Tuesday, January 3, 2012

The hollow curve, lumpers versus splitters, and arbitrary (yet useful) ranks.

Yesterday, science journalist Ed Yong asked "which genus has the largest number of living species" as a Google+ thread. He admitted that it was a trivial question of curiosity, and he's right. While the wood boring genus Agrilus (Coleoptera: Buprestidae) may have the most described number of species (~3000), this is not a particularly important or interesting question when it comes to taxonomy or systematics in general. However, the question spawned an energetic argument over the arbitrariness of taxonomic groups in terms of size and content, and I wanted to speak my piece about it.

[If you graph] the number of genera contained in any group [of organisms] against the number of species contained in those genera, using one axis for genera and the other for species, we get, apparently regardless of the group we are dealing and the quality of the systematic work done upon the group, a curve of a characteristic form. This has been called the "hollow curve of distribution". (from Ferris (1928) The Principles of Systematic Entomology)

In modern ecology the hollow curve describes the relative species abundance within a community, but it also describes the distribution of species within genera. A small number of genera will have a large number of species relative to the whole, and a greater number of genera will each have a relatively small number of species. Unlike ecology, this was an probably an artifact of human taxonomy and the continuous disagreement between "lumpers" and "splitters".
An exaggerated "hollow curve of distribution" for visualization, with numbers of species on the Y axis and number of genera on the X axis.

The general consensus in systematics is that species are real groups of organisms, and for the purposes of this post I would not like to argue over that point, or about genetics or species concepts. Likewise, monophyletic groups, that is, groups of species which share a single common ancestor, are also real. After the modern evolutionary synthesis and the description of classification as the depiction of evolutionary relationships and not overall similarity (regardless of relationship), monophyletic groups became the standard for describing taxa.

A simplified version of the Linnean Hierarchy (© Tutor-Vista)

However, taxonomic ranks (e.g. family, genus, tribe) above species and the decision to place a particular group as a genus or a tribe, for example, is completely subjective and somewhat arbitrary. The hollow curve of distribution was due to the choice of some taxonomists to make large genera ("lumpers"), and of others to whittle those large genera down to smaller genera over time, or to raise a genus to tribe and the subgenera within to genera ("splitters"). This is sometimes called "taxonomic inflation" by those who see this tactic as an attempt by the taxonomist to raise the public importance of his or her group of interest. These split up genera are all monophyletic groups, just much smaller, more manageable monophyletic groups. (ETA: other hypotheses in the comments)

Still, the Linnean ranking system remains useful in relating hierarchy. It is unambiguous that a tribe belongs to a family, that a subgenus belongs to a genus (ETA: though not necessarily to the genus you /think/ it belongs to, or including the members you would expect: see comment by Christopher Taylor below). If I tell you a species is of the Tribe Hydropsychini, it is taken for granted that the tribe contains genera, and is contained by a family (in this case Hydropsychidae). This makes it easy to organize the general reference system. Ranks also make it easier for taxonomist to communicate their phylogenies. (ETA: Again, see comment by Taylor below)

Unfortunately there are some taxonomists who are uncomfortable with the Linnean system of ranks. Among these are the people responsible for the Phylocode. The Phylocode eliminates the Linnean ranking system and replaces it with a hierarchy of unranked "clades", while species names are still described under the rules of the already established codes of nomenclature. This to me seems like hierarchical obfuscation. While it is very easy to communicate a hierarchy using ranks, it is very difficult to do so without them. Anti-ranking advocates complain there are not enough ranks to properly describe all the monophyletic groups, but the code doesn't prohibit the creation of new ranks above and within family level ranks, they just are not covered under the rules. If you want to talk about a semi-infra-tribe, or a super-kingdom, that's perfectly acceptable. Any other criticisms, like the issue of hybridism, are best addressed within the individual codes. To collapse and remake the entire general reference system just to eliminate rankings (which are useful in their own right) is preposterous. Enough people were upset over the Drosophila case that there may be rioting in the universities if Phylocode is ever successful.

So, while the arguments that taxonomic groupings are arbitrary is false, ranks are somewhat arbitrary and yet useful. The important notion is to not rely on them as indicators of importance. And that questions like "what's the largest genus" comes down to a trivial contest of lumpers versus splitters. In this case the winners are probably beetles (is anyone surprised?).
Agrilus derasofaciatus (CC Encyclopedia of Life)


beetlesinthebush said...

I didn't read through all 71 (and counting) comments on Ed Yong's thread, but I agree in principle with your summary here. I will say, however, that I have a little bit of difficulty with the idea that a superdiverse genus like Agrilus is a completely subjective result of the battle between 'lumpers' and 'splitters'. Many people have tried to split it up but failed because, unlike the genus itself, which nobody doubts is monophyletic, it has proven impossible to define monophyletic units within the genus. It's pretty easy to recognize something an undescribed Agrilus as belonging to that genus, and of the many, many species awaiting description there is no other genus to which they can be assigned when described. They are placed there not to show allegiance to lumpers and shun splitters, but because that is the smallest unambiguously monophyletic unit that can be discerned.

Obviously I don't mean to suggest that there are no monophyletic subunits within the genus (as with any genus), and it is the degree of difference between those units that determines (subjectively rather than arbitrarily) whether they rise to the rank of genus or not. How much difference is required to consititute a genus tends not to bounce around from one genus to another, especially within a family since specialists tend to focus on family-level units and reach consensus with other specialists in the same family.

I guess this is a long-winded way of saying that I think there actually is an element of reality to the "holow curve."


Christopher Taylor said...

It is unambiguous that a tribe belongs to a family, that a subgenus belongs to a genus.

Wrong. Very wrong, and for the same reasons that you talked about in this very post. There is absolutely no reason (and it has happened regularly) that someone could split a 'family' to the point that it is more exclusive than another author's 'tribe'. 'Family Hominidae' as used to include Homo sapiens and its direct fossil relatives only is a smaller group than 'subfamily Homininae' used to include both humans and the African great apes. Among tanaids, the 'families' Anarthruridae and Agathotanaididae of Larsen & Wilson (2002) included species that had all previously been placed together in the 'subfamily' Anarthrurinae. Perhaps most extreme of all, the annelid 'family' Siboglinidae of recent authors includes taxa that some had previously treated as separate 'phyla' Pogonophora and Vestimentifera.

So no, you cannot simply assume that one author's 'subfamily' is a subgroup of another author's 'family'. You can't even really assume this when comparing publications by the same author, because authorities have been known to change their preferred taxonomies. The only place where you can generally take this line for granted is within a single publication. Where it is usually redundant because you will already have the explicit context of the publication itself to provide the relative status of the taxa.

ZL "Kai" Burington said...

Thanks for the correction, Christopher. I agree that the raising and lowering of taxa and the disagreements between authors causes a great deal of these orphans, /despite/ the rules of coordination and priority in the ICZN. Perhaps I am too comfortable in insects, where the higher classification doesn't jump around as much these days as it does in other groups. At least, by a single author (and in many cases between authors), the ranking system can be used to communicate hierarchy better than a non-ranked system can. Thanks again for your criticism.

ZL "Kai" Burington said...


That's a good point. When I was writing this post, I was referring to the hollow curve, but so much of Ferris' explanation seemed to be anecdotal evidence. I've experienced what you've described with several large caddisfly genera like Hydropsyche and Cheumatopsyche. Their characters show them to be obviously monophyletic, yet there aren't any obvious characters to separate out the subgenera. Indeed, when some people have tried to separate Hydropsyche into several genera, more evidence has come back to suggest that wasn't a good idea after all.

What it seems to me is that, in many cases, the authors of old described fewer, larger genera, and that these genera are being split down into smaller groups. Some of the large genera are left, nearly unsplitable, and so we get a hollow curve.

But it could very well be that large genera are hints of recent diversifications; if they are hard to separate out, maybe they diverged only recently. The oldest Cheumatopsyche fossils are 25 million years, and there are more than 300 species in that genus, new ones described every year.

And I don't mean to say that there are these guilds of lumpers and splitters, but that there are distinct philosophies on larger versus smaller groupings that have been at odds, and that this often shapes the taxonomy of various groups. And perhaps I'm wrong about that as well.

Thanks for your comment, and your interest.

BioBob said...

I would not be surprised at all if "the hollow curve" does represent the results of evolutionary process.

As you hint, and others discuss, evolutionary theory is rife with the concepts of adaptive radiation, punctuated equilibrium, concurrent with gradual 'simplification' by extinction.