I RECENTLY READ a paleontology book that irked me by referring to the basal synapsids as “mammal-like reptiles”. This term pre-dates modern phylogenetics. It violates several rules of phylogenetics, the first being that all clades (groups of related organisms) should contain organisms sharing a single common ancestor in that clade, and the second being that clades should nest within each other, so that each more recent clade fits into its ancestors’ clades as well. In other words, clades should not be polyphyletic, but monophyletic. An example of a polyphyletic group is the pachydermata, an obsolete category containing elephants, hippopotami, rhinoceri, horses, tapirs, and pigs simply on the basis that these organisms have thick skin and hooves. Now these animals are placed into separate clades since they do not share a recent common ancestor. Polyphyletic groups are a great embarrassment to phylogenicists.
Before modern phylogenetics it was often said that amphibians evolved from fish, then the reptiles evolved from amphibians, with the mammal-like reptiles evolving from the reptiles, and the mammals from the mammal-like reptiles. In modern phylogenetics this picture is painfully misleading. In actuality the mammals and the reptiles come from two distinct evolutionary branches, and neither descended from the other–not to mention that also neither group descended from the amphibians. Yet this misrepresentation lives on.
The old model is based on sloppy definitions. A tetrapod lives chiefly in water? It’s an amphibian. A tetrapod lives on land and is not a mammal? It’s a reptile. A tetrapod has some early mammalian characters? It’s a mammal-like reptile. Again, we return to my earlier complaint about the sloppy use of the term “worm”–when you use a term so indiscriminately, it ultimately provides no useful information at all. Since evolutionary taxonomy is about discerning relationships, and sloppy terminology obscures this information, it’s important to have classifications nailed down hard.
The drawback of this is that it requires specific names for extinct organisms that do not fall into the modern groups (Amphibia, Reptilia, Mammalia, etc.) Laymen are familiar with these groups, so it seems easier to stick extinct animals into those groups when talking about evolutionary history–who wants to hear about temnospondyls, synapsids, amniotes, and cynodonts? But by misclassifying extinct groups like this, we only succeed in spreading misinformation and confusion. How is it possible that fur-bearing, milk-producing mammals could evolve from reptiles with their hard scaly skins? It’s not. But it is possible for furry skin and milk glands to evolve from a bare, glandular skin possessed by the synapsids.
So the true picture, briefly given: The early tetrapods evolved from fish and made their way on to the land. A split occurred that produced the amphibians (modern amphibians and a variety of extinct forms) along one branch, while the other led to the amniotes, organisms that produce several characteristic membranes during embryonic development, one of them the amnion. The early amniotes split into two branches, the synapsids (based upon characteristic skull features) and the sauropsids. The sauropsids include all modern reptiles and their descendants and a variety of extinct reptiles (the sauropsids contain Reptilia). The synapsids, on the other route, passed through the early therapsids, through the mammaliaforms, and eventually produced the mammals.
Also note that before the divergence of Amphibia there were multiple amphibious groups that have been labeled “amphibians”. The group Amphibia has also recently been called Lissamphibia to distinguish these organisms from their ancestors. While this works, it has the drawback that most people don’t know what a lissamphibian is. For this reason I think we should refer to the non-lissamphibian amphibians as “amphibious” or “amphibian-like”, meaning they lived an amphibious lifestyle while not being grouped among the amphibians themselves.
Because of the nesting of the phylogenetic tree, you can trace back the human descent through ancestral clades. So we are mammals, mammaliformes, therapsids, synapsids, amniotes, tetrapods. . . Things get a little strange once we reach the fish, since Linnaean taxonomy was invented first and these names carried directly into phylogenetics. It ends up we might technically be sarcopterygians, or lobe-finned fish, and teleosts, or bony fish. But before this we fall neatly into the vertebrates, chordates, bilaterians, and down on the line. We are definitely not mammal-like reptiles!
While most groups currently recognized are monophyletic, Reptilia is a paraphyletic group. Above I said that the reptiles are contained within sauropsids, but there is one group of sauropsids that are not classified as reptiles–the birds. The birds descended from theropod dinosaurs or a close relative of these dinosaurs, so actually birds are technically dinosaurs! But when taxonomists were developing Linnaean taxonomy, the modern reptiles obviously fit into Reptilia together, but birds have so many different features that they were grouped on their own in Aves. Now that we know more about extinct transitional organisms there are paleontologists who have argued about whether some dinosaurs should be placed in Aves!
Keep in mind that these routes of descent are incredibly simplified. Consider that 75% of the types of mammals that have ever lived are now extinct, include multiple major groups of mammals as distinct as the surviving mammals, the eutherians (placental mammals), monotremes, and marsupials. There are many different extinct clades. You can get an idea of some of the diversity of extinct life by scanning through the Tree of Life web project. This website has the ambitious goal of compiling accurate phylogenies for as many living organisms as possible, and will continually be updated as our knowledge advances.
[Edited to correct some errors pointed out by David Marjanović. Thank you!]