This is an important question and one that is often brushed aside. I study domestic dogs and there is an effort to consider them as a monophyletic lineage (species or subspecies), even though it seems very likely that this group is polyphyletic and that non-domestic species continue to be used in producing new breeds or forms. My feeling is that for systematic analysis "species" (and genera should be monophyletic. This may be difficult as we find out how extensive gene flow is, especially in plants.

It really is an intriguing question. I'm afraid there is no easy or satisfying answer to it, though.

I think it's also highly dependent on the species concept that you apply. If you use the phylogenetic species concept then per definition a genus should be a monophyletic clade that contains only species which derived from a common ancestor and which share synapomorphies. The problem that occurs here is that there is a discrepancy in our knowledge. We can only directly observe the species that live today but don't know much about their potential common ancestors. We therefore cannot know with certainty if species that we observe today and which we placed into one genus really derived from a common ancestor.

An issue in taxonomy is that all of these classifications (order, family, genus...) are non-biological categories. They don't exist in nature. We invented these things in an attempt to bring some kind of order into the system and to be able to give things a name. Using even more of these categories (like subgenus, subfamily, sections, tribus etc.) doesn't make things easier and also doesn't help with the monophyly problem. If you split up a paraphyletic genus into different subgenera then some of the subgenera might be monophyletic but the genus itself will remain paraphyletic.

In evolution everything is more or less gradual but our categories are fixed. It is very hard to tell what even a species is (that's why there are a couple of different species concepts out there). Just imagine an extant species today which evolves and changes in some way over time. At which time point would you say that it is a new species? This decision is pretty much arbitrary and different people would probably argue very differently here. This problem already arises at the very tip of the phylogenetic tree, at the level of species. In higher hierarchies levels such as genera or families these decisions become even more abstract.

If we apply the phylogenetic species concept then in nature everything should be monophyletic because species derive from common ancestors. The question is just whether we are able to find the correct place in the phylogenetic tree to place our artificially created categories to mimic speciation processes that happened in the past.

All of this is coming from a zoologist's point of view, though. In animals we have much less issues with hybridization of species that then also produce viable and fertile offspring. In plants that's a whole different story and makes things even more complicated.

Hi

there is nothing mandatory in taxonomy, contrary to the situation for nomenclature, for which we have ICN.

Monophyletic taxa are just a goal, inspired by cladistic paradigm and computational abbilities.

There is now largely enough evidences, at least for plants, that evolution is basically reticulate, at various scales.

Nevertheless monophyletism remains the least worst approach to taxonomy.

regards

marc

Marc Philippe is correct that there are no rules that genera or higher taxa must be monophyletic. However, classifications based on synapomorphies are more predictive than classifications based on arbitrary combinations of symplesiomorphies and synapomorphies, and so monophyletic groups are a desirable basis for forming a stable and predictive classification.

Of course, whether some group actually is monophyletic (contains all the descendants of the common ancestor) is an unknowable metaphysical issue, and a hypothesis of monophyly is only as good as the empirical evidence that supports it.

Circumscription is a separate issue that is not related to monophyly. You can have two monophyletic groups that are sister taxa and call them one genus, or two, and either choice is a judgment call that has nothing to do with cladistics.

It is important to not conflate intraspecific variability with taxonomic variability. Species do not possess synapomorphies, and so cannot be monophyletic. At the level of individual organisms or alleles, you may say that at some point they coalesce or are "descended from a common ancestor" (or two, if sexually reproducing), but those are reticulate, tokogenetic relationships, and the concept of monophyly does not have any taxonomic meaning at that level.

This is only a dogma adhered to by cladistic formalists despite its being based on - even simply logical - contradiction: all subtaxa of a monophyletic taxon must, by definition, be descendants of a common ancestor, but... no such ancestor could have existed ("no taxon can be ancestor of another taxon") because it would be (also by definition!) paraphyletic! Moreover (and more important) - contrary to what Andrew writes, cladistic classification are not only not more predictive than synthetic ones, but they are completely non-predictive at all: they "predict" only the synapomorphies on which they are based on, and even these very inexactly and irreliably (e.g. the main synapomorphy for holophyletic taxon Insecta [Pterygota] are wings in imagines - could anybody show me wings in e.g. fleas?). So, attempts to avoid paraphyly are both hopeless (anyway, any non-monotypic monophyletic taxon is by definition paraphyletic, and if split to avoid paraphyly, the resulting less inclusive taxa will also be paraphyletic!) and harmful (result in non-predictive, i.e. in fact misleading, purely formalistic classifications)!

Roman Bohdan Hołyński gave the answer in a nutshell. The bad thing about the monophyly requirement is, that by and large it has become a dogma. Any dogmatic approach to a scientific question is in itself suspect.

Be aware of the logical conflict in the monophyly concept that has been pointed out by Roman Bohdan Hołyński (an by others before him). And then, ask yourself, non-dogmatically, what do you wish to achieve. The monophyly concept is relevant for classification, and it is valuable as a method to assess the "value" of any proposed classification. Nomenclature is an entirely different concept, and serves communication. We have grown to accept that the names of organisms must reflect their classification; but when accepting this becomes axiomatic, it reflects negatively on the usefulness of names as information vectors. When we wish to address a taxon that is worth while being recognized, because it is recognizable and we want to communicate about it, it is not primarily relevant that the taxon be monophyletic. Changing names each time a new classification hypothesis is proposed impairs the usefulness of names as information vectors and, in the long term, is contra-productive.

Even when one is reluctant to abandon full congruence between classification and nomenclature, one should bear in mind that translation of the first into the second is not a strait-jacket but leaves options; and it is the duty of responsible biologists to choose the option that is least disruptive with respect to familiar usage, and is most robust under the variation in classifications, or phylogenies, that are likely, if not certain, to occur in the future.

Werner: Thanks for the extended interpretation! I of course almost totally agree - with one caution: we should be aware of the often misleading cladistic usage of the term "monophyletic" [=neither poly- nor paraphyletic], as contrasted with traditional usage [=not polyphyletic, but including paraphyletic] - to avoid confusion Ashlock (1984) introduced unambiguous term holophyly (for cladistic interpretation of "monophyly"), and I would strongly suggest to use this, because paraphyly and polyphyly are fundamentally different relations: paraphyly means dscendance from common last ancestor, when the majority of characters (not only those most apparent, used in descriptions and determination keys) are also common (inherited from that ancestor), while members of a polyphyletic taxon descended from different ancestors, and have inherited different sets of characters (even if some superficial convergences may mimic synapomorphies). In other words, paraphyletic taxa are highly informative, sharing (almost by definition) many common (inherited from common ancestor) characters with few autapomorphic differences, whereas polyphyletic are mis-informative, with their many (inherited from different ancestors) different characters masqueraded by few superficial convergences. So, wishing to achieve truly natural (informative) classification, we should indeed not "discriminate" paraphyly (as shown in my previous comment, anyway hopeless endeavour...), but polyphyletic taxa are anyway unacceptable!

Roman,

Thank you for clarifying this point. It is polyphyletic taxa that I have been concerned with. How do we resolve the issue of hybridization and gene flow that are showing up in numerous groups of animals, e.g. those discussed in Arnold's 2016, Divergence with Genetic Exchange. Is it legitimate to put a Latin binomial on a polyphyletic taxon?

Raymond: "Is it legitimate to put a Latin binomial on a polyphyletic taxon?" - in my opinion not, but we must be careful with the interpretation of which particular taxon we consider polyphyletic: biodiversity, similarities vs. differences between taxa, are the results of [gradual!] evolution, so the circumscription of various categories (e.g. "poly-" vs. "monophyletic") is never perfectly "crisp". So, e.g., occasional (even relatively frequent, even if - what is not always the case... - truly proven) gene exchange does not necessarily mean polyphyly, and it is not always easy to reliably interpret particular cases (e.g. hybridization is relatively frequent among ducks, but nevertheless interspecific differences remain clear and agreeing with phylogenetic "branching pattern"). To be frank, I have little expertise in such situations (especially in plants...): I work with groups where "reticulate evolution" has never been demonstrated, so I have never been in a situation forcing me to sseriously contemplate on such problems, but anyway we should avoid "automatic", formal decisions.

A Latin binomen applies to a holotype specimen, which by definition cannot be polyphyletic. The circumscription of the species is not a nomenclatural issue. Sexually-reproducing species are related by tokogenetic relationships (interbreeding) and do not exhibit the properties of monophyly, paraphyly or polyphyly, which are terms referring to relationships among taxa.

Your answers are all insightful and interesting, however the situation I deal with involves domestic descendants of wild ancestors, where it is obvious that several different lineages have been involved over the last few thousand years, and as far as I know there is no type specimen. I am referring to domestic dogs, which are now considered to be one subspecies, Canis lupus familiaris, which are polytypic and have been derived repeatedly in different parts of the world from different ancestral forms to such a degree that it is virtually impossible to differentiate them on a consistent basis, in fact in many cases the ancestry of any given specimen is difficult to determine. As a result we have a grabbag assemblage being given a latin trinomial (the original Latin binomial Canis familiaris was given by Linnaeus). This would not be an issue, except that on a regular basis the fates of thousands of individual animals may hang on how they are classified. This is a case where folk taxonomy seems to be trumping phylogeny. Similar situations apply to domestic cats and cattle, but these species are less contentious from a legal perspective.

Andrew: The terms like paraphyly refer to taxa, not to names, and especially not to holotypes - otherwise the very terms of poly-, mono-, para- or holophyly would be not only superfluous but simply meaningless!

If this is true, why do people use Latin binomials to describe the Taxa or the holotypes they discuss professionally?

Raymond: I do not understand your question? Everything is referred to by its name - what another possibility could you suggest? And the name is "anchored" on the holotype - but tis is a technical solution with no relevance to para- vs. holophyly.

I guess I am saying that I am talking about organisms that are placed in phyletic trees and given a Latin name that are obviously polyphyletic. I am trying to figure out if this is justifiable.

Raymond: As I wrote a week ago, demonstrably polyphyletic taxa are in my opinion (and in that of the overwhelming majority of taxonomists) unacceptable, the problem is only in reliable demonstration of polyphyly.

These creatures exist, and some investigators seems to be trying to force them into a monophyletic tree, but this is clearly not true. What should you call a species population that is clearly of polyphyletic origin as many canid lineage's seem to be?

I don't know much about the phylogeny of dogs. But reading your comments makes me assume that there are several instances where dogs were domesticated independently. If this is true and all of these separate lineages are pooled under the name "Canis lupus familiaris" then the issue that you're having is obviously created by lazy taxonomy. If the domestication of dogs happened independently a couple of times then this creates separate lineages which should get individual taxonomic names. If you pool all of them under the same taxon name, even though we know that this taxon comprises different lineages, then of course this doesn't make much sense because it refers to a functional unit but not to a phylogenetic unit. It's like pooling all blood sucking insects into one group which of course can never be monophyletic, either.

From the preceding exchange on dogs, I am led to conclude that zoology would be well served by developing a concept analogous to the cultivar notion in botany. Breeds differ in origin and in practice from wild taxa, just as breeding programs follow different paths, not strictly comparable to the processes of natural diversification by mutation and selection.

What I mean is that, for the interbreeding population of the species “dog” (wild and domesticated), it is meaningful to use the single binomial Canis lupus. But there is an eminent practical need to refer to (and therefore, to name) individual dog breeds, for which purpose their “phylogeny” is all but irrelevant. Botany has its own Code for cultivar names, the International Code of nomenclature for cultivated plants (ICNCP), which has its own governance. As cultivar epithets are often used in conjunction with “normal” botanical names of genera and species, the two Codes are of course inter-related. But the ICNCP is able to deal with practical (including economically relevant) aspects like registration authorities, legal protection of cultivar names, and the like, in a way that would be unthinkable in the complex, traditional frame of official plant nomenclature. There are likely to exist legal rules and procedures governing such aspects, presumably limited to single taxonomic groups such as dogs or bovines, and almost certainly applicable only within single countries. Perhaps someone familiar with the naming and definition of animal breeds might comment.

Well, same problem with Phyllanthus L. Recent molecular studies showed that this mega genus is polyphyletic. So, a classification is forthcoming, presumably splitting the genus into many smaller genera with many combinations. This creates taxonomic problems. Example, before the age of molecular phylogenetics, Dr. Ding Hou had split Celastraceae into a number of segregate genera. Later, he realized and admitted that he himself is sometimes unable to identify these split genera accurately based on the key prepared by him! Therefore, the conclusion is that, you have to maintain a balance and always think of morphological correlations of molecular data. In other words, you should not go for excessive splitting, nor excessive merger but maintain a balance, best suited for easy identification.

Regarding domesticated animal breeds: there are registries, such as the American Kennel Club for dogs, and individual breed registries for horses (e. g., Jockey Club for Thoroughbreds), cattle, and most other species whose gene pools are artificially subdivided by selective breeding. Many of these are recognized internationally. I don't think it would be particularly helpful to come up with a Latin trinomen for a Jack Russell Terrier or an Afghan Hound.

(Sorry that this thread has strayed so far afield from the original question!)

Monotypic genera indicates poor genetic base and recent speciation of a particular species.

Modern research is a linear process. Nature often strays from this. Gene flow is not restricted to top down. Some grasses periodically outcross outside of their normal generic limits, as well as among species in the same genus. Lateral transfer of genes occurs in many organisms. Some taxa exist as species complexes, periodically back crossing to other species in the same genus or even to other genera. Monophyletic taxa are nice when they fall into place. However, not every ancestor will belong with only one group of descendants and to therefore lump everything into one larger group would be the logical path to make everything monophyletic and simultaneously of little use for sharing taxonomic identities. There is also often conflict for taxonomic resolution among data sets for different characters including for different genes sampled from the same individuals, including from different genes. Lump or split with great care.