24 Responses to “Is everything we know about sauropod phylogeny nonsense?”

1. Valentin Fischer Says:

   
   October 31, 2017 at 1:01 pm

   Hi Matt and Mike,

   The multiplication of morphological characters might be able to the break the ecological bias in the phylogenetic relationships of extinct animals. Ecomorphological convergence can be pervasive, though but should have its limits. We somehow tested this in short-necked plesiosaurs earlier this year. Author postprint is accessible here: http://orbi.ulg.ac.be/handle/2268/211092

   TL;DR: convergence between two distinct clades of plesiosaurs is seen in both a cladistic and an ecomorphological datasets but the amount of convergence is smaller in the cladistic dataset and does not seem to perturb the signal too much (provided we recovered the true phyletic signal, obviously).

   We had the same issue with Cretaceous ichthyosaurs: what people were calling ‘Platypterygius’ is probably a collection of distinct ophthalmosaurid lineages converging to a niche of large predators. Here also, the multiplication of OTUs and characters broke this apparent clustering.

   Valentin
2. Andrea Cau Says:

   
   October 31, 2017 at 1:01 pm

   If you suspect that sauropod phylogeny is biased by ecomorphotypes, just run partition analyses, test removal of ecologically-related features, try different character weights and so on.
   This could be done even with actually published matrices.

   Also, even if taxa may converge on similar morphologies, this is clearly solved re-defining characters in a more accurate way: ecotypes result as distinct combinations of character states, and cannot form clades.
3. Anne Says:

   
   October 31, 2017 at 2:02 pm

   I keep encountering this idea that supplementary information in Nature journals is not reviewed. Does this mean that I should stop reviewing it? Because frankly, it takes 10x as long as the paper itself…
   (Of course, I did not review this particular paper. But come on.)
4. Mike Taylor Says:

   
   October 31, 2017 at 2:11 pm

   Anne, I have no idea whether it’s reviewed or not. If in fact you did review it, that’s great. But it seems ridiculous to me that in an online-only journal that needs no arbitrary length limits, big chunks of a paper get scooped out and shoved into an appendix of indeterminate status. That material all belongs in the paper — which is exactly where it would be in a sensible journal.
5. Abdré Fonseca Says:

   
   October 31, 2017 at 2:28 pm

   Hi guys,

   If we ever get to a day where we can perform molecular analysis on dinosaurs I’m sure lots things will change, just like it happens to modern animals (whales inside roofed mammals).
   But till there I think we should not “torture” ourselves with the idea that many of our groups that are similar in morphology (Brachiosauridae, Dicraeosauridae) are unnatural cause for the majority of species we just have bones and comparing they’re morphology is the only thing we can do to try knowing more about them.
   And about Morrison taxa, I think someone one day should do like Tschopp did in 2015 with diplodocidae but this time with Macronarians I’m sure a lot of new thing would came out of all that Camarasaurus material.
6. Mike Taylor Says:

   
   October 31, 2017 at 2:38 pm

   Abdré, I am sympathetic to the idea that there’s maybe not much point torturing ourselves about things that, for now at least, we maybe can’t know. At the same time, many of our ideas about sauropod evolution are dependent on the phylogeny being correct at least in broad brush-strokes. All this makes me uncomfortable.

   The good news is that, as far as I know, Emanuel Tschopp himself is doing a detailed specimen-level analysis of the big complex of things that we call “Camarasaurus“, so it’ll be interesting to see what that shows us.
7. Daniel Vidal Says:

   
   October 31, 2017 at 2:52 pm

   Well,Titanosaurs and Diplodocids used to be grouped together based on the skulls known in the past and tooth morphology. Same goes with Euhelopus and the mamenchisaurs. Same with Camarasaurus/Morosaurus, which would be grouped with Apatosaurs/Brontosaurs rather than with Brachiosaurus.

   I think it is a matter of sampling (the more specimens you get, and the more complete, the clearer the picture gets), and yes, molecular data might be a great addition to morphology, but, again, molecular data is not 100% reliable (just as morphology ain’t either).

   It is a problem we face in historical sciences: to piece events from what we have left of them in the present. I think I’m with Matt: we will know better as time passes and we get more (maybe even… better?) evidence.
8. Anonymous Says:

   
   October 31, 2017 at 5:14 pm

   If it makes you feel any better, the situation with mammals isn’t quite as bad as you make it out to be. The aforementioned groups of mammals are easily distinguished from each other, even if they do share similarities. What was the primary issue in mammalogy was the relationships of larger clades, i.e., in the case of moles it was clear that talpids, golden moles, and marsupial moles were three separate things, the question was whether talpids and golden moles were part of a larger insectivore clade (even then, it seems as though golden moles were seen as closer related to tenrecs and talpids were closer related to soricids rather than one clade of “moleyness”), and few thought marsupial moles were related.

   And even then, there are major differences in the skeleton. Talpid forelimbs look nothing like golden mole forelimbs. The flesh-and-blood appearance is more deceiving than the bones. Same goes for thylacines and canids, etc.

   So the comparable issue wouldn’t be “is Diplodocidae a polyphyletic grab-bag of taxa that belong to six or eight different groups”, but “is Diplodocoidea monophyletic or is it possible that, say, rebbachisaurs, are really related to dicraeosaurs and diplodocids or are they related to turiasaurs or within Titanosauriformes”.
9. Matt Wedel Says:

   
   October 31, 2017 at 5:19 pm

   Thanks, all, for the discussion.

   Andrea wrote:
   If you suspect that sauropod phylogeny is biased by ecomorphotypes, just run partition analyses, test removal of ecologically-related features, try different character weights and so on.

   Two potential problems with that. One, I’m not confident that we know enough about sauropod functional morphology or ecology to know what the ecologically-related features are. For example, did diplodocids rear to feed? If so, there’s a shedload of limb and vertebra characters that are probably part of that functional complex. If not, those characters might still be functionally interrelated, but for doing something we don’t understand yet.

   Second – and I am airing my ignorance here – if one did partition analyses on the Caribbean anoles, would it be possible to remove all of the ecologically-related features? Or has the morphology been so overwritten by ecomorphological factors that it’s impossible to recover a non-eco-driven signal? The level of convergence we see in some places does not give me great hope.

   To be clear, I don’t think everything we know about sauropod phylogeny is nonsense (had it been my post, I probably would have chosen a less inflammatory title), and possibly none of it is nonsense. But given how much we’ve learned about molecules and morphology in the past 2-3 decades, I’d be shocked if dinosaurs were the first big, diverse vertebrate clade for which molecules didn’t force any surprising rearrangements.

   Sauropods seem particularly ripe for this, since their limbs are so simple and their vertebrae, while complex, seem to evolve the same features over and over again in different clades. Maybe there are many paths to sauropod-hood, but only one destination. The fact that sauropods never evolved into anything other than large*, columnar-limbed quadrupeds with long tails and long necks**, despite 140 million years of attempts***, suggests that their bauplan was some sort of local maximum, with less viable morphologies in all directions. Which makes me wonder if the sauropod form was the evolutionary morphological equivalent of a roach motel: things got in, but never got back out. (I explored this idea a bit in the Fistfull of Podcasts interview.)

   (Alternatively, maybe I’m waaay wrong and sauropods did evolve into other things, and we just haven’t found those transitional fossils. There had to be at least one viable route from non-sauropod-hood to sauropod-hood so they could evolve in the first place. Maybe that was a two-way street.)

   * Yeah, I know, island dwarfs and Saltasaurus – still pretty darned big by terrestrial vertebrate standards.

   ** Yeah, I know, Brachytrachelopan – still pretty long-necked by any non-sauropod, non-bird standard.

   *** It is worth remembering that organisms vary all the time, in lots of ways. So if we don’t see those variants becoming fixed in populations, it’s probably because they’re not viable. Yeah, there are some things that are developmentally forbidden, but shorter tails or more gracile limbs are not among them.
10. Matt Wedel Says:

    
    October 31, 2017 at 5:29 pm

    Thanks, Anonymous. Our comments passed in the ether, but you raise some very good points.

    If it makes you feel any better, the situation with mammals isn’t quite as bad as you make it out to be.

    Word. I was thinking more about the anoles, the moles were just more grist for the mill.

    So the comparable issue wouldn’t be “is Diplodocidae a polyphyletic grab-bag of taxa that belong to six or eight different groups”, but “is Diplodocoidea monophyletic or is it possible that, say, rebbachisaurs, are really related to dicraeosaurs and diplodocids or are they related to turiasaurs or within Titanosauriformes”.

    Yeah, I think that is absolutely the right level of concern. Like, I’m pretty sure that Apatosaurus, Diplodocus, Barosaurus, Tornieria, etc. are a real clade. They’re so weird compared to the sauropod baseline, and so limited in time and space, that it seems impossible for them not to be. But I wonder if the, say, Diplodocinae-level sauropod clades are like Tetris blocks, and we’ll find different ways to stack them in the future.

    Brachiosaurs worry me. Brachiosaurus and Giraffatitan differ enough in weird, seemingly trivial ways that aren’t obviously connected to ecology, that I wonder if they achieved brachiosaur-hood independently.

    To be perfectly clear, I don’t think these heretical notions are currently better supported than the consensus phylogeny. But it doesn’t seem impossible that they might become so in the future, given more lines of evidence (not necessarily molecular). But as Dan Simberloff pointed out back in the early 80s, when scientists are wrong it is often because we fail to imagine a large enough universe of possibilities. It seems healthy to push ourselves to imagine what it would take to upend current phylogenies – without denying their usefulness right now, or ceasing to make progress by using them.
11. mborths Says:

    
    October 31, 2017 at 5:36 pm

    Let’s not sell our morphological analyses too short here. I will continually be amazed that molecular phylogenetics didn’t completely blow up all the detailed anatomical systematic work performed before we could sequence living critters. Lepidosaurus didn’t have to be monophyletic, and flight could have evolved multiple times in mammals leading to a polyphyletic Chiroptera.

    While wolves and thylacines look similar, no one ever thought they were actually part of the same group, even when there were just skins and skulls to work with. Sure Afrotheria was surprising by putting together African “insectivores” and paenungulates, but morphologists figured out hyraxes and elephants were in the same group without molecular evidence, and everyone was pretty confused by insectivores anyway. The older the split, the more cautious we should be of the sister-taxon relationships, but that applies to both morphological and molecular datasets.

    It’s kind of amazing a basic anatomical observation like: “those things have odd numbers of toes and those things have even numbers of toes; they must be in different groups” or “those things make a mouth from the blastopore and those things make an anus from the blastopore; they must be different groups” actually held up. There are are plenty of examples of congruence between morphology and molecules. We just like a good puzzle and focus on the areas of the tree where the two are in conflict. We should be skeptical of results, but that’s the scientific project. We should also be constantly be revising the characters used in morphological analyses to reflect new insights from developmental and physiological studies. But I think molecular studies show it’s not all a waste of energy :)
12. Jason Silviria Says:

    
    November 1, 2017 at 1:58 am

    The Losos et al. paper on anoles got the morphological phylogeny wrong because it was phenetic; it was based on overall superficial similarity of (soft body) anatomy, rather than looking at the details of muscular and osteological characters (and distinguishing different expression of the same character state from different character states or even whole characters). As other commenters have noted , mammalogy learned this lesson the hard way, but eventually the morphological and genetic data agreed with each other, and we now have a whole list of synapomorphies for Afrotheria, Laurasiatheria, etc., most of which are cranial and postcranial rather than dental. (Teeth kind of suck when it comes to analyses that require high bootstrap values, which is unfortunate since the majority of the early mammalian fossil record is jaws and teeth.)

    I don’t think you should worry about the major well-defined sauropod clades being polyphyletic. While the quest to clean up the Jurassic-Cretaceous diplodicoid and macronarian radiations is in its infancy, at least the relevant material is almost all postcrania, and much of that at least preserves the important diagnostic features. Certainly “Camarasauridae” and “Brachiosauridae” actually harbor several distinct clades, as has been repeated demonstrated, but we still have Camarasauridae and Brachiosauridae sense stricto. And Diplodocidae and Dicraeosauridae being polyphyletic? That’s seem a little far fetched.

    Also important to note is that there is a huge amount of data on dinosaur faunas from eastern North America, west Africa, Australia, and Antarctica that is currently missing. If anything, surprises like the turiasaurs, Tornieria, and Suuwassea demonstrate that western North America does not have all the answers. Remember, Pangea was still a big continent that was just starting to break up…
13. John D'Angelo Says:

    
    November 1, 2017 at 4:49 am

    Large-scale morphological phylogenetic analyses were still pretty new when molecular data came onto the scene as the main way of resolving the relationships of extant organisms. One might consider that the real lesson is that morphology-based analyses aren’t good enough yet, not that morphology-based analyses are futile. For example, in mammal phylogeny, teeth make up a disproportionately large portion of the morphological phylogenetic data. Maybe ten years from now we’ll have accumulated enough data from the rest of the skeleton and soft tissue that the phylogenetic signal will begin to reflect the true topology better.

    What’s more is that, if reading anything Mickey Mortimer has written has taught me anything, it’s that a lot of the variation in phylogenetic results is the failure of scientists to create adequate phylogenetic analyses, not the data itself being uninformative. Maybe if the average quality of phylogenetic analyses increases (which is related to the supplementary material peer review issue), the inconsistency between morphological and molecular data will decrease, and so will our confidence in analyses of taxa with only morphological data available.

    When it comes to sauropods, some sauropod clades are just too well-supported to fall apart. Flagellicaudata and Rebbachisauridae are probably among them. But really, “everything we know about sauropod phylogeny” isn’t all that much. Pretty much every possible arrangement of Brachiosaurus, Euhelopus, Andesaurus, and Saltasaurus exists in some recent paper or other. Most non-neosauropod sauropods are under-examined in a phylogenetic concept, and of course there is still nearly no resolution within Titanosauria. There’s a whole world of possibilities when it comes into sauropod phylogeny without even considering the reliability of morphological data.
14. Andrea Cau Says:

    
    November 1, 2017 at 6:07 am

    Matt,
    my argument is based on the idea that sauropods are not different from other fossil clades. I have spent 15 years in theropod phylogenetics, assembing a huge data set, mostly based on the rich phylogenetic production of the last 30 years. Theropod phylogeny is not that different from your depiction of sauropod problems: we have multiple convergences, mechanical constraints (bipedism and gigantism) and feeding/locomotory homoplasy. Even if so complex, many problematic relationships of the past have been solved. Nobody still considers tyrannosaurids as allosauroid-relatives, or elaphrosaur-like forms as ornithomimosaurs: we have realised that these were convergences due to feeding and locomotory adaptations constrained by the relatively narrow theropod bauplan. So, I am optimistic, and feel that once a big sauropod phylogeny assembling and merging the present datasets will improve accuracy. Note that a merged dataset is not just a copy-and-paste of previous matrices: it is a critical revision of all characters and character states, and an analysis of state distribution even in taxa traditionally excluded from sub-clade analyses.
    Large taxon samples also improve accuracy, as noted in the first comment by Valentin Fischer: including all taxa with unique combination of characters in fact may identify unexpected lineages, often a priori rejected by our “phylogenetic prejudices”.
    Partition analyses may help in identifying if different regions evolve (and affect phylogeny) differently from the others: this is a first step toward a resolution of problematic areas.
    In brief, all problems you raised are legit, but I see the situation not that bad or negative: we just need a more detailed taxon and character sampling. It is a hard job, but someone should do it :-)
15. Anonymous Says:

    
    November 1, 2017 at 6:19 am

    Losos actually had something to say about the likelyhood and degree of convergence in his new book, which is at least partly inspired by his work on lizards. He finds (mostly through a general anecdotal survey) that degree of convergence and phylogenetic relatedness are closely linked. That is, anoles may be close enough related to produce indistinguishable ecomorphs, but anoles and geckos or anoles and curly-tailed lizards in the same evolutionary scenario are not going to turn out the same.

    With regards to sauropods, I’m not an expert, but aren’t “camarasaurs” and “brachiosaurs” currently in a state of flux anyway? Every phylogeny I’ve seen seems to have them all over the place, with relatively low consistency. Sometimes Europasaurus is a brachiosaur. Sometimes it’s not. Sometimes Sauroposeidon is a brachiosaur. Sometimes it’s not. And it seems like it’s been almost impossible to get anything to stick to Camarasaurus.

    There’s also the Dakotaraptor/Anzu issue to consider. The Hell Creek Formation is one of the best collected formations out there, and many people assumed its dinosaurian faunal list was essentially complete (beyond identifying small troodontid or dromaeosaurs to genus). Then out of nowhere we have two taxa show up that there had been no indication of having been there before. It’s one thing to invoke absence of fossils as a possible explanation, it’s another to see it.

    “Maybe if the average quality of phylogenetic analyses increases (which is related to the supplementary material peer review issue)”

    From where I’m standing, the biggest issue with phylogenetic analyses right now is that few studies go out of their way to explain how the authors discriminated between character states, and matrices are getting so large it’s possible to hide inaccuracy through obscurity. I was checking a recent phylogenetic analysis from a Science/Nature paper and found that the terminal taxon I was interested in had been coded in this analysis had been coded as a really bad chimera: it was a polytaxic coding, but some characters were coded based on a really derived in-group (the equivalent of coding sauropodomorpha based on “prosauropods” and somphospondylians) and others are states unknown for any member of the group at all. You couldn’t even justify it by saying it was a hypothetical ancestor based on symplesiomorphies. And this paper got published and is frequently being used in discussions of the phylogeny of the group.
16. Heinrich Mallison Says:

    
    November 1, 2017 at 10:07 am

    Remember that Witzel et al (IIRC) paper that showed how few feeding-related inputs you needed to create vastly different sauropod skulls with lots of seemingly unrelated characters like nostril position etc.?
17. David Marjanović Says:

    
    November 1, 2017 at 7:32 pm

    Andrea Cau, John D’Angelo and Heinrich Mallison bring up important points.

    The main reason why the molecular analyses of placental and bird phylogeny disagreed so strongly with the previous consensus is that the previous consensus was hardly based on phylogenetic analyses at all. Even today, morphological analyses of these groups have hardly begun. What there is now is still too small, still samples different body parts very unevenly, and still contains too many correlated characters.

    In my own field, the largest data matrices are now attaining about 350 parsimony-informative characters. They should have at least 1000 – and they could have, easily, judging from how little overlap there is between these matrices! On top of that, there are whole body parts that have recently been studied but never yet been included in a matrix, there’s character correlation – including outright redundancy – all over the place, there are typos and similar accidental, unsystematic misscores all over the place, and so on and so forth.

    Morphological phylogenetics is simply a huge amount of work – much more so than the molecular version.

    This includes such minor details as the specimen numbers of the elements that make up the holotype

    I don’t understand. The paper says clearly that the holotype is UMNH.VP.26004 …?

    What if “turiasaur”, “brachiosaur”, and “titanosaur” are the sauropod equivalents? For “Caribbean island” read “continent”; for “lizard species”, read “sauropod clade”.

    Given the fact that the Middle and Late Jurassic was closer to a Pangaea situation than to a Caribbean situation, I’m rather optimistic about this. It certainly remains to be investigated how isolated Asia really was, though.

    I keep encountering this idea that supplementary information in Nature journals is not reviewed.

    Oh, that has nothing to do with the journals or the publishers. The vast majority of reviewers, as far as I can tell from experience, never take a look at supplementary information. Probably they assume a priori that it’s all just spreadsheets of measurements that can hardly be reviewed. I’m really glad that you’re one of the exceptions.
18. LeeB Says:

    
    November 3, 2017 at 5:50 am

    At least with sauropods or mammals you get lots of hard parts to look at. Consider fossil sharks where you get (mainly isolated) teeth.
    Fossil lamniformes include a number of recently extinct species, which seem ripe for study by collagen fingerprinting or some other form of proteomics.
    It would be nice to know how Parotodus, Carcharomodus, Carcharocles, Carcharodon, Isurus and Lamna really relate to each other; never mind “Alopias grandis”, Alopias and Megalolamna.

    LeeB.
19. Fossil Friday Roundup: Nov 3, 2017 – VOTE! | PLOS Blogs Network Says:

    
    November 3, 2017 at 5:07 pm

    […] of the Lepospondyli (Palaeo[Online]) Is everything we know about sauropod phylogeny nonsense? (SVPOW) A New Look for Sinosauropteryx (Dr. Neurosaurus) Plant-eating dinosaur had huge chisel-shaped […]
20. Anonymous Says:

    
    November 4, 2017 at 6:53 am

    @LeeB

    I think there’s currently an upper limit to how old you can get useable proteins out of fossils. I know a group that’s trying to get collagens out of fossils about as old as the youngest Carcharocles specimens and they’ve had no luck. Also, right now proteomics seems to be in a rather unorganized phase, similar to when DNA analyses were in their infancy and you were getting things like trees where rabbits were related to guinea pigs because we didn’t know about long branch attraction. Several of the proteomic analyses I’ve seen have produced results that not only disagree with morphological analyses, but mtDNA and the like as well.
21. LeeB Says:

    
    November 4, 2017 at 7:26 am

    There are studies getting proteins our of 3.5MYA arctic camels and 4.0MYA ostrich egg shells whereas the youngest Carcharocles and Parotodus are about 2.5MYA .
    Also a study looking at bone samples from England got material going back 1.0MYA with proteins preserved but one of their samples was from the bottom of the North Sea and was about 50KYA; it had much more sequentiable proteins than similarly aged material from on land.
    This suggests that material from the ocean bed with cold anoxic conditions preserves more protein than terrestrial samples.
    So shark teeth dredged up from the ocean bottom should be a target for future studies.
    Also teeth preserved in phosphate or manganese nodules may have undegraded proteins locked in and preserved.

    LeeB.
22. Chase Says:

    
    November 4, 2017 at 9:51 pm

    @Dr. Taylor Biogeographically speaking, I suspect that there was a Late Jurassic-earliest Cretaceous faunal mixing event between Europe and North America (perhaps through island hopping?). Something like that certainly seems to be occurring with Ornithomimosauria (I discussed this a bit in my paper regarding new Arundel ornithomimosaur specimens), and some of the oddball dinosaur, reptile, and mammal specimens from the Arundel Clay definitely make me suspect something along the lines of such a mixing (esp. the ornithomimosaur, nodosaurid, and ceratopsian specimens). From this hypothesis, it would follow that Turiasaurs and other clades spread into North America and a “turiasaur-shaped sauropod” did not independently evolve in North America. I agree with you though that we should be very cautious of homoplasy in fossils.

    @Jason Silviria “there is a huge amount of data on dinosaur faunas from eastern North America … that is currently missing.” I hope to help close this gap in understanding in these coming years. The dinosaur material of eastern North America is very, very odd, and I do feel that looking more closely at material from this part of globe will change our understanding of Cretaceous vertebrate biogeography to an appreciable extent.
23. Heteromeles Says:

    
    November 19, 2017 at 12:48 am

    Late to the party, but I wanted to add something about the “There is a semi-tongue-in-cheek biogeography “law” that states “Everything is everywhere, and the environment selects.”

    I may well be wrong, but I believe that “tongue-in-cheek” is incorrect. Rather, it was a description of what was thought to be going on with microbes before DNA sequencing became cheap enough for people to see differences. The original assumption was that, because microscopic spores are light enough to become airborne and travel intercontinental distances (based on aerial sampling from jets in the 1960s, IIRC), that microbes could get anywhere, but that where they actually grew was a function of the environment. This logic was extended to everything that reproduced by spores, including fungi, ferns, and fern-allies. It was seen as the reason why the single species bracken fern was supposedly the most widespread plant in the world, and why the same apparent species of asexually reproducing lichen is found at both poles and on mountain tops in between (a friend of mine wanted to work on this lichen, whose name escapes me, under the title “bipolar asexual disjuncts.” Alas, he couldn’t get it funded, as traveling to Greenland and Antarctica to study a single lichen was deemed too extravagant for a master’s thesis).

    With molecular methods, it turns out that this view is oversimplified. For example, bracken ferns turn out to be about 10 very similar species, rather than one widespread species, and microbes turn out to be massively more diverse and less widespread than they were thought to be before. Unfortunately, there’s little work on things like fungal biogeography and invasions. We know that fungal invasions exist, thanks to the wonderful world of pathogens and the worldwide work of ag inspectors, plant pathologists, and public health experts to catch them, but we really have almost no understanding of spore-based dispersal outside of economically important situations.
24. Mike Taylor Says:

    
    November 19, 2017 at 2:00 pm

    Thanks, Heteromeles, that’s interesting stuff!

Leave a Reply

Enter your comment here...

Fill in your details below or click an icon to log in:

Email (required) (Address never made public)

Name (required)

Website

You are commenting using your WordPress.com account. ( Log Out /  Change )

You are commenting using your Facebook account. ( Log Out /  Change )

Cancel

Connecting to %s

Notify me of new comments via email.

Notify me of new posts via email.

Δ

This site uses Akismet to reduce spam. Learn how your comment data is processed.