This is a problem I keep running into and the current approach leads to a stalemate over and over again. There are many large plant genera where identifiers would appreciate having subgenera and sections added; when there are hundreds of species, having smaller categories to capture unidentifiable observations is really helpful.
In almost every case someone at some point has come up with an infrageneric taxonomy for the genus. However it’s often decades old, and phylogenetic work in the past 20-30 years has upended any hopes of the old clades being monophyletic. In many cases the phylogeny is still unresolved enough that no one has proposed a new infrageneric taxonomy, so as far as I know the old names are still valid (although generally no longer respected by academic consensus).
POWO doesn’t cover infrageneric taxonomy in plants so it’s one of those cases where iNat curators need to use primary literature and decide on our own what to do. This summarizes most of the key points, as noted the Curator Guide only explicitly requires monophyly for species complexes for whatever reason:
What should be prioritized in these situations and why?
If there’s an older, official taxonomy for defined species groups, then we definitely should be using that until a replacement is published. This is close enough to a published hypothesis of monophyly, or something equivalent. The rule that clades on iNat need to be monophyletic is primarily a proxy rule so that people don’t add random “groups” that are overlapping, which creates problems. As long as the groups are well-defined and accepted in the literature, I think it’s fine until something better is formalized… this is why we still have Nematocera and Orthorrhapha, for instance.
Monophyly in the sense of “non-overlappingness of clades” is absolutely critical for the iNat taxonomy. Any taxon only has one parent, but one taxon can have many children. The moment that non-monophyletic species groups are added to the taxonomy based on arbitrary characters, then suddenly you have ‘species group A+B’ and ‘species group B+C’ as children of ‘Genus X’–and is ‘species B’ a child of ‘species group A+B’ or ‘species group B+C’? I hopefully don’t need to tell anyone that the characters that seem important to delineate an arbitrary species group for one person will seem irrelevant in a different context, so this is just asking for fights if we try to narrow it down to only one non-monophyletic parent species group for a species.
If we want to add a non-monophyletic ‘species group A+B’ as an option for making identifications, then some kind of additional function needs to be added to the iNat taxonomy. I guess maybe create them as some kind of ‘unofficial taxa’ like BugGuide does.
Yeah in this context I am only referring to formal taxonomy, where phylogenetics has generally replaced phenetics as a reference for relatedness in building the tree. Informal intuitive groupings of unrelated species like “carnivorous plants” or “mangroves” are a different subject which would require different functionality.
If there are two competing taxonomies published, we should obviously use the newer one if it’s been accepted, but I see no reason to stop using an older taxonomy based on phenetics if a replacement hasn’t been developed?
I think monophyly is the key requirement. If there’s an old monophyletic grouping that would be useful, then it’s probably up to curatorial discretion as in
But I would be against adding any obviously non-monophyletic groupings. I see this sometimes when some curator wants a complex for IDing in their specific country and adds a complex that isn’t comprehensive (ie, only includes a subset of species that would be in the complex because those are the only ones in that country). These types of groups strike me as somewhat “exclusive” because they aren’t useful to users in other countries/regions and preclude the use of those taxa in groupings that are useful more broadly. They also cause issues when someone from another region IDs to those taxa expecting a more comprehensive, monophyletic group (which the taxon ends up not being).
Hmm, what do you mean by “obviously” here? I agree that there’s problems if a taxon is created without checking globally what species should be included. That’s another issue that often restricts addition of infrageneric ranks (e.g. lack of literature placing any of the species in entire continents in the global south).
But if it also means not using a group if recent phylogenetic work has found the previous groupings to be non-monophyletic, that’s the main issue I’m asking about. It’s pretty common for proper monophyletic trees to have groupings that are non-intuitive from morphological or other species concept perspectives that would’ve been used traditionally.
There is also the issue that using an old taxonomy may result in a bunch of incertae sedis species that have been described since that revision. But the most commonly observed species are likely to have been described already, or have a reference in literature that places them within the taxonomy.
I share the opinion that if an older phenetic-based infrageneric classification is the best available, and is determined to be helpful for iNaturalist (based on community discussion and consideration of any drawbacks as noted above), then we shouldn’t hesitate to use it until something better becomes available.
I think the reason monophyly was emphasized for Complexes is that we are (potentially) inventing those groups out of whole cloth here on iNaturalist, and the preference is that there is some basis for them in existing sources outside iNaturalist. If an outside source is not basing a Complex on a determination of monophyly, then they should be treated with the same caution as with older formal classifications using official taxonomic ranks.
I certainly agree with this, and if I’m understanding what you are describing correctly, I think would be impossible on iNaturalist. I don’t think the system allows any taxon to have more than one parent taxon.
Personally I find subdivisions very useful. With Juncus species in Tasmania there are two groups, “leafless” and leafy Juncus. These two types are not split although there has been flags raise regarding this in the past.
It would be more useful in my mind having them split
For example, we have a taxon called Azalea Caterpillar and Drexel’s Datana Moths (Complex Datana major). It includes two species: Datana major and Datana drexelii. These two species absolutely do not form a monophyletic clade, according to any literature. I just checked the trees on BOLD, and based on barcodes, you’d need to include about half the genus Datana in a clade if you wanted to include this pair of species. But this pair of species is the most challenging pair to identify in the genus (as adults anyway; the larvae look nothing alike). So it exists, on iNat alone, as a “species complex”. It’s a useful bin to put observations in, but it’s unambiguously an “obviously non-monophyletic grouping”. There are other examples like this, but as a moth guy this is the first one that pops to mind.
Unless it’s referred to as a complex in literature, this seems like a good candidate for using an observation field to categorize. Or get someone to publish that half genus clade!
I’ve been vocal against complexes specifically in Lycosidae because few have been tested against DNA. There will be some rearrangement of species among groups when that happens, and I think it’s wiser to be patient for that and also the development of good field identification keys to use on iNat images. In the meantime I created two traditional projects for lenta and for South Texas Spiders groups instead. Sure, it’s a little more work that way but since we already know one of the lenta group species (tigana) needs to be moved into the South Texas (eventually to be a new baltimoriana group), having an extra degree of control over how things get sorted is a good thing. Plus it allows me to easily include observations with only a genus ID.
As well as the discussions linked by @upupa-epops, When should a genus be subdivided? was another recent thread along similar lines. I added a comment there sharing the rules-of-thumb I have used.
Only add an infrageneric taxonomy if you’re aware of most recent research for the genus.
Don’t bother adding infrageneric taxonomy where it would provide little value for iNat users, such as for genera with only a few species.
Use an infrageneric taxonomy that is supported by phylogeny if one is available.
If there is strong evidence that an infrageneric taxonomy conflicts with phylogeny, think carefully about the likely disruption that will be caused once better evidence means we need to revise it. It’s fine to leave the genus undivided for now.
If there is active disagreement about infrageneric taxonomy by researchers in the area, it may be best to wait for consensus to appear.
Add a flag with details of your proposed changes and solicit input from the most frequent/knowledgable identifiers.
I know I’m in the minority on this one, but I’m fully in the former camp. I see taxonomy as first and foremost a tool for humans to make sense of nature by categorizing organisms, so taxonomy should be done in a way that is intuitive and makes sense to humans. To me this is more important than phylogenetic relationships
I agree with the utility of taxonomy. The problem is that non-monophyletic taxa end up being less useful. Non-monophyletic taxa are almost always based on superficial similarities that end up being less important when all aspects of biology are considered. An extreme example of this would be grouping bats and birds in a single taxon because they share wings. You might say that’s a ridiculous example because that’s clearly not a good idea (e.g. bats have fur, lactate, etc), but the problems with lumping bats with birds are basically the same type of problem with all non-monophyletic taxa. Especially since the advent of molecular biology, monophyly has become very important, because it is the greatest predictor of similarity beyond superficial characteristics.
I don’t think I agree with this. Or maybe to state it with more nuance, it is a strong predictor but not a reliable rule. Take these bird examples; in both cases one taxon in a clade found itself in a new niche and became very different, leaving behind multiple taxa in the old niche that remained similar to each other. Elegant Terns are apparently descended from the western Atlantic subspecies of Sandwich Tern but went west and got orange bills. Western Atlantic and eastern Atlantic Sandwich Terns remain nearly identical but are now paraphyletic because some descendents of one of them happened to speciate in new conditions.
The paper Species as a Heuristic: Reconciling Theory and Practice (pdf, open) may be relevant to this topic. Their definition “avoids precluding species generated by processes that do not automatically or immediately result in monophyly, such as hybrid speciation, polyploidy, or paraphyly in the case of recent ancestor-descendant speciation.”
(Which reminds me of the catchphrase by Bateman about an orchid: “with apologies to monophyly”)
In a case like this, I think it makes no sense to insist on monophyly.
I think there’s still a big difference between this kind of non-monophyletic clade, which is generally accepted by taxonomy, and us extrapolating new groupings based on superficial results.