Mammalian Macroevolution Muddle

Over the past few days I have been reading the various blog commentaries on the recent Nature paper, Bininda-Emonds et al. (2007), “The delayed rise of present-day mammals.” The paper (1) constructed a “supertree” for virtually all extant mammal species (4,510 out of 4,554!!), (2) dated the tree using sequences from 66 genes and 30 fossil calibration points, and (3) concluded that placental mammalian orders arose before the K-T impact 65 million years ago, and that mammalian families arose a substantial time after the K-T impact.

Doing phylogeny on this scale is a substantial achievement, and the authors deserve the attention they are getting. But I am not sure that everyone commenting on the “Did the K-T event lead to modern mammals?” issue clearly understands the macroevolutionary concepts involved. [1] To illustrate, I will pick a particularly egregious, and therefore clear, example, from an ID blog. An IDist commentator named Bradford challenged the evolutionists as follows:

Where were the environmental pressures associated with:

Egg-laying mammals splitting about 166 million years ago, marsupials splitting from placentals 20 million years later, a 50 million years lull, then during the next 20 million years or so the evolution of the extant orders of mammals occurs and then the asteroid hits. Where are the environmental links corresponding to these events?

Bradford seems to think that modern evolutionary theory says that there “should” be some obvious event like the K-T impact to explain “major” events like the divergence of monotremes and marsupials+placentals, the divergence of marsupials from placentals, the divergence of the mammalian orders, the divergence of the mammalian families, etc.

But think about it for a second. There is no reason to think that these divergence events were dramatic when they happened. When the lineage that gave rise to the modern monotremes split from the lineage that gave rise to marsupials+placentals, this was very likely just another routine speciation event. Both of the daughter species in all liklihood still laid eggs. All that happened is that, millions of years after the initial split, one of the lineages lost the egg-laying trait. Attributing the later major innovation to the earlier divergence event is to give the divergence event retroactive, after-the-fact significance, significance the divergence event wouldn’t have if you were able get in a time machine and go back and watch the species diverge over a few thousand years.

In other words, divergence events are not always the same thing as “major innovation” events (probably not even usually). This fact has substantial significance in several other areas once you start to look think about it. Imagine you are a future paleontologist who is spectacularly lucky and find the fossils representing the exact two species that were the progenitors of the monotreme and marsupials+placentals lineages. My question is, would you even know what you found? If you were to use traditional taxonomy to classify the two fossils, you would probably put the two species in the same genus. If you decided to join the 21st century and use cladistics, you would know that the fossils were early mammals and that they had features shared with all living mammals. But these fossils, even though they are from the true ancestral species of monotremes and marsupials+placentals, probably would not have any major distinguishing derived characters that would allow you to place either species specifically in the monotreme or marsupial/placental branch. The distinguishing characters might not evolve for millions years, many species later in each lineage. The upshot: even if you had the exact fossils from the divergence point, you would probably miss this fact, and not date the divergence point until millions of years later.

Here’s another interesting implication that also seems difficult to “get.” Something like the K-T event could be important for the origin of modern (living today) mammal groups, but this fact could be unrecorded in the molecular phylogenies – all as a result of the completely standard and prosaic process of extinction. For sake of argument let’s assume that date is rock solid. Let’s also say that the species of this crown group dominate the Big Mean Predator niche that was formerly occupied by certain Big Mean Dinosaurs, and that after the K-T event it really was the case that this group of mammals immediately took over this niche and dominated it for the next 64 million years. If all of this were true, should we expect the molecular clock, even a perfect, inerrant molecular clock, to give us a date of 64 million years for the modern crown group?

Not necessarily. Remember that most species that have ever lived are extinct – 99% or something. This means that any group that has been sitting around and speciating in its niche for 64 million years will have experienced a large number of extinctions, probably 10 times or more extinct species than there are living species. You can only get DNA from living species, so the only way your molecular clock will give you back the date of 64 million years is if the very first diversification event after this niche was occupied – remember, two very similar species with small differences – both produced lineages that just happened to survive all the way down to the present day. If one of those two species went extinct immediately, perhaps outcompeted by its sister species, too bad. If both lineages survived for 20 million years, but some minor predatory innovation in one derived group at 44 million years led to a general advantage of the derived group, and the eventual extinction of the other lineage, then too bad, your molecular clock will read 44 million years ago. If hungry Clovis hunters hiked over to North America 12,000 years ago and ate the last representative of one of the two lineages, then too bad, you won’t get the 64 million-years-ago date. None of these scenarios, although they all give different molecular dates for the “diversification” of the crown group, would change the fact that the “key evolutionary event” that “produced” the crown group was the occupying an ecological niche left empty by the dinosaurs.

Here’s the point: the age of a crown group depends at least as much on subsequent extinctions as it does on whatever “key evolutionary events” we might hypothesize were behind the origin of a particular crown group. I’m not sure there is even a strong reason to think that dating the base of a crown group “should” reflect some significant evolutionary event. The timing of the “original diversification” of a crown group depends just as much on evolutionary events millions of years later! If we assume lineage extinction happens “at random” at a coarse level of analysis, then even a pure null model would suggest that the common ancestor our extant, suriving Big Mean Predators would date to some point after the “key evolutionary event” that produced the lineage in the first place. And the problem would probably get worse, the older the “key evolutionary event” was.

Now that I have set this up in imaginary terms, we can think of a few examples where the “key evolutionary event” and the “age of crown group” are not the same. E.g., the origin of “birds” with feathers and flight capability was before Archaeopteryx, 150 mya, but the age of the crown group – the common ancestor of modern birds – was much later (although looking at a recent paper indicates that the common ancestor of modern birds might date well back into the Cretaceous).

The expected “crown group delay” could be estimated fairly simply with modeling based on some simple assumptions about speciation and extinction rates. The problem would presumably be less severe for more speciose groups. I’m sure someone has already done all this work and probably all of these issues are well-understood amongst people who have seriously thought about cladistics and macroevolution. But I haven’t seen much of this in the blog commentaries, or even in the original Nature paper. Maybe the law of large numbers ameliorates the problems I raised. Or am I missing something?

P.S.: None of this means I am taking a position for or against attributing “the radiation of modern mammal groups” to the K-T event. What I’m trying to point out is that “modern mammal groups” are pretty much arbitrarily defined from the perspective of someone who is following evolution as it progresses through the ages without knowing what random things will happen to be the survivors tens of millions of years later. I was going to complain about the concepts of “family” and “order” as well, but you can pretty much see what I would say by referring to my “Down With Phyla!” posts.


(1) The worst error is to miss the fact that Bininda-Emonds et al. acknowledge that the fossil record unambiguously shows that mammals did radiate dramatically just after the K-T event. So it’s not as if they’re saying nothing happened after a meteor impact wiped out the dinosaurs.[1.1] It’s just that [1.2] those fossil mammal groups were not primarily ancestral to modern lineages.

(1.1) Dinosaurs, except birds. And yes, there are various debates about the exact relationship between the dinosaur extinction and the impact event. Perhaps dinos were already on their way out and the meteor was just the coup de grace. For the purposes of this post all we need to know is that many things went extinct at 65 million years ago.

(1.2) If the dating used in the study is right; one key point to remember is that fossil calibration points can only give minimum ages for divergence events, so if they find older fossils, or even if they don’t, the dates could be moved back.