Creating species complexes

So, I’ve been going back and forth about creating species complex taxa for Euphorbia. The utility of creating these complexes is extremely clear to me and there have been many times where I wished I could use them. Problematically, though, the complexes I have in mind don’t meet all the criteria (my notes followed by “-”; the last two points are excluded since they are simply procedural and not criteria for recognition):

  1. “Species complex is monophyletic (i.e. sibling groups of species)” - The phylogenies are not well resolved enough to know this with high confidence.
  2. “Complex is recognized in the literature” - This is incredibly hard to apply. Sometimes they are noted but rarely as species complexes per se. More as groups which have sometimes been given section or subsection recognition.
  3. “A named subgenus, section, or series does not already exist for the group” - Sections exist in Euphorbia. They are necessary to make sense of a genus with over 2000 species. Sections in Euphorbia are equivalent to genera in most other groups.

I will also some implicit criteria that can be extracted from the following statement: “Species sometimes intergrade and there are places on the tree of life where adding hard range map boundaries is arbitrary and/or identification to species level is often not possible.”

  1. Species intergrade.
  2. Creation of species boundaries on continuous variation is arbitrary.
  3. Identification to species level is not possible.

I wish to create the following complexes and note which criteria they meet (there are a lot more, but these three are the ones I’ve thought about most recently):

Euphorbia dentata complex:
Well known to cause problems with identification. Can be structured to include E. dentata, E. davidii, E. cuphosperma, and E. tubadenia, though E. shiediana would probably need to be added to make the group roughly monophyletic (E. shiediana’s placement is unknown but clearly closely related to the other 4). Several of these members are extremely difficult to ID without seeds. Some species express population variability that makes them difficult to separate even with seeds.
Criteria: 1. Maybe met? 2. Depends on whether you include previously recognized taxa as “recognition”. 3. Failed, obviously. 4. Probably true. The group needs more work to see if this is absolutely true. 5. Probably true. See 4. 6. Without seeds, this is often true. sometimes true with seeds.
Edit: The way I created it, it is synonymous with the E. dentata alliance in Mark Mayfield’s 1997 dissertation ( A systematic treatment of Euphorbia subgenus Poinsettia (Euphorbiaceae)).

Euphorbia cyathophora complex
Causes problems when we don’t know what the roots are like. Includes E. cyathophora, E. hormorhiza, and E. elliptica. E. pinetorum could be included, but I think it’s distinct enough to not worry about.
Criteria: 1. Phylogeny not complete enough to know. 2. Hard to say. 3. Failed, obviously. 4. Essentially true in most characteristics, especially when looking at E. elliptica and E. cyathophora. 5. Probably true. 6. True without roots unless future taxonomic work reveals more useful characteristics.

Euphorbia hyssopifolia complex
An extremely well known problem. Includes at least E. hyssopifolia, E. nutans, E. vermiculata, and probably a couple of others that I can’t remember off the top of my head (I need to review McVaugh’s work) in North America (though the two do not appear sister phylogenetically). In South America, probably includes E. lasiocarpa as E. hyssopifolia can have hairy fruits and E. lasiocarpa can have glabrous fruits.
Criteria: 1. Probably not, but the taxonomy is so fuzzy that we may not know until a complete taxonomic revision of both the South American and North American taxa is complete. 2. Sort of. Wheeler described this as the “Pandora’s box” of Euphorbia in the 40s. It’s gotten better since then, but there’s still a lot of difficulties. 3. Failed, obviously. 4. True in all ways except seed characteristics. 5. Probably not true. 6. Definitely true without seeds.


On a related note, would anyone mind if I made the criteria more explicit on the curator guide? The main reasons for creating a complex are not even explicitly made criteria. And the last two points are comments on how to make a complex, not criteria themselves. I’d change it to match what I have outlined above.

I also feel like we should discuss criterion 2. What constitutes being recognized in the literature? If we can come to a consensus, that should probably be changed to be more explicit as well.


Like most difficult issues, I sit on the fence. So I’ll just point out that species complexes are an ongoing point of contention.

A general search of species complex:

A discussion closely related to this one:

Species group as a new rank:

I think maybe the most important question is the one you asked in the second post–what is the reason of having a complex? Is there a value to the biological community–perhaps beyond just sorting things into smaller groups. What is the value of having things sorted into smaller groups which are not necessarily “real” biological groups (evolutionary lineages).


For one, even bringing the ID to Section is not nearly enough information for practical use, IMO. As Nathan says, it’s a little like bringing the ID to a genus, except even a section within Euphorbia has dozens of common species with wildly different distributions and traits.

I’m generally a lumper, but Genus Euphorbia should be like six genera IMO.


What I’m getting at is we need to spell out what the “practical use” is so that everyone understands the need. It’s kind of hard to move forward without starting there. I’m not saying there isn’t a practical use of groups that may or may not be “real” biological groups (evolutionary lineages, as currently understood). We could create a group of “things that fly” of “plants with red flowers”, but we don’t.

For conservation, tracking range changes, or other things I suppose. Unfortunately I haven’t used iNat data for scientific ends myself, but I know many people here have. Imagine if you wanted to study red oaks, but half of all observations were only marked “Quercus”. I am guessing a little here. Nathan or someone like @spurgegkr could give better examples, probably.


I disagree on failing criteria 3. I would think that it means whether upper classification exists for the group of species you want to include. In this case you want to divide the higher group further, so I’d think it is a pass.


Have my vote for more Species Complexes and infrageneric-supraspecific things, even at the cost of relaxing criteria.

I keep reading things like “identify to the level you’re comfortable with”… Whenever one is comfortable proceeding further to some intermediate ‘group of species’ (backed by at least some ‘pro’ literature), perhaps iNat shouldn’t stand as much in the way. Don’t be content with broad genus ID, squeeze every last drop of refined info from those obs, dammit.


I am pretty convinced that complexes should answer to practical issues. I mean that point 4, 5 and 6 are particularly fitting. At the same time they provide important benifits such as allowing to ID observation to a rank that is close to the species and to search for organisms that are closely allied.
On other hand, complexes may be often lacking in literature but I think that many of us can provide a taxonomic view supported by common sense and by a vast and renowned knowledge. In this light, maybe, demanding at any cost that a complex has already been cited in literature could sometimes turn out to be detrimental or unpractical.


This is most easily captured in criteria 4-6, and especially 6. Beyond the problem of identifiability, there are three characteristics of the complexes that would be helped by aggregating them into a complex:

  1. These are some of the most commonly observed species.
  2. These often don’t or can’t be IDed to species-level (e.g., I don’t feel comfortable adding a species-level ID) because they are difficult to ID.

The reason for creating this is the same reason any taxon should be created: it allows for easy sorting. Allowing these groups to be treated as separate entities means it’s easier to focus on learning how to identify them, communicate the difficulties with identifying them, and exclude them from a search if you want to focus on something else. This last one is particularly important in the case of sect. Anisophyllum and sect. Poinsettia. The E. dentata and E. cyathophora complexes each make up about 1/3rd of all sect. Poinsettia. In other words, there’s only 1/3rd of all sect. Poinsettia observations that don’t fall within these complexes. With the E. cyathophora complex, at least we can confidently assign a species throughout most of its distribution. With the E. dentata complex, you pretty much have to evaluate everything at any location throughout its range. The E. hyssopifolia complex is only about 9% of the section (about 9,000 observations), but is extremely well known as being a problematic group. Additionally, these estimates at least slightly underestimate the problem as these groups almost certainly have more observations IDed only to section-level (i.e., they’re proportion can’t be calculated because they are not IDed). I don’t think it would surprise me if the actual percentage of the E. dentata complex was closer to 35-40% of all sect. Poinsettia.

The cost is additional taxonomic complexity which translates to a steeper learning curve. And when there is not a 1:1 correspondences with the literature, it gets difficult to convey information to users wanting to learn. I get it. However, these three groups (at least the E. dentata and E. hyssopifolia complexes) have constantly been a problem when I try to ID things. I can’t ignore them because I know how to ID them and I know that a ton of them are on the border of identifiability. If I wanted to focus on them, I could get them to species so I can’t just hit reviewed and move on. At the same time, I can’t get them out of my way by other means so I just have to let them stay in the way.


So the main reason would be for communication and preliminary sorting. I’m not understanding this part of the description: “Species sometimes intergrade and there are places on the tree of life where adding hard range map boundaries is arbitrary…” Do both of those just mean “sometimes we cannot place a specimen into a species taxon”?

I’m being devil’s advocate, don’t have a strong position, and just contributing (hopefully meaningfully) to any decision whatever it may be. I think most folks agree that there are costs and benefits of opening up the species complex category to any group that’s difficult to distinguish–hence the ongoing discussions. There are lots of instances of it being difficult to place specimens into a species taxon, so we could start to see lots of species complexes created for the convenience of sorting rather than for identifying real biological entities (monophyletic groups).


I think this may be the case with many and maybe most species complexes. This is a group of species that are confusing. This may be because they are closely related but it might also be because they are morphologically similar and relationships have not been resolved.

The curators guide says “If a “principal species name” is not established in the literature, use the earliest published species name for the name of the complex.” This implies that it may be in the literature but doesn’t have to be named in the literature. Any group of species that are easily confused and recognized as such could be a species complex. If there is already an equivalent or almost equivalent section or subsection, use it instead. Of course, maybe the sections would complicate iNat too much (if you put them all in) and a few species complexes would not. Perhaps that should be considered.

The curators guide should be edited to say that you should add these sections, assuming they makes sense, before you would have a species complex but perhaps only if these sections are useful.

If you can easily sort a few taxa from a gigantic group into a smaller group, I’d say there is huge value in that. You can say this is one of a few species and not one of the other 2000 species. That shows that some messy species fit together in a small group that someone should look closer at that group.

I would be wary of the intergradation part of these. In the genus Malacothamnus that I studied for my PhD, there are lots of zones of intergradation between groups I would not consider species complexes. They just easily hybridize. I think this is where the monophyletic groups come in. It should be a monophyletic group but that monophyly could just be a morphologically based hypothesis as a huge amount of DNA-based phylogenies are unresolved or haven’t even been attempted.

Bottom line, if a species complex would be useful to both the experts and some iNat users, I think they are worth considering if the possible costs have been considered and wouldn’t be too high. I would personally rather have more levels of division than less as it helps move IDs along but I would also leave most of those levels to the experts and just focus on the simpler levels I am familiar with.


You might also just consider observation fields when things could get messy. And maybe that should be put in the curators guide under the complexes section as an alternative to consider for experts that want their own sorting that others might not want. This is what I did for the Malacothamnus intermediates.

1 Like

Yeah, that’s hard for me to make sense of too (wording of the curator guidelines). I interpreted as taxonomy is sometimes arbitrary (vis. gray zone of speciation, De Queiroz 2007).

Also, I should note that species do not have to be monophyletic. You can have perfectly good species that are paraphyletic. This is well established in systems that speciate through peripatric speciation.


I don’t have time to read all this right now but I have some (controversial) takes on this, that I feel should be “the way”:

  1. a species complex should normally only be used for a group of three or more taxa within a large genus. Case-by-case for species pairs.

  2. species complexes should be possible to make even if there is no literature citation or source saying that they exist. Cryptic species groups and overlapping identification traits are not rocket science, it should be something that we can interpret for iNaturalist as identifiers and/or experts, and should be able to use and put into practice as we see it necessary.

  3. species complexes should only be made when there are a significant enough amount of observations that would benefit from the change. If your complex only has 3 observations, or even only 10, and there’s only twice that many for the whole genus, it probably isn’t necessary.


I agree. The unresolved nature of the tree I think is sometimes the reason why people consider species a part of a complex. As such, I think it’s a poor criterion.

This one took a bit of thinking, but I think I agree with your interpretation. If this is to be implied, I think the guidelines should probably be explicit about it.


I agree. With how their worded now, I think the last point (criterion 6) dealing with identification is probably the main criterion that makes sense in the context of iNaturalist. Point 5 really only makes sense if you’re reading a species delimitation study and don’t know how to implement the results of that study. Point 4 is fine, but it is implied by the other points.

1 Like

I completely agree! I feel like these shouldn’t be the only criteria. I like 5 and 6 in my original post.

I don’t know who ultimately should make this decision–staff or some kind of consensus among curators. Any curator can edit the wiki, but it reads more like an official document and I wouldn’t be comfortable editing it to change the criteria. Between this and the other species complex discussion linked to above, now is probably a good time to ask the staff if they’d like to make a decision. Regardless, it definitely needs to be reworded at the very least.

1 Like

I say all this as someone who uses complexes for groups like this series of closely related west Asian Argiope, which form a cryptic group. Apparently, no spider researcher has officially said “this is an indistinguishable group of species and is officially a complex”, and so a lot of people have tried to remove it saying it needs a citation to be valid.

However, the issue is that complexes are something that are more readily used by non-experts (including citizen and “amateur” scientists), and so often times there is no official citation for a useful complex. Some experts use them. Some experts do not, or relegate them to subgenus, or nothing at all.

I firmly believe in complexes based on their purpose and utility. The act of a citation has no significant impact on a complex’s credibility when it comes to managing data for useful purposes like on iNaturalist.


Yeah, I had a very hard time parsing this statement (curator guideline wording, not mine), but ended up with the last 3 criteria. I was trying to preserve a few more elements from the original wording, but your wording probably works just as well. It’s an effective criterion of which the three I gave are a subset.

1 Like