Thursday, March 19, 2015

What is the age of a species?

We had journal club today, and I realised that something that I had always considered quite obvious and logical is not as obvious and logical to everybody else. The question here is: how old is a species?

If you are like me, you will visualise the phylogenetic tree of the species and its closest relatives, point to the moment where it diverged from its sister group, and say that that is its age. See the following diagram:

We know today of the existence of species A, C and D, and if we reconstructed the phylogeny I would say that the divergence time of A on the one side and the ancestor of C and D on the other side, marked with the black line and the number 1, is the age of species A.

However, there is a little snag: What if there were side branches on the tree of life that we do not know about because they are now extinct? In the hypothetical case above, there existed a species B for some time but then it died out. So really the age of species A should be the divergence time from B, indicated with the purple line and the number 2. But if we only know about A, C and D, we can at least say that A is at the most as old as indicated by the black line.

So far so good. What I learned today is that there are apparently people who believe that the age of a species is where the gene copies it is carrying or the family lineages inside it coalesce back in time. This is indicated by the red line and the number 3: the yellow gene tree shows that all extant gene copies (of that gene) in species A are derived from an ancestral copy that existed at time 3.

This time would often be much closer to the present than times 2 and 3. In the case of us humans, we are usually inferred to have diverged from the chimpanzee lineage a few million years ago, but the last female and male common ancestors of all of humanity - the individuals from whom we have all inherited our mitochondria and Y chromosomes, respectively - have been inferred to have existed a few ten to hundred thousand years ago. And indeed some people seem to assume that that is then the age of our species. (See also my earlier dissection of the Doomsday Argument.)

When I was reminded of this concept today I couldn't believe that this would make sense to any biologist. Mike Crisp pointed out the first major problem: this coalescence point constantly moves through time as gene lineages and family lineages proliferate and die out again. As one can see even in my little diagram above, there were several earlier times at which the same would have been true. So how can the age of the same species constantly be corrected downwards as it gets older?

Worse, if speciation events are spaced closely in time and effective population sizes are large, it is well possible that coalescence time might actually be in an ancestor of species A instead of itself; this is known as incomplete lineage sorting or ancestral polymorphism. In other words, species A would then be older than its existence as an independent lineage. Surely that is immediately recognisable as nonsensical.

However, Mike also pointed out that (as I would add: in cases where it is after the lineage divergence time that we can infer), this coalescence point provides a lower limit on the age of a species. So it must then at least be as old as that, and it is at most as old as the divergence time from its living relatives.


  1. I think we should try and drop entirely the concept of a taxon's age. The whole terminology implies that taxa, species or any other, are born at some point. I don't think this is a constructive way to think of taxa. To be sure lineage splitting can lead to the formation of multiple lineages that might one day come to be recognized as distinct taxa. But the process whereby those lineages accumulated differences to the point of being recognized as distinct did not all happen at the cladogenesis event. The process of "speciation" is drawn-out in time and it seems a mistake to pretend that it corresponds to the geographic isolation that precipitated lineage splitting.

    As for the coalescent criterion, that seems silly. The expected coalescence time for a set of alleles in a population is dependent only on the population size. Does it make sense that the age of a species is determined only by its effective population size? I think not.

    In general, the question of when species begin and end (when not by extinction) seems entirely subjective and better avoided. I suggest we view species (and other taxa) as synchronic entities that do not, in any helpful sense, have ages.

  2. I also find that thinking of species asynchronically quickly leads to contradictions and absurdities, but that is more because people tend to see species as static bundles of morphological characters.

    Therefore I am less concerned about talking about the age of lineages and taxa in a more general sense. Yes, speciation is a drawn out process. But let us say it takes 300,000 years - if somebody is discussing the history of a group across 30 million years, on that scale it still seems like close enough to instantaneous to not matter very much.

    At any rate it doesn't matter for many of us who are mostly interested in the present time-slice...

  3. Agreed on this: "it doesn't matter for many of us who are mostly interested in the present time-slice"

    But I am not sure I buy the argument that because the "window of speciation" is short relative to the inferred age of a species, that species age is a valid concept. The problem for me is all kinds of evolution happens even after the lineages get to the point where, if we had been there, we would first have recognized two species. What makes the events before then "speciation" and those afterwards just "evolution," when they are really all the same? Speciation is not really a process, but a post hoc explanation of a pattern. I admit I am being a bit picky here.

    I also suspect that taxon age is sometimes used a sneaky way to capture a vague idea of "primitiveness" - something which causes no end of problems. I just reviewed a paper that made a big deal about studying "the oldest mammalian family" (or was it order? animals schmanimals!). Does that make you as uncomfortable as it makes me?

  4. Well, we may disagree here, but I see a difference between speciation - a lineage split - and for example anagenetic evolution, where a lineage stays one lineage while changing over time. What is silly is the attempt to delimit anagenetic species along that lineage.

    What makes me uncomfortable about 'the oldest mammalian family' is more the insinuation that taxa of the same Linnaean rank can be compared meaningfully. I guess I'd say 'earliest diverging lineage of the mammals'.

  5. In case you are wondering how I come to be commenting: I am spending a few months in Melbourne, hosted by Dan Murphy. He sent me a link to your blog, and I could not resist commenting. Hope you don’t mind!

    We still don't agree - but it's still fun. Cheers.