tagging a few people who might be interested in discussing the addition of subgenera and sections for Salvia:

@kai_schablewski @vandalsen @alex_abair @xanergo @jrebman (others?)

(also note: https://www.inaturalist.org/pages/curator+guide#nodes)

Hello. After reading the points about adding additional taxonomic nodes to Salvia, I have the following observations:

Subgenera
There are 11 Salvia subgenera relatively well established which could be added, however:

a) The subgenera cannot be easily recognized based on external morphological characters, floral dissections are needed in most cases, also in combination with geographical information; for example, sometimes Old World salvias (belonging to different subgenera) being grown in the Americas have been identified as species of subg. Calosphace (a group endemic to the continent, with some weedy members introduced worldwide)... a specialist might immediately identified this mistake based on its familiarity with American salvias, but a non specialist would not have the tools to determine the subgenus without a dissection of the flower

b) There are not updated checklists for all the subgenera, so it might be tricky to make the initial assignment inside the system; the Old World subgenera Glutinaria, Heterosphace, Salvia and Sclarea could be the more problematic on this issue... since these embrace many Asian species, it would be necessary counting with specialists from this region in order to complete the taxonomic assignments into subgenera

Sections

a) There is not a natural sectional classification for Salvia yet. Especially, the most speciose subgenus, Calosphace (embracing most of the American salvias), is very unstable at this taxonomic level

b) The available sectional classification for subgenus Calosphace is that published by Epling in 1939, and continuously improved up to 1968... there is available a list in which every Calosphace species is assigned o a section, but even if this would be used as a provisional classification (considering it has been demonstrated to be unnatural), it includes about 114 sections, which is excessive and poorly informative, at least for iNaturalists interests.

So, I would recommend not adding more taxonomic nodes at the moment. I guess it would be really informative or interesting for those genera that have been recently transferred to Salvia (Rosmarinus, Dorystaechas, Perovskia, Zhumeria... etc.)... but, in general those have few species, sometimes geographically restricted or already recognized with a vernacular name, as the Russian sages (Perovskia)

What do you think?

(1/3)

I went searching for this flag and would like to revive it.

I would be strongly in favor of adding all subgenera while expunging/excluding all sections. I think it's important to make the distinction between these ranks, as subgenera are relatively stable by now, but sections (as mentioned by @xanergo) are largely pre-modern.

I think it's worth pointing out that the current situation is one of partially curated nodes, not one of all species descended directly from the genus. As such, leaving a small number of species incertae sedis is not ideal but is still far, far better than the current situation which treats 98% of the genus as incertae sedis to accommodate traditional segregate genera (726 species and nothospecies grafted to the genus, including component species of the S. verbenaca complex, and 14 grafted to subgenera). As such, adding the remaining subgenera and classifying as many species as possible is not conceptually different than the current state, just cleaner. Of many solutions, the current state is among the worst. The complexity of Salvia inherently tests the cleanliness of the ideal iNat schema, but I would argue that this makes it even more important to attempt parsing the genus intelligibly under the iNat philosophy - shying from it is admitting defeat.

I would disagree that the obscurity of subgeneric characters is material here, and posit that the misidentification of cultivated plants of nonnative subgenera is also immaterial. Many or most users will be identifying at species rather than subgenus, so for them the subgeneric characters are moot - and even between species, misidentifications are common. I don't know how data on this could be gathered, but my impression is that infrageneric divisions are primarily employed by users with some familiarity of the group (and then only sparingly), as the naif finds them too esoteric. Furthermore, if even subgeneric placement is uncertain, the identification can still be knocked back to genus.

(3/3)

As I see it, the major argument against adding subgenera is the remnant of species that cannot be easily classified. However, we have that situation already and it's by far skewed to unclassified taxa. Classifying larger parts of the genus into subgenera both

a) clarifies scope of the problem of unclassified species, and
b) makes doing so a much more tractable task (say, 40 species instead of 726).

It's certainly not ideal to have mixed rank daughters of the genus in the iNat philosophy, but species incertae sedis do happen. The Proteaceae: Persooniae example on the Curator Guide isn't necessarily germane as here we would be intentionally creating a comprehensive schema with known lacunae, rather than an entropy-driven misclassification. Calosphace is by far the largest subgenus, but its circumscription is not particularly ambiguous, so grafting those species to the subgenus would be time consuming but not contribute to the conceptually difficult remainder.

I only know a few Salvia species, so don't have much sense for the current strength/validity of the subgeneric taxa. But I agree there is a good argument for completing the subgeneric classification that has been started here, as far as it can be taken at least. Curation will remain incomplete at that level as has been pointed out, however, so if that seems less desirable, then the alternative would just be to eliminate the current subgeneric classification and graft everything back to the genus. In principle I'm always for a more informative (node-rich) classification, but practical considerations may counter-weigh.

It does sound like an attempt at Sectional subdivision would be premature.

If the consensus ends up being in favor of subgenera, and someone can post or link the most current classification (e.g., checklists) for that level, I'm happy to work on a subgenus or two and get their species moved.

In that case it sounds like we should probably abandon subgenera in the iNat taxonomy for now, until all of them have decent checklists.

There's about 650 names in the lists I'm sharing in the DropBox folder. About 600 Calosphace, the complete subgenus, 19 Audibertia, also complete, and about 30 corresponding to Dorystaechas, Meriandra, Zhumeria and Rosmarinus. The approximate numbers of the other subgenera are: Glutinaria 100 spp, Heterosphace 50 spp., Salvia 70 spp, and Sclarea 120 spp. The estimated number of Salvia is about 1000 species.

At present I have 28 (3.8%) of taxa already on iNat that I don't have a strong subgenus for. This encompasses 213 observations: 15 in Europe and the Mediterranean (including 2 misIDs in South America), 16 in central Asia, 19 in the Levant, and 163 in South Africa. That's 0.01% of all Salvia observations on iNat.

Another issue I hadn't picked up on before was that there is still not a valid subgenus name for "Heterosphace". Not sure if it's allowable or possible put it in quotation marks, or to leave it as sections Heterosphace and Salviastrum, since those two clades are so easily separated out.

As something of a naif (a very committed one), I have found the Sectional node in Pelargonium very helpful, which is why I first raised the issue with Salvia. Admittedly, the former classification system is complete and appears morphologically coherent to non experts, although the same can't be for natural and naturalised artificial hybrids, but, yes, I think any partial attempt is better than none for all the reasons @ajwright gave. : )

I don't know if it helps, @ajwright, but IC Hedge included several of the unassigned African species in your spreadsheet to his Species-Group L, which he said was part of Sect. Heterosphace Benth.; sect. Neosphace Briq. The species included in that group are: S. nilotica, S. aurita; S. scabra; S. obtusata; S. triangularis, S. tysonii; S. repens, S. runcinata; S. stenophylla; S. schlechteri.

His Species-Group G, part of Sect. Hymenosphace Benth., included S. dentata, which Bentham called S. crispula. (Similarly, S. obtusata Hochst. was known to Bentham as S. marginata E.Mey., S. runcinata L.f. as S. monticola Benth., and S. triangularis Thunb. as S. tenuifolia Burch. ex Benth.)

[Hedge IC. A revision of Salvia in Africa including Madagascar and the Canary Islands. Notes from the Royal Botanical Garden, Edinburgh. 1974;33:1–121]

You will find S. garipensis on iNaturalist and within South Africa but POWO names that species S. gariepensis, which is what Bentham unilaterally renamed it.

You have S. disermas and S. radula under Sclarea. I don't know of any research on S. granitica. Below is all IC Hedge says, and it's the most I can find. @chris_whitehouse?

Although S. granitica was first collected as long ago as 1838, exactly 100 years were to pass before it was re­collected, not this time from the type locality but more than a hundred miles to the north. Krauss made the original gathering on the mountain Babylon's Tower in Caledon and, in 1938, Pillans the second one from Clanwilliam division. The latter collection remained undetermined until during the course of this study I was able to see a specimen of the Hochstetter species and identify the Clanwilliam plant. To both Bentham in the Prodromus and Skan in Flora Capensis, S. granitica was "an imperfectly known species" both because the original description was so scanty and because they never saw the type specimen. Although I have now seen four separate gatherings of it, all of which are virtually identical, and there is no doubt that it is an extremely distinct and interesting species, no clear answer can be given to the question of its affinities. It appears, as does the unrelated S. muirii, to occupy a very isolated position among the southwest Cape species with no connection at all with the shrubby species, more or less confined to this region, and very little affinity with any of the several species of the previous species-­group L which are unrepresented within the area of S. granitica or, for the most part, within the southwest Cape at all. Although in floral characters, S. granitica is unlike any of the species of species-­group L, it does share with many species of that group the character of a creeping rootstock.