Zofia Kielan-Jaworowska

Although superficially resembling a coffee table book, Mammals From The Age Of The Dinosaurs is nothing of the sort. There are no color pictures of reconstructed mammals frolicking in the footprints of sauropods; in fact there is only one reconstruction in the book, and it’s a line drawing. Still, there’s a whole lot of stuff of interest.

*The Mesozoic is a long time

The earliest known mammals (which needs some clarification, which I’ll get to eventually) are Early Late Triassic (Carnian, if you please). That means that most – two thirds – of mammalian history takes place in the Mesozoic, not the Cenozoic (the putative Age of Mammals). The Cretaceous alone is longer than the Cenozoic. The Triassic date makes mammals almost contemporary with dinosaurs.

*The K-T extinction was almost as hard on mammals as it was on everything else

Depending on which taxonomist you believe, there were somewhere between 8 and 26 mammal groups at the subclass level in the Mesozoic. Four of those – Monotremes, Multituberculates, Marsupials, and Placentals – made it across the KT boundary (multituberculates went extinct sometime before the Oligocene). Even in these groups, survival was hit and miss – there are three known marsupial families in the Cretaceous and one in the Paleocene – but the one in the Paleocene is not one of the three from the Cretaceous. (However, see below for various problems with taxonomy).

*Mesozoic mammals are small and nocturnal

The average Mesozoic mammal is 5-10 grams, about the weight of a modern shrew. Not until the Cretaceous are there any known that are as much as 10 kilograms, and that’s only two fossils. Brain casts, when known, indicate relative enlargement of smell, hearing and touch areas, suggesting little things scurrying around in the dark eating bugs. (When skulls are available, there are sometimes extra foramina in the snout, suggesting nerve connections to vibrissae; perhaps hair first developed as “feelers” rather than insulation?) There are a few groups with grinding molars that might have been seed eaters. It’s generally assumed that they were homeothermic like extant mammals; there’s some evidence based on bone studies that therapsids, the probable ancestors of mammals, were homeothermic.

*Mesozoic mammals are mostly teeth

Tiny little teeth, obviously. The Purbeck Limestone is reported to have over 800 mammal specimens – all of which are isolated teeth. Sometimes taxonomic units all the way up to the family level are created on the basis of isolated single teeth; sometimes not even complete teeth. Famous vertebrate paleontologist Alfred Romer commented on his colleagues’ penchant for mammal teeth, complaining that you would think that was all there was to the animal – that teeth begat other teeth, and so on.

*But, still, teeth are what is important

Teeth are the animal’s connection between the skeletal system – which is almost always the only thing preserved as a fossil - and the outside world. Mammals share heterodont teeth with their predecessors and presumed ancestors, the therapsids (specifically the cynodont therapsids). Next time you’re at a museum, compare the teeth and jaws of a Dimetrodon (the sail back thing) with an Tyrannosaurus or Allosaurus or whatever big theropod carnivore is on display. The therapsid Dimetrodon has incisors and canines and premolars and molars (they’re shearing molars, not grinding molars); the theropod just has teeth, all more or less the same. This is not to say dinosaur teeth are simple; they’re complex enough in internal structure, with multiple kinds of material that wore away in such a fashion to give effective cutting or grinding surfaces (early paleontologists had a problem figuring out how huge sauropods ate enough to maintain their bodies with such small heads and “simple” teeth; compare the heads of an elephant and an Apatosaurus). It turns out dinosaur teeth are somewhat more complicated than originally thought – but still nothing like mammal teeth.

*Taxonomy is, as usual, a problem

There’s a difference between theoretical and operational taxonomy; specifically, the Biological Species Concept first proposed by Ernst Mayr defined species as reproductively isolated populations. The problem is that nobody actually identifies species that way; instead a species is what a taxonomist specializing in that group thinks it is. This causes no end of problems with Creationists, the Endangered Species Act, and local politicians trying to regulate what kind of things people can keep as pets; species are nowhere near as hard and fast as the nontaxonomist lay public think they are – leading to things like Preeble’s Meadow Jumping Mouse, the Eastern Cougar, the Dusky Seaside Sparrow, and the Mount Graham Squirrel; the “kinds” and “baramins” of creationists; and towns trying to legislate against “exotic” small cat “species”. If it’s hard with species, it’s even harder with higher taxonomic units – at least there’s a definition for species; there’s nothing for anything above (that’s not to say there isn’t a definition for Class Mammalia – there is, it’s not what you think it is, and we’ll get to it in a minute) but that there isn’t a definition for the taxonomic unit “Class” that works across groups of organisms – i.e., is Class Hexapod somehow the same level of organization as Class Mammalia.

*And the realities of fossils make it more of a problem

The three extant mammal subclasses (or infraclasses, or however you want to group them) are Prototheria (monotremes – extant platypus and echidna, and “Prototheria” isn’t actually a recognized taxonomic unit any more but it fits here), Metatheria (marsupials), and Eutheria (placentals). All three are defined based on their reproductive systems; monotremes have a cloaca – a single opening where the reproductive, genital and urinary tracts leave the body (the name “monotremes” means “one hole”). Marsupials have a marsupium – a pouch, where essentially embryonic young are held until they are old enough to function on their own. Placentals have a placenta – a structure that allows the embryo to remain in the womb until ready for birth (although the placenta provides nourishment to the embryo, its most important function is an immune system barrier – to keep the mother’s immune system from seeing the embryo as a foreign body). None of these features are identifiable in the fossil record. Assigning fossils to one of these groups is based on – guess what? That’s right, teeth (extant monotremes have no teeth as adults, but there are embryonic teeth so there’s something to compare). For that matter, the popular definition of mammals as a class depends on having mammae (monotremes don’t have those, either, but exude milk from the skin) and hair, neither of which is likely to fossilize (although I’m doing an experiment with an old milk carton in the back of my fridge). Thus, the actual dividing line between cynodont therapsids and mammals is: mammals have a jaw hinge formed by the dentary condyle and the squamosal glenoid. However, even this depends on having a dentary (the lower jaw bone) and enough of a cranium to see the jaw joint; thus the actual operation distinction between mammal and nonmammal fossils is based on – teeth.

*Because taxonomy is a problem, fossil mammal taxonomic groups proliferate

Mammologists, even with extant groups, have always been notorious “splitters” – creating subfamilies and supergenera and magnorders with cheerful abandon. It’s bad enough if you have a whole specimen, but it’s even worse if all you have is the ubiquitous fossil teeth. That is probably the reason there are so many mammal taxonomic groups in the Mesozoic; a lot of them are based on single molars or at best a couple of molars in a dentary. Molars have lots of “characters” - cusps and valleys and roots and so on, so it’s fairly easy to decide that your particular fossil molar is different from all other fossil molars and thus deserves its own species – or genera, or family, or whatever. That might possibly contribute to the apparent extinction of so many mammal groups at the KT boundary – perhaps many of them are not legitimate taxonomic units.

*But cladistics comes to the rescue – sort of

The whole point of cladistic taxonomy, as we’ve discussed before, is it recognizes that evolution works, and that every new species results from a single splitting from a previous species. That means a cladogram – the cladistic equivalent of the Linnaean “tree of life” – has, if you want to be strict about it, no taxonomic units other than species – no phyla, classes, orders, families, or genera. This was a major reason for the almost fanatic resistance of traditional phyletic taxonomists to the introduction of cladistics; you had spent your whole life elucidating the taxonomy of some group – Artiodactyla, say – and now some cocky hotshot from the American Museum of Natural History comes along and says there isn’t really such a thing as Artiodactyla. (It didn’t help that a lot of the early cladistics promoters were Incandescent Excretory Orifices). As often happens, the cladists were right. What that means in practice is there are either no higher taxonomic orders, or there is a taxonomic order for every split in a cladogram – and that there are no comparative taxonomic units across groups of organisms. The fact is, though, it’s still useful to talk about “orders” and “families” of organisms, and the cladists haven’t provided any alternative. The authors (Zofia Kielan-Jaworowska, Richard L. Cifelli, and Zhe-Xi Luo) try to make everybody happy by providing two taxonomic trees for mammals – a cladogram to the order level, with Linnaen taxonomy below; and a conventional Linnaean classification. It’s initially hard to believe things were as diverse as claimed. If you consider just extant placentals – placentals are an infraclass in both the cladistic and Linnaean schemes – you have things ranging from blue whales to bats to tigers to shrews to cows to me. Yet, supposedly, there are ten groups at the subclass level – one step up from infraclass – known only from sparse fossil material – mostly, again, teeth – and all of which would have looked remarkably similar if you saw them scurry by. Placentals don’t even get a subclass to themselves; they get grouped with Metatheria (marsupials and deltatheroida) in the subclass Boreosphenida. (As the “boreo” prefix indicates, both placentals and marsupials originated in the northern hemisphere – well, what’s now the northern hemisphere at least. Marsupials, in fact, evolved in North America – the oldest known marsupial fossils are from Utah. This explains why you see all the dead possums in the road; they’re just trying to get back home). There’s a corresponding Australosphenida, which includes monotremes and the Ausktribosphenida – these last, as you might guess, are known mostly from teeth. The Monotremes are only known from Australia, except one Paleocene platypus from Patagonia – this doesn’t really bear on anything but it’s so alliterative I had to include it. The other subclasses are eupantotheres, symmetrodontans, allotheres, eutriconodonts, shuotheridians, morganucodonts, and sinoconodonts. I suspect what’s going on here is similar to the problem Stephen J. Gould had with the Burgess Shale fauna; few things fit into extant groups leading Gould to propose that there were multiple new phyla in the Cambrian and evolution somehow worked different then. Well, Gould never really accepted cladistics; when cladists got a hold of the Burgess Shale fauna, they were able to build perfectly logical cladograms that accounted for all the observed characters – but which didn’t conserve the traditional Linnaean groupings. So it is with Mesozoic mammals – the apparent diversity of placental mammals is all based on a basic body plan – which was not shared with those Mesozoic groups.

*Sinoconodon is a better “missing link” than Archaeopteryx

Sinoconodon rigneyi comes from the Early Jurassic of China (illustrating that the fossil closest to the ancestral condition is not necessarily the earliest one; as mentioned above, the earliest known mammal fossils are from the Upper Triassic). I got in an online discussion with a creationist who kept bringing up various things he thought were support for Creationism; in particular, he would post links to various news articles that suggested paleontologists couldn’t tell if a particular fossil was a reptile or a bird, intending, I suppose, to indicate paleontology couldn’t be trusted if it couldn’t make that distinction. Well, that’s sort of the point; according to evolutionary theory you will always have a condition where whatever arbitrary method you use to group organisms doesn’t work anymore. All species – and, therefore, any taxonomic unit above the species level – begin with a split; an organism is born which is of a different species than its parents (note that the assignment of the organism to a new species can only be made by hindsight). (As I understand it, the Catholic Church, having accepted Darwin, has a similar doctrine regarding the soul – at some point or another in the Pleistocene someone was born who had a soul – but whose parents didn’t have souls. Initially disturbing but over the years I’ve met a number of human-like creatures who didn’t have souls, so the scariness wears off). So with Sinoconodon (or, to be more correct, a distant direct ancestor of Sinoconodon – it was a mammal but Mom and Dad were still therapsids. That brings us back to the “What makes a mammal” problem mentioned earlier; going by extant mammals, most people would say hair and milk; those with a more anatomical bent might say a single bone in the lower jaw, three bones in the middle ear, confluent external nares, and diphyodont teeth. As mentioned above, though, the hard and fast mammalian condition is that jaw joint between the dentary condyle and the squamosal glenoid – and Sinoconodon has that. Did it have hair? Well, maybe; as mentioned the cynodonts were probably homeothermal and thus may have had hair too, thus it isn’t a distinguishing mammalian condition. How about milk? Probably not; we’ll get to the reason below when we talk about teeth. Single bone in the lower jaw and three in the middle ear? Nope, Sinoconodon has a dentary, alright, but the articular, prearticular, and angular (which will eventually become the middle ear malleus, incus, and stapes in more derived mammals) are still attached to the dentary (although they no longer form part of the jaw articulation). Confluent external nares? In cynodonts there’s a process of the premaxillary bone that separates the nares; in mammals there’s just cartilage – except for Sinoconodon. Diphydont teeth are deciduous – the young are born toothless, then get “baby teeth” that fall out and are replaced – but only once. Therapsids – including cynodonts – have polyphydont teeth; they are born with teeth, and they are continuously replaced throughout their lifetime, usually in an alternate pattern. Again, back to the museum with the theropod skeleton. Note the tooth pattern – every other tooth in a Tyrannosaur jaw is larger than its neighbors. That’s the polyphydont alternate replacement pattern; it ensures at least half the teeth in the animal’s jaw are new. So why are mammals different? Well, any nursing mother, be she placental or marsupial or even monotreme – would tell you why it wouldn’t be a good idea to have the babies or joeys or puggles born with teeth. There’s another reason, having to do with development; mammalian skulls grow proportionally larger during development than those of other animals. If you didn’t replace teeth the initial set would end up being too small for the eventual skull size. As you can probably guess by now, Sinoconodon had polyphydont teeth – and, therefore, probably didn’t nurse its young. Or yelped a lot.

*Which brings us back to teeth

And why not replace the teeth more than once? Well, if you remember back to you childhood, when your permanent teeth came in they sometimes felt a little strange in your mouth until they wore in for a while. Polyphydont teeth are nonocclusive; they don’t match up in the upper and lower jaws such that they mesh when the animal bites. They can wear into a meshing pattern, but they don’t start out that way. Other than Sinoconodon other ancestral mammals (at least all those where there’s enough jaw material to tell), even though they have dyphodont teeth, don’t have naturally occlusive teeth; they had to wear into a matching patterns. Eventually mammals do develop naturally occlusive teeth, but it went along with being dyphodont – if you replace the teeth too often and in an alternate pattern, you always have at least half your teeth new; an advantage if all you are going to do is bite and swallow, but a major handicap if you want to chew – you end up with teeth that don’t quite fit. And that, of course, is the big deal with mammals and why fossil teeth end up being so important – mammals can chew. When we got the extra strong jaw joint we lost some of the reptilian ability to swallow stuff whole but traded it for the ability to chew and gnaw. Herbivorous dinosaurs apparently depended on gizzards to do their chewing; their jaws just aren’t set up to do it. They couldn’t gnaw either; there are no dinosaur beavers.

*Multituberculates probably really are a major separate group

They are pretty abundant from the Late Triassic through the rest of the Mesozoic and so are one of the better known groups, and as mentioned they are one of the four major mammalian groups to make it past the KT extinction. Unlike most of the Mesozoic mammals, they seem to have been herbivorous or at least omnivorous rather than insectivores. They are generally convergent on rodents; small to medium sized things with prominent incisors and chewing molars. There are a couple of big differences, though. Rodents – just about all extant mammals, in fact – have a skull that is compressed laterally; it’s high than it is wide. The multituberculates are the other way around, they have broad skulls compressed vertically. That, in turn, has something to do with the jaw musculature – multituberculates chew backwards; in the “power” stroke the lower jaw moves backwards rather than forwards or sideways as it does in all other mammals (except for the poorly known gondwanatheres). You can tell this from the way the muscle attachments worked and from the wear patterns on the teeth; ever since I read this I’ve been watching my back yard squirrels to see how they chew. A little bit is known about multituberculate reproduction; there are some skeletons that include the pelvis. The pelvic opening is very small; more like marsupials than monotremes or placentals; it’s therefore probable that multituberculates were viviparous and had a marsupium or something similar to host the embryonic young. Analysis of the limbs suggests multituberculates had a sprawling stance – the upper limb bones left the shoulder and pelvis parallel to the ground, like lizards – but that they were powerful jumpers. If you’ve ever watched squirrels progress across a lawn, they “saltate” – move in a series of little jumps. Multituberculates seem to have done the same thing, except at much steeper angles – an almost froglike 30-45° angle for each jump; both the hindlimb pattern and the way the vertebrae connect seem to indicate this. So, yes, I’ll concede that backward-chewing marsupial frog mammals are sufficiently weird to get their own subclass. It’s been suggested that because of the probable reproductive mode the marsupials and multituberculates share a common ancestor, but it doesn’t seem to work out.


Well, sorry about the much longer than normal review but there is a lot of cool stuff here. Unfortunately this isn’t a book for the semiserious paleontologist; I had to spend an awful lot of time looking stuff up and I still am unsure whether I understand most of the details of mammalian cranial evolution. I found the book’s organization problematical; after an introduction to mammalian taxonomy it dives right into a long section on the temporal and geographic distribution of Mesozoic mammals. Unfortunately this assumes the reader is already familiar with the topic; it doesn’t do much good to talk about where to find Cretaceous eutriconodonts if you won’t have a clue what eutriconodonts are for several chapters. The authors consistently use European marine stage names for the Mesozoic; admittedly some of these are international now. However, they also use “Liassic” for early Jurassic, which is not internationally recognized. They do the best they can with illustrations – there are plenty of them, but it’s hard for a paleomammology novice like myself to figure out things like the shape of a parietal bone based just on a drawing. I wonder if this is an application for 3D printing? A library of scaled fossil bones that you can print, so you can actually handle a comparison series and see how they differ. There’s an extensive bibliography; one section on primary sources and one on general reading, which is laudable. There’s also an appendix which has all the characters used to make the mammalian cladograms, so you can do your own if you want. Pretty expensive; the only reason I have it at all is it was ½ off list at the GSA convention – and even then it was pretty steep.… (altro)