted by sbrobeson 6 months ago

I vaguely remember the prospect of S. mexicana in NY being touched on in Werier's recent catalog. Give me a day or so to check this before you commit.

Posted by ddennism 6 months ago

Also the atlases need to accommodate known southern intro(s?) of S. sempervirens before this change should be committed.

Posted by ddennism 6 months ago

I'm in no rush to commit (quite the opposite, I've become very wary of committing changes, such that other curators often go ahead of me to use things I've drafted), so take what time is needed to research or point out necessary edits.
is there a good source for southerly introductions of S. sempervirens?

Posted by sbrobeson 6 months ago

Cassi, thank you for including me on this thread that includes important taxa in the range of where I live and work. I am not an Asteraceae expert, but I am very familiar with these plants.

The name Solidago mexicana already exists at the species level in the taxon backbone with 2,565 observations so named (1,212 RG). There are 10,882 observations with the name Solidago sempervirens (6,957 RG). Analysis and mapping based on these observations as currently named already reveals the north/south distinction.

It is true that a subset of the 10,882 Solidago sempervirens are actually Solidago mexicana and any mapping or analysis based on these will be in error to the extent the subset are incorrectly named.

Is the purpose of this split/switch to "fix" the subset of Solidago sempervirens observations that are actually Solidago mexicana but currently named Solidago sempervirens?

I submit that the wisest course to remedy this problem is to go through each of the suspect observations and manually assign the correct name based on the character states provided by experts in their descriptions and keys. The characters to distinguish these two taxa include leaf size, inflorescence size, capitula size and number of ray and disk flowers, any and/or all of which should be discernable in the vast majority of observations. An expert should be able to correctly identify all but a handful that are too blurry. Geography could then be used to get 100 percent of the observations identified to one or the other taxa. Two or more experts working together could get them all back to Research Grade. This will ensure that observations of "real" Solidago sempervirens and Solidago mexicana that may be out of their expected ranges will be correctly identified and mapped, ensuring the research integrity of iNaturalist data. No one ever said that good taxonomy wasn't tedious much of the time.

Forgive me if I'm misunderstanding the process and result of this proposal, but changing the assigned names of these problem observations mechanically in a batch change based solely on geography is not good scientific practice for obvious reasons.

All classifications are hypothesis and should be continuously tested with new evidence. One of the most marvelous things about iNaturalist is that it accumulates a treasure trove of objective, verifiable data to do precisely this.

Posted by danielatha 6 months ago

@danielatha -
"Is the purpose of this split/switch to "fix" the subset of Solidago sempervirens observations that are actually Solidago mexicana but currently named Solidago sempervirens?"

Usually taxon changes like the one @sbrobeson proposed above are meant more to restore the sense of an earlier ID moreso than blanket "fix" old IDs, although that is a side-effect. I agree that those blanket "fixes" can be dangerously circular, which I assume is the obvious reason you refer to. But sometimes it's the least-worst option, because we also have a duty to try to preserve the sense of previous ID'ers.

For example in this instance:
This taxon change was first floated by cassi almost 4 years ago. At that time, "S. mexicana" had only very recently (< 1 mo.) been added to the iNat taxonomic backbone and had very few observations. In that context nearly all southern "S. sempervirens" identifications were probably S. mexicana. Even if an ID'er thought a given plant on, say, the Gulf Coast was S. mexicana (or would have thought so if they were aware of this taxon), they would have been forced to use "S. sempervirens" because the other name was simply unavailable on iNaturalist.[1] More likely, they were simply using the wide sense of S. sempervirens, which included the mexicana populations. The taxon change, had it been committed very soon after the instatement of "S. mexicana", would have represented our best effort to preserve those earlier ID'ers intention to indicate an ID of Solidago sempervirens sensu lato in any area of range overlap (by bumping up their ID to the subsection rank) or switching it to S. mexicana.

But now we're left with kind of a mess because the taxon change was not committed immediately after S. mexicana was added. IDs of "S. sempervirens", particularly those left in the intervening four years could now mean:

S. sempervirens sensu lato
S. sempervirens sensu stricto (if occurring far south, then essentially asserting that they are introduced / otherwise adventive)
S. mexicana but it wasn't available
Some alternative species concept of S. sempervirens that doesn't admit a substantive difference at any rank between the supposed mexicana and sempervirens s.s. populations.

[1] This is complicated by the existence of subspecific taxa within S. sempervirens at that time, however.

Posted by ddennism 6 months ago

This all assumes that the range is so well-known that there will be presumably very few mis-applications of the new taxa based on location.

From my impression of these taxa, though, this doesn't seem to be the case.

Semple's latest map (I think - I don't have access to their cytogeo paper from 2022):
https://www.phytoneuron.net/2016Phytoneuron/73PhytoN-SolidagosempervirensComplex.pdf (page 9)

is mostly the same as the one from his website, referred to above. It does show one LA introduction of S. sempervirens sensu stricto, and in the text, they refer to other southern introductions of S. sempervirens s.s. in Mexico.

They also note that midwestern highway-borne S. sempervirens are "atypical" - and, unfortunately, they are apparently atypical in one of the few ID characters they provide to distinguish between S. mexicana and S. sempervirens s.s.

They disparage the use of some of the ID characters used by earlier authors, finding that, particularly in the case of floret number characters, that there was so much overlap that the characters were essentially useless.

There's also: "in a few cases final identification to species of some specimens in the area of range sympatry of S.
sempervirens
and S. mexicana in North Carolina to Maryland was difficult". (page 9 - caption)

In Semple's latest key only one character is given at the couplet separating S. sempervirens from S. mexicana:
https://www.phytoneuron.net/2019Phytoneuron/48PhytoN-Solidagouliginosa.pdf (page 41)

"2. Upper leaves somewhat reduced from mid and lower leaves [to S. sempervirens and others]"
"2. Upper leaves much reduced from lower leaves [to S. mexicana and others]"

Elsewhere, differences in leaf width, inflorescence shape and capitula size are also given. I'm not sure that these characters are really discernible from photos, particularly if the experts working from specimens could not assign several specimens all the way from NC to MD. Maybe they are.

TLDR - I think our best course of action is to forego the taxon split and just try our best to apply these muddy characters when possible. There are too many unknowns with respect to the real ranges of these taxa (southern intros and northern extent of S. mexicana - the checklist I talked about above cites it all the way up in Long Island!) to use an atlas-informed split.

Posted by ddennism 6 months ago

I appreciate the additional information and perspective of someone that knows the Goldenrods and the details of this particular case better than I do. Full disclosure, I had not read Semple's Phytoneuron paper. I was perhaps naieve in assuming that the characters and values in Weakely's key were a consensus of current thinking and reliable as presented. And maybe they are, but now I know the situation is more complex than just two couplets in a key. Considering the comments above and a quick review of Semple's paper, I am less optimistic even experts could apply names confidently to most of the photo observations in this complex. Nevertheless, they may try and since both names are available for use, future observers and identifiers can apply the name of their choice. I agree with the first sentence of Daniel McClosky's last paragraph directly above.

I have worked with herbarium specimens all my professional life. I have a deep respect for the traditions and practices that we inherited from our predecessors and I think they have served us very well. Observations in iNaturalist are not exactly analogous, but I try to apply the same basic principles, procedures and courtesies that I would with a specimen.

That being said.... Resolving what we think are problems with the identification of past specimens or observations by changing the actual name applied by someone else is not good practice in any situation. When choosing a lectotype for a name when the original author designated more than one specimen or none, it is recommended (if not mandated) that the lectotypifier use only objective, verifiable, internal evidence as left by the original author to chose among the materials that were available at the time to the original author. No one would dream of removing an annotation made by the original author because we thought they were in error or they lacked some external evidence. I know from years of being an editor of the taxonomic journal Brittonia that many lectotypifications have been rejected in peer-review because the lectotipifier said "The original author should have known about this piece of information" or "If they had know about X, they would have done Y".

Correct me if I'm wrong, but it seems to me that this split (and others like it) would actually change the name applied by someone previously to something else. So if the observer or an identifier applied the name Solidago sempervirens, in some cases that name would change to Solidago mexicana or possibly one of the other taxa in the complex, as if they had chosen that name themselves. That's called a correction. In herbarium and taxonomic practice, the only acceptable correction of prior work is to "correct" the spelling of a name to facilitate information retrieval.

A far more honest, transparent and scientifically sound course is to simply newly identify the specimen/observation with any name we think is more right and leave the original name unaltered. That is how it has been done in herbaria for over two centuries. If it is a machine driven identification based on an algorithm, that should be made clear and explicit. Annotating the specimen or observation in this way remedies any and all harm that could come from a specimen carrying a "bad" name, no matter how egregiously wrong it might be.

This process may be tedious and time consuming, but that is the nature of taxonomy. I make it a rule for myself never to apply a name to an observation or specimen that I have not actually examined myself. I still make mistakes, but fortunately, most I catch early in the process. And because I actually saw the specimen/observation myself, I can learn from my mistakes.

Posted by danielatha 6 months ago

"That being said.... Resolving what we think are problems with the identification of past specimens or observations by changing the actual name applied by someone else is not good practice in any situation."
"Correct me if I'm wrong, but it seems to me that this split (and others like it) would actually change the name applied by someone previously to something else."

This might be a partial misunderstanding. The system always preserves the record of the previous ID. Users can remove their IDs, not the system. It simply adds an additional ID induced by a taxon change, and this is accompanied by a note explaining that the system added this ID, not a user. It then deprecates the previous ID for the sake of iNat's agree-disagree conflict resolution scheme, the idea being that we know, logically, that the old ID had to have been used in a different sense (that of the previous taxonomic scheme).

I don't like this either, by the way. My view is that taxon changes like the ones above are an unfortunate logical necessity given iNat's (1) use of a single, rigid hierarchy as a taxonomic scheme. (e.g. there's no room for a "sensu" system) and (2) desire for an automatic conflict resolution scheme (i.e. the "community ID" system). To see why this is a logic problem for iNat, consider the consequences of leaving up, unaltered, the previous IDs added by users of a different taxonomic scheme. We would quickly lose track of what the names meant, as they were added under any number of previous taxonomic schemes. This would destroy the ability of the agree-disagree system to function, among other major site consequences.

My personal preference (like yours, I think) would be something that functioned more like annotations on herbarium specimens. In a digital context I guess this would basically mean open-ended, time-stamped text tags. This would be a major change to the design, organization, and function of the website.

In any case, it sounds like you are opposed to all taxon changes that result in any change to any user's ID. It's a great discussion for the iNat forum where more people will see it, but it might be getting outside the scope of this particular issue between S. mexicana and S. sempervirens s.s. I hope you might post some of your thoughts there.

Posted by ddennism 6 months ago

I think it's best to proceed with the split. I understand Daniel's concern, but this is the mechanism that iNaturalist gives us for changing the circumscription of taxa, as we are doing here (S. sempervirens s.l. to S. sempervirens s.s. and S. mexicana).

Observations in the area of known range overlap (around the Chesapeake) will be bounced back up to sect. Maritimae and will have to be manually examined anyway. The range map is recent from Semple, so I don't think there's any reason to think that it's grossly inaccurate. Obviously someone diligently examining observations after the split might turn up additional out-of-range specimens, but the casual identifier to the north or south of the zone of sympatry is likely to identify any seaside goldenrod as their "local" taxon even if it doesn't quite fit, so I don't think this would make the situation significantly worse. Going through with the split will give us a more coherent circumscription of the two taxa than we have now, rather than waiting for a perfect reidentification of every observation (which will be very difficult to pull off anyway due to inactive identifiers and so on).

Posted by choess 5 months ago

Thank you Chris and Daniel for the additional comments. I still think the preferred method to solve the "problem" is to create the entity Solidago mexicana in the taxonomic backbone and manually transfer the appropriate observations to it based on an expert analysis of the empirical evidence. The taxonomy page for each taxon should be updated as well. Other than the time and effort involved (which may be lacking), I don't see how this method is incompatible with the iNat taxonomic and identification scheme. And I don't see how that method compromises the community id system or the iNat taxonomic concept scheme. Theoretically, every observation would have a new identification of one or the other name and anyone so wishing could agree or disagree with that new identification. If they are unsure about the current scheme, they could simply refer to the taxonomy pages for those taxa This would avoid the very real issue of someone's identification (including the observer themselves) being changed without their knowledge or consent.

But I acknowledge that takes time and effort that may not be available and many are unaware of past and current taxonomic species concepts. And I acknowledge that the current circumscription of these taxa in iNaturalist is flawed.

With the above reservations, I do not oppose the split. It will be instructive for me to watch the process and the results.

Posted by danielatha 5 months ago

It compromises the community ID system because it conflates different senses of "Solidago sempervirens".

Posted by ddennism 5 months ago

I agree with the principle that we want to assist the community ID system by using the mechanism that we have available, partly (again) because undoubtedly some past identifiers may have intended what we now understand to be S. mexicana. (I can't offer as much on the particulars of this case as I'm no goldenrod expert!)

and, I think, besides -- it will still remain an option for us and for the original observers to go back and re-identify observations even as the same species they were listed as before. committing a split doesn't mean setting anything in stone for any observation (or group of observations)!

in case it was unclear -- S. mexicana is already present in the iNat taxonomy. a "taxon split" does not create any new taxon entries, it can only be used to redistribute observations among known/existing taxon entries.

Posted by sbrobeson 5 months ago

@danielatha your method is definitely not wrong! It used to be the only option before iNat added atlases, and I've done it that way myself. (Phegopteris excelsior is essentially sympatric with P. connectilis, so manually looking for and reidentifying it was the correct way to go; an atlased split would have just blown up a lot of P. connectilis IDs to no purpose.)

It's really a numbers game: the larger the number of observations to be re-determined, the more likely it is that some will not get determined to the current taxon circumscription due to observers leaving, not responding, etc. When there's a very large number of observations and the taxa are mostly allopatric, the atlased splits probably get more observations to the correct taxon than an exclusively manual redetermination. But your points about what careless taxon changes can do are well taken.

Posted by choess 5 months ago

any sense that this split should proceed or not?

Posted by sbrobeson 28 days ago

Add a Comment

Sign In or Sign Up to add comments