split into |
tagging in some frequent observers / IDers / experts:
@kalelevin, @grnleaf, @seanblaney, @wanderingeden, @alexuscomplexus, @sekihiker, @ajwright
@ajwright - I think I grabbed that from the old Russell (1955) treatment:
"Viola macloskeyi subsp. macloskeyi is found only in California, extreme western Nevada, and in Oregon, where it meets the much more widespread V. macloskeyi subsp. pallens [now V. minuscula], which occurs from northernmost Quebec and Labrador south to the Great Smoky Mountains in Tennessee and North Carolina and west to the northern and middle Rocky Mountains, the Pacific coast states, and British Columbia."
IIRC, Kim Blaxland said something similar on her website.
If you're aware of a better (or conflicting) source, please let me know so I can update the atlases (or reassess whether it is wise to do an atlas-informed split). The type of V. macloskeyi is from Mt. Hood, which is awfully far north.
I think it's possible that the western populations of V. minuscula may be something else, but it makes more sense to me to lump them into V. minuscula until the question is resolved. That's following Ballard et al. (2023), who do currently place them in V. minuscula:
"Western North American populations are tentatively included [in V. minuscula] but are not typical and require reexamination."
I want to look at the full Hitchcock & Cronquist treatment for the NW, but the 1976 printing of the field manual gives the ranges as:
"BC to Al[ber]ta, s in mts to s Cal [...] var. macloskeyi"
"BC and W[ashingto]n to Atl[antic] coast, Can to se US, and in RM to Colo [...] var. pallens"
So, in my estimation there are a few things preventing split, at least a simple and botanically correct one:
Russell's concept of V. macloskeyi excludes the type of that name, which as you note was collected outside the range given by Russell, and which the protologue explicitly states has "margin obscurely crenate-serrate".
Based on Northwestern sources, it's not clear whether the characters that separate V. macloskeyi auct. non F.E. Lloyd from V. minuscula form a cline.
Whether or not there's a morphological cline, morphotypes appear to be interfingered in the northern Rockies and NW, hence Ballard et al.'s uncertainty about the western extent of V. minuscula.
In general, this seems to be another case of there being no detailed work on the northwestern violets (@kalelevin notwithstanding!), and no clear answer to the many taxonomic gray areas. Unfortunately, this split implicitly recognizes a botanically untenable concept of V. macloskeyi, which is rather beyond our capabilities to fix - @kalelevin do you know if there's a basionym from the Sierra, and whether or not you could publish on elevating it?
I don't see why Russell's concept of V. macloskeyi necessarily excludes Lloyd's type. Mt. Hood is at the northern end of Russell's stated range, not outside it. His morphological concept also matches - I guess you could quibble about how obscure "obscurely" should mean, but I personally wouldn't consider that a conflict. Neither, apparently, did Marcussen et al., (2012) (who found that V. minuscula and V. macloskeyi weren't even sister taxa!); Ballard et al. (2023); nor Kew's POWO. We kinda have to follow POWO whenever possible.
That's interesting (and disappointing) that the field manual gives a much larger range for V. macloskeyi var. macloskeyi. If it's really true that there are so many instances of V. macloskeyi in the strict sense so much farther north than Russell's range (I couldn't find any myself on iNat!) then that will mean bumping many more observations up to Stolonosae. That'd be a bummer. Does the Field Manual draw different morphological differences between the varieties than Russell did between his subspecies?
No matter what, I still think the prudent thing is to make a split in agreement with the taxonomic opinion iNat defaults to (POWO). If some future publication finds that Russell's concept of ssp. macloskeyi has the wrong basionym or whatever, then we can easily make a one-to-one taxon change to reflect that. If some future publication finds that Russell's circumscription is more seriously in error, and the taxa need to be remodeled in some other, more complex way, then we can accommodate that too. The proposed taxon change doesn't destroy much taxonomic granularity, unless we have to draw the range of V. macloskeyi way farther north, overlapping considerably with V. minuscula.
Just as a reminder, the status quo: At least two entities are lumped together in a mega-species ranging from the Dominican Republic to Nova Scotia to British Columbia to southern California, despite strong molecular evidence showing that the Oregon-Californian pops aren't even particularly closely related to the rest.
Do you have a non-Cronquist source for the existence of interfingered populations of V. macloskeyi sensu stricto north/west of the Columbia River?
I just went back to Russell (1956, Am. Midl. Nat. 56:491-503), and you're right, I had missed his X in Fig. 9 for the type specimen of V. macloskeyi, the northernmost of his stations. In his 1955 paper, he mentions having seen the type at NY (this sheet: https://sweetgum.nybg.org/science/vh/specimen-details/?irn=702290), which has 14-16 teeth per leaf.
Additionally, he takes pains to point out (1956: p. 499) that there is "a strong north-south cline" in the variation between his V. macloskeyi and the now-V. minuscula, which he attributes to hybridization. As such, Russell (1956) does not support a clean geographic split between the two, and in fact explicitly states that such a split does not exist. Further, his Figs. 10, 11, and 12 imply that the type of V. macloskeyi is partway (rather, most of the way) along that cline, and absolutely not part of the morphological pole in the more southerly California populations. This may require that the southern form be separated at equal rank, but again and agreed, that is outside the scope here.
Marcussen et al. (2011) recovered fairly low divergence within the Plagiostigma violets, and use a network reconstruction method with low citation numbers (30 on Web of Science, so I'm less familiar with its ins and outs). Their experimental design was well-suited to their target question, but recently-diverged species relationships in tetraploid Plagiostigma might best be considered by-catch in need of more targeted analysis before their genomic history can be considered sorted out. Additionally, their sampling of "V. pallens" consists of three samples from New England, and their sampling of V. macloskeyi was one sample from the Sierra Nevada, so 1) their analyses cannot speak on the cline and putative hybridization presented by Russell (1956), and 2) their results are indistinguishable from sampling-dependent geographic structuring which may not hold up in central and western V. minuscula.
I understand the utility of using a global aggregator as a standard, and the technological and social utility of having strong data bins. Here, though, the bins don't appear to exist where they meet geographically - if there are data on iNat to contradict Russell (1956), they ought to be quantified and published, which may be difficult without a standardized (pressed) leaf presentation. The upshot appears to be that when, not if, the local (eastern North America) change which has been propagated to the global (POWO) database is implemented here, all of the cline indicated by Russell (1956) and not examined by Marcussen et al. (2011) will need to remain at Stolonosae.
Thank you for bringing up the 1956 paper! I had totally missed that; I was working off the assumption that his 1955 treatment summarized his overall view of the genus. But having now read the 1956 paper - it still does. I've now altered the atlases to slightly expand the zone of overlap (whether this is actually a cline or not^) reflected in that paper.
https://www.inaturalist.org/atlases/85451
https://www.inaturalist.org/atlases/85455
I still don't see evidence of V. macloskeyi sensu stricto north of the Columbia River, so luckily it seems like the range of overlap can still be contained to mostly Oregon.
As for the Marcussen paper (2012) - why do you need the PADRE network reconstruction to see the phylogenetic structure they found between V. pallens and V. macloskeyi? Isn't this presented in the ML GPI tree earlier in the paper, where they recovered a similar non-sister relationship on both homeologs? I agree that it's frustrating that they didn't include a northwestern sample of V. pallens, but I do feel that this is still strong evidence that at least eastern V. pallens is not conspecific with CA V. macloskeyi - which necessitates a split.
As for the type of V. macloskeyi - those margins look obscurely-crenate to me, which matches Russell's concept of subsp. macloskeyi. It looks like there are some Greene names for Californian plants (Viola anodonta and Viola parnassifolia) that could end up being the substitute for V. macloskeyi if Lloyd's type is found inapplicable for whatever reason later.
So - are we in agreement that the zone of overlap is represented on the atlases?
.
.
.
^It's true that Russell said he found a N-S cline of phenotypes between his subspecies in his 1956 paper, but his pheno-cline maps aren't that convincing to me. Most of them show little to no variation in phenotype across the range where he mapped his subspecies as co-occurring, but then a sudden occurrence of ssp. macloskeyi-only traits well within its Californian range. If you overlay these steep "clines" over his map of sampling locations, you'll see that most of these clines were actually informed by apparently very few samples with intermediate traits. They're just too steep. Why so much phenotypic uniformity over the region of supposed subspecies co-occurrence? This could be interpreted as evidence of rampant admixture, as Russell seemed to think, but it's weird. It's like there's a stabilized Oregon-form, but this isn't reflected in his subspecies assignments. The only pheno-cline that unambiguously supports a smooth N-S cline along the zone of sympatry is Fig 15 for leaf crenation/tooth number. However, even this one is difficult to interpret, because he omitted all the northern subspecies data from that figure. What a weird thing to do! I assume that adding-in the northern subspecies data to that figure would only add more contrast to the cline, making it more obvious - so why didn't he? Maybe the rest of the data is in his dissertation, I don't know.
By the way, if I can read into your responses, I share your frustration with this change. I would've also preferred to wait to break up V. macloskeyi. Unfortunately, V. minuscula was added by another curator prematurely, and without the necessary concomitant break-up of V. macloskeyi. We can't have parallel taxonomies on the site. I'm trying to seek out the path of least disruption here, which means minimizing the number of IDs that get bumped up to Stolonosae.
I communicated with Dr. Ballard about the problem of overlap, and drew the atlases somewhat differently to accommodate possible overlapping ranges according to his notes from vouchered specimens for V. macloskeyi sensu stricto. (See atlas pages for details - there were no large changes). This should accommodate some, but not all, of the various "clines" reported by Russell.
I edited the first paragraph of the taxon change description above to reflect this.
I think the solution you landed on will be fairly close to the real signal in future studies. My reading of the literature and material is that almost all Northwest specimens are truly part of the V. minuscula bin, the only sticking point is the Oregon Cascades and the type of V. macloskeyi.
My point with the PADRE and GPI results was that they can't test Russell's cline hypotheses, although like you I'm growing steadily less convinced of those:
I agree that there are multiple questionable things in both Russell papers, and plenty of logical leaps that aren't necessarily supported. Specifically looking at figures 9-11 in the 1956 paper, the isometry lines don't even line up with where the specimens are... It looks like the lamina length lines don't even include the southern California populations - which are shown as 6 specimens in the Transverse Ranges, except that he declines to enumerate specimens in 1956 and instead refers to the 1955 paper, which lists only one specimen from the Transverse Ranges (from two herbaria). It is unclear from an admittedly quick reading if he subsequently sampled more, if he mapped individuals within a gathering, or what. I agree that the hand-drawing of contours can't separate variation in geographical average morphology vs. variation between specimens. It's a shame we don't have the data to analyze in a modern GIS, but his dissertation is available if you want to ILL it :P https://primo.lib.umn.edu/primo-explore/fulldisplay?docid=UMN_ALMA21417657620001701
All of these hand-drawn maps look like they were drawn partly based on the data and partly to keep the lines and labels separate for readability, even if that went against the data distribution.
(Also strangely, one of Russell's vouchers for V. minuscula in Washington is from a very strangely low-elevation habitat and might be V. palustris s.l.. I'll have a look at the specimen next time I'm at WS. The location is no longer accessible, unfortunately.)
In Figs 14 & 15 it looks like he's excluding V. minuscula entirely, which means that there are only 3 Oregon gatherings included - one from right near the California border, one from near the Washington border, and one roughly halfway between (Klamath Co, the type from Mt Hood, and Crater Lake, respectively). Fig 14 seems to show that the Oregon specimens (at least the northern ones) are different from V. macloskeyi s.s. in a way that makes them similar to V. minuscula. Fig 15 is confusing in that it shows the California morphological pole as having as many as 20 leaf crenations, but the 1955 paper indicates that California material should have no more than 8 crenations. Fig 15 also intimates that the type gathering of V. macloskeyi has greater than 40 crenations per leaf.
Interesting, Fig 10 and 11 seem to imply that northwestern material, including the specimens he maintains in V. macloskeyi s.s., out-minuscula*s *minuscula - that is, it's even more morphologically disjunct from Sierran material than Eastern populations.
In short, my reading of the 1956 paper is that the Oregon gatherings included in V. macloskeyi s.s. are, in Russell's own dataset, separate from V. macloskeyi and referable to V. minuscula. I cannot see a reason why they were retained in the same taxon as the Sierran material. It is very frustrating that Russell presented no scatterplots of the two taxa together for any character.
As for the type of V. macloskeyi bearing obscurely-crenate margins, the crenations of V. minuscula should also be obscure, just more numerous.
It appears that the resurrection of V. minuscula Greene is partly predicated on Marcussen et al. (2012) referring their Sierran sample to V. macloskeyi, and that neither that investigation nor Ballard et al. (2023) have presented evidence that the type of V. macloskeyi is conspecific with the Sierran material. I'd be interesting to hear if you have more insight into this.
My conclusion based on the available data, which is admittedly well beyond the scope of iNat, is that the widespread northern and eastern taxon should be called V. macloskeyi and that the California taxon should be called either V. anodonta or V. parnassifolia (priority between them not having been yet established; interestingly both names referring to the nearly entire margin). They appear to interfinger but not necessarily intergrade in the Siskiyou Mountains and Modoc region of the California-Oregon border.
The upshot here is probably that dramatically fewer observations than I had feared need be referred only to the Stolonosae. Most of the Oregon High Cascades material which is currently being retain as V. macloskeyi here on iNat clearly shows fairly toothy leaves, and I am skeptical that any are referable to the Sierran taxon.
P.S. Have you read Russell's poetry?
I think we're of similar minds here. My first impression was the same as yours re: the identity of the type of V. macloskeyi vs. the entire-leaved taxon in California, just based on the presence of crenulations on that specimen. And you're right to count them - there are an awful lot of them for an entire-leaf taxon. It's only one trait, though, and I think the depth/shape of the "teeth" may end up being important; I'm thinking of something parallel to the distinction made between teeth shape to distinguish between V. incognita and V. minuscula in Michigan Flora, for example, at couplet 14 here ¯\_(ツ)_/¯. IDK - maybe it's conspecific with the eastern taxon, maybe it's the same as the Sierran taxon, maybe it's a third thing. Too many possibilities and not enough information.
But, Harvey says that he has looked at a subset of herbarium specimens to confirm some county-records of V. macloskeyi sensu stricto, and his very-preliminary conclusion was that the Sierran taxon does, in fact, extend into Oregon and maybe even the southern tier of Washington counties. (He offered this northern extent of the range in response to my asking for his advice on drawing range maps with maximal overlap, so I assume he's erring on the side of being too-expansive here). But he also cautioned that this is all preliminary. He also said that he's come across several specimens of white-flowered V. palustris misidentified as V. macloskeyi/pallens. I suppose those could have confounded earlier studies.
I have not read Russell's poetry - thanks for that! What a fascinating guy - I had no idea.
Please feel free to mark the toothy-leaved plants you referred-to above into V. minuscula / V. palustris as appropriate. No reason iNat can't help inform whoever tackles this problem in the future.
@ddennism if you're going to create a taxon relationship for V. macloskeyi sensu stricto, then the relationship for sensu lato needs to first be deleted so sensu stricto can be linked to POWO