Wednesday, December 07, 2005

Species is as species does... Part I - Variation vs. Speciation

People often ask me about species in the fossil record, and they are often posing some good questions. This will be the first of a three-part series of entries that I am going to do on this topic.

"How do we know that two fossils - often very similar - are different species?" The especially inquisitive person will often point out an array of different dog skulls and likens it to the similar array of hominin skulls. "How do we know that these alleged hominin species can't interbreed?" The short answer is that we don't know. The difficult part is explaining why it doesn't matter all that much.

Much of the problem seems to stem from an attachment to the Biological Species Concept (BSC) as canonical truth and that knowledge of breeding compatibility has everything to do with reconstructing phylogeny. The BSC states that a species is a population of organisms that can interbreed and produce fertile offspring. It's easy to see why it causes alarm, because it obviously can't be applied to fossils. It's also invalid when applied to asexual organisms (of which there are a great, great many).

In a somewhat concerned tone, this person might also state their belief that, were the different breeds of dogs to be extinct and their skulls discovered by palaeontologists today, we might well classify them as different species. Oh the horror! They would be so dismally wrong and never even know it! Wouldn't this cause dramatic consequences for our phylogenies of animals? The answer is that it's unlikely because of the way phylogenies are reconstructed (as opposed to the way people think they're constructed).

Palaeontologists employ a modern method of systematic biology called cladistics. Cladistics ignores characters unique to individual species because these don't tell us anything about relationships. For instance, if we're trying to figure out how a cow, a trout, and a human are related, the fact that a human has an opposable thumb tells us nothing about how it is related to the cow or the trout, as neither have it. Similarly, the fact that all three have a backbone is useless (except in setting the three apart from invertebrates, but we haven't stipulated that). Instead, cladistics focuses on character that are unique and shared between broader collections of species (i.e. four limbs with digits are uniqu and shared between humans and cows, to the exclusion of the trout).

But first, let's consider how the assumption of species might affect our methods of determining relationships in the first place. Let's imagine that dogs have been extinct for several million years and that palaeontologists have discoverd a bunch of different dog skulls. They've named each distinct type of skull a new species. Suppose that the result of their cladistic analysis looks like this:



Systematists working on extant (i.e. not extinct) animals also use the very same techniques that palaeontologists use (but will employ molecular data, not just morphological data wherever they can). Assuming that the phylogenies are the correct ones, they should get a result that looks like this:



The interrelationships should be identical. The differences are only in the very tips of the branches (breeds vs. species). The assumption of 'species' or 'breeds' is trivial, because the results are the same. The question, however, is would the same result be obtained were it not already known that dogs are interfertile?

The modern complement of dog breeds are (barring some mechanical difficulties) interfertile. They are, apparently, also interfertile with wolves. If we look at an array of dogs, we can see that a key character unites all of the domestic ones: the raised frontal (or 'forehead'):



This clearly sets them apart from foxes (top) and coyotes (bottom):





(by the way, all these great skull pictures are from this fantastic site)

What should be immediately obvious is that, despite the differences between the breeds of domestic dogs, their recent common ancestry with each other is general character that they share in common to the exclusion of other canids and carnivores. This is why modern systematics focuses mostly on shared similarities, rather than on separating them by differences. Because of this fact, domestic dogs (as we know them) would probably have been classified as a single genus, even on the basis of their morphology. Perhaps they would have been classified as distinct species, but as a total group they would have been recognized as distinct from all other canids. Historically speaking, this is identical to the true phylogeny. This is not to say that we'll always have the right answer. Rather, it is to say that by being sufficiently prudent about our classification, we should still be able to uncover the broad-scale patterns of evolution, at the very least.

For palaeontologists, it will never be possible to determine interbreeding relationships. But as long as we don't take the BSC as canon (and we shouldn't), then there is no problem. Naming a new species in palaeontology simply reflects being able to point to a distinctive morphology - preferably if we can show that populations of fossil animals can similarly be categorized. If we're concerned about whether or not two different fossil species (or perhaps even genera!) are interfertile, we are hopeless. However, we should still be able to uncover relative common ancestry. In the diagrams below, we can see that the trees (evolutionary relationships) are identical. However, the taxonomy is quite different.



So a species is as species does: for the purposes of palaeontology, it identifies a range of morphological variation. Since species/breeds are always nested the same way within genera (more inclusive groups of variation), it doesn't really matter if we call them species, breeds, or varieties in fossil animals. As long as they're within the same genus (and don't take the BSC to heart), then we're saying exactly the same thing about the course of evolution. Indeed, there is no question that all the different skulls of the different dog breeds would be classified in the same genus, along with wolves based on morphology.

I'll follow this up in Part II of this post, where I will discuss another nagging question: the classification of dogs and the Tazmanian 'wolf'. If we didn't know it from living specimens (though, thanks to man, it is now extinct), would it not have been classified as a dog?

2 comments:

RPM said...

Cladistics ignores characters unique to individual species because these don't tell us anything about relationships. For instance, if we're trying to figure out how a cow, a trout, and a human are related, the fact that a human has an opposable thumb tells us nothing about how it is related to the cow or the trout, as neither have it. Similarly, the fact that all three have a backbone is useless (except in setting the three apart from invertebrates, but we haven't stipulated that). Instead, cladistics focuses on character that are unique and shared between broader collections of species (i.e. four limbs with digits are uniqu and shared between humans and cows, to the exclusion of the trout).

Actually, cladistics or parsimony based approaches won't work on any of these examples because you need at least 4 taxa for cladistic analysis. With 3 taxa, you only have one unrooted tree (or 3 rooted trees) and all characters are either shared by all taxa or unique to a taxon.

Martin Brazeau said...

Hi RPM,

Thanks for your comment. You're very right about the nature of cladistic statements and taxon number. However, I wasn't really applying cladistic analysis to the taxa in that example. The example doesn't postulate a relationship statement, but aims instead to explain the concept of character distributions.

When I moved into dogs, the notion of broader taxon inclusion and outgroup comparison became implicit in statements such as:
"despite the differences between the breeds of domestic dogs, their recent common ancestry with each other is [implied by a] character that they share in common to the exclusion of other canids and carnivores."
(Upon looking back at that post, it appears that I didn't proof very carefully and the part in square brackets got chopped out -- sorry!)

In short, the essays were never intended as a cladistics tutorial, but instead set the reader up to look at distributions of characters, rather than differences or overall similarity. The first essay is about how the differences wouldn't necessarily amount to a radically different statement of relationships, even if we called the terminals species. The second essay is more about the perils of overall similarity judgements.

Cheers,
Martin