I guess in some fossil taxa, phenetics are still the way to go. Fossil taxa of foraminifers, dinophytes/-flagellates etc. without extant members won't leave much choice.

I think the summit of phenetics was over because the arbitrarity of phenetics is not overcome easily. The aim of  classification is to establish a natural system and  cladism and evolutionism  would be more consistent to the aim philosophycally.

What a rather bizarre question. Cladistics is a methodology, evolution a scientific theory and phenetic characters are just that: morphological characters. Cladistics is just one of many methods currently employed to resolve phylogenies, and without a doubt molecular character sets are much more powerful than phenotypical ones. Both are, however, complementary to one another when trying to resolve phylogenies. Any phenotype is, after all, an expression of its genotype!

Phyologenies ideally reflect true ancestries and thus the evolutionary history of a species. Methods and character sets employed to decipher the way how a particular group of organisms have evolved are part of studying evolutionary processes. Ideally, modern taxonomy aims to resolve as accurately as possible the evolutionary history (= phylogeny) of species. Nomenclature should ideally reflect this, but nomenclature also serves the simple purpose of providing a reference framework for biology in general to name and thus distinguish different species.

The antagonism that your question implies is, in my opinion, an erroneous way of looking at evolution. It implies that one specific method using only one particular set of characters is superior to any other. Nature does not make such a distinction. Science, aiming to be objective, should not either.  Taxonomy, the attempt to name species according to how they related, is just that: an attempt. In that effort any taxonomist strives for the best approximation of reality. In the end, what we believe to be able to recognize as a species is only a hypothesis, a scientific approximation of the truth, using the characters and methods available at the time.

I fully agree with Frank's outline. There is nothing more to say.

I, too,  fully agree with Frank's outline. There is nothing more to say.

I respectfully submit that Frank has missed the point of the original question. Phenetics was a methodology, not a description of characters. It was the grouping and classification of organisms based on overall similarity, as opposed to Hennig's phylogenetic systematics (which became known as cladistics) which used special similarity (synapomorphy) to group and classify organisms. Those were the competing paradigms and while Hennig's basic ideas certainly won out over phenetics (see D. Hull's great book Science as a Process), phenetics is back on the rise. With the accumulation of massive genomic data sets, some folks are turning to phenetics as a way to analyze these data sets. 

I agree with Brian. Phenetics (numerical taxonomy) was a methodology developed by Sneath & Sokal in order to give a more scientific (objective) approach to traditional comparative morphology. Hennig proposed his methods of Phylogenetic Systematics [= "cladistics"] as an improved alternative to these more general similarity (clustering) based methods. His major innovation was to determine whether shared characters were likely due to convergence (homoplasy) or common ancestry, and if genuine (ancestal) evolutionary characters were inherited unchanged ("symplesiomorphy"), and thus not necessarily an indication of recency of ancestry, or the more informative shared derived character states ("synapomorphy"). This establishes polarity in a phylogeny, with only synapomorphy indicating recency or common ancestry. There were also differing philosophical commitments driving the debate. The principle of parsimony was originally considered the only way to determine the best tree (the simplest/shortest tree, ceteris paribus. Popperians viewed such falsifiable approaches as the only true (philosophically sound) mark of a scientific approach to evolutionary relationships. The reason there is now "nothing more to say", is because cladistics ultimately won the war, but the battles were bloody (and often fought over relatively unimportant distinctions between the different factions - much like the various sectarian intrigues among christian denominations since the reformation). With the advent of molecular methods and the vast amounts of data unleashed in the genomic era, dogmatic approaches have succumbed to a much deeper and more subtle appreciation of evolutionary theory in phylogenetics (as well as much more sophisticated mathematical models employed in the latest software tools). In addition to this, the hypothetical nature of the "one, true tree" has been proven beyond doubt, and parsimony shown to be an often unreliable simplification. Ironically, phenetic methods have not died out, but are now used only as simple approximations (clustering methods such as neighbor joining), in order to "seed" more laborious search algorithms such as Maximum Likelihood and Bayesian Inference methods.

Taxonomy (systematics) is thus the intersection between Phylogenetics and Nomenclature. The "best" phylogeny represents the "scientific" part of taxonomy - the most current understanding of evolutionary relationships; whereas nomenclature represents our attempt to fit the (largely arbitrary aspects of) ranking and naming within the framework of Linnean Classification. This distinction is incredibly important, because it too has led to much confusion and controversy. In order for a classification to be "natural" and "accurate", it must reflect current understanding (best hypothesis) of evolutionary relationships. Every scientific name is thus a placeholder for a scientific hypothesis based on an underlying body of phylogenetic inference. There are no competing "phenetic" taxonomies, as such a thing would imply some other (pre-Darwinian) explanation for homology. However, traditionalists who disagree with strict "Cladistic" (monophyletic groups only) approaches to Nomenclature feel that paraphyletic taxa should not only be tolerated, but preserved. No modern taxonomists I know of actively promote known polyphyletic taxa.

I am sorry but Frank has missed the point of my question, as Brian said. There is a lot of confusion between cladistics vs cladism, phenetics vs pheneticism. I am not asking about methodologies for infering phylogenies, but about the philosophical attitude toward classification. Phylogeny and classification are logically separable things, even if they are treated together by most modern systematists.

Christopher gets the point, he replied "No modern taxonomists I know of actively promote known polyphyletic taxa". This is my question. Pheneticism explicitly allows polyphyletic taxa. Do you know someone promoting this view today? Or who was the last one?

Thanks Damien. The short answer is no. And in reality, the whole idea of polyphyletic taxa is a cladistic concept. Pheneticism (if such ever really existed - the only adherents I can even imagine would be creationists/"Intelligent Design" adherents) would never have admitted to the value of classifying unrelated organisms for any but accidental and pragmatic reasons.

Dear Damien. I appreciate the clarification. "Cladistics = methodology". "Cladism = school of thought", right?

Out of curiosity:

Could you perhaps clarify, what you mean by "evolutionism"?

Also, please excuse my ignorance about authors who you cite "promoting polyphyletic taxa".

Why would anyone in his sane mind do that? The term "polyphyletic" literally mean "of multiple ancestry". Why would anyone suggests that organisms with different ancestry are related?

Where phenetic classification systems still exist, or at least a hybrid phylo-phenetic classification system, are in prokaryotic taxonomy.  This is because pure cladistics miserably fails when applied to these organisms.  Cladistics inherently presumes that characteristics are inherited strictly “vertically” in a lineage.  The relatively large amount of lateral genetic transfer in prokaryotes create many “false synapomorphies” that make purely cladistic taxonomy an exercise in futility.  The current practice for prokaryotic taxonomy could be described as “numerical taxonomy fused with 16S phylogenetics with a dash of whole genome DNA similarity.”  Finding a workable species definition in prokaryotic taxonomy for the genomic era is still a work in progress.

I agree with Christopher, I am unaware of anybody trying to recognize polyphyletic taxa, or even paraphyletic taxa for that matter. If classifications are information storage and retrieval systems and phylogeny is the information we want to store and retrieve, then only monophyletic groups work. If I were to guess where acceptance of poly/paraphyletic taxa occurs, it might be with microbiologists simply because those systems are full of reticulations making monophyly perhaps a too difficult thing to infer with confidence. 

@Frank, evolutionism = evolutionary taxonomy, i.e. a school of thought allowing paraphyletic taxa because they represent macroevolutionary leaps (about this, see my last paper "A formal analysis of phylogenetic terminology: Towards a reconsideration of the current paradigm in systematics").

Back in the 60s-70s, Sneath and Sokal promoted a classification scheme based on cluster analysis of phenotypic characters whose aim was explicitly not to represent phylogeny at all. It was supposed to be a purely pragmatic classification for information storage and retrieval. However, like Brian said, the information stored and retrieved was not the phylogeny but only the phenotypes. So it resulted in polyphyletic taxa in cases of convergent evolutions or of reversions. They just didn't care about mono/polyphyly, phylogeny was a distinct matter for them.

So I am just looking for people alive still promoting these ideas, or else for the last dead people who did.

Damien, that is an interesting historical question. I poked around a little and as recently as 1996, Priest and Auston in their book Modern Bacterial Taxonomy, advocated phenetic classifications over phylogenetic classifications. 

Thank you Brian, this is an interesting reference indeed ! The fact that these people are still alive tells us that the debate is not yet an historical question.

Hi Damien! I would dare to say (admitting this answer will be disgusting to some) that if there are adherents of pheneticism today, they don't have a meaningful scientific production. In its very origin pheneticism was interested in grouping and separating organisms without considering phylogeny. This is its "original sin" and its fatal flaw. Well, this is my opinion. All the best!

Neighbor Joining is a phenetic method.  So, if you use it to make "phylogenetic" trees, you are, whether you recognize it or not, a pheneticist.

All (at least all known or imaginable to me) methods of phylogenetic reconstruction are based on characters (i.e. on phenotype - even if the characters are DNA sequences), so in this sense all - not only Neighbour Joining - are "phenetic". But all those currently in use (including Neighbour Joining) use analysis of character similarities and differences as (admittedly imperfect but the only available) evidence of degree of phylogenetic affinity, while true (Sneath and Sokal's) phenetics did not bother with phylogeny, and so, in fact, did not include the character analysis (did not discriminate between homologies and convergences, to say nothing about apo- vs. plesiomorphies) at all. And this is a principal difference. To my knowledge, at least in zoology declaredly phenetic approach is a history (though Brian is right that "With the accumulation of massive genomic data sets, some folks are turning to phenetics as a way to analyze these data sets"). On the other hand, I am rather astonished with Brian's being "unaware of anybody trying to recognize polyphyletic taxa, or even paraphyletic taxa for that matter" - I, e.g., am one of those considering acceptation of paraphyletic taxa as logical, theoretical and practical necessity (see my "Fallacies and false premises" and several other papers downloadable from ResearchGate, in which you will find references to many other recent authors); in practice paraphyletic taxa are "tacitly" widely accepted by taxonomists in most groups of organisms - contrary to cladist's claims, domination of cladistics in Vertebrata is rather exception than rule in other organisms).

Very well put Roman! I would only clarify my view that, although purist "Cladism" (in the strict sense) does not dominate current (macro-organismal) taxonomic practice, we all operate within the dominant paradigm of phylogenetic systematics in the broad sense (cladistics with a "small c"). Absolutist Hennigian monophyly is a hubristic delusion, and the necessity of grappling with unavoidable paraphyly, from species to kingdom, indeed humbles us; but thankfully, pervasive epochal extinction has made this the exception rather than the rule.

Thank you for all your answers.

Roman Holynski, I didn't know your work, these are great articles against cladism. It's a shame that they were not properly published. I encountered the same difficulties recently.

Andrew Bower, NJ and other derived distance methods (probabilistic or not) result in a phylogram, not a phenogram. Taxa are not grouped according to their overall similarity, unlike UPGMA. So their use don't make people pheneticists.

I agree with John and Isuppose that phenetic classifications systems exist in prokaryotic taxonomy. The field of preference of phenetic classifications remains microbiology. However, I think we have to distinguish two questions: how to classify the species (building a taxonomy) and how to constitute some classes with a maximum of resembling characters. From my viewpoint, phenetic species is a statistic tool to constitute units or classes of similarities. So we can think that this has more to do with statistics or biometrics than taxonomy (stricto sensu). Descent and fertility relations are ignored.

Hi all, i invite you to to read that paper. Regards

https://www.researchgate.net/publication/273718023_Phenetic_Analysis_of_Morphological_and_Molecular_Traits_in_Acanthaceae_Juss?ev=prf_pub

Conference Paper Phenetic Analysis of Morphological and Molecular Traits in A...

This seems indeed interesting, thank you.

It is worth pointing out that phenetic approaches do have their uses in species delimitation studies to identify discrete phenotypic entities, either to confirm the distinctness of candidate species identified through other approaches (e.g., barcoding) or as a first pass when no molecular data are available. 

Some might take the view that DNA sequence analyses are phenetic analyses dressed up in cladistic methods and terminology because the character states themselves cannot be initially defined as derived or not without constructing the tree first because homology of bases between taxa cannot be identified where genes are of different lengths. Matching of bases requires algorithms that give the best overall fit by making the genes the same length - essentially creating homologies that are not empirically observable in nature using a phenetic method. Then the matching bases are used to create trees that then by default define the homologies. At least that is my impression and a view shared by others, although still others would presumably disagree. What I see as the phenetic nature of DNA sequence analysis may explain why there is so often a mismatch between sequence and morphological trees - as exemplified by the mismatch in human and great ape phylogeny.

Dear Damien,

As to your question: "Are there still adherence to phonetic classification today?"

Perhaps, for you will be interested in this work.

My best regards,

Irina Foronova

John, it is not actually the case that molecular phylogenetic analyses are intrinsically phenetic, although the confusion is understandable. This view seems to be due to conflation of a couple of different steps in molecular phylogenetic method. It's true that the starting point is based on an assumption of homology, but this is no different from morphological character assessments (except in the methods used - DNA extraction using the same primers, etc.). In fact, it is not the case that morphological homologies are a priori observable through a "lens" of cladistic practice. Apparent homologies must be established before phylogenetic "polarity" (plesio-/apo-) can be established. Similarly, the issue of assigning homology to individual bases is overcome during alignment, and is not directly related to the problem of searching for the "best tree". It is true that alignment of individual bases (and indel "gaps") can be ambiguous, and this is overcome by algorithms that search for the optimal alignment (analogous to parsimony in that it may not represent the "true" evolutionary/biological alignment, which cannot be absolutely determined due to lack of information). The trees are generated during a separate step using different methodologies, and do not define homologies. The homology debate has been conflated here with a debate that occurred between different phylogenetic schools of thought on whether or not to treat Maximum Likelihood and Bayesian Inference methods as truly phylogenetic (some purists would only allow parsimony methods such as PAUP). This is an entirely separate issue, and is due to the need for phenetic seed trees (NJ etc.) in ML & BI.

Christopher,

I appreciate that my views on this matter are not shared by the majority and nothing I may say here will change that. Yes, in morphology one may assess each character state INDIVIDUALLY with respect to an outgroup and therefore restrict character analysis to those character states that are unique within the group being analyzed. Yes – optimal alignment is a phenetic process where individual bases are matched according to an overall ‘best’ fit between gaps and substitutions. So in the latter one starts a molecular analysis with phenetic characters – characters defined by overall similarity. There is the additional problem of not knowing which base state is necessarily derived. In morphology one at least has a comparative basis for assigning the derived state. Of course the best supported cluster of derived character states is the one chosen as most likely closest to the actual phylogenetic sequence. One may do any number of computations with sequence characters, but I rest on my opinion that whatever computational clustering technique is used, one essentially ends up with a phenetic result and to me that at least provides a coherent reason for why extremely robust morphological cladistic results do not always agree with molecular sequence results. But, all that said, I recognize that just about everyone thinks differently about this.