Living Words: the politics of taxonomic objects

| 5 Comments | 3 TrackBacks

During the first world war, according to a hoary old tale, the message was sent from the trenches to the command post behind the main front of the British Army from soldier to soldier. The message was sent as Send reinforcements, we're going to advance, but was received as Send three and fourpence, we're going to a dance. As things are transmitted, they change. School children are often introduced to this as the game "postman".

Change of things passed on seems to be inevitable. In biological history, of course, we call this evolution; but elsewhere we just call it change. Like all of culture, science changes too. Ideas are formed at one time for a reason and become different, or are abandoned, or we give them fresh life in ways that we didn't expect before.

Some of these ideas are the taxonomic ranks that go by the name of Linnaean. Linnaeus - also known as Carl von Linné (1707-1770) - was a Swedish botanist in the 18th century. He is responsible for a wave of enthusiasm that swept Europe for collecting and naming species of plants and animals, and he set up the tradition from which Charles Darwin ultimately sprang to form evolutionary theory pretty much as we now have it.

Linnaeus proposed five ranks, in his massive Systema Naturae, the tenth edition of which (1759, exactly a century before the Origin of Species) is the standard reference for species and genus names. If Linnaeus names it in that book, and someone else named it elsewhere, before or after, Linnaeus' names take priority. He named, for example, Homo sapiens.

Incidentally, these binomials, as they are called are the genus and species names. The species name is called the epithet of the species, and they are always printed in italics, by convention. The genus name is often abbreviated as a single letter, once it has been spelled out in a paper or book, and species epithets are always lowercased. So, for example, we have T. rex and H. sapiens, but never Sapiens or Rex.

Now, genus and species are logical terms based on the categories of logic developed first by Aristotle. They simply mean the general term and the specific term in categories. In that logic a species can be a genus for an even more specific term, but Linnaeus made them fixed ranks at the bottom, so that what we call a species in biology is what a traditional logician would know as an infimae species or "lowest specific term".

The ranks above them, though, are our focus today. Linnaeus, in big letters an inch or so high, begins with a political analogy: all reality is an Empire of Nature. Imperialism is a dirty word now, but in those days the most noble political organisation was, of course, the Roman Empire. The Holy Roman Empire still had some cachet too, although Linnaeus and his nation were firmly Lutheran, and he had trained in Protestant Holland. Nature is an empire, and it is ruled, of course by the Emperor, God.

The other ranks in his taxonomy were:

Regnum (Kingdom),
Classis (Class),
Ordo (Order),

And below the species he had another logical object - the Variety (varietas), which is an "individual" in logical terms, and not a class term. When you get to varieties, you have gotten to individuals, quite literally in Linnaeus' system.

Any botanist or zoologist will notice that we are missing two ranks. In the Linnaean taxonomy that ruled unchallenged (mostly!) for nearly two centuries, there are two other ranks that Linnaeus did not name. And one of them is particularly important in recent evolutionary discussions: phylum.

All these terms are political terms too: kingdoms naturally, to an eighteenth century naturalist (the term biologist being independently coined around 1800 by Trevarius and Lamarck), fall under an empire. Classes are obvious - there were the aristocracy, the middle class or bourgeois, and of course the peasantry or rustic class. And "order" is a name used for religious bodies as well as knighthoods ( social rank Linnaeus himself received some years later).

Linnaeus had a general term for these groups, no matter what the rank - and it was a military one; phalanx, the term for the highly organised ranks of the Greek armies that conquered the world and left Europe in its thrall throughout the Enlightenment and afterwards. The term did not catch on, and we had a number of proposals for "neutral" terms for taxonomic groups - one of the best known being taxon (from the Greek for division, around 1927), and another being deme from the Greek for "people", in the sense of population. Deme came later to mean a local breeding population, despite the best attempts of its coiner.

As Linnaeus' phalanxes marched on through the new science of biology, under the two kingdoms of plants for botany and animals for zoology (and minerals, but again Linnaeus was unsuccessful with his mineral species - they didn't seem to make sense, and eventually his classification scheme was abandoned for the current scheme of igneous, sedimentary and metamorphic rocks, etc.), it became clear that Linnaeus' scheme was not enough.

He thought that there were a relatively small number of species, genera and the other taxa (or phalanxes), each defined by a small number of characters - he wrote in an appendix to one of his works that all plants were definable by three characters with 60 variant forms each, giving a maximum of 603 = 216,000 maximum of plant species. We now have millions described.

So at a point early in the nineteenth century it became necessary to increase the number of ranks to accommodate this increase of species. Two terms were introduced. One  is due to a predecessor of Linnaeus, Joseph Tournefort (1656-1708), who proposed that things be organised into "families". The term Family was interposed between Order and Genus, in keeping with Linnaeus' conception of the divine society of nature. But the other term - phylum - is the invention in 1817 of Cuvier, and it has come to have a distinctly evolutionary meaning.

Cuvier noted that animals (the kingdom) fell into four major groups that were higher than Orders. He called them, in French embranchements, or "branches", but with not the slightest hint that these were historical branches. His main opponents in that respect were Lamarck and his follower Geoffroy, who had proposed a scale of evolution. In his Regne Animales, Cuvier called these embranchments by the Greek word phylum (phulon, meaning a race, tribe, or class; more politics).

Modern readers may know that Stephen Jay Gould felt that there was sufficient difference in phyla to pose an evolutionary conundrum. In Wonderful Life (1991), Gould proposed the reality of bauplans, a German loanword meaning "builder's plans" or "blueprints" as we would say. These bauplans constrained later evolution, he thought. Others note that this is true of any level or rank in the taxonomic scheme. But there has been an even more radical revision of Linnaeus' ranks of quasipolitical organisms.

Classification is the process of putting things into classes, or "kinds". At the time Linnaeus wrote it was thought that these classes were "thoughts in the mind of God", a view that can be traced back to the teacher of Aquinas, Albertus Magnus. But today we think of these classes as ways to summarise the properties of natural things (or as convenient ways to order things, such as books, but we'll leave the conventional aspect to one side for now).

This granted, that classifications should be "natural", biologists have found that there are an indefinitely large number of levels or ranks, and that these do not match up from group to group. Under a methodology of classification now used called cladistics (from the Greek word for branches), there are as many levels in the tree of classification as there are speciation events between the outgroup and the particular species. There are no fixed ranks.

Oddly, this is more like the classification scheme of traditional, medieval, logic, although they too thought there were a set number of levels in God's mind. But this was due to Aristotle's metaphysics rather than his logic.

So, instead of Linnaeus' serried ranks of armies of organisms arranged neatly and hierarchically into divine patterns or battle formations, we have classifications as hypotheses that can now be tested by different characters or by adding other taxa into the analysis. This is more scientific. Is it less political?

No, it isn't. If there is any science that is political, it is taxonomy. Taxonomists become highly charged with emotion and verve when their preferred ideas are under threat. It is not hard to see why - apart from the fact that change means, as it does to all scientists, that they have to relearn their speciality, it also means that their own work under the now-outdated model is worth less than it was. Scientists live and die (as scientists) by the value of their work. Remove that, and you remove the very thing that drives science - credit. Credit is, as philosopher David Hull has argued in a seminal work, Science as a Process (1989), the "fitness" of the evolution of science.

Recently, this has come to a head with a proposal known as the Phylocode, which is intended to replace Linnaean classifications with something based on the phylogenetic (evolutionary) history of the species. There are those opposed and those in favour, and the two groups are treating this with all the gusto required by the Kissinger dictum (Academic politics are so vicious because the stakes are so low, which he took from someone else in a fine irony). I watch one botanist break down in tears when he responded to Brent Mishler, of the University of California, Davis, arguing in its favour. Why is this?

The Phylocode applies on the assumption that we must name groups according to how they evolved. To that end, it assumes that cladistic trees, or cladograms, are accurate and reliable representations of the evolutionary history. The Linnaean ranks, now at around 18 levels what with Tribes (another political term!) being introduced, and sub-tribes, super-orders and the like now installed, simply cannot cope with the complexity of actual evolution as determined by cladistics. But not even all cladists agree with the Phylocode. Gary Nelson, a famous opponent of what has become known as process cladism (the view that cladograms equal evolutionary trees directly), has argued against it in seminars and conferences. Gary was my PhD supervisor, so I must declare my interest here.

Gary's and others' (such as Norman Platnick at the AMNH where Gary used to be curator of fishes) argument against Phylocode is this, and it relies on that conventional aspect of classifications I ignored above: one primary task of classification is to allow specialists to talk to each other, and to impart some shared information in doing so. A specialist in spiders agreed when I spoke to him about this, noting that if someone says a spider is a Mygalomorph, then he knows it is a hunting spider with retractable downward facing fangs, which probably hunts alone at night, and is not a web-spinner. The Linnaean classifications carry information that the phylogenetic classifications do not.

Note, however, that Gary and his colleagues believe this is a matter of convention and ease of communication. These classifications are not natural classifications, and conclusions about the organisms cannot be drawn from them - you cannot, for example, compare the rates of evolution of higher taxa like an Order, because there is nothing to compare - Orders are artificial. Someone as far removed from Gary's pattern cladism (the view that cladograms mark patterns in the ways traits are distributed, and need further interpretation to become evolutionary trees - in fact they test evolutionary trees) as John Maynard Smith, who died just recently and will be sorely missed, agreed - taxa - all taxa in his case - are just conveniences to biologists.

And this is something Linnaeus himself admitted. He did not think that what we call the Linnaean system was anything more than a convenient way to teach and communicate. He made a few sketches for a natural system, but it didn't work any more than any of the other natural systems (including Tournetfort's) did, until cladistics came upon the scene.

So now we are here - no longer do we think the world is organised by God into neat hierarchical ranks of armies, so much as we think that evolution is the cause of things looking treelike. From an analogy with the political realities of human society, we use a metaphor of organic growth and differentiation, one we get directly from Darwin himself.

Will Phylocode win out? If it does, what are the implications? Science is not organised like Linnaeus' phalanaxes either - it is a democracy and a marketplace of ideas. Like deomcracies and marketplaces throughout history, it is a terribly messy and inefficient way to test ideas and find the value of things, except, as Churchill noted, for every other system ever tried. People have tried to force science to behave in a civil manner and to remain on economic or political target, but every time the result has been moribund and corrupt research.

Phylocode will win out if it makes the science more reliable, easier or generates some research that would not otherwise be done, and which is effective. This is always why ideas are adopted in science. It is a democracy, but a democracy of fierce pragmatists. They do not construct the reality they study, despite what the social constructionists may think, so much as have it thrust upon them. I think Phylocode will fail; but I might be wrong.

In one way this is like the argument librarians have between the merits of the Dewey decimal system or the Library of Congress system. Similarly, with computers, it may not matter. We might be able to maintain both, so long as at the specific level they coincide. But defining units of anything in science is at least partially a matter of being forced to do it by the evidence, by the data. Cladistics works for constructing hypotheses about relationships; it need not work as a way of teaching taxonomic traits or for communicating. Time will tell and in the end, it will depend on what works best. We shall see how these battles, and this politick, resolves itself in the future.

In the end, the dance of science, popular science and pseudoscience, and the politics and warfare of the ballroom floor, goes on, if you want my fourpenneth' worth...

A good treatment of Linnaeus is:

Stafleu, F.A. (1971) Linnaeus and the Linnaeans. The spreading of their ideas in systematic botany, 1735-1789. Oosthoek, Utrecht.

Gould's book is:

Gould, S.J. (1991) Wonderful life: the Burgess Shale and the nature of history. Penguin; Penguin Books Canada, London: Toronto.

The specifications and papers on the Phylocode proposal are available at <> and Gary's arguments against it are at <> with links to other objectors such as Platnick. Discussion can be found here: <>.

Thanks to Wesley Elsberry and Ian Musgrave for constructive criticism and spell checking.

3 TrackBacks

Taxonomy from Keywords on April 27, 2004 3:10 AM

For much of history, the science of biology was the science of classification. As can be seen by the text accompanying this image (from a web page on "New Zealand polychaetes"): Schmarda was the first significant investigator of New Zealand polychaetes... Read More

The Panda's Thumb spends most of its time swatting down creationism, especially in its shiny new guise of 'Intelligent Design Theory'. But the truly interesting stuff is real biology, and the Thumb takes time to discuss this as well. The Read More

Living Words: the politics of taxonomic objects John S. Wilkins posted Entry 171 on April 26, 2004 04:50 PM. Trackback... Read More


Two comments John.

1) You say that Linnaean ranks are artificial; thats true only in one particular. While there is considerable debate about the nomenclatural side of this, there is very little debate about the taxonomic; that is, everyone agrees that taxon names are clade names, regardless of how we define those taxa. Thus, while the inclusiveness of any particular taxon in the Linnaean system is arbitrary beyond inclusion of the type at the rank of family (and totally arbitrary for other ranks), there is a very real sense in which “orders”, “families”, and so forth DO refer to real biological entities. They refer to clades, just as (X and Y), (X not Y), (X in Y) and all the rest under the PhyloCode do, only, they’re not defined with explicit reference to such. Im very, very strongly in favor of phylogenetic nomenclature, but i dont think you’re giving the traditional system credit for its ability to adapt, however poorly, to our understanding of evolution.

2) In a very Benton-esque (2000) paragraph, you say that phylogenetic nomenclature throws utility to the wind. The weight that certain critics place on this argument seems to me a result of a mistaken notion that clades defined with direct reference to ancestry are somehow unstable in terms of their content. This, of course, applies equally well to Linnaean ranks which suffer content changes both with shifts in phylogenetic hypotheses (a necessary result of phylogenetic taxonomy), and with subjective disagreements based on no evidence at all. For example, Rowe & Gauthier (1992) document all the clades to which the name “Mammalia” has been applied (they count 10), despite the widespread agreement about most aspects of synapsid phylogeny. Other cases abound.

While this synonymy exists under the current PhyloCodeless regime for those that use phyolo nomen, the code will do away with all of that. It will be impossible to use the name “Mammalia” in two different phylogenetic senses – a practice perfectly acceptable under the ICZN an its botanical and bacteriological counterparts. To the contrary then, i’d say that phylogenetic nomenclature is actually MORE utilitarian because its much, much more stable (in terms of actual names, definitions of names, AND content).

I recommend that everyone check out the replies Cantino, de Queiroz, Bryant, Lee, and Donoghue have been putting out there in response to the (misguided and/or confused, IMO) critics, particularly Bryant & Cantino’s 2002 “A review of criticism of phylogenetic nomenclature: is taxonomic freedom the fundamental issue”, in BR 77: 39-55.

Sorry for being so long-winded. ;)

Thanks for the response, GFA.

1. Linnaeus himself said that Linnaean ranks were an artificial system. As to how artificial they are now, well it depends on who you listen to. Not all taxonomists insist on a group name being a clade - Mayr doesn’t (and hence all those who still follow his rules), and Ken Kinman is insistent on allowing paraphyletic groups in his “eclectic” system. I would say that neither of those systems name natural groups except by accident.

The use of Linnaean ranks by cladists is unobjectionable to me. But then we do have the arbitrary nature, phylogenetically speaking, of what levels of the cladogram to dignify with a name. So, natural but arbitrary…

2. I don’t mean to say that clade-based names throw utility to the wind. Of course a good many such names are useful, and are chosen to be useful. But they are not useful because they are clade-based, but because that clade was chosen for its utility. As it happens, Linnaean stability is legendary, nay, mythological. I looked at Pinnipeds a while back (though the Mammal Species of the World) and found that at the specific level there was around a 40% change since Phoca was named by Linnaeus, and around 60% turnover of generic names (from poor memory; I must dig up those results sometime and follow them through). I’d love to see this followed through and analysed properly.

It seems to me that Phylocode will indeed be easier to apply once only. But given the instability of the cladograms (which of course they ought to be if they are sensitive to new evidence), this will possibly cause an explosion of names, each one stable but no longer used. This relies on what it is that we want classification to do. (Of course, this might all be crying “wolf”, and the end result will be just a whimper.)

I would rather think that “classification” applies to two distinct class-concepts. One, the clade-based one, applies to phylogeny, and is empirical. The other, the grade-based definitions of old and Mayr, are conceptual and conventional. Attempts to make grades equal clades are doomed to fail in cases where evolution has not neatly recapitulated ontogeny or an analog of it. When they coincide, then fine. But I think that grade-classification, or gradification, is something different and of a different utility (mainly communication and teaching) to cladification (as Mayr calls it).

1. I guess the point I was trying to drive home here is that taxonomic and nomenclatural issues are often incorrectly conflated. Which groups we name and how we define those names are two separate issues. Linnaean nomenclature is compatable with a phylogenetic taxonomy; the groups the names refer to can be perfectly natural, even if the way they are defined is arbitrary. As you said, natural but arbitrary.

You’re perfectly right about Mayr and some of the others, of course. I suppose I was being slightly hyperbolic. However, its beyond despute that MOST systematists these days are of the phylogenetic sort, and will, at the very least, totally avoid polyphyletic taxa and paraphyletic ones as much as humanly possible, even if there are some pheneticists and “evolutionary taxonomists” running around. Interestingly, Mayr’s “monophyletic” taxa rule out polyphyly as well, so he was halfway there.

I have no doubt about von Linne either. After all, he had no concept of monophyly.

2. The proliferation of new names is another common objection to phylogenetic nomenclature. There is, first of all, no need to name each and every clade. Certain clades (in particular nodes with monotypic stems, poorly supported ones, etc) are in fact recommended to NOT be given a name. Secondly and most importantly, while a certain phylogenetically defined taxon may be found to be more or less inclusive (it can never be found to not apply to any clade, unless of course its one of those “context specific” ones), there are ways to minimize such shifts. For example, crown groups with poorly resolved ingroup relationships can be given stem-modifed node definitions such that whichever groups ends up being the most basal, the name wont inadvertently refer to a clade less inclusive than the one intended. This is really the same thing as node with tons of specificers: I want to refer to all extant birds to the exclusion of paleognaths, anseriformes and galliformes, but including only, say, gaviiformes and passeriformes in my definition may make my name refer to a much less inclusive clade. If, for example, they were actually sister groups, i’d be leaving out an awful lot of extant birds. Instead, we just include gaviiformes, passeriformes, and ALL the rest. Whatever the change in ingroup relationships, my name will refer to the same clade.

Similarly, we can just refrain from including certain species or types as specificiers if we are uncertain about a form’s position relative to some internally well-resolved clade. There are adaptive definitions (X not Y or if such a clade doesnt exist, A not B) and context specific definitions (X and Y, provided it does not include A and B) as well. Whichever we use, these well-worded definitions actively counter the need for those sorts of changes.

Thus, in a phylogenetic taxonomic context, our options are a system in which there are no rules for clade synonymy and homonymy (I can capriciously apply new names to clades with prior names, or apply the same name to different clades), no rules governing how inclusive a clade must be (the content of any non-family rank taxon is arbitrary, and at the family, only requires the inclusion of the type), rank-based changes in name spelling which can result from changes in phylogeny or subjective disagreements as above (family “idae”, subfamily “inae”, and so forth), etc – or, a system that explicitly ties clades to names, that cannot result in name spelling changes, that minimizes shifts in content, etc.

If, on the other hand, its really about “evolutionary taxonomy” verses phylogenetic taxonomy – about naming clades as opposed to naming grades, there is a whole host of new problems. How “useful” are the arbitrary, unbiological grades that create confusion about phylogeny? Donoghue & Cantino (1988) describe the shock of two researchers when they discovered that the genus Heterogaura was derived from within Clarkia, precisely because they interpreted both names as reflecting phylogenetic realities.

And consider some of the potential practical consequences. Suppose some compound of medicinal value is found in a plant species. Because the plant is so rare and difficult to breed, attempts are made to locate it in other species of the same genus and family. But if the “genera” or “family” in question is poly/paraphyletic, and if the biochemists conducting the research mistakenly believe the groups are accurate representations of relatedness, they could waste valuable time and money exploring the wrong paths. If professional systematists can do it, lay people and pharmaceutical biochemists sure can!

Then there is the added subjectivity of grades. Phylogeny gives an objective measure against which taxon content can be reliably judged; to the extent that groups are not made on its basis, they are inescapably subjective in that measures of similarity must necessarily reflect characters or groups or organisms that particular systematists wish to emphasize.

Indeed, I cannot think of a single way in which paraphyletic taxa can convey any sort of useful information clades cannot, and thats with all their problems to boot.

BTW, can you tell us who the weepy botanist was?

No, I can’t. But if I could, I wouldn’t. There’s been too much personality in recent taxonomy for the taste of this sub specie aeternitatis-swilling philosopher…

The rest I guess we are in agreement, in terms of principles, at any rate. At least with respect to “evolutionary” grades. But let’s not revisit the monophyly “debate” hey? :-)

I enjoyed reading this. My small foray into biological taxonomy so far involved freshwater snails and ostracods. You know that you are ‘cutting edge’ when your ‘hot’ citations are from the 1800’s.

Hoewever, I did use Jardine & Sibson, and Sneath & Sokal in my disertation. The problem has always been (and still is), “How do we identify a smallest unit/taxa.” and “How do we organize taxa into larger units/groups.” When I teach this I try to present both the “old school” descriptive taxonomy and the notions behind cladistic approaches. This is all a generalist can do until the specialists sort themselves out.

About this Entry

This page contains a single entry by John S. Wilkins published on April 26, 2004 4:50 PM.

KwickCode is Here was the previous entry in this blog.

Johnson’s burning scientific issue is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.



Author Archives

Powered by Movable Type 4.361

Site Meter