Saturday, March 30, 2013

Biological species and related concepts

This post is part of a series on species. Read the previous parts here and here.

The Biological Species Concept (BSC) is perhaps the most famous of them all. In science, it is one of the things evolutionary biologists Theodosius Dobzhansky and (in particular) Ernst Mayr are best known for. It is basically what most non-scientists would also come up with when asked what a species is, unless they have never given the topic a thought before: species are breeding groups, i.e. groups of individuals that form a reproductive community.

Many colleagues will tell you that the BSC is very popular with zoologists but not so much with botanists, supposedly because it works well with animals but not so much with plants. And indeed I know many well accepted plant species that can interbreed, at least potentially. But that is of course circular reasoning: should they then really be separate species? Maybe botanists are simply splitting species too much.

On the other hand, the idea that some interbreeding or even potential interbreeding already makes two populations conspecific may be a caricature of the BSC as promoted, for example, by Coyne & Orr's book Speciation (which, by the way, is a fantastic resource for those interested in species); really the BSC is often understood to allow for some rare gene flow between biological species.

At a minimum there appears to be a spectrum from stricter to looser interpretations, which is one of the reasons why it makes sense to discuss the BSC in the strict sense together with related concepts, even those using somewhat indirect criteria. In order of decreasing strictness, one could argue that two individuals are members of different species...
  1. only if they cannot be crossed experimentally and/or the hybrid resulting from such a cross is completely sterile
  2. if they could be crossed experimentally but they would usually not voluntarily interbreed in nature
  3. if the hybrid resulting from crossing them is pretty infertile although it could potentially, rarely, back-cross
  4. if they occasionally exchange genes in nature but only so rarely that there is very little impact on the gene pool and the evolutionary tendencies of the two species
Surely many widely recognized species both of animals and plants would have to be lumped together under the first, strictest view (even mules can exceptionally get pregnant, so horses and donkeys would be conspecific under this definition), and it seems plainly absurd to base the decision on species boundaries on the success of crossing experiments in captivity or by artificial pollination if two putative species never actually exchange genes in nature.

However, the other views also have their problems. The second raises the question of what mechanisms of isolation are permitted. If we are talking about animals keeping their two putative species pure by selecting only conspecific partners or about plants with a system to reject alien pollen then we have the Recognition Species Concept, okay; but what about two groups of organisms that could easily interbreed but never do so because they occur on different land masses? And what if we can guess that they will be reunited as soon as the next ice age lowers sea levels enough to reunite the two land masses? In that case, we should probably argue that the ability to potentially exchange genes would be a good reason to lump them into one species.

Finally, #3 and #4 allow some occasional gene flow, which means that here the BSC grades into the phenetic and cluster concepts to be discussed in a future post. We would then in effect not ask whether the two species interbreed but instead whether their genetic composition is different enough or if there are few intermediates between them, and those are questions answered with cluster analyses or phenetics.

This consideration also introduces another aspect that makes many colleagues uncomfortable about the BSC, or even leads them to conclude that species do not exist or are arbitrary constructs: can we provide a hard cut-off for how much interbreeding is allowed? It is obvious that unless we are dealing with allopatric speciation after long distance dispersal or perhaps speciation through polyploidization, interbreeding does not abruptly stop from one generation to the other. Instead, it gradually tapers off. This means that the time-slice we inhabit will sometimes cut through well-defined lineages on the phylogenetic tree, which we will then call unproblematic or "good" species, but it will also cut directly through some ongoing speciation events, and then those of us who want a clear yes or no are going to be disappointed.

Indeed that is the core argument advanced by an American colleague who promotes the idea that species do not have a special reality. Interbreeding decreases gradually over time, thus the species level is arbitrary, or so he says. When I discuss this issue with students, I present something like the image below, and ask them whether they think that red and yellow are arbitrary or useless concepts because there is orange between them. Of course not. Another example would be age. Because we do not turn from toddler to adult within one second, adulthood must be arbitrary, and so obviously we can let a six year old steer a car or handle a loaded firearm, right? No, I assume we can all agree that the concept of adulthood makes sense even if there is a grey, or, to stay with the image, orange, area between it and childhood. Likewise, "different-species-ness" defined as the inability to interbreed is non-arbitrary even if it come in degrees. Where nature is a bit fuzzy, biologists should accept that fuzziness.

Perhaps the oddest cases from the perspective of degrees of gene flow are, by the way, ring species. Those are groups of populations that show a pattern again not dissimilar to the red-orange-yellow image here but over a geographic instead of a time dimension: all populations along the geographic range can happily interbreed with their neighbors, but if you bring the two ends of the spectrum together, they can't. That means that according to the BSC, red is the same species as orange, and orange is the same species as yellow, but red is not the same species as yellow. Sometimes nature can be really fuzzy, but there is probably no species concept that can deal with ring species any better than the BSC.

However, the BSC has a few other drawbacks or limitations that I have not even mentioned yet. The most obvious one is that it is no help with asexually reproducing groups. Of course, the simplest solution there is to argue that the concept of species simply does not apply to them. To a certain extent I would agree; certainly the "agamospecies" of Taraxacum, Hieracium or Rubus are in no way comparable to real, sexually reproducing species, and in larger asexual groups of organisms it would be trivial to have a phylogenetic classification all the way down to individual organisms. But not only would that appear unsatisfactory to many people who want to group everything into species (not that that is a real argument), there are also some really problematic cases such as some species of daisies that have otherwise indistinguishable sexual and asexual populations.

Another limitation that may not occur to everybody is that the BSC is completely useless for any time-slice except the present. We cannot cross fossils to see if they belong to different biological species. And even in the present crossing experiments are rarely practical. Can you blame the authors of the Flora of Central Africa for not conducting crossing experiments between all the putative species of the various groups of plants they have to treat in the flora? Of course not. Yet they still have to make a judgement call on species boundaries. In that sense the BSC may be nice in theory but as a guideline for empirical research it is comparatively unhelpful.

Summary (Biological species and related reproductive community concepts)

Grouping criterion

The common theme of this family of concepts is to view species as reproductive communities. Depending on the specific concept, individuals belong to the same species if they can produce fertile offspring, if there is evidence that they would do so under natural conditions, if they would recognize each other as potential mates, or similar. Different concepts in this family are of varying strictness with regard to the level of gene flow that is still permitted between two populations before they are considered conspecific.

Ranking criterion

Same as grouping criterion.

What is it good for?

Works only in one time-slice (from an asynchronous perspective, all of life on earth is a reproductive community, that is the logical consequence of common descent). An advantage of the BSC and most of its relatives is that the concepts are generally empirical. However, the real test would in many cases be crossing experiments, and those cannot realistically be conducted in all cases where decisions on species limits have to be made, and especially not in the case of fossils. Biological species are therefore perhaps best seen as a concept for theorizing or as an ideal that one tries to approximate by using other concepts in practice. For example, under the assumption that significant interbreeding between two populations should lead to the observation of numerous genetic or morphological intermediates, one can use a phenetic cluster concept and expect that it will arrive at the same answers as the experimentally more demanding BSC would.

Are species real?

Yes, because reproductive communities are obviously real - a swallow can interbreed with other swallows but not with a baleen whale. But there are two caveats: there are fuzzy areas such as ring species or other ongoing speciation processes where it is difficult to decide if there is still one species or already two, and asexually reproducing organisms do not belong to species under these concepts.

Are species a special rank?

Yes, under these concepts there is a significant difference between species and higher or lower levels of biological diversity. Again, this does not apply to asexually reproducing organisms.


  1. The late ichthyologist Luis Rivas coined two useful terms, more explicit than sympatric and allopatric. Syntopic means you catch them together in the same seine haul. Allotopic means you have to make two seine hauls to get them both. Syntopic means they encounter each other on a regular basis, allotopic means they live in the same area but don't usually come into contact.

    The BSC works best when populations are syntopic, and worse when they are allopatric. Hybridization experiments, properly done, can falsify the hypothesis that allopatirc populations are conspecific. But are not necessarily as convincing as supporting the hypothesis that allopatric populations are conspecific, because the real test is what will they do if they every come together in the wild.

    I got into ichthyology with an aquarist background, so I have done a far amount of hybridization experiments, and at one time, was very familiar with the extensive literature on fish hybridization. There is a lot of fascinating stuff in the fish literature.

    I described what is now Gantholebias hoignei, as a separate species from G. zonatus. An aquarist friend in Venezuela had been collecting them, and never found them together. G. hoignei lives in shaded pools, G. zonatus out in the savanna.

    I went down and collected some. I had no trouble breeding them in identical aquarium setups. (I raised several generations of both.) However, heterospecific pairs ignored each other. This supported my conclusion that I had two species in hand, rather than one. We later found them together in an area cleared for agriculture, with no sign of interbreeding. We found that their breeding behavior in the wild was quite different. Also their karyotypes are such that production of hybrids (by stripping eggs and sperm, perhaps) is extremely unlikely. DNA analysis does show them as sister species.

  2. That sounds pretty straightforward then - but again, it is not practical to do crossing experiments in all cases, not least because not all plants and animals can even be cultivated / held in captivity.

    As for syntopic versus sympatric, a couple of months ago we discussed a paper on putatively sympatric speciation in limpets in our journal club. One of the take home messages of that discussion was that sympatry or allopatry are very much a question of the spatial scale.

  3. I'm comfortable that sympatry means the ranges overlap on a range map, and allopatry means they do don't. Populations can be sympatric without ever encountering each other, so syntopic is a useful term.

    Most of my fish taxonomic and systematic work has been done on fishes which I could keep and raise in captivity. Just because I never quit being an aquarist,I suppose. When we were working in Belize, I bred several species of cichlids so I could preserve series of fry and juveniles for comparative material to identify fry and juveniles in our collections.

    Rivulus hartii males generally have colored anal fins, and females do not. However, over a fair sized geographic area, this is just reversed. Maybe we have isolating mechanisms or something. I got material from four disjunct populations, including the one with color reversal, and did heterolocality crosses, among all localities, with homolocality controls. I found that heterolocality crosses consistently produced significantly more eggs and viable fry than homolocality crosses. I was involved with other things and said, well it is all R. hartii, and went about my business. It would have been interesting to figure out what was going on.