Normally the dinosaurian world is rocked by a new fossil -- the biggest, fastest, or toothiest. But the latest dinosaur research threatens to change our understanding of how dinosaurs evolved at a much deeper level, and blow aside 130 years of agreement on the topic.
A new paper published in the journal Nature suggests that scientists need to reorganise the major groups used to classify dinosaurs. This means we may have to revisit what we think we know about the first dinosaurs, what features they evolved first, and where in the world they came from.
The way we classify dinosaurs goes back to the 19th century. In 1887, Harry Govier Seeley, a classic, hard-working Victorian palaeontologist, divided dinosaurs into two major suborders based primarily on their hip structure. Saurischia comprises the flesh-eating theropods such as Tyrannosaurus and the ponderous, long-necked sauropodomorphs such as Diplodocus. Ornithischia comprises all the rest, including the two-legged Iguanodon, and the armoured, four-legged Stegosaurus, Triceratops, and Ankylosaurus.
This ordering of dinosaurs has stood the test of time for 130 years, weathering the onslaught of cladistics in the 1980s, when palaeontologists began using computers to analyse and categorise groups of animals based on features that pointed to a common ancestor. There are now thousands of diagrams (cladograms) of dinosaur subgroups, and ever-growing data matrices, that closely document the anatomical features of each species.
The new paper completely disrupts the consensus over Seeley's categories. The researchers ran a cladistic analysis of 457 characteristics across 74 species (that is a data matrix of 33,818 bits of information recorded from skeletons). They concluded that, based on 21 unique characteristics of the fossils, the theropods were more closely related to the Ornithischia group and should be moved into that category. This would create a new group named Ornithoscelida and leave behind the Sauropodomorpha.
The trick in cladistics is to find a unique anatomical feature that evolved at a specific time and can indicate a particular subgroup. For example, Seeley noted that the hip bones of ornithischians were arranged with pubis and ischium running backwards (superficially, like modern-day birds). Meanwhile, the hip bones of saurischians (including theropods) matched other reptiles, with pubis forwards and ischium back.
This suggests the two groups split from a common ancestor and evolved different hip shapes. This was a massive anatomical change or novelty, and palaeontologists until now have assumed that it happened only once in evolutionary history. Grouping the theropods with the ornithischians suggests that the hip change occurred later and raises the question of whether some early theropods had this feature.
The researchers also suggest that the new analysis can reset our understanding of where dinosaurs originated and what their diet was. The classic view was that the first dinosaur was a carnivore living in what is now South America. The new analysis makes this more of an open question and suggests they might have evolved as omnivores in the northern hemisphere.
Tree of life
None of this changes what we know for sure about what dinosaurs evolved which traits and when. But the key point is that accurately depicting the tree of life matters. If you care about modern biodiversity, it's important that all species are not equal. Some are more distinctive than others, possessing more unique features, and having a longer independent history. Working this out requires an accurate tree.
On a broader scale, getting the tree right affects our calculations of rates of trait evolution, extinction and post-extinction recoveries. We will never find the very first dinosaur but we can establish some things about it by estimating the ancestral states of different species from a correct tree.
We invest enormous efforts into constructing testable systems for categorising different species, and their size is increasing as computing power grows. When I ran my first cladogram in 1982, I had to use punch cards on a mainframe computer, and I could include only ten or 12 species and 50 or so characteristics.
Today, I was able to run all the data for this new paper through my desktop computer and get an answer in 33.21 seconds, while writing this article at the same time. Recent publications have sported trees of all 10,000 species of birds, and even summary trees of all life. The dream is to run such trees with all 1.5m named species, using data about both genes and physical shape.
Is this new paper the true answer for the evolutionary origins of dinosaurs? The data we have is riddled with question marks, and so the algorithms still struggle to calculate the one true tree. This is no criticism of the researchers, just a statement of the practicalities. We don't know yet whether we can see the wood for the trees.