A very important and truly wonderful paper in Nature described a tour-de-force analysis of the Mammalian Evolutionary Record, and draws the following two important conclusions:
- The diversification of the major groups of mammals occurred millions of years prior to the KT boundary event; and
- The further diversification of these groups into the modern pattern of mammalian diversity occurred millions of years later than the KT boundary event.
[This is a repost from gregladen.com]
The KT boundary event is the moment in time when a ca. 10 km. diameter object going very fast hit the earth in the vicinity of the modern Yucatan, causing the extinction of the dinosaurs (and almost everything else larger than a microwave). It has been suggested that this event resulted in (allowed for) the subsequent diversification of the mammals, presumably because the earlier extinction event opened up previously filled niches, into which the mammals evolved, and possibly because of dramatic climate change that occurred with this event.
One of the reasons that this study is important is that it seems to falsify this long-standing hypothesis.
I have a number of comments on this paper. Before I make them I want to say that I have absolutely no strong feeling on which of these ideas is either likely or important. I'm not coming at this with a particular agenda regarding this evolutionary pattern of mammal evolution. The KT boundary question is not directly in my area of research (I'm post middle Miocene). I do have one agenda-like perspective, however, that I want to lay out in the beginning: Larry Moran points out that the evidence for close connections between climate change and evolution seems to weaken with every new study, and that we have to simply start believing that the connection is a falsehood. I disagree with this conclusion, but in a way that is not meant to preserve the old "climate change - evolution" link. I agree with Larry that the link as I've ever seen it described and tested has failed to find support in the data. However, I do not believe it is correct to say that there is no relationship between climate change and speciation. I think that the way the link has usually been postulated is incorrect, and that when it comes to understanding the environmental change-speciation link we are dumb little babies (so far). I'll discuss this in more detail below.
The following figure is from the Nature paper, and shows the new mammal phylogeny. The dotted circle is the KT boundary. Notice that the major taxa are shown to have emerged before the event, and the rise of numerous subsequent species after it. The lineages shown on this graph represent nearly every single living mammal, but no fossil species are used since this is a molecular phylogeny. Click on the figure to get a larger copy.
The (Possibly) Real Importance of this Paper
Never mind the KT event or the test of the hypothesis that mammalian radiations are linked to it. This paper provides what appears to be
the best current phylogeny of the living mammals at the higher taxonomic level, significantly better than anything we have had before. One could argue about some of the details (one has to pick "a" tree for this kind of study, and there are alternative possibilities) and the resolution of the given tree is poor for the more recent radiations (I'm personally looking for the best current rodent tree and I'm afraid this one won't do). However, attempts to put the major groups together have been difficult and often unsatisfying, so it is great to have a new, most up to date version.
However, the reader should not assume that this tree is the final accepted phylogeny of the mammals. Other research teams will come out with critiques and there will be revisions, I'm guessing sooner rather than later. Watch for it. What will be interesting is to see if the critiques involve the overall pattern described here in relation to the KT boundary.
I am not certain that we are ready to believe calibration of molecular phylogenies.
The timing of events on this tree are estimated by using molecular clocks calibrated by various bits of the fossil record. The authors were fairly careful in doing this, not using a single clock, but adjusting for suspected rate variations among different clades. The fossils give minimal dates for divergences .... so if you have a fossil that has an undoubted featured of a certain clade, and the fossil is dated, then you know that the split between that lineage and it's nearest molecular relative predated that fossil date.
I have not done a thorough analysis of the supplemental data supplied with the paper, so please understand that I do not have a specific criticism of the way in which the molecular data are calibrated. My gut feeling is that the authors did a great job. However, I do have an overall problem with molecular calibration and I want to preach caution.
The initial diversification of the living superorders and orders is set in this study at 93 million years ago. An acceleration of diversification is set startling at the "Early Eocene" (the paper does not give a date for this, so I'll set it at 54 million years ago).
Taking 65 million years ago (the KT event) as a benchmark, the earlier date would have to be moved towards the present by 43 percent in order to "fit" the data with the KT boundary. I other words, if you KNEW that the initial diversification happened at the KT boundary, then you would have to adjust all of your molecular data by 43 percent, or putting it yet another way, if all you had was the molecular phylogeny and the date of the KT event, you would have to have confidence in your molecular calibration sufficient to believe that you could not be wrong by 43 percent. Forty-three percent sounds like a lot.
I guarantee you that most molecular biologists will say that 43 percent is a very large number. I also guarantee you that a LOT of fossil people will say that 43 percent is NOT a large number, and that there have been many cases where the molecular calibration is off by a factor of two.
For example, over the period of several years, the molecular data for the split of humans and the other apes had the following pattern: In the 1970s, there was one group saying 5 million, other groups saying much more, like 6-7. Over time, the "long view" groups revised and revised until finally they were also saying something close to 5 million, or even a little less. Then a fossil was found in Ethiopia that dated to just under 5 million and it looked like a good candidate for an australopith at the boundary between a last common ancestor and early hominids. So everybody was pretty darn happy with 5 million years.
Then, suddenly, more fossils started to show up and now we are looking at likely hominids closer to 6.5 or even 7 million years. The hominid-ness of the earliest fossil is somewhat in dispute, but frankly, the nay-sayers are probably wrong ... the hominid-ape split probably dates to between 6 and 7 million, closer to 7.
So, the calibration of the DNA systems used to date the human-nonhuman ape split, a topic that has received considerable attention, has a fudge factor of 30 or 40 percent, depending on how you look at it.
Now, some of you are already thinking: Right, sure, but the early date in this study (93 million years) must be based on dated fossils! You can't move the molecular estimate of the timing of a split between lineages to a point in time AFTER the existence of fossils demonstrating the split! Yes, you would be correct about that. So now the question is, are the early diversifications (the ones around 93 million years) linked to fossils that demonstrate the split? Again, I have not looked at the specific cases used for this calibration, but I believe that there are fossils of early mammals, indicating these splits, dated to well before the KT event. However, the first appearances in the fossil record of terrestrial mammals is hard to estimate, and the earlier in time one looks the more likely one underestimates the age of these events. If anything, I would guess that the 93 million year date is an underestimate, and that these splits really happened a bit earlier.
Is the Eocene (and Later) Estimate for Diversification Wrong?
Let's say that the 93 million year date is an underestimate by 15 percent. If we recalibrate the entire tree based on this guesstimate, then the later diversification (said in this paper to be Early Eocene) would move from 55 million years to 63.5 million years ago.
Eocene and later diversification is not about the KT boundary
If it is true that there is a post-Paleocene (Eocene and later) diversification of living mammals, then this does not mean that mammals did not diversify in the early Paleocene, just after the KT event.
Studies looking at just the fossils did not disappear on the publication of this paper. I'll give you one very handy and excellent example: John Alroy of the Smithsonian has a paper looking at early appearances of mammalian taxa in North America in relation to time. The following figures are from his paper: http://www.nceas.ucsb.edu/~alroy/Paleocene.html
Figure 1. North American mammalian diversity, origination (new appearance) rates, and extinction rates through the late Cretaceous and Cenozoic. Data are based on multivariate ordination and standardized sampling of faunal lists. (a) Standing diversity. Y-axis is logged to show the lack of either a log-linear (exponential) or asymptotic (simple logistic) pattern; instead, an offset between two logistic curves at about 65 MYA is indicated. (b) Origination rates. (c) Extinction rates.
Figure 2. Trends through time in North American mammalian body mass distributions. All species falling into each 1.0 MY-long bin are considered. (a) Mean body mass. (b) Standard deviation of body mass.
These fossil dates do not require calibration in relation to the geological column ... they ARE the geological column. There is no ambiguity about the relationship between a spike in mammal species novelty and the KT event, at least in North America (where the direct events of the KT event may have been the most severe, by the way). The conclusion from the molecular data appears to be incorrect from this perspective.
Is the Later Diversification a Later "Event" or an Artifact of the Pattern?
It is possible that for any molecular record of sufficient diversity (a record of several higher taxa) and time depth, there will emerge "waves" of diversification that are partly determined by actual splits between species and partly determined by the patterns of extinction such that the apparent timing of the diversification a) has little do to with the actual events and b) follows along behind the "present" in fits and starts. Let's look at an example, at least of "a".
Now, I'm not going to do the actual work on this, I'm just going to mentally walk you through it. Imagine we wanted to estimate the time of diversification of the dinosaurs from the molecular data. We determine the clade of living forms that includes all possible dinosaurs. This would be the Archosauria, which includes the dinosaurs, the birds, and the crocodiles. Unfortunately, since we are basing this on living forms (using DNA) we can only sample the crocodiles and the birds, no dinosaurs.
The resulting analysis would show an initial diversification (the bird/croc split) way earlier than the presumed dinosaur radiations (I'm assuming there were multiple!) and another, much later radiation (the birds). We would be like the three statisticians hunting rabbits ... we might be able to convince ourselves that we hit the rabbit, but the rabbit would be laughing at us.
It seems to me that the fossil record shows a diversification of mammals hard on the heels of the KT boundary, and the living mammal molecular reconstruction either has some calibration issues or is tracking a different phenomenon. Or both.
Even if the later radiation is "the" radiation... was KT unimportant? (Is environmental change really unrelated to speciation?)
This is a bit of a philosophical question, but it gets to the issue I mention above about the relationship between climate change and speciation. Larry is not going to like this, and I heartily look forward to his comments if he has chance to make some.
Free oxygen in the atmosphere is essential to much life on this planet. But it did not always exist. Had the biological processes that resulted in atmospheric oxygen not happened, all of the organisms that depend on it today would not exist. So, that ancient event - a clear example of environmental change -- has a lot to do with all later evolution.
That is a very indirect link, making it trivial to the question of a connection between environmental change and speciation.
In terms of numbers of species as well as biomass, living ungulates mainly depend on widespread grasslands. Widespread grasslands emerged as a feature of the environment during the Miocene. The radiations we see of ungulates could not have happened were it not for the appearance of these grasslands (an environmental change). That is a less trivial link. It is not likely (as has been attempted) to find specific events ... a particular dessication event, the closure of the Panama land bridge, etc. etc. to a particular radiation of the ungulates. And there is more than one radiation, likely. But the grass-ungulate link is less trivial than the oxygen-nearly everything link.
The rise of a particular evolutionary novelty and a particular climate event or environmental change is perhaps unlikely to have happened, and if it did, it would be hard to see. The history of speciation is not clearly linked to specific environmental changes to the extent that would be necessary for this to be our main explanation. This is complicated by the fact that many of the major "events" we see are rare, but we know that at least in the last few million years (and probably at various other longish periods of time in the past) orbital geometry running on cycles of tens of thousands of years has to be accounted for. But there is a very large scale link, and sometimes that link is closer and better fit historically and functionally than other times.
It is wrong to say that "environmental change causes speciation" and leave it at that. But it is also wrong to say that "environmental change is unrelated to speciation," because there is a range of "trivial" to more direct connections between environments and adaptive patterns. I believe that we are not in a position to describe a pattern or to develop a strong theory in this regard at this time. Persistent belief in a simple environmental change-speciation link has probably, in retrospect, wasted a lot of our time and energy, but that is how science works. We need to move on towards a more nuanced and meaningful set of models.
It was the Birds Fault!
OK, so let's say the fossil record is wrong, and we must simply believe the molecular record. Mammal radiation did not occur right after the KT boundary. Why?
Well, I want to re-emphasize that it probably did, so "why" may not be a valid question. However, there is one thing I'd like to throw into the works. I am about to make a number of enormous logical leaps, so hang on to the safety rope.
Some clades experience increase in body size over time. This does not necessarily mean that small forms disappear, but rather, the range of body sizes across species in a clade increases. In a way, you can think of this as diversification of potential niches, because (at least for terrestrial mammal) there is a link between body size and several important aspects of "niche" running from diet to predator-related issues to nesting, etc. So one thing you might expect is for there to be a link between increase in range of body size and increase in species diversity.
One system that may drive body size increase is the predator-prey relationship, whereby prey "outgrow" various predators, but predators also increase in size, over evolutionary time. It seems that dinosaurs at various time and places experienced predator-prey "arms races" in body size. It seems that this also happened with mammals.
Now, in the Paleocene, after the KT boundary, it is my understanding that few large (like, mammoths and such) critters were to be found, and that terrestrial ecosystems were dominated by largish avian predators. These Killer Big Birds (much like the Big Bird from Sesame Street but with a larger beak and a much coarser attitude) were probably the main predator on early post KT terrestrial mammals. But it seems (and I may be totally wrong here) that there was not an arms race for body size. Maybe a little one, but the mammals did not grow enormous during this period. That was to occur later, starting in the Eocene.
Why? Well, my hypothesis is this: Growing large has costs and benefits, but other adaptations do as well. The costs of growing large include slower reproductive rate and greater vulnerability in relation to the food supply, for instance. What if mammals that were being preyed on by Big Bird were able to adapt to this predation in a different way than getting larger? There are of course many ways to adapt to predators other than to outgrow them, and even when we do see large body size emerging we also see other things happening at the same time (like being fast, being cryptic, or being hard like a walnut).
Specifically, I hypothesis that the benefits of large size after subtracting the costs of large size were less than the benefits of some other strategy, and that strategy is one that would only work for a mammal being eaten by a bird. Perhaps the slightly different thermodynamics of mammals and birds was exploited, different diurnal patterns of energetics, or locomotory patterns. Fill in the blank: Mammals, with the special mammal feature of X reduced predation by birds, who were limited by Y by using strategy Z, such that Z is NOT increase in body size.
When Big Bird was replaced by mammalian predators, then we have mammal preying on mammal, and this discordance between X and Y (allowing for adaptation Z) did not apply. Large body size still has it's benefits, and that becomes a major mode of adaptation, so we see the Eocene beginnings of a mammalian radiation involving body size increase and diversification of species. (This could also explain why rodents have not all grown to be larger than snakes and hawks.)
ADDED LATER: There is an excellent discussion of this research on Panda's Thumb.
Alroy, John. nd. The Fossil Record of North American Mammals: Evidence for a Paleocene Evolutionary Radiation. http://www.nceas.ucsb.edu/~alroy/Paleocene.html
Bininda-Emonds, O.R.P., Cardillo, M., Jones, K.E., MacPhee, R.D.E., Beck, R.M.D., Grenyer, R., Price, S.A., Vos, R.A., Gittleman, J.L., and Purvis, A. 2007. The delayed rise of present-day mammals. Nature 446: 507-512.
It should also be noted that the study was focusing on extant mammal groups; there was a radiation of mammals after the K/T event that left no living descendants are therefore are not part of the supertree (the authors, if I remember correctly, recognize this).
This paper has come under fire, though, particularly in a paper recently published in PLoS. Another study by Wible et al. published last year also suggests that the fossil evidence does not support the hypotheses of the supertree paper.
As for the "terror birds," some were likely predatory, but the issue is still controversial (I don't think that the large birds were somehow keeping the mammals down, at least not at anything more than a very localized scale). I'll have to read up more on that, though, as I don't want to be so vague.
Overall there is a lot of skepticism of molecular clocks among paleontologists, and from what I've seen in the wake of this paper the paleontological evidence is at variance with the model of mammalian divergence proposed in the supertree paper.