In 1957, Alfred Sherwood Romer (1894–1973), the leading vertebrate palaeontologist of the 20^th century, wrote, “One of the most important steps in the evolution of vertebrates was the “invention” of the amniote egg… Its appearance marks the beginnings of the history of the reptiles and the potentialities of evolution of the great groups that are dominant today, the birds and mammals. The evolution of the amniote type of development was a necessary antecedent to the true conquest of the land.”

In the Devonian and Carboniferous, the first tetrapods to evolve limbs from fishy fins were broadly amphibious in habits, requiring water to feed and breed, as in modern amphibians such as frogs and salamanders. Coupled with waterproof skin, more agile skeletons, proper biting and holding jaws, the amniotic egg was a ‘private pond’ in which the developing reptile was protected from drying out in the prevalent warm climates, and enabled the Amniota to move away from the waterside and dominate terrestrial ecosystems.

This has been the textbook view ever since, despite dissenting voices. Many biologists had surveyed the diversity of reproductive modes across living amniotes and noted that lizards and snakes can switch between viviparity (bearing live young) and oviparity (laying eggs). Sometimes, different species within a single genus show both behaviours, and genomic studies show in fact that reversals from viviparity to oviparity can occur, and indeed do occur among the Squamata (= lizards and snakes). In explorations of the ancestral state among squamates, it has been debated whether their first representatives laid eggs with hard, mineralised shells, eggs with unmineralised (‘parchment-type) shells, or even bore live young.

The amniotic egg had seemed such an asset of the Amniota (= reptiles + birds + mammals) that it would be crazy, in evolutionary terms, if they ever abandoned it, or at least had not started out with it. The amniotic egg of reptiles, birds, and mammals shares key features of the external shell, the various extraembryonic membranes that provide protection for the developing embryo and collect waste fluids, as well as the yolk sac. But palaeontologists have noted over the years how many extinct amniote groups produced live young – this includes all the marine reptiles of the Mesozoic (ichthyosaurs, plesiosaurs), as well as possibly the marine mesosaurs of the Permian.

Then, we discovered that the aquatic choristoderes of the Mesozoic also showed examples of both egg-layers and viviparous parents. Lead author Baoyu Jiang of Nanjing University showed that some species of choristoderes did both, mimicking the squamates apparently in being able to flip-flop between both reproduction modes. Then, Mark Norell of the American Museum of Natural History showed in 2020 that even the dinosaurs originated as forms either laying parchment-shelled eggs or bearing live young. This shocking result was partly based on the absence of evidence: before the Early Jurassic, about 190 million years ago (Ma), palaeontologists find no tetrapod eggs of any kind, and after that point they occur in abundance. Hard-shelled eggs are fossilizable, and their absence is suggestive, but not conclusive. Norell and colleagues though demonstrated the case through an ancestral states analysis, a technique used to infer the nature of ancestors based on our understanding of the evolutionary relationships of a set of species and the distribution of traits among them. By analysing egg types across dinosaurs and close relatives, Norell and colleagues pointed to late evolution of the hard, mineralised shell.

We extended the study over 51 fossil species and 29 living species to explore the likely ancestral states in each major clade, and then all the way down to the origin of amniotes 320 Ma in the Carboniferous. We also implemented cutting edge evolutionary models in our ancestral states analysis that more adequately capture the evolutionary process underpinning complex traits, such as the amniotic egg. Among the living clades, birds, crocodilians, and turtles show hard-shelled eggs at their origins, whereas mammals, squamates, and extinct marine reptiles show viviparity at their roots. The so-called hard-shelled egg of the living monotremes (platypus, echidna) is in fact only lightly mineralised and not necessarily typical of the Mesozoic mammals, some or most of which might have been viviparous. The new discovery about choristoderes, as well as a recently published example of viviparity in a Triassic archosaur points to viviparity at the root of the clade Archosauria (including extinct forms, dinosaurs, pterosaurs, birds, crocodilians).

Therefore, the three great branches of Amniota, namely Mammalia, Lepidosauria (Squamata and relatives), and Archosauria show viviparity and extended embryo retention at their roots, so pointing to these conditions in the first amniotes. Extended embryo retention (EER) is widespread among vertebrates and refers to cases where the developing young are retained by the mother for a lesser or greater span of time. In Squamata, EER is common and variable, and the young can be released, either inside an egg or as little wrigglers, at different developmental stages, and there appear to be ecological reasons for EER in these cases, relating to temperature and food supply.

So, our work, and that of many others in recent years, has consigned the classic ‘reptile egg’ model of Romer and the textbooks to the wastebasket. The first amniotes had evolved EER as a means to protect the developing embryo for a lesser or greater amount of time inside the mother, so birth could be delayed until environments become favourable. Whether the first amniote babies were born in parchment eggs or as live, snapping little insect-eaters is unknown, but this adaptive parental protection gave them the advantage over spawning earlier tetrapods.