What if Amphicoelias fragillimus was a rebbachisaurid?

October 21, 2018

An important paper is out today: Carpenter (2018) names Maraapunisaurus, a new genus to contain the species Amphicoelias fragillimus, on the basis that it’s actually a rebbachisaurid rather than being closely related to the type species Amphicoelias altus.

Carpenter 2018: Figure 5. Comparison of the neural spine of Maraapunisaurus fragillimus restored as a rebbachisaurid (A), and the dorsal vertebrae of Rebbachisaurus garasbae (B), and Histriasaurus boscarollii (C). Increments on scale bars = 10 cm.

And it’s a compelling idea, as the illustration above shows. The specimen (AMNH FR 5777) has the distinctive dorsolaterally inclined lateral processes of a rebbachisaur, as implied by the inclined laminae meeting at the base of the SPOLs, and famously has the very excavated and highly laminar structure found in rebbachisaurs — hence the species name fragillimus.

Ken’s paper gives us more historical detail than we’ve ever had before on this enigmatic and controversial specimen, including extensive background to the excavations. The basics of that history will be familiar to long-time readers, but in a nutshell, E. D. Cope excavated the partial neural arch of single stupendous dorsal vertebra, very briefly described it and illustrated it (Cope 1878), and then … somehow lost it. No-one knows how or where it went missing, though Carpenter offers some informed speculation. Most likely, given the primitive stabilisation methods of the day, it simply crumbled to dust on the journey east.

Carpenter 2018: Frontispiece. E. D. Cope, the discoverer of AMNH FR 5777, drawn to scale with the specimen itself.

Cope himself referred the vertebra to his own existing sauropod genus Amphicoelias — basically because that was the only diplodocoid he’d named — and there it has stayed, more or less unchallenged ever since. Because everyone knows Amphicoelias (based on the type species A. altus) is sort of like Diplodocus(*), everyone who’s tried to reconstruct the size of the AMNH FR 5777 animal has done so by analogy with Diplodocus — including Carpenter himself in 2006, Woodruff and Foster (2014) and of course my own blog-post (Taylor 2010).

(*) Actually, it’s not; but that’s been conventional wisdom.

Ken argues, convincingly to my mind, that Woodruff and Foster (2014) were mistaken in attributing the great size of the specimen to a typo in Cope’s description, and that it really was as big as described. And he argues for a rebbachisaurid identity based on the fragility of the construction, the lamination of the neural spine, the extensive pneumaticity, the sheetlike SDL, the height of the postzygapophyses above the centrum, the dorsolateral orientation of the transverse processes, and other features of the laminae. Again, I find this persuasive (and said so in my peer-review of the manuscript).

Carpenter 2018: Figure 3. Drawing made by E.D. Cope of the holotype of Maraapunisaurus fragillimus (Cope, 1878f) with parts labeled. “Pneumatic chambers*” indicate the pneumatic cavities dorsolateral of the neural canal, a feature also seen in several rebbachisaurids. Terminology from Wilson (1999, 2011) and Wilson and others (2011).

If AMNH FR 5777 is indeed a rebbachisaur, then it can’t be a species of Amphicoelias, whose type species is not part of that clade. Accordingly, Ken gives it a new generic name in this paper, Maraapunisaurus, meaning “huge reptile” based on Maraapuni, the Southern Ute for “huge” — a name arrived at in consultation with the Southern Ute Cultural Department, Ignacio, Colorado.

How surprising is this?

On one level, not very: Amphicoelias is generally thought to be a basal diplodocoid, and Rebbachisauridae was the first major clade to diverge within Diplodocoidae. In fact, if Maraapunisaurus is basal within Rebbachisauridae, it may be only a few nodes away from where everyone previously assumed it sat.

On the other hand, a Morrison Formation rebbachisaurid would be a big deal for two reasons. First, because it would be the only known North American rebbachisaur — all the others we know are from South America, Africa and Europe. And second, because it would be, by some ten million years, the oldest known rebbachisaur — irritatingly, knocking out my own baby Xenoposeidon (Taylor 2018), but that can’t be helped.

Finally, what would this new identity mean for AMNH FR 5777’s size?

Carpenter 2018: Figure 7. Body comparisons of Maraapunisaurus as a 30.3-m-long rebbachisaurid (green) compared with previous version as a 58-m-long diplodocid (black). Lines within the silhouettes approximate the distal end of the diapophyses (i.e., top of the ribcage). Rebbachisaurid version based on Limaysaurus by Paul (2016), with outline of dorsal based on Rebbachisaurus; diplodocid version modified from Carpenter (2006).

Because dorsal vertebrae in rebbachisaurids are proportionally taller than in diplodocids, the length reconstructed from a given dorsal height is much less for rebbachisaurs: so much so that Ken brings in the new version, based on the well-represented rebbachisaur Limaysaurus tessonei, at a mere 30.3 m, only a little over half of the 58 m he previously calculated for a diplodocine version. That’s disappointing for those of us who like our sauropods stupidly huge. But the good news is, it makes virtually no difference to the height of the animal, which remains prodigious — 8 m at the hips, twice the height of a giraffe’s raised head. So not wholly contemptible.

Exciting times!

References

 

Advertisements

40 Responses to “What if Amphicoelias fragillimus was a rebbachisaurid?”

  1. Dale mcinnes Says:

    Um …. exciting …. yes, but look at your size estimates. Eight meters to the hips is not twice the head height of a 6 meter giraffe. Or am I missing something in the translation?

  2. Mike Taylor Says:

    Eight meters to the hips is not twice the head height of a 6 meter giraffe, but it is twice the head height of a 4 meter giraffe, which is more typical. (There is in fact no record of a 6 m giraffe.) But, sure, insert the ratio you prefer: 36% taller than the world-record giraffe, for example.

  3. Dale mcinnes Says:

    No record of a 6 meter giraffe? I find that hard to believe. I’m thinking of a certain zoo specimen. Of course this would substitute as a ‘trophy’ specimen. 4 meters seems on the down side of an average bull.

  4. Mike Taylor Says:

    I’m not a giraffe expert. I’m just going on what Wikipedia tells me, which is that “The tallest recorded male was 5.88 m (19.3 ft) and the tallest recorded female was 5.17 m (17.0 ft) tall.” Unfortunately, both of the primary sources that it cites for this are books, and I don’t expect to easily find them online.


  5. Zach Armstrong previously (in 2014) reconstructed fragillimus as a giant rebbachisaur based on his skeletal reconstructions of Limaysaurus and Rebbachisaurus itself;

    https://www.deviantart.com/palaeozoologist/journal/Was-Amphicoelias-a-rebbachisaur-440611550

    His reconstruction is around 2 meters shorter, well within any sensible error bars when working with material this fragmentary (and most of that seems to come from the neck and tail).

  6. Mike Taylor Says:

    Yep: Ken acknowledges Zach’s prior work in the paper.

  7. william dale McInnes Says:

    5-6 m for a really big giraffe sounds more reasonable(17′-19′).

  8. Mike Taylor Says:

    Seriously? We have a radical reinterpretation of what was quite possibly the single largest land-animal of all time, and you guys want to talk about giraffes? Well, OK. Show me a 6 m tall giraffe in the literature.

  9. Andreas Johansson Says:

    What does the rebbachisaur reconstruction mean for weight estimates?

  10. Mike Taylor Says:

    Carpenter (2018) says nothing about a revised mass.

    Eyeballing the comparative reconstructions of the diplodocine and rebbachisaurid versions of the animal, it looks to me that there is not that much difference in the sizes of the torsos — almost all the length difference is accounted for by the long neck and tail of the diplodocine. But we know from experiments in GDI such as my own on Brachiosaurus and Giraffatitan (Taylor 2009:803) that the torso accounts for something like 70-75% of the total mass. So I’m thinking that the new mass might not be that much different from the old — whatever that was! — and could still quite possibly have reached 100 tonnes.

    Carpenter (2018: caption to figure 7) says the Limaysaurus silhouette is based on that of Paul 2016, which is The Princeton field guide to dinosaurs. I don’t have that, but I think it’s the same book as Paul’s (2010) Dinosaurs: A Field Guide, which I do have. His Limaysaurus reconstruction is on page 187, accompanied by a telegraphic entry giving the length as 15 m and the mass as 7 tonnes. I’m not sure how much work went into those numbers, but if we take them at face value, and consider Marapuunisaurus as identical to Limaysaurus but at twice the linear size, that would make it eight times as massive, or 56 tonnes. Paul tends to estimate masses at the light end of most other palaentologists, so we might think that this figure, too, is on the light side.

  11. Andreas Johansson Says:

    Thanks:)

  12. Andrea Cau Says:

    Sorry for the self-promotion: although Carpenter (2018) mentions a 2014 blog post on this topic, he ignored (I assume he was not aware of) my 2012 blog post suggesting a rebbachisaurid status (and a re-size) for A. fragillimus: https://theropoda.blogspot.com/2012/01/amphicoelias-fragillimus-e-un.html

  13. Mike Taylor Says:

    Ha! Well, nice work, Andrea! When I was offered this paper to review, the very first thing I thought to myself was “I’ll have to remember to ask him to cite Zach Armstrong’s DeviantArt post” — then it turned out that I didn’t need to, because the first version of the manuscript I saw already contained that acknowledgement. Sadly, it seems neither Ken nor I was aware of your even earlier work. Sorry about that.

  14. Mike Taylor Says:

    Going back and reading your post now …

    I see a typo: “Cope and Mook (1921)” for Osborne.

    And then …

    Graaaghaghhg! I evidently not only read, but actually commented on your post back in 2012, but evidently forgot all about it. I’m so sorry.

  15. william dale McInnes Says:

    For what it’s worth, these giraffes are indeed some 4′ taller than average specimens of their species.

    The tallest giraffe ever recorded was a Masai bull named George. George stood at 19ft and his horns almost grazed the roof of the 20 ft high Giraffe House when he was nine years old. He arrived at Chester Zoo from Kenya in January 1959 and died in July 1969.Jan 4, 2016

    The tallest Giraffe in the world ever recorded was about 19 feet tall and was called George. It was kept in Chester Zoo in Cheshire, England back in 1959.
    Currently the tallest Giraffe is Zulu which is just over 19 feet tall. It is kept in Folly Farm zoo in Saundersfoot and Tenby, Pembrokeshire, England.

  16. Andrea Cau Says:

    For those interested, I scored A. fragillimus in the data matrix of the Tataouinea 2015 monograph (Plos One), following Carpenter (2018), and it turned out as, alternatively, sister-group of Europasaurus, sister-group of Haplocanthosaurus, or sister group of Kebbashia (the rebbachisaurine-limaysaurine node).


  17. […] which is 99.9% white, living my comfortably middle-class life, mostly thinking about things like sauropod vertebrae and European cheeses. I know that there are more important things. But Doctor Who, at its best, […]


  18. Dr. Cau – I have updated my post to give due credit to your prior work, as unfortunately I was unaware of it. At least this way if people check out the link Carpenter acknowledged in the paper, they will see a link to your post right away!

    Best wishes,
    Zach

  19. Andrea Cau Says:

    Thanks, Zach. It has to be remarked that your argument was more detailed than my suggestion.


  20. Here is the link to the pdf:
    https://www.utahgeology.org/giw/index.php/giw/article/view/28

    Zach, I sent this to you in a message in Deviant Art

    Ken

  21. Vahe Demirjian Says:

    The conclusions regarding the new size estimates for and classification of Maraapunisaurus outlined in Carpenter’s new paper make sense for a number of reasons:

    1. Sauropods with tall dorsal neural arches were unknown in Cope’s time, and Cope’s argument that increased femoral length was correlative to a taller dorsal neural arch hasn’t been borne out my rebbachisaurid discoveries, because all rebbachisaurids have the same femoral length as known diplodocoids from the Morrison Formation despite have taller neural spines than those of diplodocids.
    2. Lingwulong highlights the dearth of knowledge about early diplodocoid evolution, and Maraapunisaurus represents the first pre-Valanginian rebbachisaurid, helping to further reduce ghost lineages for Rebbachisauridae and Dicraeosauridae in the pre-Kimmeridgian-Tithonian.
    3. Suuwassea is currently regarded as a dicraeosaurid, while Haplocanthosaurus is recovered as a basal diplodocoid by Whitlock (2011), and Maraapunisaurus is the third non-diplodocoid from the Morrison Formation.

    I think anyone ought to see how Maraapunisaurus compares with Haplocanthosaurus.

  22. Mike Taylor Says:

    Andrea:

    For those interested, I scored A. fragillimus in the data matrix of the Tataouinea 2015 monograph (Plos One), following Carpenter (2018), and it turned out as, alternatively, sister-group of Europasaurus, sister-group of Haplocanthosaurus, or sister group of Kebbashia (the rebbachisaurine-limaysaurine node).

    Well, that’s pretty crazy! I’m struggling to see what would have drawn it towards Europasaurus or Haplocanthosaurus. Do you happen to have the synapomorphy lists from that analysis?


  23. I’ll probably start a big fight, but here goes. The result of Andrea shows that the NEURAL SPINE of Maraapunisaurus is similar to THOSE of Europasaurus, Haplocanthosaurus or Kebbashia. Given that the neural spine alone is NOT the entire animal and does not even represent the entire animal, the results really do not reflect phylogenetic relationship. Sorry, but to me this represents one of those misuse of phylogenetics. If we are really trying to solve phylogenetic relationships, then we need to apply it to more than one bone (and a partial one at that!). If the point is to determine what taxa the neural spine is like, then only the neural spines can form the matrix. Everything else is extraneous and misleading since statistically they will exert a “pull”.

  24. Mike Taylor Says:

    Well, hang on, Ken. The neural spine of Maraapunisaurus is all we have to work with. (In fact, a drawing of the neural spine). Isn’t that just as true of the kind of assessment you made of it as it is of a numeric phylogenetic analysis? In either case, we are necessarily reaching a conclusion about the whole animal on the basis of a drawing of a part of one vertebra. That’s just the way it is. Why would that be more of a problem for a cladistic assessment than for yours?

  25. Kenneth Carpenter Says:

    Like I said, I’ll start a fight :-) . In my analysis, I compared morphology for morphology of the spines. The fact that the closest spine was that of Rebbachisaurus is why I concluded Maraapunisaurus was a rebbachisaurid. I did not consider the rest of the Rebbachisaurus skeleton. The data sets out there for sauropods includes a lot more than the neural spines. Phylogenetic software can and has been used on non-biological morphologies, e.g., archaeologists have used it to determine relationships among pottery and projectile points (there is a book that pioneered this usage: https://uofupress.lib.utah.edu/cladistics-and-archaeology/ ). This usage shows that the software is really showing relationships among physical (morphological) features, and that is what is really being done in paleontology. In the case of the neural spine, only the neural spines have any bearing, the rest of the morphologies skew the results and is really untestable until more of Maraapunisaurus is found.

    I think we forget that the results in phylogenetic analysis is really pointing out relationships among morphological features to which a taxonomic name has been hung.

  26. Mike Taylor Says:

    I’m still not following your point, Ken. In your analysis, you compared morphology for morphology of the spines. Good. But in Andrea’s numerical analysis, he compared morphology for morphology of the spines, too — since there was nothing else for his analysis to compare the morphology of the Maraapunisaurus spine to. Why do you consider that invalid if yours is valid?

    I think the real argument in favour of your approach is quite different: it’s that the cladistic analysis can only take into account specific characters that have already been coded for the other taxa, whereas your by-eye analysis can take into account other features that have not been numerically coded.

  27. Kenneth Carpenter Says:

    Andrea said he used the data matrix of Tataouinea published in 2015. That would be https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0123475#sec039. There are 346 characters in that data matrix, the vast majority of which have nothing to do with the neural spine. These other characters are going to exert a statistical pull and the poor Maraapunisaurus neural spine is just going for the ride because it is statistically overwhelmed. IF Andrea only used the neural spine portion of the data matrix, then the results only point to a relationship among neural spine morphologies of the sauropods, which does not necessarily translate into phylogentic relationships based on the whole animals.

  28. Mike Taylor Says:

    I don’t see that at all. How can the non-NS characters in the matrix exert any pull on the Maraapunisaurus spine?

  29. Ivan Says:

    Very interesting article, also congrats to all of those who reached similar conclusion independently.
    It seems Dr.Carpenter is shooting around it to finally get this truly colossal beast properly sized :)
    Hope the next one will catch it.
    I think there are some problematic implications in Maraapunisaurus being scaled isometrically from the very derived Cretaceous rebbachisaurids. In particular I cannot imagine it 8 m tall at the hips and stretching its neck to low browse (as the comparison with derived rebbachisaurids would suggest).
    AFAIK, rebbachisaurids were adapted to a particular niche of low browsing, so they got small, grew complex teeth batteries and reduced the neck length. I don’t thing this can apply to such a big sauropod.
    It reminds me the iso-scaling of the derived Saltasaurus to estimate the size of the primitive Argentinosaurus, it didn’t work.
    It’s not that I don’t accept the rebbachisaur status for Maarapunisaurus, the proofs look convincingly.. I just think the actual animal might be more similar to primitive diplodocoids in proportions (given its age and phylogenetic position), in this respect the Apatosaurus or Haplocanthosaurus shape might be a more likely start for a scaling up.
    I know we do not have a single bone to work with, so probably this is just an hypothesis difficult to test

  30. Mike Taylor Says:

    You could be right. We think of the extreme pneumaticity of rebbachisaurids as a derived character that evolved some way down the tree from their split from haplocanthosaur-grade diplodocoids. But perhaps the pneumaticity was intially a size-related feature that merely failed to get unevolved when rebbachisaurs became smaller.


  31. […] a comment on the last post, Mike wrote, “perhaps the pneumaticity was intially a size-related feature […]

  32. Andrea Cau Says:

    Dr. Carpenter, I am the first that concludes that the position of a taxon scored for just the few data available from the posterior view of a single drawing of a neural arch is quite weak.
    That said, we can argue the same for ANY taxonomic interpretation of A. fragillimus, including the whole arguments used in your paper. What makes a qualitative comparative analysis of a few features better than the quantitative analysis of the SAME features?

    Note also that the analysis did not dismiss a rebbachisaurid status (Khebbashia is a node inside Rebbachisauridae) but just found other two alternative positions equally valid based on the scarce amount of data. This is perfectly plausible given the amount of data available. I suspect that it is just due to some potential autapomorphic combination of features in that neural arch, or, alternatively, that some feature was not interpreted correctly and mislabelled as a different (this may be true for some of the laminations depicted in the only available figure of A. fragillimus).

    Dr. Carpenter, a comment on how phylogenetic analyses work: note that in a phylogenetic analysis the unknown scores (the “?”) have no active effects on the placement found, which is exclusively determined according to the actually scored characters. So, the actual placement of A. fragillimus is not biased by the unknown data, but by the potential conflict among the scored elements.

    In conclusion, I merely tested what happens to a neural arch with the features listed in Carpenter (2018) when analysed phylogenetically: it may be a rebbachisaurid, but also other kinds of neosauropods. And this is, conservatively, the best conclusion from a single neural arch visible in a single view of a sauropod specimen lost over a century ago.

    There is no reason for further discussion: we need new fossils, because the status of that taxon is really weak based on the mere lost specimen described in that short and very old paper.

  33. Kenneth Carpenter Says:

    Andrea, well, I am handicapped in any response since you closed the door: “There is no need for further discussion…”

    Peace

  34. Andrea Cau Says:

    I am even open to discuss… but what could we say of scientifically valid beyond what you wrote in your recent paper? It is surely the most detailed study on A. fragillimus we have, but in absence of an actual fossil (or additional evidence of the original fossil) we cannot argue anything else beyond just discussing the distribution of the key characters among sauropods.
    For example, there is no way to objectively test if the controversial phylogenetic position is due to mis-codes, instead of due to wrong interpretation of the figure, instead of a mere absence of actual features in what was a taphonomically-damaged specimen.

    Sincerely, I feel that any further discussion on a lost fragment of vertebra (which so many paleontologists disagree in most of its details) is, sorry for my tone, beyond actual paleontology and too speculative for being taken seriously.
    We cannot develop a discussion on key details that are mostly based on subjective assumptions and cannot be tested objectively.


  35. Very well, Dr. Cau, Please compare the posterior dorsal neural pedicle and spine of Europasaurus in Carballido et al. (2014), Fig 4C with the neural pedicle and spine of Maraapunisaurus as drawn by Cope. Except for the single, narrow postspinal lamina, there is not much in common between the two. There is certainly no apomophy that unites the two as sister taxa. Europosaurus certainly lacks the pneumatic neural spine, tall neural pedicle, and festooned spinopostzygapophyseal lamina, apomorphies that Maraapunisaurus does shares with Rebbachisaurus. Please explain how you can conclude that Maraapunisaurus COULD BE a sister taxon of Europasaurus? We ignore the other two taxa for now.

    Carballido, J.L. and Sander, P.M., 2014. Postcranial axial skeleton of Europasaurus holgeri (Dinosauria, Sauropoda) from the Upper Jurassic of Germany: implications for sauropod ontogeny and phylogenetic relationships of basal Macronaria. Journal of Systematic Palaeontology, 12(3), pp.335-387.

  36. Andrea Cau Says:

    With your reply, it seems you have not understood the point of my final reply. We can discuss whatever element you like, but all is just based on the interpretation of Cope drawing. In fact, as I wrote, I did not test “A. fragillimus” but “Carpenter (2014)’s interpretation of A. fragillimus”, which is just a subjective and non-testable interpretation of a drawing made in the XIX century and cannot be controlled to any actual fossil available.

    My comment was epistemological, more than paleontological.

    In any case, I have checked the apomorphies of the three alternative positions of A. fragillimus in the analysis of Carballido et al. (used in the Tataouinea monograph).

    1- as sister taxon of Europasaurus:
    -postspinal lamina rough and extended for whole spine
    -high neural arch pedicel compared to centrum (assumed due to proportions relative to zygapophyses, and one of the rebbechisaur features you suggested)

    This positions requires 2 rebbachisaur-like convergences in A. fragillimus (festooned spdl and dorsolaterally directed transv. proc).

    2- as sister taxon of Haplocantosaurus:
    -high neural arch pedicel (as above)
    -dorsolaterally directed trans. proc.

    This position requires festooned spdl to be convergent with rebbachisaurids and postspinal lamina convergent with Europasaurus.

    3- as a rebbachisaurid:
    -high neural arch pedicel (as above)
    -festooned spdl
    -dorsolaterally directed trans. proc.

    In short, the three alternative positions differ mainly on how a few features homoplastically distributed among neosauropods are placed in the tree.
    And this is exactly what we have to expect for a taxon based on such very limited amount of data.
    Could you dismiss that the pneumatic differences between Europasaurus and A. fragillimus are just due to size difference? Could you dismiss that lamination difference between Haplocanthosaurus and A. fragillimus is just due to random variation in highly homoplastic vertebral features?

    No, you cannot, bacause with such a limited amount of data any alternative position is not statistically significant to the others.
    And given that these features cannot even be directly verified on the fossil but are just based on our faith on the accuracy of Cope’s drawing (something that I can may even agree with, but that in any case is a subjective and non-testable assumption, so not a robust scientific datum), we cannot argue anything robust beyond the mere: “It is similar to rebbachisaurids, but we cannot exclude other placements in Neosauropoda”.

    Again, I am sorry for my reply, but if the vertebra is lost, any argument on its status is very speculative and subjective. And I say this even if I consider the rebbachisaurid hypothesis intriguing and plausible.


  37. Dr. Cau,
    Thanks for your lengthy reply. I have no problem with this more detailed explanation, but even you have to admit your previous explanations were too brief (especially your first) leaving me to guess what you meant.


  38. In reading back through all of the comments I see that I was not always clear either. My apologies to all.

  39. David Marjanović Says:

    If the point is to determine what taxa the neural spine is like, then only the neural spines can form the matrix. Everything else is extraneous and misleading since statistically they will exert a “pull”.

    That might happen in a parametric analysis (maximum likelihood or Bayesian inference), where statistical approaches are used to try to compensate for missing data (as a byproduct of trying to compensate for long-branch attraction). Parsimony, however, doesn’t do any of this.

    So, the characters that can’t be scored for Maraapunisaurus will influence, in a parsimony analysis, what the rest of the tree looks like. If you will, they will influence which sites are available for M. to land on. But they cannot influence at all which of those sites M. will end up in. Missing data is not a problem for parsimony analysis; they cannot create a false signal.


Leave a Reply

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

WordPress.com Logo

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

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

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

Facebook photo

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

Connecting to %s

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

%d bloggers like this: