15 February 2012

Amending the PhyloCode: The Species Problem

Earlier I mentioned a proposal by Cellinese, Baum, and Mishler to make a major revision to the PhyloCode, removing pretty much all mention of "species". In this post I'm going to take a high-level look at some of the proposed changes.

Independence from Other Codes

Currently the PhyloCode relies on the rank-based codes for rules concerning species. The PhyloCode does not cover species themselves, but it allows them to be used as specifiers, although, per Note 11.1.1, the implicit specifier is the type specimen. Thus, the rank-based codes are necessary for indicating which is the type specimen of a species.

The idea of removing the dependence on rank-based codes is an appealing one. After all, the rank-based codes are themselves independent. The proposal does away with the dependence by not allowing species to be used as specifiers, and by removing any mention of "species" as components of a clade.

My take.—Independence is a fine ambition. But I don't think it's a necessary one. The PhyloCode already must acknowledge the rank-based codes, as it allows for names to be converted from names governed by the rank-based codes.

I do support removing mention of "species" as components of a clade. The proposal changes "organisms, populations, or species" into "organisms or populations" in multiple places. This works for me. (And many consider species a type of population, anyway.)

Only Specimens and Apomorphies as Specifiers

Under this proposal, definitions that currently use species as specifiers would instead have to use specimens. Thus, instead of defining, e.g., Mammalia as, "the clade stemming from the last common ancestor of Ornithorhynchus anatinus (Shaw, 1799) and Homo sapiens Linnaeus 1758," it would have to be defined as something like, "the clade stemming from the last common ancestor of British Museum of Natural History (Zoology Collection) 1979.2184 and Uppsala domkyrka: ossa Caroli a Linné."

My take.—This doesn't effectively change anything, since the species names are currently just used as a shorthand for their type specimens. But it's a nice shorthand! Using collections and access numbers (the human type specimen is an exception here) makes definitions difficult to read and open to deleterious typos. (If I had written "1978.2184", I would have anchored Mammalia on a skink!) The rank-based codes overall do a good job of linking binomials to type specimens, so why not use this?

Of course, the rank-based codes do not do a perfect job. Many species named early on do not have type specimens. This is a problem that I would like to see addressed in the PhyloCode, although I'm not sure what the best solution is. Perhaps the best way is to disallow use of species that do not have type specimens, unless a neotype is assigned under the appropriate rank-based code at the same time.

Chresonyms, not Binomials

The proposed changes would allow the conversion of species names into clade names. Currently this is prohibited by Article 10.9. Cellinese et al. argue that "species" is just a rank like any other rank, and thus the names of species should be available for conversion.

This introduces a difficulty, though. Under all rank-based codes, species names are binomials, where the first part is a generic name, unique within the code, and the second part is an epithet, unique in combination with the generic name, but not necessarily unique on its own. Under the PhyloCode, all clade names must be uninomials. Cellinese et al. follow the lead of Article 21 in considering species names as the epithet plus a citation. In other words, a chresonym. Thus, the species name "Homo sapiens", which was authored by Linnaeus in 1758, could be converted to a clade called "sapiens Linnaeus 1758".

For consistency, Cellinese et al. propose that all clade names be chresonyms, rather than uninomials, as in the current version. This introduces some major issues, since a chresonym does not have a single orthography. The name "sapiens Linnaeus 1758" could be rendered "sapiens C. Linnaeus 1758" or "sapiens Linnaeus, 1758" or "sapiens Linnævs 1758" or even "sapiens L."

I note that they could have followed another route here. The PhyloCode already allows for the conversion of some binomials! Recommendation 10F shows how to convert subgeneric names. For example, the subgeneric binomial "Homo (Homo)" could be converted into a hyphenated clade name: "Homo-Homo". But unlike rank-based binomials, these names are fixed (i.e., changing the genus/superclade does not change the name). A similar approach could be taken for converting species names into hyphenated clade names, e.g., "Homo-sapiens".

My take.—This is the most problematic part of the proposal, as far as I am concerned. I think they should have followed the lead of Recommendation 10F rather than Article 21. The latter is constrained by trying to be orthographically consistent with the rank-based codes. But, if you are converting the name into a PhyloCode name, there is no need for strict consistence, as Recommendation 10F shows. Changing the basic nature of clade names so that they can have varying orthography would be a bad move in my opinion.

Are Species Special?

There is disagreement over the assertion by Cellinese et al. that "species" is just a rank. Some workers consider species to be a special type of taxon (e.g., lineage segments), or at least feel that the PhyloCode should allow for this approach (as it currently does).

My take.—I'm sympathetic to the idea that species are not really special, but I don't think they approximate well to clades, except in cases where there is no descendant species. This is effectively true for extant species, but only effectively. (Who knows how many species our own will give rise to?) And it's definitely not true for certain extinct species (e.g., Homo ergaster). Of course, this goes for genera as well (e.g., Australopithecus), and their conversion is allowed (although probably not recommended in the case of Australopithecus). But there's another problem.

Taxonomic units are important to systematics, and the only codified way we have of referring to them right now is to use specific or infraspecific names. It's not a good system (binomial nomenclature has major problems), but it's something. I worry that if we use species names for clades instead of units, we're going to miss having names for units.

That said, I'm a bit on the fence here.

Impact on Implementation

The PhyloCode is to be implemented when it is published along with Phylonyms: a Companion to the PhyloCode, a.k.a. the Companion Volume. Work on this volume has been ongoing, with submissions using the rules in the current draft. As of the beginning of this month (February 2012), 255 contributions were in some stage of review, with 85% of them externally reviewed and 32% accepted for final publication. But because the proposed changes would not allow species to be used as specifiers, all of these entries would need to be revised.

My take.—This seems to be the most incendiary issue. Personally I would rather get the code implemented sooner, with the option of looking into species name conversion at a later date. Implementation was originally meant to take place last decade; we've waited long enough.


  1. Discussions like this are one reason I can't take the phylocode seriously. Of all the things we should be doing, renaming perfectly usable names connected to lots of existing data is not one of them. We have enough problems in retrieving biological data without introducing new ones.

    It seems to me that the one genuine insight in phylogenetic nomenclature is the notion of defining clade names with reference to the least common ancestor of two or more taxa. Simple and straightforward (better yet, computable). But to think that it is worthwhile to attempt to rename all of life suggests a certain lack of perspective.

  2. It's not "renaming", though, it's associating an existing name with an actual definition. Granted, there is some modification in the edge cases of cross-code homonyms — where it actually fixes a major problem in the current system — and subgeneric binomials — where the modified names are pretty easily backwards compatible: s/([A-Z][a-z]+)-([A-Z][a-z]+)/$1 ($2)/). (And it's also worth pointing out that subgeneric names are only converted as hyphenates when the epithet is not unique.)

    I also disagree that binomials are "perfectly usable". Binomials are a kludge and have always been a kludge. (Ditto for mandatory rank suffixes.) A taxon's name should not change when the taxonomy changes.

    That said, I agree that this proposal has the potential to be quite disruptive, and I think it will fail to be passed for that reason. It does have a couple of good points, though, and I attempted to highlight those.

    1. I don't have a particular problem with bionomials themselves, they are a solution to problem of vastly more taxa than there are single word (non-nonsense) names. Having two names greatly increases the space of possible names (from n to n**2). The (misguided) suggestion to use chresonyms is another attempt to solve the same problem.

      I do agree that the names shouldn't change when taxonomy changes. For me this is the lesson of the Drosophila melanogaster case. I'm quite happy for a taxon called "Drosophila melanogaster" not to be grouped with other things called "Drosophila". I'm not sure that there is anything about existing codes of nomenclature require us to change the name, it's a convention taxonomists have adopted. Change that, and much of instability in nomenclature disappears. We'd just need to accept that inferences about relationships based on a name (e.g., congeneric species must be closely related) may be false. We already know to be cautious when interpreting a taxon name literally (e.g., it might be misleading about where it lives or what it looks like), so I don't think this is a big leap.

    2. Well, you're right that binomials themselves are not the problem — they're really just qualified names, after all. And if the namespace were just a namespace, as you suggest, they would not be a problem (which is why hyphenates would also work).

      But they aren't just namespaces. They are names of taxa, and when the species is moved out of that taxon, its name has to change. You can't move Drosophila melanogaster out of Drosophila and still call it "Drosophila melanogaster". (You also can't have overlapping genera, e.g., you can't expand Drosophila to include Sophophora, not without making Sophophora a subgenus and therefore no longer a namespace.)

      Even worse, in some cases you have to change the spelling of the epithet. If we move Sinosauropteryx prima into Compsognathus, we have to rename it as Compsognathus primus! The local names aren't even the same!

      That said, you're right that chresonyms would be even worse.

    3. To be clear, your system would work if we accepted that genera could be polyphyletic. But then we lose the utility of those names, which is another disaster unto itself.

    4. "You can't move Drosophila melanogaster out of Drosophila and still call it "Drosophila melanogaster"."

      I think Rod's point is that: yes, you can, actually. That certainly seems to be what people are in fact doing.

      "To be clear, your system would work if we accepted that genera could be polyphyletic."

      But we already know that not all genera can be monophyletic. (More precisely, if they are, then they can't partition life. They must either overlap or leave some individual not in any genus.) Is it such a leap to accept polyphyly as well as paraphyly?

      (I am not sure how I feel about this myself ... just trying the notion out. If you think of genus just as "the bit of a species name that comes before the space", maybe it's not a big issue.)

    5. P.S. Can't you just turn your CAPCHA off completely? It's a real barrier to getting an actual discussion going, and Blogger already knows who I am.

    6. Well, okay, Mike, yes you can do it, but you're in defiance of the zoological code if you do so. I mean, if you want to be an anarcho-taxonomist, yes, you can do anything you want.

      I didn't see an option for it but I'll look again.

    7. "Well, okay, Mike, yes you can do it, but you're in defiance of the zoological code if you do so."

      Are you sure? I don't recall anything in the Code about monophyly of genera.

    8. I didn't say anything about monophyly (which of course it does not mandate and never has — the principle of taxonomic freedom). I said that if you don't include Drosophila melanogaster in Drosophila (i.e., whatever taxon you call "Drosophila", be it mono-, para-, or polyphyletic), then you can't call it "Drosophila melanogaster" (by ICZN rules).

    9. Right, but once you accept paraphyletic and polyphyletic general -- which, implicitly, traditional taxonomists have done since forever -- surely the only real definition of the genus Drosophila is "The union of all species named Drosophila *"?

    10. Well, yes, but some people would like to use genus names such as "Drosophila" for actual useful taxa. (Maybe you're missing that point because in your field almost every genus is monotypic, but entomology is quite different in that respect!)

    11. Incidentally, this issue would "bubble up" to the family level as well, making those names useless as well. (And in botany, mycology, bacteriology, etc. it would bubble up even further!)

    12. You're right, of course, that my Mesozoic dinosaur focus does give me a terribly skewed view on broader zoological nomenclature. In my field, species names are uninomials, really; they are merely spelled with a space in the middle.

      Still, I think my broader point stands if fruitfly-jockeys want to keep calling The Animals Previously Known As Drosophila melanogaster by the name "Drosophila melanogaster", who are we vertebrate palaeontologists to tell them that they can't? In the end, nomenclature has to serve the people who use it.

    13. Err, but it's the ICZN saying they can't (at least, if you want to use "Drosophila" to refer to a taxon that doesn't include D. melanogaster). The ICZN commissioners backed this up with a decision, but it was already the case before any decision was made. (And I don't think any of the commissioners are vertebrate paleontologists, but I'd be willing to bet a fair number are entomologists.)

      I agree on the point about nomenclature serving the users. But that's the point — the rank-based codes don't do a great job, thanks to the fact that they mandate name changes when the taxonomy changes (for both multinomials* and ranks with mandatory suffixes).

      * polynomials???

  3. What Rod said.

    Every time we (I am as PhyloCode advocate and Companion Volume author) propose another one of these Let's Chang ALL The Things! schemes, we harm the Code's reputation in the eyes of the undecided and unconvinced.

    Really. Just stop.

  4. David Marjanović16 February, 2012 10:31

    Of course, the rank-based codes do not do a perfect job. Many species named early on do not have type specimens. This is a problem that I would like to see addressed in the PhyloCode, although I'm not sure what the best solution is. Perhaps the best way is to disallow use of species that do not have type specimens, unless a neotype is assigned under the appropriate rank-based code at the same time.

    I seriously think the best solution is to keep ignoring the problem. As long as no type specimen is actually needed in practice, there is no problem. When the need arises, people will designate a (neo)type, and that specimen automatically becomes the specifier according to the current draft.

    This automatism is also very useful when the type changes, like when a lectotype is selected from a syntype (it's not all that rare that a syntype is found to consist of specimens from different species!) or when a type is lost and replaced by a neotype.

    The current draft just says "the type" and lets the rank-based codes figure it out.

  5. "The current draft just says "the type" and lets the rank-based codes figure it out."

    Right. Which is perfect. Excellent software design, encapsulation and delegation.

  6. Come on, Mike, you know the danger of relying on a third party API....

    Maybe you guys are right, though. Come to think of it, someone would have to check that all species used in all submitted Companion Volume entries do in fact have type specimens, and then designate a neotype wherever there isn't. There's enough work to be done already, and this is a rather minor issue that could easily be fixed retroactively.

  7. The difficulty with allowing the rank-based codes continue to exist is to permit the ranks that they endorse to continue to have meaning, or merit. When newer codes that can operate without them actually instead parasitize them to function, this creates a system of interdependence that becomes harder to remove. Moreover, it creates a scientifically viable system of nomenclature and reference (PC) and forces it into a system that isn't scientific (ICZ/BN). I'd like the ability to go back to using Oviraptoroidea at some point (this doesn't advocate I will, but have the ability).

  8. "The difficulty with allowing the rank-based codes to continue to exist ..."

    *SCREEECHCHCH* (That was meant to be the sound of a needle being snatched off a record.)

    Woah, there! Are we now going to prohibit the ICZN from existing? That is ... impressive. Perhaps overreaching just a little, though.

    There is a lot of unreality floating around here. At present, the ICZN is used by maybe a million working taxonomists around the world. The PhyloCode is used by perhaps a dozen. Now I want to change that latter statistic as much as anyone, but come on. It's an ICZN world. Any tactic we come up with that begins by alienating a million taxonomists is a non-starter.

    BTW., Mike, your CAPCHA has got a whole lot more hostile just now. Can't you put it back how it was, or better still, get rid of it? (We don't need it at SV-POW!.)

  9. None of these codes are about science. They're all about nomenclature. Even though the PhyloCode is based on the idea of ancestry and descent, it says nothing about how you come up with your phylogeny.

    You actually can use Oviraptoroidea right now, as long as you use it for a taxon that doesn't include Caenagnathus (=Chirostenotes).

    It's not feasible to create a PhyloCode that covers taxonomic units (such as species) right now. The matter was looked into and it was agreed that it would be a Herculean effort that wouldn't give us a much better system that the current codes. The PhyloCode has enough of an uphill battle as it is.

    I think it's okay, at least for the near future. The processes of alpha taxonomy (organizing life into taxonomic units) and beta taxonomy (organizing units into higher taxa) are distinct enough that it's possible to have different codes governing the nomenclature for each.

  10. I didn't change my CAPTCHA settings at all! Must be a Google thing.

  11. As I said in response to the original thread on this, just get the Phylocode out. Not only don't we need to rewrite Phylonyms, we need to tell the authors who haven't gotten drafts in that they have a VERY limited time to do so, or else they won't be included.