Monday, June 24, 2013

Asynchronous species concepts: internodal and composite

As indicated before in this series about species concepts, most of them apply only to contemporary organisms or more generally to those existing in the same time slice. That becomes clear quite quickly if we try to apply them in an asynchronous fashion, i.e. through time.

The Biological Species Concept (BSC), for example, sees species as breeding groups. Because we humans do not interbreed with desert oaks, and indeed would find it hard to do so, we are clearly separate species. If we let our gaze drift over all of evolutionary history, however, it becomes clear that there must have been an unbroken chain of individuals connecting me, the desert oak I photographed on the Great Central Road in 2010, and some common ancestor the two of us had sometime a few hundred million years ago. In other words, seen through deep time all of life on earth is a breeding community, and thus the BSC would fail to cleave the diversity of life into species.

As another example, the Genotypic Cluster Species Concept (GCSC) identifies species as clusters of individuals in some morphological or genetic analysis that have no or few intermediates with other such clusters in the same analysis. Again, because all organisms on the planet appear to be parts of the same great tree of life, and because evolutionary change happens gradually through the change of allele frequencies in populations, there just is no place along the branches of the tree of life where there are "no or few intermediates". Just as the BSC, the GCSC would not work.

Of course, if you see species as breeding groups, you might immediately conclude that applying species concepts through time must be absurd. Still, there are those who have tried to formulate ideas on how to make the word "species" work in this context. There aren't many, and one of them I have already discussed before is Willi Hennig's Internodal Species Concept, so I will keep it short on that one. A few more words are then needed for the newer alternative.

Internodon Species Concept (ISC)

Perhaps because he was a zoologist, Hennig (1966) approached the species issue coming from the BSC, and it can be assumed that his interest was then to make it compatible with cladistic thinking. As explained here in more detail, he pointed out that organisms have different relationships with each other: predominantly tokogenetic within biological species, predominantly phylogenetic (tree-like) from biospecies to biospecies. In other words, all sexually reproducing organisms on the planet can be envisioned as having a position in a tokogeny, a network of ancestors and offspring, and then when you "zoom out" you see that this network is merely a branch of the big tree of life, isolated from other such networks, or biospecies, through the absence or near-absence of interbreeding.

These branches do, however, connect with each other when you go back in time along the tree of life, and the moment where they connect or, seen from the past towards the future, diverge, are the speciation events. It is clear that under the BSC, that is the moment from when on two diverging lineages are separate species, but where is the border between both of them and the ancestral species? Well, here the ISC simply says: at the same spot. Every species starts with a speciation event and ends with a speciation event, so that every species is an internodon on the species tree.

The Internodon Species Concept: Although the red flower lineage remains morphologically unchanged, it is counted as a new species after every speciation event.

As discussed in my previous post on the matter, this has a very counter-intuitive consequence. If a species splits in two reproductively isolated populations, they are both new species distinct from the ancestral one even if one of them is morphologically indistinguishable from it. That probably happens quite often. One can easily imagine a single seed of a widespread species on a continent being carried onto a remote oceanic island. Released from competition but faced with a new environment, and carrying only a minuscule part of the great genetic diversity of its ancestral population, the new population on the island may evolve quite rapidly to become a distinct species. But, one could argue, the ancestral species on the mainland has not changed at all.

But a cladist would perhaps reply: yes, it has. It has diversified into the species on the island and the remainder on the mainland. The ancestral species does not live on in the morphologically identical mainland species but instead in the mainland species and the island species. That is the logic behind the ISC. But as indicated above, it leaves many people dissatisfied. That is why the idea of composite species was invented.

Composite Species Concept (CSC)

To put it into the simplest possible terms, the Composite Species Concept of Kornet & McAllister (2005) is essentially like the ISC except that it considers internodons that are morphologically indistinguishable to belong to the same species. In the case of the above mainland species, the ancestral species would then be considered to be the same species while only the island species is a new one.

The Composite Species Concept: Here the red flower lineage is the same along the entire branch because we do not see any obvious morphological differences between the ancestral red flower at the base of the tree and the contemporary species that is sister to the black flower one.

We could now envision the same to happen repeatedly, with three different islands, and then we would have a species phylogeny with one long branch of four internodons all belonging to the same mainland species and three side branches arising from it belonging to the individual island species 1 to 3. That might give us pause, because it means that this mainland species is "paraphyletic" to the three island species. Is that allowed?

There are certainly different opinions on this, but a reasonable case can be made that such a view is compatible with cladistic practice. As mentioned on this blog before, the point of Phylogenetic Systematics is to group species into monophyletic supraspecific taxa, not to make species monophyletic. In fact under its standard definition, the term monophyletic can simply not be made to even apply to sexually reproducing species in the first place.

If you are familiar with cladistic methodology, you will notice another thing about composite species that appear "paraphyletic" to several other species. Those other species all differ in at least one character from their ancestral composite species (i.e. they each have at least one autapomorphy), otherwise they would not be considered a separate species under the CSC, and by necessity all the internodons of the ancestral composite species do not differ morphologically from each other, otherwise they would be considered different species under the CSC.

In other words, there are no synapomorphies we could identify to group one internodon of the composite species with one of the other species. And that means that in actual practice, we have no way to infer even only the sequence in which the three island species diverged from it - we only pretended we had it when we considered our mental model of what might happen in evolution. In an actual cladogram resulting from parsimony analysis, we would really only see a polytomy with the ancestral composite species sitting directly on the internal node and the three other species sitting on branches of varying length depending on the number of autapomorphies they have acquired.

How a cladogram of the above relationships would look like. Because there are no synapomorphies along the red branch, we cannot infer the sequence in which the green, blue and black species diverged.

It is then also in this sense that a composite species cannot really be considered paraphyletic, because the character information to decide that isn't there; at best, it is metaphyletic. If there were information to group part of it with one of the other species, we could just as well use that same information to break the composite species up into two small, homogeneous ones, and indeed the CSC seems to demand we should do so, which would again take care of the "paraphyly".

These considerations lead us to the problem with the CSC, however. Another way of describing the concept is as a typological or morphospecies concept applied asynchronously, and consequently it comes with the same weaknesses. The most important of these is that it is entirely unclear how big a character difference is enough to consider two organisms to belong to different species. Slightly different leaf shape? No hairs on the lower leaf side? Different flower colour? If not, why not?

All in all, both asynchronous species concepts considered here are clearly not very practical. Their main use is as a theoretical model or conceptualization when discussing evolution and speciation.

Updated on 25 June 2013 to correct several typos, a missing link and similar minor oversights.


Hennig W, 1966. Phylogenetic Systematics. Urbana: University of Illinois Press.
Kornet DJ, McAllister JW, 2005. The Composite Species Concept: A rigorous basis for cladistic practice, pp. 95-127 in Teydon & Hemerik, Current Themes in Theoretical Biology, Springer.


  1. There is, in addition to cladistic speciation events, the concept of patristic speciation. This is when a species A evolves over time into species B without a vicariance event. I think paleontologists have not abandoned the concept. One can still use the biological species concept to separate A and B. All that is needed is a time machine and a big grant.

    There is a situation somewhat like your last illustration in stickleback fishes. There is a widespread armored marine species which has budded off, at different times, very similar unarmored freshwater species.

    The biological species concept produces logically testable hypotheses of relationship. Often the tests are not practical to perform. Other times they give inconclusive results. However, if one finds that A and B cannot successfully hybridize, then that is good evidence they are separate species.

    I think a basic aspect of the biological species concept is the postulate that individuals who engage in interspecies mating attempts have reduced fitness. I haven't seen much on this in the literature.

  2. You raise several interesting issues but I think I cannot entirely agree. Patristic speciation (which I know as "anagenesis") still has the same problem for the BSC as all others, that all of life forms one breeding group if seen through time, there are simply no breaks. Yes, A and B will be different biospecies, but there will always be a C that is precisely in the middle between them, and a D and E that are precisely in the middle between A and C and C and B, respectively, until there is only a big mush.

    Note that maybe I am a bit biased here because I also consider the same observation to be a major argument against the practicability of "evolutionary" classification.

    It seems as if people have very different ideas about what the BSC entails, from "any interbreeding at all makes two populations the same species" to "a bit of interbreeding is allowed as long as it does not have a significant impact on the character of the two populations". Reduced hybrid fitness would be a demand somewhere in the middle of those two positions.

  3. I'm not talking about reduced hybrid fitness, although this is an aspect of it, but rather reduced fitness of the individuals who participate in interspecific matings. If you can get access to Volpe's papers on toad hybrids, for example;

    I no longer have access, but I took population genetics from Volpe around 1964, and this was discussed. As I recall, the Audbon Park studies involved mark and recapture of individual toads. None of the females who mated with the sterile male hybrids were ever recaptured. I think this included females who produced hybrids were also not recovered. Unfortunately the study could not be followed up because of construction in the park.

    The "any interbreeding at all makes two populations the same species" is, I think, a naeve and uninformed view. The a little hybridization here and there is OK view has to be looked at in light of the situation. For example the Delco and Hubbs studies of a hybrid swarm of red x blacktail shiners that formed during disruption of the stream by highway construction. The swarm disappeared almost immediatly when construction was complete and the stream returned to normal. Observations such as this give one pause in interpeting successful production of fertile hybrids in the lab.

    I wonder if your problem with patristic speciation is a view of speciation as a rapid and definitive process? I think one cannot know if one is observing allopatric speciation in the field. Suppose I see two somewhat differentated allopatric populations, clearly closely related in terms of common ancestry, am I observing allopatric speciation. I think if there are two species, should they come into contact, they will retain their specific identies. Now what may happen to the two populations in question? One or both may go extinct. If they persist, they may never cone into contact, but if they do, they may maintain their identity, or they may merge is such a way that no evidence of their separation is preserved. So, I think that one can only infer an allopatric speciation event, after having observed two related species. I've thought about this a good bit, and discussed it with colleagues, and my vew seems reasonable.

    In the allopatric example, if I do hybrid studies and can reproduce both populations under the same conditions, but cannot get hybrids, or fertile hybrids at least, I would conclude that speciation had occurred. If I get fertile hybrids then I have to think about it some more.

    Please excuse spelling errors, if any, spell check not working.

  4. Ah, I misunderstood your point about interbreeding individuals, sorry.

    Yes, the very strictest interpretation of the BSC appears to be a straw man, but there is no lack of similar misunderstandings among colleagues in biology. The way I was taught the BSC I also assumed for a long time that (a) it demanded zero gene flow and (b) therefore it is obviously stupid, at least when applied to botany.

    The thorny issue of allopatric vs sympatric speciation is not really what my remark was about. It is simply that most species concepts demand there to be some kind of "gap", be in morphological variability or between breeding groups, and along the branches of the tree of life there simply isn't any gap, to the best of our current understanding.