Theories of speciation

Continuing on from my last post, let's consider the modes of speciation that are called into account for the existence of species.

Here is a list taken from Sergey Gavrilets, which I put in my most recent paper in Biology and Philosophy (2007).

Vicariant – divergent selection and stochastic factors like drift after division of a population by extrinsic factors such as geographical changes;

Peripatric – a small subpopulation, mostly isolated, at the extreme of the parent range. The idea is that it will have both a non-standard sampling of alleles, and also be subjected to divergent selection pressures in extreme environments (for that species);

Centrifugal – central populations that carry a sample of many alleles that become isolated through, say, “island” formation (such as the mountain “islands” in the Amazon);

Punctuated equilibrium – the appearance of relatively rapid speciation and subsequent stasis as the population reaches equilibria of alleles. In my opinion, this is inappropriately included here, for it is a “pattern” rather than a “process” (or “event”) of speciation, and as such can be caused by any of the other scenarios/modes;

Chromosomal speciation – the rearrangement of chromosomes, either by duplication or insertion, fission, fusion or inversion;

Hybridization – the fusion of two genetic lineages, usually from distinct species, including allopolyploidy. In allopolyploidy the genetic complement of two species is paired up by a loss of secondary division, giving a symmetrical set of chromosomes;

Reinforcement – once hybrids are of lowered fitness for whatever reason, selection will tend to reinforce separation of the gene pools (for example, a hybrid rock and grass dwelling lizard might be less able to survive in either environment as well as the “pure” lines);

Competitive – this is Darwin’s scenario. Members of a species that are in strong selection for a limited resource may result in specialized forms that are thus in less competition with the ancestral forms that make use of some other resource;

Clinal/ecotonal – Gavrilets calls it “speciation along environmental gradients”, where limited migration and selection leads to aggregation of forms at the terminal ends of the cline;

Host shift – this is the case of the Rhagoletis fruit flies mentioned above, that Stuart Berlocher (1999; 2000; 2002) and colleagues have studied. Host fidelity replaces geographic isolation;

Runaway sexual selection – this is secondary selection by mate choice of polygenic traits (Lande 1981).

In that paper, I locate each of these modes as a unique coordinate or region in a three-dimensional space, the axes of which are:


Gene flow – the rate of migration between populations, or the amount of genetic material exchanged in a mating event, from none to 50%;

Selection – the degree to which selection is endogenous to the organisms (such as climatic selection, say on the Souay sheep or the finches of the Galápagos Islands), or is intrinsic to the organisms (such as mate choice or immunological compatibility); and

Stochasticity – which is basically whether selection is directional or stabilising, or whether the rate of change of the genetic constitution of the population is due to genetic drift and other stochastic "forces".

Now, suppose we have a species of flowering plant, say. It buds off a new species due to climatic and pollinator adaptation, with no exchange of genes between populations. In binary terms, its coordinate would be 0,1,1. That is the explanation of that species in terms of how its speciation occurred. If it is in sympatry (shares a geography) with its parental species, it must have adaptations that prevent competitive exclusion, and so remaining a species, rather than merging back into the parental metapopulation, is due to selection for those niche adaptations.

This, it need not be said, will not be true of species that occupy other regions of "speciation space". The plant has its own explanation, and its own phenomenal reality. We know it is a different species simply because introgression and hybridisation are inhibited, and competitive exclusion does not occur because parent and child species occupy different points of the fitness landscape.

But is this sort of species (which may have formed in isolation from the parental species) required by any theory of speciation or evolution, or biology in general? [And generalising, is any type of species so required?] The answer is, no. Simply put, if it merged back into the parental species, it would not be identified as a species itself, but a variety (and maybe a fairly ephemeral one at that).

What we have is a post hoc phenomenon. If it is distinguished by its genetic and ecological properties, however acquired, then it is a species. If it isn't, then the issue doesn't arise. This is a bit like weak anthropomorphism - any universe that lacks humans isn't going to be discussed by them, but you cannot make the inference that there is something special about human occupied universes simply because we are here talking about them. Similarly species - if it is separate from other species (in its own unique and contingent manner) then it is a phenomenon that demands explanation, but if it doesn't, the theories of population genetics, developmental biology, ecological interactions and the like all continue on apace.

Post hoc explanation is not an illicit move in science; it can't be, because one of the major roles of a scientific theory is to explain what is observed. But that doesn't imply that the phenomena are theoretically significant. Just that they interest us enough for them to call for an explanation.

There is a school of thought in the philosophy of science that treats ontology, the set of objects that one thinks exists, as basically the bounded variables of the theory of that domain. In this case, species would be theoretical objects if they were such variables of a theory. But they aren't. So we need to establish what sort of ontology they, and other phenomenal non-theoretic objects, may have.

Consider planetary orbits. They were observed and debated for a very long time before Newton proposed that there was a general physics that accounted for them (and made predictions about them). But in so doing, Newton demoted these orbits from theoretically important objects to special cases of larger and more universal physics. "Planetary orbit" is a special kind of astrophysical dynamics, one which aperiodic comets, entire star systems, and even entire galaxies all obey. Even if no orbits actually existed (and we can perhaps envisage this in some universe) under this physics, the movements of objects would be still covered by Newtonian dynamics.

Likewise species. They obey, and when they occur are post hoc explained by, the biology of populations, interbreeding, selection, drift, and so on. But they are not themselves theoretical objects, any more than planetary orbits are in physics. [I know - the "nebular hypothesis" of planetary formation requires that bodies in the disk of accretion form more massive objects by gravatational attraction, but if there were no such disks, they needn't.] Species occur, and are explicable in a multiplicity of ways, but they do not follow formally from any theory of biology.

My general characterisation of species was that they are the nexus of coalescence of genes, haplotypes, parent-child lineages and so on. In abstract terms, species are these coalescences that are distinct from other such coalescences, something I have called the synapomorphic conception of species (in my 2003), and each and every one has a general set of properties and modes of speciation, and a unique set of these that only they have (the synapomorphies, or shared characters, which are causally active in maintaining separation).

So that's why I say species aren't theoretical objects, but are phenomenal ones. As a take-home exercise to the reader, try to imagine under which conditions organisms like ours wouldn't form species at all...


Wilkins, John S. 2003. How to be a chaste species pluralist-realist: The origins of species modes and the Synapomorphic Species Concept. Biology and Philosophy 18:621-638.

———. 2007. The dimensions, modes and definitions of species and speciation. Biology and Philosophy 22 (2):247-266.

More like this

It occurred to me as I was chatting to a friend (KiwiInOz) that I actually have a philosophical method. It comes as a surprise. I thought I just meandered along, but as I yet again did a semantic space diagram to outline the issues (in this case in biodiversity measures that my friend and I are…
This is a repost of a piece I wrote for The Panda's Thumb in March 2004. I add it here to put it in the Basics series. It is, wrote the Roman poet Horace, fit and proper to die for one's homeland. The word he used for homeland was "patria" (dulce et decorum est pro patria mori), and the word has…
If somebody asked me to write a short essay giving an overview of my favourite topic, the nature of species, I doubt that I could. I can write a long essay on it (in fact, several) but it would be excruciatingly hard to write a short one. For that, we need a real writer. Carl Zimmer is the guy. He…
Here is a working list of species concepts presently in play. I quote "Concepts" above because, for philosophical reasons, I think there is only one concept - "species", and all the rest are conceptions, or definitions, of that concept. I have christened this the Synapormorphic Concept of Species…

The physicist Schrodinger in an essay about evolution suggested a possible mechanism for speciation that might come into play in certain circumstances. It might be the same as "clinal/ecotonal", but I'm not familiar with that term.

Schrodinger's idea is that under certain circumstances, random expressed variations can lead to segregation. For example, consider a species living at a boundary between two environment, say where the mountains meet the lowlands. Random variations among individuals may make some individuals better suited to live in the mountains, while other individuals may be better suited to live in the lowlands. Individuals will tend to spend more time in the areas where they are better suited.

Over time, the distribution of genes in the population will tend to be correlated with geography: there will be a higher percentage of mountain-favoring genes in the mountains than in the lowlands and a higher percentage of lowlands-favoring genes in the lowlands. The occasional lowlands-favoring gene that appear in the mountains will tend to get weeded out, either by those individuals dying (normal natural selection) or by those individuals leaving the mountains to seek their fortunes in the lowlands. Eventually, the filtering process will produce a great enough genetic variation between mountain-dwellers and lowland-dwellers to produce two different species.

This is the "reinforcing selection" model of speciation. Hybrids are neither as good as one form or the other due to their lowered fitness. But this model leads to hybrid zones where the ecotype is not strictly one or the other, or shifts over time back and forth.

Thanks for this posting; it gives one something to chew on. I confess to coming away from the previous one without a really clear conception of what question it was answering -- in spite of the question having been explicit. But at this point I have a much stronger idea of what a species is conceptually.

Thank you for this posting. Trying to conceptualize species as theoretical objects has caused me great difficulty. I will consider a phenomenalist framework going forward.

I have not read Schrodinger (sorry to say). The post above is almost exactly the argument I presented for sypmpatric speciation in a Population Genetics term paper in 1967. I had a fish population and a mosaic environment in a lake. If I can think of the idea also, this suggests it has a certain obviousness.

I don't think any proposed speciation process is surefire.
Because I have been mostly concerned with recognizing species, I have come to think more about speciation events rather than speciation processes. I think speciation is generally allopatric in the species I worked on. And I have come to the conclusion that if I were to encounter a speciation event in progress out in the field, I would be blissfully unaware of it.

If a speciation event involves a small isolated population, It seems to me that selection may be less important than drift, random events, etc, so long as the population remains small. Selection is a statistical happening, and statistics do not predict well with small populations. Perhaps it is better to be lucky than good when you are not numerous.

By Jim Thomerson (not verified) on 05 Sep 2007 #permalink

Jim, yes, this is in many ways a formal statement of What Everybody Actually Knows (as opposed to what they say). I have bene particularly influenced by the work of Murray Littlejohn, whose work was at its height in the 1960s.

My point is that this is not an either-or situation. It need not be Allopatry-Only or Sympatry-Only, but we should recognise that these are not polar opposites but rather alternative cases in a given theoretical space. And it is interesting that some regions just aren't occupied in my rough and ready classification (I'd love someone to put figures to this).

And of course I argue that species are effects, not units, of evolution.