Friday 2 September 2022

Should we allow paraphyletic genera?

This post is a rather theoretic one, and some might wonder what it has to do with the main topics of my blog. However, the question if we should allow paraphyly in some cases in taxonomy is relevant for the naming of species, some of which are in the focus of this blog. 

For those who are not familiar with the term “paraphyletic”, it describes when a group is not a natural group in the phylogenetic sense, a group that has a common ancestor but does not include all descendants of this common ancestor. Paraphyletic groups are avoided in modern taxonomy because treating them as if they were the same as mono- or holophyletic groups (such that have a common ancestor and include all of its descendants) is comparing apples to bananas. To illustrate that, I take my favourite vertebrate group as an example: dinosaurs (I am actually quite a big fan of Mesozoic paleontology ever since I was a kid). It is nowadays very well-established that birds descend from non-avian dinosaurs, so that they are part of the Dinosauria clade. Why? Because Velociraptor is closer to birds than Tyrannosaurus, and Tyrannosaurus is closer to birds than a Triceratops. If birds weren’t dinosaurs, Velociraptor isn’t a dinosaur either. But in this case, Tyrannosaurus would not be a dinosaur either, because it is closer to birds and Velociraptor than to Triceratops. And so on. Consequently, birds are dinosaurs because they are nested on the dinosaur branch of the phylogentic tree. Dinosaurs themselves are part of what has traditionally been called reptiles. That means that birds should be reptiles, because birds are dinosaurs. That sounds crazy when comparing a lizard with a duck, but that’s evolution. A crocodile is closer to birds than it is to lizards, and a Tyrannosaurus is closer to birds than a crocodile. Thus, the group of reptiles in the classical sense is not a natural group because there is that decision to regard some members of this clade not as members of this group for historical reasons that go back to the time of Linnaeus. This is called paraphyly. Therefore, Reptilia is not the same as Aves (an impression that can be created by the traditional rank system of four tetrapod “classes”), rather there is the Sauropsida clade, that includes a common ancestor and all its descendants, living and extinct. On this clade, there is the Aves clade, which includes all birds. Reptilia, however, is not a clade. It is a collective term for all sauropsids that are not birds, including not very bird-like animals such as lizards, and extremely bird-like animals such as Velociraptor mongoliensis. Since paraphyletic groups make the taxonomical system much more arbitrary than it already is, they are discarded in the modern (phylogenetic) systematics, otherwise one would compare apples and bananas. Systema naturae came 101 years before Darwin’s On the Origin of species and our knowledge of extinct organisms. 

Another problematic reminiscence of the original taxonomy that came before our knowledge of evolution are “ranks” for clades. That is the hierarchical system of species, genus, family, order, class and so forth. Nobody was ever able to define these ranks objectively and universally – how can we know that Hominidae, Tyrannosauridae and Canidae are of the same “rank”? There is no definition. Aves, for example, is considered one of four tetrapod “classes”, descending from reptiles, which were considered a “class” themselves. Now we know that Reptilia is not a natural group and most researchers use the name Sauropsida instead, but Sauropsida and Aves cannot both have the same “rank” as Sauropsida is a much more inclusive group. The same problem exists with any rank. For example, the Raphidae (dodos and its close relative Pezophaps solitaria) was downgraded to a subfamiliy (Raphinae) because it is obviously nested within the Columbidae (pigeons), so both cannot be the same “rank”. Therefore, “ranks” make taxonomy even more arbitrary and only worked back in Linnaeus’ time when we had no knowledge of evolution and extinct animals (as a side note, I wonder if Linnaeus would have questioned his own system if he would have had the possibility to classify living Velociraptor or Brachiosaurus, which are obviously between the “classes” of reptiles and birds in his understanding). “Ranks” for clades are not considered to be of any factual relevance in modern systematics anymore. 

This brings us to genera, which are ranks as well. However, genera cannot simply be ignored as they are perpetuated by the binominal nomenclature in biology. Each name for a species is composed of a genus epitheton and a species epitheton. Some species share the genus epitheton because they are considered members of the same “genus” (for example Panthera leo and Panthera tigris). A genus is based on a type species, and all the other species are assigned to that “genus” when they are considered “similar enough” to the type species to be considered members of the same “genus”. The problem is, however, nobody ever defined how to measure this similarity objectively and universally and how much of that measured similarity is “similar enough”. At least not that I am aware of. But we would need something like that in order to create a consistent concept of a “genus” that is necessary because it is relevant for the naming of species. The perception of what is “similar enough” also shifts with time. Back in the 18th century, a “genus” was about as inclusive as is a “family” is nowadays – at least in some cases (take elephants: back in the 18th century, there was Elephas maximusElephas africanus and Elephas primigenius, which are nowadays considered three different genera, ElephasLoxodontaMammuthus). If that was not problematic enough, there is the evolutional process going on. Species evolve into new species, the phylogenetic tree continues to grow. And as a consequence, there is inevitably the case that one genus evolves into another. On the cladogram, the resulting “genus” is paraphyletic. For example, the Haast’s eagle Harpagornis moorei is nested within the “genus” Hieraaetus, the two Bison species nest within the “genus” Bos. But saying “that genus is paraphyletic” may be not entirely correct, as a “genus” is not a clade but a “rank” that is often congruent with a given clade. A genus is defined as all species that are “similar enough” to the type species and the type species itself. Nobody defined a “genus” as a clade, so it cannot actually be paraphyletic. So, is Harpagornis moorei “similar enough” to the type species of HieraaetusHpennatus? Are Bison bonasus and Bison bison similar enough to the type species of BosB. taurus/primigenius to be considered members of that genus? The answer is entirely subjective, and there is no way to objectively and consistently measure “similar enough”. Perhaps Harpagornis moorei is more distinct from Hieraaetus pennatus than Bison bison is from Bosprimigenius. Including Harpagornis moorei into Hieraaetus means that also a hypothetical descendant of H. mooreiwould have to be included into Hieraaetus because it nests on this clade. That means that till the end of all days all species that would have evolved from Harpagornis or another member of the Hieraaetus clade, however distinct from the Hieraaetus type species they may be, have to be included into the “genus” Hieraaetus to avoid paraphyly. And considering that each genus descends from another genus, right down to the last common ancestor of all organisms classified under the rules of the ICZN, all animals are necessarily members of the same, ancestral “genus”, if a paraphyletic “genus” is to be avoided. That means that all genera are essentially synonymous. This makes the concept of a genus meaningless, if we are consistent with the approach that a “genus” should never be paraphyletic. But a “genus” is a rank and not a clade, and ranks are based on arbitrary subjective decisions based on similarity to a defined type species, and not on clades. They are often congruent with clades, yes. And regardless of how we name the species, Bison bison is a member of the Bos clade, and Harpagornis moorei is a member of the Hieraaetus clade (by the way, the same problem also goes for species. There is no definition of a species that works universally, but if we look at the individual level, all species in the history of evolution are paraphyletic on a cladogram, because some individuals assigned to the ancestral species will necessarily closer to the new species than to the earliest member of the ancestral species. Looking at the time scale, the concept of a species is artificial and arbitrary, and species are definitely not clades). Another way to avoid genus paraphyly instead of lumping is splitting. For example, Loxodontaafricana and Loxodonta cyclotis are “similar enough” to be considered a member of the same “genus”. But it turned out in recent genetic research that L. cyclotis is actually closer to Palaeoloxodon than to L. africana. How to avoid the paraphyly? If L. cyclotis is not “similar enough” to Palaeoloxodon, it might be assigned to its own “genus” because it is closer to Palaeoloxodon, but not similar enough to be a Palaeoloxodon species. That might work when looking only at Loxodonta and Palaeoloxodon. But all genera evolved from another genus. There will always be species that are closer to the descending genus than to the type species of a given genus. As a consequence, each species would end up with its own genus, making the concept of a genus moot. So, both approaches to avoid genus paraphyly, both lumping and splitting, would, when applied consistently to all organisms, make the genus as a rank useless – the problem is more fundamental, because a rank is treated as a clade here. 

So, considering that the concept of a “genus” as a rank is arbitrary and subjective, and the concept of a “genus” as a clade makes the term absurd because all living organisms would either be members of the same ancestral genus or each species would be its own genus so that no genus is paraphyletic, what should we do? At the moment, the way it is, it is inconsistent. See the example with Harpagornis. The concept of a genus, however, is tied to the binominal nomenclature. If we change the binominal nomenclature, we will have to change the names of millions of organisms. And naming a species with two words is more definite than just one word. But we could do something intermediate: erecting a new first epitheton for each species that is not a type species of an already existing “genus”. That would also be a lot of work, but it would be more consistent and compatible with the evolution of species and the cladistic system. In some cases it would work easily, when the species epitheton is a latin name on its own. Take Panthera as an example. I was unable to find out what the type species of Panthera is, but it would work in all five species: ParduspardusLeo leoTigris tigrisOnca oncaUncia uncia. Taking the Bos clade, it would work as well, because a number of synonynomous genera or subgenera have been erected or some species epitheta are just a latinized version of the name of the species: Bos primigeniusGaurus gaurus (gaur), Bibos javanicus (banteng), Novibos sauveli (kouprey), Poephagus mutus (yak), Bonasus bonasus (wisent, bonasus is just another name for bison, by the way) and Bison bison(American bison). For a lot of species, new first epitheta would have to be erected, especially for fossil taxa. That would be a lot of paperwork, but let us be honest, each scientist dreams of naming a new taxon, so that “Name, Year” will be mentioned next to the name forever. So, I think that many would not mind and take that opportunity to immortalize their own name by naming a new taxonomical name. And since it is not the description of a new genus that includes several species and has to be differentiated from other genera, no profound diagnosis is necessary because the species it refers to already has been diagnosed. 

I know that this is a radical approach, and it would take decades until it is established and the rules of the ICZN would have to be changed. But it would be consistent, less arbitrary, and compatible with evolution. As already mentioned, Linneaus came 101 years before Darwin, and taxonomy has already reacted by discarding paraphyletic groups. Maybe the next little revolution that is necessary to transfer taxonomy into the 21st century is to abandon the concept of a genus. The alternative is, if taxonomy is supposed to be consistent, to allow “paraphyletic” genera (which are not really paraphyletic since they are not clades). 

 

9 comments:

  1. This is basically a rehash of the "Phylocode" debate that ran for years with fairly high temperature. Eventually, the phylocode proponents found that the only way to get this done was to create an entirely new system, which to no-ones surprise failed to attract general interest.

    The main problem on radical propositions like this is that the role of nomenclature is not to perfectly reflect phylogeny (that's what we have cladograms for), it is to be a system that allows us to keep track of a large number of species and groups of species. The main users of nomenclature are biologists in other fields than phylogeneticists, to them it is of little concern that the occasional genus turns out to be paraphyletic, it is far more important that the naming is fairly stable. Various forms of phylogenetic nomenclature gives us a plethora of names for any nook and cranny in the evolutionary tree, but not an adequate overview.

    There's of course a perfectly acceptable way around this. The statements "birds are dinosaurs" and "birds evolved from dinosaurs" expresses the exact same phylogeny, the only real difference is that the first emphasizes the phylogenetic continuum, the other the radical anatomical and ecological shift. One is really not inherently superior, though the latter has the advantage allowing us to use an old and messy, but very practical way of naming organisms.

    ReplyDelete
    Replies
    1. I was a big fan of the PhyloCode as a teenager some ten to twelve years ago, I consciously did not mention the PhyloCode in order to avoid the impression that I want to discard the whole established system of the ICZN, which will have no chance of being taken seriously in many disciplines of biology. As a side note, I have the impression that certain elements of the PhyloCode are already established in some fields, like dinosaur paleontology where it is the rule and not the exception to only work with taxa that are properly defined in the way suggested by the PhyloCode. I think that is a very clear and practical way to handle it. See the controversy around Artiodactyla vs. Cetartiodactyla. If there was a proper definition for Artiodactyla, which can only result in a clade that happen to include Cetacea (no matter if node-based, branch-based, autapomorphy-based etc.), it wouldn't be a matter of preference if one uses Cetartiodactyla or an Artiodactyla into which whales are included. The same situation with Crustacea vs. Pancrustacea. I have no problem with insects being land-adapted crustaceans, whales being even-toed ungulates that lost their hooves and live underwater and birds being dinosaurs that learned to fly. That's evolution. I don't understand how using paraphyletic groups is practicable, maybe it is practicable when ignoring extinct taxa and looking only at the living world of today. The "four classes of tetrapods" for example really only work if one ignores the extinct taxa and therefore millions of years of evolution.
      Moving back to genera, I think my proposition doesn't even require a change of the rules of the ICZN, because there is no definition of how similar two species have to be to be considered members of the same genus. In a dinosaur forum someone wrote about 14 years ago it depends on how you "calibrate your genericometer", which is a way of putting it that is as funny as accurate. If some workers decide their "genericometer" is calibrated that narrow that each species is its own genus, there can be no objective way of saying that is wrong. Using the names Leo leo and Tigris tigris instead of Panthera leo and Panthera tigris is, as far as I have read the ICZN, not violating the rules of the ICZN.

      Delete
    2. I could tell your were a Phylocode proponent ;-)

      Taking years to redo rename all species isn't really a practical option. It will mean tossing out centuries of painstaking work and leave the field in limbo for years, likely decades until a new system has been established, with not guarantee that a new system _will_ be established.

      This is one reason PhyloCode failed. Another reason is that the extremely strict formula suggested only is obviously superior when viewed from a narrow POW of computational phylogenetics. In any other field it offers no substantial benefit over the existing codes, at least not benefits worth risking decades of instability over.

      I work in a natural history museum, and pylogenetic nomenclature is really only widespread in a single field: Dinosaur studies, which account for many peoples misconception that they are the only game in town. The idea that "most researchers use the name Sauropsida instead" is really only true within fields of dinosaur studies. Other fields mostly stick to traditional nomenclature, though the "you can name any node" idea has wider application as it can be quite useful.

      That whales evolved from artiodactylans and insects from crustaceans is as you said just evolution. The only relevant question is whether we have a language to describe these evolutionary events. I maintain that your description and mine are both perfectly serviceable and both describe exactly the same phylogeny. Mine have the slight benefit that it will also be understandable to the layman though.

      Delete
    3. I think also other fields than dinosaur studies tend to try to avoid paraphyletic groups nowadays, otherwise names like Pancrustacea or Cetartiodactyla would not exist. I prefer to see phylogeny as a big phylogenetic tree, with f.e. one branch being Dinosauria, and birds being on this branch, and one branch being Artiodactyla, with whales being on that branch. After all, those names are just names - the approach "it's an artidacylan because it has hooves, if it doesn't have hooves its another group" is, in my opinion, stuck in the 18th century. Hooves, in the case of artiodactyla, are just one trait that can be present, were likely present in the LCA of the group and can also be lost in some members of the group. As for the understanding of laypeople, I don't think it is all that relevant. Science is sometimes hard to comprehend to non-scientists, see quantum mechanics for example. I think the phylogenetic approach can even be beneficial in teaching people about evolution, while the Linnaean approach neglects the fact that evolution is a gradual process and all species are intermediate forms in a gradual flow, and not just Archaeopteryx and lung fishes.
      But apart from this more general discussion, I would be curious what your take on to the question of paraphyletic genera is. It seems to me that you prefer to allow paraphyly in genera, am I right with that?
      Of course it would be a lot of work to rename a large portion of all species that are recognized so far, and would make the work of distinguishing different genera moot. But that would be the consequence of genera being poorly defined, or actually undefined, in the first place.

      Delete
    4. True, it's not that people _want_ paraphyletic groups, it's just that a number of the more obvious ones happen to be (crustaceans, artiodactylans, reptiles, amphibians, Australopithecus etc).

      My job is actually to teach evolution (I'm a lecturer at the Nat.Hist.Mus. in Oslo (we've got a nice auroch skull btw, I can send you some pictures). Teaching groups from kindergardens to master students, I find the Linnaean approach works best. A strict phylogenetic approach means spending time on teaching people names, rather than teaching them evolution.

      The idea that traditional nomenclature ignores groups, doesn't take gradual evolution into consideration etc are all truisms you'd often hear from the Phylocode crowd, but with no actual basis. No-one seems to have a problem with mammals evolving from reptiles and reptiles from amphibians and so on. After all, names are just names, to the people I meet the revolutionary scenarios and mechanisms are behind these quite dramatic shifts are for more interesting than whether Morganucodon is a primitive mammal, a stem mammal or a therapsid. Why mammals became mammals is far more interesting.

      Delete
    5. I have nothing against using words like reptile or others as vernacular terms, which they are, but not as groups, because they aren't groups but arbitrary collective terms. I think there is a difference between a collective term and a clade that reflects an actual evolutionary relationship. It may be the case that kindergarden kids might at first not understand why whales are articodactylans etc. but science should not base itself on what kindergarden kids understand - physics doesn't spend much attention on that aspect either. Anyway, its probably a matter of preference, but I have the impression that the majority nowadays tries to avoid paraphyla, in the case of genera we have no other consistent option than accepting paraphyly or radically changing the way we name organisms.
      BTW, I would very much like to see photos of the aurochs skull at Oslo, you can send it to daniel.foidl( )itmed.at - many thanks!

      Delete
  2. Slight correction: Loxodonta cyclotis is not closer to Palaeoloxodon antiquus than to L. africana. That conclusion was drawn from mitochondrial studies, and nuclear studies have since revealed the similarity to be due to hybridization. https://www.pnas.org/doi/abs/10.1073/pnas.1720554115

    ReplyDelete
  3. This comment has been removed by the author.

    ReplyDelete
  4. What if we went the other way around?

    Make it (species) [subspecies, if needed?] (genus), and the rest can be a code of descent from the hypothetical first organism.

    Even if some new genetic analysis reveals somewhere in the distant past, X12 was descended from Y20, instead of W99, the most recent parts of the code remain the same, as well as the most distant.

    ReplyDelete