Holodiscus is currently a taxonomic mess, even in POWO (I'll explain). No two modern treatments of the taxa north of Mexico agree on how to treat the five taxa that have been recognized there (discolor, dumosus, glabrescens, microphyllus, sericeus).

To illustrate, I've posted the following 1-page PDF chart showing how various modern floristic treatments (Flora of North America, Jepson Manual, Intermountain Flora) compare against each other, and with POWO and iNaturalist. (The recent Oregon and Pacific Northwest checklists seem to follow FNA for what they cover.)

https://drive.google.com/file/d/1roJ78y4OGYTrO6_FfXEWFFPvszbAxE3h/view?usp=sharing

One thing to note in that chart is that POWO includes dumosus (in different combinations) under both H. discolor and H. dumosus. So clearly their treatment is not yet ready for prime time.

Flora of North America would be the only other complete treatment for the North American taxa. Should we deviate from POWO and adopt FNA for now, or wait for POWO to get things cleaned up first?

If we do deviate, the resulting taxon changes will potentially affect thousands of observations, so we should try to get some community consensus first. I am @ tagging the following folks as a start, but feel free to @ tag others in your comments who might have input. (And of course, no need to respond if this doesn't interest you.)

@aaronliston
@ahowald395
@ajwright
@aspidoscelis
@cyberflora
@deantaylorbotanist
@dhwilken
@gentilcore
@jdjohnson
@jrebman
@rick_williams
@rojosmojo
@sedgequeen
@sganley
@stevejones
@tmessick
@twainwright
@walterfertig
@wisel

To my knowledge, nobody is working on Holodiscus, let alone with the fine-toothed comb it deserves.

Ley's 1943 monograph is not a great help, ignoring apparent interesting variation and making much of what doesn't seem at first glance to be important, but it will be interesting to see if her ideas are vindicated with modern methods after the lumping craze of the 60s through the 90s. Additionally, the variety H. dumosus var. cedrorus was recently published.

In my opinion, there is a clear need for much new primary research on the genus, and debating it extensively here to me verges uncomfortably on trying to do that research without requisite data. No treatment currently appears adequate for the genus.

So, the question to me is, what is the best solution while we wait, knowing that every option is bad?

To me, FNA seems the best treatment currently, and POWO poorly reasoned. (POWO in general is suspect for North American plants, pursuant to some apparent patterns in the treatments they choose to follow.) However, I am not sure that making any change that will affect a large number of observations, when doing so will not provide more clarity to the community or to workers (as in this morphologically difficult complex), is justified when all options are unsatisfactory.

I don't have a lot of experience with the taxonomy of this group but from working on the Lupinus lepidus complex last year, I'm sure that POWO doesn't keep up with the FNA treatments. I think POWO is generally great but they don't have infinite resources to keep up with all of the taxonomy changes around the world.

From a data management perspective, it's a lot easier to maintain the finer divisions now and lump them together in the future if genetic data support it than it is to lump everything now and manually separately them later if genetic research supports that instead.

I just saw this flag in a comment after trying to fix a whole bunch of observations called H. dumosus in CA as all the treatments I looked at said only H. discolor was in California except for one variety of H. dumosus, which was recently put in H. discolor. Anyway, key to California Holodiscus from 2020 below. Whether it is right or wrong, I don't know and it sounds like no one else knows either.
https://www.mapress.com/pt/article/view/phytotaxa.477.2.13

Yeah, and it should be noted that the DNA sequences they use to make their decision to consider them all one species often can't distinguish between closely related taxa, but I guess it confirms they are likely at least somewhat closely related.

That does add another wrinkle... I looked at the protologue and Spiraea discolor var. dumosa does appear to possibly be an illegitimate name. It may hinge on 1) Watson's use of synonymy to indicate prior circumscription vs. typification, as was sometimes done in America and still in fossil works, and 2) his apparent preference to not use autonyms?

I'm now wondering why we're not using the basionym Spiraea ariifolia Sm. (-> Holodiscus ariifolius (Sm.) Greene), since I can't immediately find a reason why it's not valid or legitimate...

Riiiiight, S. discolor is a Pursh name, 2-3 years before Smith.

I think I see the issue now. S. dumosa was a Nuttall manuscript designation. Torrey & Gray did not accept it in their flora, only citing it in synonymy under S. ariaefolia var. discolor, and so did not validate the designation. However, this SHOULD mean that S. dumosa Nutt. ex Hook. is not an illegitimate homonym, since no valid homonymous name antedates it - however, it is illegitimate on the grounds of including the type of S. discolor, since Hooker accepts Torrey and Gray's circumscription of S. ariaefolia var. discolor (which include's Lewis's specimen) even though he does not cite the Lewis specimen himself.

This means that Watson erected a new name in S. discolor var. dumosa, but one which cannot be combined at species rank in Spiraea because the Hooker name blocks it (it is not blocked in Holodiscus). Its legitimacy is also debatable itself since it is possible that Watson used it where we would use the autonym - he cites the Torrey & Gray circumscription like Hooker, but it's less clear in this case because he correctly uses Pursh's name for the overarching species. So, its combination H. dumosus (S. Watson) A. Heller may also be illegitimate - and it unfortunately antedates H. microphyllus Rydb. by six years.

It's probably resolvable as I can only assume someone has previously done work explaining Watson's working methods. I think it probable that all use of any instance of "dumosus" will have to be discarded.

I noticed that POWO has now updated its treatment of Holodiscus (https://powo.science.kew.org/taxon/urn:lsid:ipni.org:names:1041427-2), eliminating dumosus and considering everything north of Mexico to be discolor with 4 varieties: cedrorum, discolor, glabrescens, microphyllus. Oddly, they consider glabrescens to only occur in California, which contradicts FNA and others. H. dumosa is synonomized with discolor at the species level instead of the variety level, which may require some finessing of any taxon swaps.
Personally, I would be fine with iNat adopting this as an internally-consistent scheme, although it does not include var. sericeus and adds cedrorum, which some may find objectionable.

Sounds reasonable to me

Thanks all, I went ahead and committed the swaps.

Checking the POWO Holodiscus records, I found Holodiscus discolor var. dumosus Maxim. ex J.M.Coult. included as a synonym in Holodiscus australis A.Heller, not as any variety of Holodiscus discolor. The iNaturalist, however, does not contain H. australis, neither suggests any replacement for H. discolor var. dumosus.
Should it be synced with POWO again?