The rules I follow are: 1) preserve traditional nomenclature for groups as long as they are monophyletic. 2) only name taxa that require new names (e. g. converting a traditional named "group" that is paraphyletic into two or more clades, one of which may need a name). 3) Use the widely-accepted code of nomenclature for your organisms (ICZN, ICN for botany) and not the "Phylocode" which does not have any formal standing in biological nomenclature and is not used by anyone other than its sparse but vociferous proponents.

Perhaps a few personal rules that can be generalized: (1) Don't bother about ranks. (2) Don't introduce new clade names. (3) If you think you have to introduce new clade names follow some simple somewhere defined concept (like phylocode -http://www.ohio.edu/phylocode/) and name only clades that appear to be well supported (high bootstrap/ decay values etc.).

Usually to label the nodes we simply use an alternance of letters and digits. Otherwise I don't understand your question/problem.

Classification using cladistics will be less subjective, that classifications based in a few selected characters, but there is no absolute objectivity (as far as we are selecting characters and taxa for our hypotheses; and in several cases there are confronted hypotheses of relationships). I am very fond of the use of cladistic in my field (fossils therapsids), but I don't get cladistic hypotheses with good enough bremer support in the groups obtained to feel comfortable adding new names to groups I am not sure they will survive (or in some cases I am pretty sure they will not survive) future analyses.

I'm not sure that entirely objective decisions are possible or even desirable. Assuming you want your taxa/named groups to be monophyletic, then that rules out a lot of groupings, but it doesn't prescribe any. Names are practical labels that allow human beings to recognise things, so you should name clades that are biologically meaningful and ideally also have a clear morphological identity, while at the same time avoiding any more disruption to existing naming systems than is strictly necessary. The ranked Linnean system does unfortunately sometimes force you into naming things that shouldn't really be named in order to avoid paraphyly. Non-ranked systems such as the phylocode don't, but in many/most organismal groups nobody uses them. There are supposedly objective rules that can be applied to naming (sizes of groups, quantifiable degrees of separation etc), but they tend to lead to bad names that don't represent the things that people wish to refer to. Don't be afraid of asymmetry (e.g. naming monspecific taxa if they are genuinely highly distinct and isolated), and don't newly name anything unless you are pretty confident it is monophyletic and isn't going to make another taxon paraphyletic if new evidence comes along.

I agree that we should try to be as objective as possible but we can neve be 100 % objective in our decissions.

Concerning the naming of clades:

I think it is better to use a term previously establish (even if you have to redifine it) than to introduce a new one. New names for clades should only be given if the support for the clade is high and no (more or less) similar group to the one you try to name has been named before.

The rules I follow are: 1) preserve traditional nomenclature for groups as long as they are monophyletic. 2) only name taxa that require new names (e. g. converting a traditional named "group" that is paraphyletic into two or more clades, one of which may need a name). 3) Use the widely-accepted code of nomenclature for your organisms (ICZN, ICN for botany) and not the "Phylocode" which does not have any formal standing in biological nomenclature and is not used by anyone other than its sparse but vociferous proponents.

When you need to decide at which level in your phylogeny you want to put some nomenclatural level (like genus): chose a level that is 1) well supported (and probably has many characters, since these are linked) 2) is easily recognizable (so easy characters rather than difficult to access; in plants: preferably also when sterile; for animals: maybe also in juvenile stage)

Thanks already to everyone for the many helpful and interesting answers! Because it was not clear to everybody, what I wanted to know, I added a further explanation to the question.

The way I understand your problem it would be better if you carried out your analyses first and then try to be objective, maybe try to find more or different characters for examble. This way if you have any doubts you will be able to reflect on them specifically. As for the the labeling problem the other answers should help you in your decisions.

Good luck!

My suggestion regarding the problem/example: Leave out all established taxa names that don't work any longer, use the others (with a similar or only slightly deviating meaning) and don't introduce new names (that might have only a short life span) but leave new clades unnamed pending further studies.

A counterquestion: Why bother about names, nomenclature and taxonomy if there are more interesting activities (like science)?

Michael B. - I also think that phylogeny is more interesting than nomenclature and taxonomy, but the reason for bothering with the latter is that we can't do anything with biodiversity (conserve it, monitor it, understand interactions within it) unless we have commonly understood names for biologically meaningful units.

Cladistics cannot be objective because cladistics cannot determine true monophyly. Cladograms are not monophylograms except by definition, and that is no way to do science. If two nodes are the same taxon, then splitting between those nodes leads to evolutionary (not phylogenetic) paraphyly.

As you know Richard, your view relies on your assumption of evolutionary stasis and a preference for interpreting shared characters at distant nodes (particularly chronologically distant ones) as nearly always being due to shared inheritance, even when parallelism/convergence fits the data much better. Your definition of monophyly (which is very much a minority view) is effectively a formalization of this assumption.

Neil:

Lot's of loaded words in your message. E.g. my supposed "assumption" of evolutionary stasis--actually evolutionary stasis is widely accepted and is the basis for taxonomy. Species gotta stick around, generating similar progeny for thousands or millions of years to have the distributions they do, and clearly (from my own studies, in prep.) pupping off one, two or three or more daughter species without anagenetic change to the ancestral species. One should never interpret shared characters at different nodes as anything much since cladistic data sets are not usually weighted by observation of conservatism of traits. Traits do not evolve, in any case, species do, and cladograms only serve to help taxonomists organize complicated relationships of taxa. When the same TAXA are at distant nodes, they imply a shared ancestral taxon of the same name. Parallelism is okay, but excuses of convergence are not parsimonious when a shared ancestral taxon is right there in the data set, so why propose an unknown, undefined, unnamed, and undemonstrable entity as a "shared ancestor"? My definition of monophyly is that of a majority view. You are assuming everyone is a cladist. Hey, you ducked my purposefully irritating assertion that "cladograms are not monophylograms except by definition..." Instead you implied that my theories were assumptions, a raw, bald-faced tergiversation. Is this because you secretly agree with me that cladistics cannot determine, alone, monophyly? (If you define monophyly as a clade because all clades are defined as monophyletic, we might as well stop the exchange here.)

Richard, I'm not denying evolutionary stasis, just the tendency in your methods to assume that in any given situation it's the best explanation for a particular pattern. Of course the scenario you describe happens - sometimes you will get a series of speciation events, with one of the lineages at each successive bifurcation retaining (for practical purposes at least) all of the definitive ancestral features. But all we have (usually) is the extant taxa and, if we're lucky, the tree that explains their relative relationships in terms of historical lineage splitting events. You don't decide what taxa are present at what nodes by assuming a priori that certain characters (the ones that define taxa of course..) are conservative, rather you find out what characters are conservative by reconstructing/modelling their evolution on a tree that has ideally been constructed using as much information as possible, and in the process discover whether similar organisms at different nodes share particular characters /because/ they are conservative, or because they have arisen more than once.

But let's say that we have a situation where a unique and distinct taxon has arisen from within a large and widespread group, thus rendering that group (minus the distinct taxon) paraphyletic (in the cladistic sense). Why is that a problem? It's only a problem if we want to name the larger group in a way that excludes the distinct taxon. The only good reason to do that is because you are using a ranked taxonomy, which says that if you name your distinct taxon at a particular rank (say as a genus) you are obliged to name any larger taxon that includes it at a higher rank (say family). This usually isn't sensible or practical, so if you want to only name cladistically monophyletic taxa you have to split up your larger group into smaller genera or else include your distinct taxon in the larger group without naming it separately. In a non-ranked system, you would just name the larger group (including the distinct taxon) and also name the distinct taxon, without having to name anything else. I'm not necessarily saying that non-ranked systems such as the phylocode should be universally adopted and certainly not saying that it's practical to do away with the established codes in the short to medium term, just that the problem isn't naming cladistically monophyletic groups, it's a fundamental incompatibility between monophyletic naming and the ranked systems we are stuck with for historical reasons. Your solution, i.e. redefining monophyly to encompass groupings that don't include all of the descendents of a common ancestor, strikes me as ultimately entirely forced by this historical contingent rather than being a way of naming what I would consider to be natural groups.

Try to include all species of a genus into matrix (see publications by Ounap et al on Lythria). Use two or three subspecies if a genus is monobasic. Check the tree for conflicting characters and try to change the polarization

[email protected]

Hi, may I dare suggesting the inclusion, whenever possible, of type species in you analyses and the careful examination of their phylogenetic relationships before you take any formal decision? As trivial as this may seem, I could find no mention of this in the discussion which followed your question.

Micheal B. - At the end of the day we still have to deal with the (evolving) rules of taxonomic nomenclature. This is part of the scientific work too. You may produce excellent scientific papers but if your nomenclature is outdated or incorrect, well, you are adding to the confusion.

In my experience, the two goals of taxonomy are biological information and effective communication, which includes taxonomic stability. Everyone has to choose the correct balance and the scale of groups tends to be taxon-specific. Remember that any changes you make are simply a proposal to the scientific community, so you make a strong case for whatever nomenclature you think is correct and it may or may not be adopted. Make sure you follow all the ICZN/ICBN rules, including reporting synapomorphies. I split a formerly paraphyletic group of snakes into three genera in a 2009 paper posted here on ResearchGate, and discussed some of this because the splitting was controversial. You're welcome to email for further discussion.

@ Allyson: A sentence from the abstract of your snake paper:

"The monophyly and distinctness of these groups warrant recognition at the generic level, and we propose a new taxonomic arrangement to reflect these findings"

contains some assumptions which may not be shared by many taxonomists and evolutionary biologists, like

"We believe that 'genus' is a meaningful/ well defined systematic category,

We think that a certain degree of morphological or other dissimilarity should be rewarded by the erection of new genera and other taxa of higher systematic categories.

We think that an established taxonomy should not only be corrected according to new results (minimal solution) but that taxonomy should always closely adopt fashions in phylogenetics/ systematics (highly invasive solution)."

The simple (minimal) solution would have been: one genus, several species (that group into several well defined but unnamed subgeneric monophyla). I think the only good argument for producing new names might have been the aim to reduce confusion.

Michael: You have a good point in the abstract, and you are correct about the minimal solution, but this paper supports my point that taxonomic decisions must be made based on the biology of the organisms. The genus I revised was paraphyletic with regards to a very divergent genus in terms of genetics, morphology, and life history. All named groups have distinct geographic ranges and life histories. The genera were named based on the totality of information, not simply a few synapomorphies. I disagree that taxonomy closely adopts fashions, nor that it should.

In addition to endorsing Andrew Brower's rules, I think that monotypic genera are not at all useful. When one finds two monotypic genera to be sister taxa, I would suggest seizing the opportunity to put both species in one and the same genus.

Neil:

Sorry to be tardy in response to your well-considered comments.

Yes, I consider evolutionary stasis to be the best explanation in any given situation when the alternative is an invented unobserved and superfluous ancestral taxon. Such unparsimonious invented unnamed taxa as "shared ancestors" are REQUIRED in cladistics because otherwise the cladistic method cannot establish monophyly. If two sequential nodes are of the same taxon, then splitting between them creates evolutionary paraphyly, which everyone agrees is to be avoided.

Your rejection of a reasonable assignment of conservative or, alternatively, a labile status to morphological traits flies in the face of 250 years of morphological taxonomy. You must admit that identifying "conservative" morphological traits as evolutionary tracking elements has been quite successful. Certainly identifying traits as conservative using a technique that cannot determine monophyly except by definition ("every clade is monophyletic") is ridiculous.

Regarding phylogenetic paraphyly, this is a problem only for phylogeneticists. One taxon generates another or a few, sequentially, or in molecular systematics, two or three molecular strains pup off an ancestral strain. Cladistics nests them. There is no problem with that as far as naming taxa is concerned. It is utterly false that taxa must be circumscribed by contiguity or clustering on a cladogram, which I repeat a third time cannot determine evolutionary monophyly.

Richard: I think our problem is that we have different ideas about the reality of taxa. As far as I am concerned organisms are related to each other by virtue of common origin, not because of the attributes they have or what taxa they are deemed to represent. Taxa are just named groups, and however we choose to define them this can have no bearing what the most recent common ancestor of any collection of organisms was or what its attributes were, or what other organisms also share that ancestor. It is not helpful to confuse the study of evolutionary history with the naming of groups of organisms, despite the fact that I and most people these days believe that naming should be consistent with historical relationships, i.e. common ancestry. When you cite the use of conservative traits over the 250 year history of morphological taxonomy, how are you implying that we should use such traits today? To help establish historical evolutionary events, or to define named groups? If the former, how would you explain the way we should go about this without using the words "taxon" or "taxa"?

Neil: Common origin is okay as a way to judge relationships, but cladistics can only estimate common origin in cases of true sister-group scenarios, e.g., speciation with pseudoextinction of the ancestor or when the two lineages are equivocal in helpful nonphylogenetic traits. When the ancestor remains in stasis after speciation, however, cladistics fails. The two lineages have no common ancestor (the ancestor is the same as one of the "sister groups") and any supposed common ancestral branch is imaginary. Since allopatric (peripatric) evolution is common, I think most morphological cladograms are largely imaginary. Since molecular lineages are actually molecular strains, and other molecular strains of the same taxon turn up all the time in other parts of the cladogram (as "paraphyly"), molecular systematics is also mainly nonsense. I refer you to my recent book, A Framework for Post-Phylogenetic Systematics, available on Amazon, priced inexpensively for students' pocketbooks.

A cladogram is not a phylogram. For the basic three-taxon cladogram (A(B,C)), there are five possible evolutionary paths. For example, speciation via a peripheral, small population that changes quickly because of a founder event, while the "sister species" remains evolutionarily static, is a very likely, perhaps very frequent, scenario, and it implies that the node for the "ancestor" represents no ancestor at all, but merely the point in (relative) time when the peripatric species diverged, leaving the main population largely unchanged genetically, ecologically and morphologically.

This should pose no problem. The monophyly requirement holds for taxa above the species level. Paraphyletic species or "paraspecies" do not falsify the cladistic paradigm. See Crisp.

Regarding the "rule" of Juan Schmitter-Soto above about lumping two monotypic sister genera into one: yes, absolutely, I agree in general with this rule, but only as long as it makes any morphological sense. If the combined genus is really hard to circumscribe, let alone be separated from other genera, then they are better kept apart. Another reason not to lump them might be that the difference (either molecular, morphological or both) between the taxa is at least as large as that between related genera.

Dear all,

as a paleontologist (although not a specialist of cladistics) I am interested and I have forged advices about the current massive use of cladistics (and cladograms) applied in taxonomy.

As emphasized by Juan Schmitter-Soto any single cladogram can be the basis for a number of phylogenetic trees. The problem is that in current practice almost everybody uses raw cladograms directly as a basis for phylogenetic inferences (and including for taxonomy). This is a big problem from my point of view. As has been remarked in comments above, the hypothesis that a known species is ancestral to another is not, and cannot be rendered by raw cladograms. Nevertheless, data often cannot rule out such a hypothesis, and there are many lines of evidence that such a scenario is widespread in evolution. In other words perhaps 10% of times a known fossil species indeed is ancestral to another (fossil or not) species (see Foote's papers for instance). Cladograms cannot render this, and since people use directlty cladograms, they automatocally and artificially rule out these ancestries.

A solution would be that people 'tranform' cladograms into more interpreted cladograms, which would become (several) more "phylogenetic" trees. Among others, it would include representing the possibility of placing a species at an internal node, as a direct ancestor of other taxa.

I like the idea of "cloudograms" in representing the multiplicity of phylogenetic hypotheses rather than a single "consensus" tree. Still, cloudograms are sorts of cladograms, and the step toward phylogenetic trees if not crossed in the vast majority of publications.

For my PhD thesis (which is a specimen-based phylogenetic analysis in paleontology, so morphology based, and lots of incomplete specimens), I tried a numerical approach. I basically counted the recovered 'autapomorphies' or 'synapomorphies' of two ends of a dichotomy, and summed them to get a number of morphological changes needed to get from one end to the other. I counted the changes between specimens of the same species (based on earlier publications corroborated by the recovery of them as sister groups in my own analysis), two species from the same genus, and two genera. I considered one pair of genera as the best example for generic distinction because they were also known from two different continents, thus also geographically separated. Like this I was able to establish a 'changes count', which I then applied to all other dichotomies, in order to see of specific or generic separation of the two specimens or clades involved is justified.

The absolute number of changes is only applicable to this specific analysis, as it depends on how many characters you have in the analysis, and on the systematic history of the group (because the counts were established based on earlier identifications of specimens, species, and genera).

Like this, I hoped to get rid of some subjectivity, but keep changes in taxonomy as low as possible, and I think it was quite successful. The study is now in preparation for submission, so please if you have any more questions or comments or concerns, let me know. I think this is an important question to address. We do need some more objective base for taxonomy.