Limusaurus – four fingers and a whole mess of homologies

OK, so Limusaurus was, bizarrely, a basal herbivorous ceratosaur from the Middle Jurassic of China, but what is all this manual homology stuff that seems to be exciting everyone? Well, it is quite long and complicated, so bear with me. You have no idea how many rewrites that papers went through as most of the people on the paper at one time or another completely rewrite the section from scratch in an effort to get all the information in and keep both the other authors and the referees happy.

First off, if you are not familiar with the terminology of the hand of reptiles, I suggest you go and read this little bit I did on pterosaur fingers which should explain it all and help to keep everyone on board (and some of the other parts of that post will also prove useful). I try to avoid technical terms on here if I can to make things clearer, but it is there for a reason – to keep things simple and unambiguous and that will be very important later on!

Next we need to plough (as quickly but accurately as possible) through the sordid history of avian and theropod digit homologies. Modern birds have just three fingers in their hands (despite their extensive modification as part of the wing), namely digits II, III and IV. We know this because we can look at embryos developing where all five digits are initially present and then both I and V are lost later on. This follows the expected pattern of digit loss as seen in other tetrapods where digit V is the first to be lost and then I whenever the fingers start to reduce.

Theropod dinosaurs early in their evolution have all five functional digits in the hand but this number rapidly falls to four and then by the time we get to allosaurs and other theropods it’s just three while some derived tyrannosaurs famously go down to just two. I should add that this gets a bit more complex as often the digits are lost (that it, the phalanges and unguals) but the metacarpals are retained. Tyrannosaurus rex for example while having only two functioning fingers, does have at least the remnants of a third metacarpal bone in the hand, so it kinda has both two and three hand bones simultaneously depending on how you want to count them.

This last point is also quite important since this kind of thing helps us track how digits are changing and being lost, if an lineage suddenly goes from three fingers to two it might be hard to say which of those three was lost. When there is still a bone in the hand retained however, it is obvious which one has gone. Similarly, we can often track digit reductions and losses by the way in which the phalanges either shrink in size or are actually lost in various species. The phalangeal formula might go from say 2-3-4-3-X to 2-3-4-2-X and then 2-3-4-X-X in successive species suggesting a progressive loss of the fourth digit. Finally, some digits (like the human thumb) are pretty distinctive, so it’s usually pretty obvious when that digit is lost, or still retained. In short, we (potentially) have several convincing ways of documenting changes to a hand and which digits are going what and when in various species or clades.

Derived theropods have always been though to have digits I-III. That is first of all they lost digit V (OK, normal) but then bizarrely lost digit IV next. This is hardly impossible, but it is certainly an odd thing to have done when compared to other tetrapods. We have long thought this because as we look along the various successive theropod lineages we see what appears to be a reduction and loss of digit V and then a loss of IV. This is coupled with the retention of a short, robust and slightly twisted and therefore somewhat thumb-like first digit. In short, digit I is specialised and does not seem to change and V goes and then, we assume, IV. This is problematic since birds, as we have said, have digits II-IV preserved. If that is the case then at some point there must have been a shift from retaining digits I-III to II-IV at some point during theropod evolution. A possible solution to this issue has been the proposal of a ‘frame shift’, the idea that a simple genetic change could make the ‘frame’ of I-III move over to become a II-IV. Odd though that sounds, experiments in chickens suggest that it is certainly possible, though of course it remains impossible to demonstrate in theropods and leaves us with a rather unsatisfactory solution of just having to conclude that the frame shift probably just happened somewhere in theropods but we’ll never know where or when.

So into this maelstrom steps, or perhaps it’s better to say ‘reaches’, Limusaurus. Where and how does this little guy fit into the situation? Well, I should warn you that it does not necessarily clear things up, but more like complicates things a fair bit. However the hand of Limusaurus and our hypothesis for how the theropod hand MAY have evolved is certainly novel and worthy of consideration.

The hands of Limusaurus. Between the two we get a very good idea of what a complete hand would look like.

The hand of Limusaurus (like other ceratosaurs, and no great surprise given its phylogenethic position) has four metacarpals, with a missing 5th. So far so good, but hereafter things get complex. Digit I appears to have no phalanges and the metacarpal is heavily reduced, digits II and III each have 3, and the number on digit IV is unknown (i.e. the phalangeal formula is 0-3-3-?-X). It therefore looks like Limusaurus has gone the more ‘normal’ route of digit reduction losing V and then reducing (one would assume on the way to losing) I while keeping (and admittedly reduced IV). Its position right at the base of ceratosaurs suggests that this is not a derived characteristic but one that may be primitive for the ceratosaurs and tetanurans as a whole inherited from a common ancestor. If that is the case then this would mean that the derived theropods we long though had digits I-III actually had digits II-IV, the same as birds, and that discrepancy suddenly goes away.

But what about that modified thumb of the theropods? I hear you cry. (Possibly). Surely this implies that digit I gets lost OK, but if later theropods *still* have a modified 1st digit, doesn’t this imply that this characteristic was reacquired in digit II? That’s a bit odd surely, and ‘convenient’ that a specialised digit is lost, then a digit next to it reacquires those characteristics? It *would* explain how palaeontologists had misinterpreted a transition from I-IV to II-IV as being I-IV to I-III, but how likely is this?

Metacarpals of theropods in (top Limusaurus, the L-R, Dilophosaurus, Guanlong and Deinonychus). Note how the placing of metacrpal IV (labelled 2 in the figure) is more parsimonious is one assumes a II-III-IV in the latter two taxa.

Well, this is the more intriguing part. Limusaurus, while having a reduced metacarpal I (with no phalanges remember, there’s no digit there) the digit II metacarpal we do have is both robust and twisted, like a ‘classic’ theropod digit I. Now it is true that it is neither as robust, nor as assymetrical as those others, but these features are there and as an animal which may represent the beginnings of a change, then they can easily be forgiven for not being overly specialised at this point (digit I has not even been fully lost yet remember). In other words, it really does appear as if the specialised digit I is on the way to being lost AND digit II is taking over and adopting the specialisations that traditionally make out a digit I. It firms up the general hypothesis really quite substantially at this point as now we have evidence for I going, II becoming a ‘I’ and a matching homology between birds and derived theropods. There are also some other similarities in the digits between both basal and derived theropods which when but in the context of this proposed hypothesis do appear to support a II-IV identity of even derived theropod metacarpals (something supported by a modified phylogenetic analysis we put in the paper). There are, however, and perhaps inevitably, more problems.

First of all digit IV is certainly reduced significantly in Limusaurus and indeed we see it being reduced in other ceratosaurs and more basal taxa. In order for this hypothesis to be correct therefore, it does require that this digit is secondarily enlarged in later theropods to become a fully functioning digit again. This is not actually a huge issue as it would only need to return to its original length / number of phalanges, but it does still represent a reversal (from initial large size in the ancestral condition, to a reduction, then secondary reacquisition) and thus does add complexity to the new hypothesis which is not an issue for the I-III concept.

Secondly, this is unlikely to be the whole story since there are other theropods more derived than Limusaurus that have four functioning digits such as the basal tetanuran Xuanhanosaurus and we would certainly not expect this to have a missing digit I and regained IV and V. Other derived theropods also show changes in phalanx number and structure of the digits in the manus. However, this is not necessarily a problem for the new hypothesis in some senses as this variation shows that the hands of theropods are to a degree plastic and can be modified and also some of these transitions are just as much an issue for the I-III hypothesis as they are for the II-IV hypothesis. They must be considered however and not ignored, in conjunction with the assessment of both hypotheses.

Theropod phylogeny to show the shift of the digits as hypothesised in the paper.

Overall what this really means is that we now have competing hypotheses for the homologies of derived theropod digits. No longer are we reliant on the frame shift to explain the difference between theropod hands and that of birds. I must stress that this new hypothesis is just that, a hypothesis, and hardly absolute in any terms. I do think it has great potential and quite a lot of supporting evidence in its favour already, and of course this is just the start – there is probably more evidence out there (for either hypothesis) in the literature and specimens already that simply needs to be mined out and analysed (and yes, we are already working on new papers examining and evaluating other lines of evidence). There is plenty more to come for sure and I imagine much of it will be protracted and heated as much of dinosaur research, and especially that associated with birds, often is. Certainly this hypothesis does explain a few previous discrepancies if at the expense of creating a few more, but that’s OK, evolution is not a linear progression, but a series of branches and changes and reversals. One last point to make is that even if the hypothesis as it stands directly as a result of the evidence presented by Limusaurus (in terms of the positions and timings of the changes in the manus) turns out to be incorrect it still provides a putative mechanism for the change (i.e. perhaps the same kind of shift may occur elsewhere in the theropod tree). Having multiple possible hypotheses for any given situation is rarely a bad thing and in this one case at least, we have been notably deficient for a considerable period of time, despite knowing of the serious problems presented by the differences between theropods and birds.

In short Limusaurus and the accompanying hypothesis are associated with the following evidence:

1. The reduction of digit I

2. The modification of digit II to resemble that of a traditional digit I

3. This corrects for the discrepancy between avian digits and the I-III hypothesis, but

4. A secondary reacquisition of digit IV as a functioning digit.

Now we just need a five fingered tyrannosaur with a modified digit V and we can start the whole thing again.

Xu, X., Clark, J.M., Mo, J., Choiniere, J., Forster, C.A., Erickson, G.M., Hone, D.W.E., Sullivan, C., Eberth, D.A., Nesbitt, S. Zhao, Q., Hernandez, R. Jia, C., Han, F., & Guo, Y. 2009. A Jurassic ceratosaur from China helps clarify avian digital homologies. Nature 459(18):940-944.

Share this Post