An Evolutionary Nested Hierarchy tells us more than the Functional Categories implied by Biblical "kinds"

Post of the Month: October 2011

by

Subject:    | If the Archeopteryx is the best Evolutionary Biology has to offer then the Creationists and IDers have won already
Date:       | 05 Oct 2011
Message-ID: | vrt*05WOt@news.chiark.greenend.org.uk

But more importantly, we already have one important way of telling whether an ancient theropod could fly or not : look at their wings and feathers !

We have a whole lot of fossil bird-like theropods and theropod-like birds who have clear wings and flight feathers that make it obvious they could glide. I don't think any of the bird-like theropods have shoulder structure that show they were capable of flapping flight, but neither do a lot of the theropod-like birds. (the dividing line between "bird-like theropod" and "theropod-like bird" is rather arbitrary I'm afraid; whether a given fossil will be referred to as a "bird" or "dinosaur" or "theropod" varies a lot between papers). There are also fossil birds or theropods whose wings make it clear they are secondarily flightless.

It is nonsensical that a fossil bird or theropod would have a wing morphology that's adapted to a kind of flight they don't have the lung capacity for. Those things co-evolve. So basically if you want to see which fossils could fly and to what extent don't look at the lungs (which you can't see anyway), look at the wings.

> We also have categories of things that go across all the classifications.
> There are mammals that can fly, there are mammals that live in the sea, and
> there are mammals that live on land.

Yep. Note that "flying", "living in the sea" and "living on land" are functional classifications, not trait-based classifications. Functional classifications don't yield a nested hierarchy; they don't yield much of a classification at all really.

We can try. Say we start by classifying where the animals live : we've got animals that live on land, and animals that live in the sea. Within the animals that live on land we can separate them into those that live in the forest, those that live in the desert etc, while within the animals that live in the sea we can separate them between benthic animals, bottom-dwelling animals etc. That's promising. Then within the animals that live in the forest we can have those that fly, those that climb and those that crawl.

But wait ! We also have animals that fly, climb and crawl in the desert ! We even have animals that fly and that crawl in the sea. We could just as validly start out with animals that fly, crawl and climb and then subdivide them according to where they live.

And I haven't even gotten into diet !

So functional classification is a mess. How about trait classification ? Well, it we look at animals that fly, some groups stand out immediately : bats, birds, insects and flying fish have very different wing structures. And not only do they have very different wing structures, they have very different physiologies. Even if we classify them according to differences in their DNA we find the same four groups : bats, birds, insects and flying fish.

So if we try to enlarge our trait-based classification, we find that the DNA, physiological and structural characteristics of bats are also found in many other non-flying animals; let's call them "mammals". Those animals have widely different lifestyles, but "under the hood" you can see that while they have organs performing different functions, those organs have similar structures : thus you get a bat wing resembling a dolphin flipper more than it resembles a bird wing ! And those structures are found nowhere outside of this group. There is no bird that has a bird physiology and a mammalian wing structure. Similarly, you can see that there are many flightless animals that share so many traits with flying birds you might as well call them "birds" too, like penguins and ostriches. Same with insects and flying fish (we'll call that last group "teleosts").

So okay, we've got mammals, birds, insects and teleosts; that's not very impressive, we had land and sea animals earlier which worked nicely too (fuzzy borders notwithstanding; there are always fuzzy borders). Can we classify them further ? Well, when comparing those four groups we can see that mammals and birds are much more like each other than the others. Aside from DNA similarities mammals share traits like lactation or bone structures that aren't found in any other group; conversely birds share traits like avian wings or avian lungs that aren't found in any other group either. However mammals and birds do have traits in common that aren't found in insects or teleosts : lungs, for example, or leg bones. They are also more similar in their DNA, even non-functional parts of it. Let's call their group "tetrapods". Again, we see that no other group possesses the traits that are characteristic of the tetrapods. You can go up - notice that the tetrapods and the teleosts are more like each other than either is like insects, and you can put them in the group "vertebrates", and you'll see that no animals outside that group share vertebrate traits. There is no such thing as an animal that's half-insect half-bird. There is not even an animal that's a bird with the insect version of a particular non-functional gene sequence. There *are* transitional forms between adjacent groups, because the dividing line can be fuzzy, but there *aren't* overlaps between non-adjacent groups.

Putting it another way - the interest of classification is to find a short-hand way of knowing something's characteristics. So if I see a movie classified among "anime" I'll know a number of things that movie is, and a number of things it isn't. The more like a nested hierarchy a classification is, the more useful it is because the more information gets packed into a single category. Movies don't make a nested hierarchy, so knowing something is "anime" tells me that it's animated, that it's from Japan, that it has a certain style and is more likely to follow certain tropes - but that's about it. If I want to know what kind of story it tells for instance I need to invoke another group, like "horror". "Horror anime" is the overlap of two different groups "horror" and "anime"; I couldn't guess it's horror just from knowing it's anime, nor could I guess it's anime just from knowing that it's horror.

Similarly, if you tell me an animal lives in the sea all that tells me is that... it lives in the sea. And that it can swim, I guess. Although sponges can't swim. "Lives in the sea" doesn't even tell me whether your animal breathes air or water. Heck, it doesn't even tell me whether it's an animal in the first place !

So functional classifications don't get you very far.

On the other hand, if I tell you something is a mammal, you'll know : - that it feeds milk to its offspring - that it has certain ear bones - that it has a certain type of lung - that it has internal fertilization (because all mammals are also amniotes) - that it has four limbs, or the developmental pathways for them (because all amniotes are also in Tetrapoda) - that it breathes air - that it has a vertebral column (because all tetrapods are also vertebrates) - that it has a skull (because all vertebrates are also craniates) - that it's a heterotroph (because all craniates are also metazoans) - that it's multicellular - that its cells have a nucleus and undergo meiosis (because all metazoans are also eukaryotes)

Now the nested hierarchy of life isn't perfect, mostly because of functional convergence (if you needed proof that function doesn't yield a nested hierarchy) and horizontal transfer of genes (which is mostly an issue in bacteria, and plants to a lesser extent). But as nested hierarchies go it's quite good.