Skip to main content

Dinosaurs in decline tens of millions of years before their final extinction - new paper in PNAS

There is no dispute that non-avian dinosaurs went extinct at the Cretaceous-Paleogene (K-Pg) boundary, most likely owing to a large asteroid hitting the Earth, but what has been debated for decades is whether dinosaurs were reigning strong up to the end of the Cretaceous, or whether they were already in decline and were on their way out only to be killed off by the asteroid impact 66 million years ago (Ma).

That is the question that Mike Benton, Chris Venditti and I hopefully helped resolve with our new paper that came out electronically Monday in PNAS.

The paper is pretty straightforward, and we provided lots of details in the supplementary information, so it shouldn't be a difficult read. Please do have a read. Below I provide a brief summary.

1. Previous studies
The majority of previous studies dealt with counting the number of dinosaur species in geologically defined time bins (such as geological stages), charting the resulting diversity curve through time, and making evolutionary inferences based on the peaks and troughs in the curves and what sort of Earth history events they correspond with. With this approach, there were two camps: those that argued that dinosaur diversity remained stable if not increased towards the end of the Cretaceous; and those that argued that dinosaur diversity was in decline at least for the last 10 million years before their final extinction 66 Ma. More recently, sampling corrected approaches came to dominate the literature and the emerging picture, while complicated seemed to favour an interpretation for near-stasis in dinosaur diversity through time, or that dinosaurs were not in any noticeable long-term decline.

However, the problems we see with such previous approaches (regardless of sampling correction or not) are that:
  1. They are measuring standing diversity, which is a product of evolutionary dynamics, and do not study the dynamics themselves, e.g. speciation and extinction.
  2. They are summary statistics at the time bin level (spanning millions of years), and mask vital evolutionary signals

2. What we did
We looked at speciation dynamics through time. To do that we counted the number of speciation events along each phylogenetic path using a phylogeny of dinosaurs (Benson et al. 2014) and modelled its relationship with phylogenetic path length, or time elapsed since the origin of Dinosauria. We took a phylogenetic mixed modelling approach in a Bayesian framework using MCMCglmm. This is basically a regression analysis but taking phylogenetic non-independence (the effects of shared ancestry) into account - there is a whole literature on why it's necessary to account for shared ancestry, the most prominent still being Harvey and Pagel (1991). Modelling the relationship between numbers of speciation events along a phylogenetic path and its length in time basically equates to modelling "diversification" (or more descriptively, net speciation) through time.
In our paper, we present three statistical expectations of speciation through time (Fig 1): A) a linear increase; B) a slowdown (saturation) curve; and C) a downturn curve.
Fig 1. Theoretical models of speciation through time
A linear increase in speciation through time (Fig. 1A) is observed when speciation rate is constant and above extinction rate. This model is the null model (the expectation). A slowdown in speciation through time (Fig. 1B) has been observed widely in extant phylogenies (Moen and Morlon, 2014), so this is a sensible alternative model. There are lots of explanations for why speciation slowdowns occur in phylogenies, but the real reason still remains largely unknown. A third option, which can only be tested using phylogenies with extinct tips, is when speciation increases towards an asymptote but does not saturate (as in the slowdown curve) and starts to decline (Fig. 1C).

We then appropriately modelled the different models of speciation through time by varying the model formulae in the phylogenetic mixed models.

Since we use an approach based on regression analysis, we can include additional predictor variables to the model. This means that we can statistically test whether external factors, such as environment, ecology, body size and sampling bias can explain the patterns of speciation through time. Environment and ecology are straightforward as to why we'd want to test for their effects. Body size and sampling bias can be regarded as potentially confounding factors, or external forces that might be biasing our results. Dinosaurs (especially bird-line theropods) show a propensity for decreased body size through time, ultimately leading to birds. Small body size in turn has been argued previously as a source of fossilisation bias, that smaller species don't get preserved as fossils as much as larger species. We wanted to test whether body size can explain patterns of speciation through time.

Sampling bias is a trickier business and a lot has been discussed in the literature. We wanted to make sure that our results were not biased by missing data or underrepresentation of the true diversity of dinosaurs. I think I'll save the details of sampling biases and my thoughts on it for another post, but here the only thing that is important is that we used various measures of sampling to test whether sampling bias can explain patterns of speciation through time. For measures of sampling, we used the number of fossiliferous formations for each geological stage, the number of fossiliferous formations a given species in known from, the number of fossil collections a given species is represented in, the number of valid known species in a given subclade (there are more species named than are represented in our phylogeny, so this represents known underrepresentation), and fossil quality scores (the state of preservation; species poorly preserved as fossils may have close relatives that were not preserved).

3. What we found
We found that the downturn model was the best model (Fig. 1 C) to explain the data in dinosaurs as a whole (Fig. 2A), but even better when the three major clades Sauropodomorpha (blue), Theropoda (red) and Ornithischia (green) were allowed to have separate evolutionary dynamics (Fig. 2B). Ornithischia was further subdivided into, Hadrosaurifomes (light green), Ceratopsidae (light blue) and other ornithischians, owing to the observations that the former two subclades showed different patterns in the data to all other dinosaurs.
Fig 2. Speciation through time in dinosaurs. A, The linear model (orange) compared to a curved model (dark grey) in Dinosauria as a whole. The curved model is significantly better than the linear model, implying that speciation rate slowed down and eventually surpassed by extinction rate at ~24 Ma before the K-Pg boundary. B, The model is significantly improved when different clades are allowed to have different evolutionary dynamics. The downturn is exemplified in sauropods (blue), theropods (red) and non-hadrosauriform, non-ceratopsid ornithischians (green). Hadrosauriforms (light green) and ceratopsids (light blue) do not show any signs of slowdowns in speciation rate. Translucent lines represent uncertainties in the model parameter estimates.
The 5-Group model (Fig. 2B) shows that the downturn (the point in time when extinction rate surpassed speciation rate) occurred much earlier (~ 50 Ma before the K-Pg boundary 66 Ma) than estimated in the Dinosauria model (~24 Ma). The scale of the downturn itself is exemplified in the 5-Group model compared to the Dinosauria model. This discrepancy most likely owes to the non-homogenous dynamics amongst dinosaur groups, e.g. linear increases in speciations in Hadrosauriformes and Ceratopsidae.

The downturn is statistically significant as the posterior distribution of net speciations per Myr is clearly negative with none crossing over zero (Fig. 3B, dashed horizontal), or when speciation = extinction. This is strong evidence that the speciation curve does indeed downturn rather than slowing down towards an asymptote, in which case the posterior of net speciation per Myr would overlap considerably with zero.

This means that dinosaurs were going extinct much faster than they could replace with new species (Fig. 3).
Fig. 3. Net speciation per millions of years (Myr). A, net speciation per Myr is represented as a colour gradient on branches of a phylogenetic tree. Branches with warm colours have higher net speciation per Myr while darker colours have lower net speciation per Myr, with the darkest ones having negative values (extinction > speciation). Most parts of the tree show higher speciation early in evolutionary history, except for Hadrosauriformes (light green) and Ceratopsids (light blue) in which net speciation per Myr gets higher with time. B, net speciation per Myr for the five groups of dinosaurs, Ceratopsidae (light blue), Hadrosauriformes (light green), non-hadrosauriform, non-ceratopsid ornithischians (green), Sauropodomorpha (blue), Theropoda (red). Dashed horizontal is where net speciation per Myr is zero (speciation = extinction) and any value below this line represents extinction > speciation.
We also found that sea level had a significant (but small) effect on speciation; an increase in sea level was associated with an increase in speciation. This is consistent with what Horner et al. (1992) argued about the role of marine transgressions facilitating the appearance of transitional species (or speciation in process) - though most of their transitional species were subsequently described as distinct species... However, sea level did not explain the downturn in speciation.

Ecological niche saturation (available niches get filled up to maximum) is a common explanation for speciation slowdowns (diversity-dependence or density-dependence [see Rabosky, 2013]). The idea is that as a clade expands, member species fill up available ecological niches until there are none left, leading to a slowdown in net speciation through time. We find no evidence of this, however. The number of potential competitors at clade level is not significant as a variable and does not explain the downturn.

Similarly, body size does not explain speciation dynamics or the slowdown-downturn. This is not surprising as in order for the downturn through time to be solely explained as a consequence of increasingly poor fossil preservation owing to smaller body sizes, there must be a systematic decrease in body size towards the present, which is not what we find.

Out of the many sampling metrics we tested, only valid species count was a significant predictor variable. This means that the degree to which the phylogeny is under-representing known species is an important factor to account for. However, inclusion of this variable does not change the pattern of a slowdown-downturn in dinosaur speciation; the downturn is not because of known under-sampling in the phylogeny.

4. How we interpret the results
First we have to bear in mind that we are talking about numbers of speciation events through time (or net speciation rates = diversification rates), and not about observed apparent numbers of species in any given time interval (e.g. geological stages as most previous studies have measured). This means that we can look at the processes that led to apparent diversity through time. However, this also means that direct comparisons with previous studies in terms of looking at the shapes of the curves may not be informative. Remember, standing diversity (observed number of species, corrected or not) and speciation rates are two distinctly different things.

With that being said, there are a few interesting things to note here.

First, all three major dinosaur groups have high rates of speciation early in their evolutionary histories, mostly in the Late Triassic to Early Jurassic at the latest. Dinosaur clades expanded quickly during that time. This is concordant with the fossil record in that we see similar/closely related Late Triassic to Early Jurassic dinosaurs almost globally. For instance early prosauropodomorphs like Plateosaurus, Unaysaurus and Euskelosaurus or theropods like Coelophysis are known from multiple continents but look very similar, e.g. Coelophysis bauri from US and Coelophysis rhodesiensis from Africa are now considered congeneric.

Second, the initially rapid clade expansions do not carry on but start gradually slowing down from the middle of the Jurassic (Fig. 3B) towards a peak in speciation some time in the Early Cretaceous (Fig. 2B) at which point speciation rate is surpassed by extinction rate and net speciation per Myr goes into the negative (Fig. 3B). That a slowdown in speciation occurs is not that surprising but that it started so early in dinosaur evolution, is kind of surprising. The conventional and most common explanation for a slowdown is that as a clade expands, available ecological niches fill up, and it becomes progressively more difficult for new species to appear owing to the ever reducing number of open niches. The resulting speciation curve is that of Fig 1B, a saturation curve. Speciation slowdown occurred early in dinosaur history, and this would otherwise fit well with an Early Burst model of speciation.

However, dinosaurs reach an asymptote and shift over to an actual decline (extinction > speciation), so a simple early burst model (homogenous process across the entire tree) is not applicable for dinosaurs. The downturns observed in the three major clades (minus Hadrosauriformes and Ceratopsidae) are robust to various biases that we could think up of, including the only bias metric that was statistically significant, the known degree of phylogenetic underrepresentation (valid taxon count), which is a new measure that I came up with. We thought we'd done a pretty thorough job of dealing with sampling bias (but not so according to an angry commenter on the AMA we did on Reddit. What's all this about a red flag?), so we are pretty confident that we are not observing some artefact.

That dinosaurs show a prolonged period of downturn in speciation through time, means that they were going extinct faster than they can generate new species, making this the strongest evidence yet for a long-term decline in dinosaurs before the K-Pg boundary.

What we could not determine with any level of certainty is what caused the decline (we discuss this in the paper though). The following is pretty speculative but my personally preferred interpretation has to do with the supercontinents breaking up into more or less the modern configurations with vast bodies of water separating them, leaving less opportunities for dinosaurs to migrate to new corners of the world, eventually leading to new species. I would have loved to have accurate geographical range data but as with everything palaeo, the data is not that great - a lot of taxa are only known from a single locality so that's just one set of palaeo-coordinates.

Another interesting thing we see from our results is the rise of hadrosauriformes and ceratopsids. This result 'seemingly conflicts' with previous studies by Brusatte et al. (2012; 2015) in which hadrosaurs and ceratopsids were found to be decreasing in species and morphological diversity. However, I only said 'seemingly conflicts' because as I've mentioned above, we are looking at speciation dynamics not standing diversity (species count or morphological variation), so these two results can coexist. The obvious interpretation is that while speciation rates are high, these clades have low numbers of species at any given time interval (or that speciation turnaround is also quite high). Hadrosaurs and ceratopsids are known to be distinct and limited in taxonomic overlap across geography and through time, meaning that speciation can occur horizontally by geographic segregation and vertically by temporal separation, but at each time interval standing diversity is low. This goes for morphological diversity as well. Hadrosaurs and ceratopsids are variations on the same theme - speciation can readily occur with very minor differences overall to the anatomy except for notable differences in cranial ornamentations.

One potential reason for the successes of these two subclades while other clades were in demise, is the acquisition of 'key' morphofunctional features, namely the continuously growing dental batteries in both groups, but additionally, pleurokinesis in hadrosauriforms. Such functional innovations made them capable of efficiently processing food items, allowing them to consume vast quantities of vegetation.

That Mesozoic birds were also in decline is another surprising result, given that modern birds are extremely successful and speciose with roughly 10,000 species. Having said that, a recent paper suggests that seed-eating allowed birds to survive the K-Pg mass extinction (Larson et al. 2016), and this may provide further insights into how a clade in demise can make a recovery - first they need to survive the extinction event, but they also need to survive the post-apocalyptic world (what if seeds were the only edible things left!).

5. What we can take home from this
Hopefully our study demonstrates the importance of studying the processes (speciation dynamics) rather than the outcomes (numbers of species), which is only really possible by using phylogenetic information. Statistically accounting for phylogenetic non-independence is also crucial. By doing both through the use of PCM, we can start seeing evolutionary dynamics that we could not previously observe, allowing us to get further insights into how Mesozoic dinosaurs went extinct.

6. Why should we care how dinosaurs went extinct?
We found that dinosaurs were vulnerable to mass extinction when drastic environmental change (such as those caused by the asteroid impact) occurs, because they were losing the ability to replace extinct species with new ones. This implies that any group of animals that are under prolonged periods of high extinction rate can undergo mass extinction should there be a catastrophic event. We live in a time when species are undergoing unprecedented levels of extinctions. This means that if some major catastrophe hits, then it is highly possible that whole groups of animals be completely wiped out off the face of the Earth. Such catastrophes could be anything, from another asteroid hitting Earth to simply allowing climate change to progress at its current rate triggering cascades of environmental disasters. One thing we need to distinguish is the difference between continued species loss (what we're seeing now, and what dinosaurs went through before the K-Pg boundary), and mass extinctions when whole clades go extinct (like what happened to non-avian dinosaurs, pterosaurs, plesiosaurs, ammonites etc at the K-Pg boundary; imagine ). I don't know about you, but I'd like to live in, and leave my children, a world with cephalopods, birds, mammals, and a whole load of other groups of organisms living in it.

7. In the media
This paper surprisingly got a lot of media attention. PNAS released a press release but both Universities of Reading and Bristol coordinated their respective press packages.

We worked really closely with Reading Uni's Press Office and even made a short promo video:

Here are a couple of cool summaries of our paper:

Quartz (made by Michael Tabb)

BBC Radio 4 (video by BBC News Illustrated, with Pallab Ghosh reporting)


Popular posts from this blog

R for beginners and intermediate users 3: plotting with colours

For my third post on my R tutorials for beginners and intermediate users, I shall finally touch on the subject matter that prompted me to start these tutorials - plotting with group structures in colour.

If you are familiar with R, then you may have noticed that assigning group structure is not all that straightforward. You can have a dataset that may have a column specifically for group structure such as this:

B0 B1 B2 Family
Acrocanthosaurus 0.308 -0.00329 3.28E-05 Allosauroidea
Allosaurus 0.302 -0.00285 2.04E-05 Allosauroidea
Archaeopteryx 0.142 -0.000871 2.98E-06 Aves
Bambiraptor 0.182 -0.00161 1.10E-05 Dromaeosauridae
Baryonychid 0.189 -0.00238 2.20E-05 Basal_Tetanurae
Carcharodontosaurus 0.369 -0.00502 5.82E-05 Allosauroidea
Carnotaurus 0.312 -0.00324 2.94E-05 Neoceratosauria
Ceratosaurus 0.377 -0.00522 6.07E-05 Neoceratosauria
Citipati 0.278 -0.00119 5.08E-06 Oviraptorosauria

Top 10 scientifically important theropod dinosaurs of all time (off the top of my head)

I thought I'd do a fun post for once. And since list based articles are the norm for fun on the internet, I thought I'd do one on dinosaurs, but given that I know most about theropods, I've decided to restrict my list to theropods (...maybe in a future post, I'll do other clades).

My ranking is based mostly on scientific importance so it may not reflect awesomeness, and it is obviously subjective as to how I rank importance to science. For instance, interesting discoveries or unique palaeobiology are ranked relatively low compared to wealth of information and data or completely revolutionising our understanding of the evolution of theropods.

So here are my top 10 scientifically important theropod dinosaurs of all time (off the top of my head)

10. Megalosaurus

Being the first dinosaur to be named, Megalosaurus automatically deserves a spot on this list, but given the fragmentary nature of known fossil specimens, and being mostly useless as a meaningful source for biologi…

The difference between Lion and Tiger skulls

A quick divergence from my usual dinosaurs, and I shall talk about big cats today. This is because to my greatest delight, I had discovered today a wonderful book. It is called The Felidæ of Rancho La Brea (Merriam and Stock 1932, Carnegie Institution of Washington publication, no. 422). As the title suggests it goes into details of felids from the Rancho La Brea, in particular Smilodon californicus (probably synonymous with S. fatalis), but also the American Cave Lion, Panthera atrox. The book is full of detailed descriptions, numerous measurements and beautiful figures. However, what really got me excited was, in their description and comparative anatomy of P. atrox, Merriam and Stock (1932) provide identification criteria for the Lion and Tiger, a translation of the one devised by the French palaeontologist Marcelin Boule in 1906. I have forever been looking for a set of rules for identifying lions and tigers and ultimately had to come up with a set of my own with a lot of help fro…