From the FNA treatment, and

http://www.phytoneuron.net/2017Phytoneuron/18PhytoN-Solidagospeciosacomplex.pdf

S. speciosa (s.l.) --> [S. speciosa (s.s.), S. jejunifolia, S. pallida]
S. speciosa var. speciosa --> [S. speciosa (s.s.), S. jejunifolia]
S. speciosa var. jejunifolia --> S. jejunifolia
S. speciosa var. pallida --> S. pallida
S. speciosa var. rigidiuscula --> S. rigidiuscula

S. speciosa var. erecta --> S. erecta (currently exists in our taxonomy)

S. rigidiuscula and S. jejunifolia taxon changes drafted here:

https://www.inaturalist.org/taxon_changes/84482
https://www.inaturalist.org/taxon_changes/84483

S. pallida is already in the iNaturalist system.

S. speciosa subsp. speciosa to S. speciosa sensu stricto might be tricky.

Plus, many things in S. speciosa in the current sense will probably need to move - maybe can use an Atlas for this, at least in part. The new species are mostly allopatric, but not completely, and maybe not completely known.

(all this is from the Semple website at University of Waterloo and papers cited therein)

Regarding S. speciosa subsp. speciosa:

According to the old FNA treatment, this subspecies includes the varieties rigidiuscula and speciosa (var. speciosa including the jejunifolia entity).

Based on the locations of some of the observations of S. speciosa subsp. speciosa already on iNat, some people really are using the subspecies in the umbrella-of-varieties sense.

Therefore, the taxon change should be:
S. speciosa subsp. speciosa --> S. jejunifolia, S. rigidiuscula, S. speciosa (s.s.).

Drafted here:
https://www.inaturalist.org/taxon_changes/86154

[EDIT: 3/2/23 - after a surprise, unilateral taxon change by another curator, this change is no longer possible, so will be deleted. It's a shame - some people were clearly using subsp. speciosa in a manner inconsistent with the undiscussed taxon change (e.g. for then-var. jejunifolia, etc.); I don't understand why it was committed. I'll have to try to go through those former IDs - there aren't too many of them - and try to discern the users' actual intention, and correct if necessary.]

Thanks!

The main change is drafted here:
https://www.inaturalist.org/taxon_changes/86155

In that paper it appears that Semple has data that could be used to test the hypothesis that these are separate species, but does not do so. Interpreting discriminant analyses in taxonomy is difficult, because what you want to know is usually, "Are these distinct species?" while the analysis tells you, "If we assume that these are distinct species, can we use this data to tell them apart?"

An extreme example to illustrate the problem is: suppose you have a single character, ranging from 0 to 1 and measured to the nearest 0.1, with perfectly continuous variation. There are a dozen individuals at 0, a dozen individuals at 0.1, a dozen individuals at 0.2, a dozen individuals at 0.3, and so on. You draw a line across that variation and separate the individuals into two groups, e.g., "0.3 or less is group A, 0.4 or above is group B", and run a discriminant analysis. The analysis should tell you that the two groups are distinguishable with 100% accuracy. The answer to, "If we assume that these are distinct species, can we use this data to tell them apart?" is, "Yes, perfectly!" That answer will be the same no matter what value is used as a break point to create two groups. Now suppose you assign a few individuals with values of 0.4 and 0.5 to group A, and a few individuals with values of 0.2 and 0.3 to group B and run a discriminant analysis again. The two groups will still be distinguishable, but the accuracy will be lower, presumably 90-something%.

Basically, 100% accurate assignment tells you that your groups do not overlap, but does not tell you whether or not you have distinct species. Less than 100% accurate assignment tells you that your groups overlap, meaning that the putative species are not distinct in the data used in that analysis. That doesn't necessarily mean that they aren't distinct species, but if you were weighing the evidence for and against on a balance scale, you'd put that analysis on the "against" side.

I interpret Semple's paper as providing evidence that these are not distinct species.

Not sure if you want to maintain a complex, the split entities are now encompassed by subsect. Erectae so we may not have to. I just swapped into S. jejunifolia, which means we should be entirely lined up with POWO. Is a split still going to be committed?

Reviewing this process, the state- and province-level atlases overlap to such a large extent that I think the next step is to assemble county-level atlases instead. This will be a lot of work, but I think the benefit of having fewer observations bumped up to a confusing taxon like "complex" or "series" would make the work worth it.

What do others think? Is such a split advisable or a waste of time? If advisable, is the above scheme a good one?
@rayrob, @aarongunnar, @evan8, @dustygram, @wildlandblogger, @eknuth, @tuq, @kjsytsma

The county-level data exists only, AFAIK, as it would be translated from the yellow-shaded range maps that Semple draws. (i.e. by superimposing his drawings over a map with county boundaries)

Atlases are now ready for the split-up of S. speciosa sensu lato.
The new atlas links are here:
https://www.inaturalist.org/atlases/30151
https://www.inaturalist.org/atlases/30152
https://www.inaturalist.org/atlases/30153
https://www.inaturalist.org/atlases/30157
and are based on maps here:
https://uwaterloo.ca/astereae-lab/research/goldenrods/classification-and-illustrations/solidago-jejunifolia
https://uwaterloo.ca/astereae-lab/research/goldenrods/classification-and-illustrations/solidago-pallida
https://uwaterloo.ca/astereae-lab/research/goldenrods/classification-and-illustrations/solidago-rigidiuscula
https://uwaterloo.ca/astereae-lab/research/goldenrods/classification-and-illustrations/solidago-speciosa

I also created a species complex for just these four species, even though this does not correspond to "series Erectae". I think we can get away with a smaller complex, composed only of the segregates of the old S. speciosa, by citing the Semple paper above.

I will commit the split after a few days pending no objections.