Sunday, 20 August 2017
Mice, Mice, and More Mice
In general, though, what happens is that scientists measure a number of different features from the animals that they want to study, and compare which ones have the most in common. These days, these are most likely to be physical measurements, such as fine details of the shape of the skull or teeth, if at least some of the creatures we're looking at happen to be fossils. Otherwise, it's much more likely that the things being compared are stretches of genetic code. The further apart two animals are, evolutionarily speaking, the more differences there are likely to be, and if we pick a gene that both animals possess, and that can accumulate a reasonable number of changes without stopping working altogether, we have a good basis for comparison.
Of course, that's a considerable simplification of what's actually being done. The same mutation might, after all, occur twice at different points in the family tree. Or a mutation might be reversed, through the random genetic lottery. And we need to compare lots of different mutations across the gene to build up a proper picture. So drawing the trees isn't simply a matter of counting up the differences, it requires some significant computing power, running lots of different possible patterns of relationship, to find out which one is the best fit to the data. And, even then, there are different ways of doing the fit, using different mathematical methods.
Still, computers developing in the way that they are, it's the sort of thing that's progressively becoming easier than it used to be. There are entire journals dedicated solely to analysing these sorts of trees, and they are by no means short of material to fill their pages. (You can also see, from the links on the left-hand side of this page, just how much this sort of work has expanded over the last quarter century or so). But, even in a well-studied group like the mammals, there are always refinements and expansions of existing knowledge to be made.
At least when it comes to mammals, most of these studies are performed on small groups of very closely related species. Or, since there's no reason why you have to stop at species, even on subspecies, or yet smaller groupings - in this post from 2011, for example, I looked at a study using very fine detail to uncover how black bears moved across North America. The advantage of doing studies on such relatively small groups is that we can be pretty confident all the animals are related, and we don't need to look at a vast number of individuals to get a decent picture - a few from each species (or whatever) will do.
A more recent example of this sort of study, for example, considers not even a whole genus, but just a group of closely related species within a single genus. Specifically, it looked at the common shrew (Sorex araneus) and its ten closest relatives, which are found across the mid to high latitudes of the Northern Hemisphere. The advantage of using genetic markers is that, making some assumptions about the rates of genetic change, and using fossils of known age as calibration points to check those assumptions, we can not only tell which species are more closely related to one another, but we can even make a good guess as to how long ago they last shared a common ancestor. Which, in turn, allows us to make some conclusions as to where, when, and how key points in their evolutionary history occurred.
We can also see how the coming and going of the various Ice Ages split the shrews left behind in the Old World, giving rise to various species across Europe and Asia, and, how some species that today share the same geographical area are not, in fact, each other's closest relatives. Not only that, but the advantage of using multiple individuals from each species is shown by the fact that the Caucasian shrew (S. saturnini) of the Caucasus and northern Turkey turns out to fit in two different parts of the tree, and so might not even be a single species. (Some of them appear to have cross-bred with a different species at some point, while others didn't; whether that's enough to qualify the resulting descendants as specifically distinct is a matter of definition).
But what if we want to look at a larger group of animals? The problem here is that, to analyse a full sample of all the different species can become prohibitively expensive, and the sheer volume of the analysis required can tax even modern computers, tying them up for long periods of time. There are, however, two main ways round this problem.
The first is take a limited sample, analysing just representative species from the various sub-groups within the larger array of animals you are looking at. For example, you could take samples from representative species of deer, antelope, giraffe, camel, and so on, to determine how those groups relate to one another. In some respects, this may nor be perfect, since a given species may be less 'representative' than you think, but, in general, it's a very useful tool for analysing relationships at a level high above that of the species. It's this sort of study, for instance, that allows us to work out exactly what a red panda is.
The second approach is to take analyses previously conducted on smaller groups and stitch them together, using points where they overlap and are in broad agreement. Since, in practice, not all studies on different groups are going to be conducted in exactly the same way, this may require that you make some assumptions about them being "close enough", but, again, it's a useful tool, and can be relatively cheap - so long as there are enough overlaps, of course. This type of study can be particularly useful where we want to get a broad overview of an exceptionally large group of organisms, perhaps to see general trends across the entire tree of life.
Both of these approaches are useful, and, in any event, they are often the only methods that are practical. But sometimes, scientists are able to take the third option, and just analyse a huge number of species and see what comes up. It's expensive, and time consuming, but it's not inherently impossible, and every now and then, it does happen. In March of this year, for example, a study was published covering every species of cat-like carnivore, showing not only the relationships between the various kinds of cat (which, as one might imagine, had been extensively studied already) but all the different kinds of mongoose, civet, and so on. This leant support, for example, to a proposal that the carnivores of Madagascar first got there by crossing an ice-covered sea during a period of global cooling, rather than having rafted across open water.
The bigger the group of animals we want to study, however, the harder this gets. The largest of all mammal families is the Muridae, or mouse family, which includes literally hundreds of species of rats and mice. It is part of an even larger group, the Muroidea, or "mouse-like rodents", within which its closest relative is the Cricetidae, or hamster family. The great majority of species in the latter group, which happens to be the second largest mammal family, are themselves, to all intents and purposes, either rats or mice in common English parlance. Between them, the two families contain over one quarter of all named mammal species, rising to about 30% of the total when you include the various other kinds of muroid.
one was published. It conducted fresh analyses on museum samples from no less than 900 species of muroid rodent. Which, it should be noted, is still not much more than half of the 1,600 or so that are known to exist... but suitable samples simply don't exist for many species, and, besides, 900 is still enough to make a pretty huge tree.
And, no, I'm not going to try and duplicate the whole of it in one of my diagrams. You can see the full thing here, if you want. Instead, I'm just going to include the top level of the tree, showing some of the major groups, which you can see to the left. But what does this vast dataset - including around 17% of all known mammal species, by my count - tell us about rodent evolution?
The main bulk of it, represented by the upper seven lines on my abbreviated version, together constitute a group called the Eumuroidea, which roughly translates as "truly mouse-like rodents". These then, are the most mousey of the mice, and the most ratty of the rats. At the top of the tree (as I've drawn it; the order is, of course, arbitrary) is the vast Muridae family. The first split within that occurs during the Miocene, around 17 million years ago, when the gerbils and their kin split from the 'true' mice and rats. The former group splits about a million or so years later, with the origin of the first true gerbils. The gerbils as a whole turn out to be a bit of a mess, confirming an earlier study that showed our current and official classification system for the group is basically wrong.
Also within the murids, the study particularly focuses on the 'true rat tribe', the sub-group that includes the familiar brown rat and its relatives. These first arose around 9 million years ago, in the late Miocene, when they split from the harvest mice - which, given their small size, really don't look all that similar. A number of diversifications follow as the rats spread out to colonise Eurasia, Africa, and, eventually, Australia and New Guinea. This rapid creation of new species means that the genus Rattus itself is relatively young (at 2.8 million years old, it would have arisen around the dawn of the Pleistocene) compared with those of the house mouse and field mice, which have been around for over 6 million years.
One single species, the crested rat of East Africa, is, as expected, not closely related to any other member of its family. At times, though, it had been thought to be a cricetid, so at least this confirms it's in the right place.
Moving on to the cricetids, then, there is a rapid burst of new forms appearing around 14 million years ago, with all of the five existing subfamilies popping up in the space of about a million years. That was a time of sudden global cooling, so it's plausible that the changes in the climate caused hardship for the mice of the time, separating them and spawning new groups, many of which ended up in the Americas. Interestingly, while the voles can, on this chart, trace their ancestry back around 13 million years, that's followed by millions of years of nothing happening. Then, around 7 million years ago, during another climate shift, one that favoured the foods they prefer, the voles suddenly diversified and spread across the world. It's likely significant that the first fossils clearly identifiable as being voles come from not much later than this.
The remaining groups are much smaller, and, for the most part, rather more obscure to non-zoologists. First up are the nesomyids, which I've labelled as "Malagasy rats, etc." These appear to have entered Madagascar early on in their history, although at no more than 16 million years ago, rather later than many other studies have suggested. This means that the "etc." of my tree - that is, the nesomyids of mainland Africa - are an actual group, not merely a collection of assorted lineages that had already begun to split before some of them got stuck on the large eastern island.
The obscure mouse-hamsters of central Asia are next down the tree, confirming their status as animals that don't appear to have changed much since their first appearance, perhaps because they breed so much more slowly than other mouse-like rodents. Below them are the spalacids, deep-burrowing animals long considered separate from the "truly mouse-like rodents". They first appear, according to this genetic evidence, at around the very beginning of the Oligocene, if not earlier - roughly half the way back to the time of the dinosaurs. After the three subfamilies diverge in the early Miocene, nothing much happens to them for millions of years, with the splits between the ancestors of the various modern species all being relatively recent, although often pre-dating the Ice Ages.
Finally, standing unrelated to absolutely anything else on the 900-species chart, and having done so for a whopping 45 million years, there is the Chinese pygmy dormouse (Typhlomys cinereus). Although, since these sorts of study always include out-groups of unrelated animals to check for this sort of thing, we can at least say that it is a muroid (and not, say, an actual dormouse). As it happens, it is thought to have one reasonably close relative, but the researchers couldn't get a sample of that to study, so we're none the wiser on that front.
Taking a step back, though, to look at the whole of the chart, and the estimated dates when all the groups appeared, despite some individual peaks here and there, all seem to smooth out. That is, there is no one point at which mouse-like rodents suddenly increased in diversity, something we often see in other groups when some particular niche opens up. Instead, there is a just a general acceleration; a few new groups early on, then increasingly many as time passes. More mice build up and up, breeding and diversifying in their madly mouse-like manner. Until, today, they are 30% of all mammal species.
Why that should be, when other animals have had similar chances, still isn't clear. But whatever it was, it's something that hasn't changed much in the 50 million years or so since vaguely mouse-like rodents first appeared.
[Photo by "Pethan", from Wikimedia Commons. Cladograms adapted from Mackiewicz, et al. 2017, and Steppan & Schenk 2017.]