Tetrapod Zoology

The Matamata is an incredible animal. A morphologically bizarre, highly cryptic, aquatic South American turtle, it’s equipped with a super-specialised wide, flattened skull and a host of peculiar features that allow it to engulf fish and other prey in deft acts of rapid suction.



Surprisingly large (up to 1 m long), it has a very long, thick neck, and a proboscis that it uses as a snorkel. But of course you already know all of this because I covered it in depth in a series of articles published here during 2010 (if you need a refresher, see the links given below). My aim in this fifth and final article is to wrap up the matamata series and basically cover all the stuff I haven’t so far. Believe me, some of it is very neat…

We start with some nomenclatural and taxonomic stuff (don’t worry, it’s brief). First of all, what is the scientific name for the Matamata? Throughout these articles I’ve been using Chelus fimbriatus, as have many previous authors. Alas, as argued by Cadena et al. (2008), this is – shock horror – incorrect. Well, it’s incorrectly formatted, anyway.

The generic name Chelus is apparently feminine; ergo, any species names attached to it have to be feminine too. Chelus fimbriatus should thus really be Chelus fimbriata. One of the fossil matamatas (read on) also needs a name change: while often called Chelus columbianus, it should really be Chelus columbiana. As Cadeno et al. (2008) note, a few authors (Eugene Gaffney among them) have been aware of this and have been using the names correctly, but most haven’t. It seems that Cadena et al. (2008) are correct, so I’ll make an effort to use the correct species names from hereon (I did note the use of fimbriata previously, but said that it was incorrect. Oops). Thanks to Mark Hollowell for bringing this to my attention.

How many modern matamatas?

Moving on, you’ll have noticed in the preceding matamata articles that I’ve treated all modern matamata populations as belonging to the single species C. fimbriata. This is the generally held opinion – that is, that there’s but a single species. But is there? Actually, a substantial amount of morphological variation is observed between the matamatas of the Amazon and those of the Orinoco.


In a study devoted to this variation, Sánchez-Villagra et al. (1995) showed that the two differ in carapace shape (sub-rectangular in Amazon animals; oval in Orinoco ones), plastron colour (dark in Amazon animals; light in the Orinoco population), and neck pigmentation (Amazon matamatas have two prominent black bands running along the underside of the neck, separated by a light brown or reddish area. No such patterning is present in the Orinoco population). The two also differ in their growth style, with the carapace becoming larger (in relation to the plastron) faster in Orinoco matamatas than Amazon ones (Sánchez-Villagra et al. 1995) [adjacent images from Sánchez-Villagra et al. (1995)].

These differences are consistent across ontogeny and seem to indicate that two readily distinguishable forms are present. If you want to, you could regard them as separate species – as suggested by Sánchez-Villagra & Scheyer (2010) – or subspecies. A number of specimens (from an unknown location) show a mosaic of Amazon-style and Orinoco-style features. Maybe the two forms are morphs that grade into one another. Alternatively, perhaps the ‘mosaic’ individuals come from a hybrid zone existing between two recently diverged relatives. But that hypothetical hybrid zone could also be between two highly distinct relatives that diverged a long time ago – we just don’t know. Yet. Sánchez-Villagra et al. (1995) noted that their study was preliminary and that more data was needed. They also said that the distinct Amazon and Orinoco forms would be named in time. So far as I can tell, the variation present within living matamatas has yet to be studied further and the two forms haven’t been named. Do say if you know otherwise.

The extinct ones


Having mentioned the possibility of different matamata species, did I mention the extinct ones? No, I deliberately saved this bit until last. Two definite fossil matamatas are known: C. lewisi from the Miocene Urumaco Formation of Venezuela, and C. colombiana, first named from the Middle Miocene Villavieja Formation of Colombia but later reported from the Upper Miocene of Estado do Acre, Brazil and the Lower Miocene Barzalosa Formation of Colombia. Both species were named by Roger C. Wood in 1976, though the C. colombiana holotype had actually been discovered as early as the mid-1940s (Wood 1976) [that holotype is shown here; from Wood (1976)]. The Lower Miocene records make C. colombiana the oldest verified matamata (though read on). Note that the Brazilian records of C. colombiana were reidentified as C. lewisi by Cadena et al. (2008).

‘Small’ fossil matamatas from various Brazilian localities (as in, similar in size to living matamatas) have been intimated to represent an additional species, but the possibility that they are young specimens of C. lewisi and/or C. colombiana also exists (Lapparent de Broin et al. 1993). Plio-Pleistocene matamata fossils from Brazil were said by Wood (1976) to perhaps represent a new species, though this hasn’t been followed up so far as I can tell.

Anyway, so, the oldest matamatas are from early on in the Miocene (somewhere around 20 million years ago). However, Broin & de la Fuente (1993) identified a single cervical vertebra from the Upper Cretaceous of Argentina as that of a possible stem-matamata. That’s not impossible – after all, there’s good evidence from fossils that the chelid radiation was well underway by the latter part of the Cretaceous – but better evidence is definitely needed before we can be sure that the matamata lineage really extends back that far (the phylogenetic position of matamatas relative to other chelids is controversial: see the ‘long, fat neck’ article). Possible close relatives of matamatas (namely Lomalatachelys neuquina from the Santonian Bajo de la Carpa Formation Argentina) have also been identified from the Cretaceous (Lapparent de Broin & de la Fuente 2001).


Both C. colombiana and C. lewisi share with the living matamatas the presence of three carapacial ridges composed of prominent convexities (five along the dorsal midline, and four on both of the side ridges), though the convexities on those ridges are lower and more homogenous in C. lewisi than in the others. C. colombiana has approximately parallel lateral carapace margins while the carapace of C. lewisi flares out posteriorly and the lateral edges flare outwards. Others features that allow the species to be distinguished include details of shell bone shape and the form of the pelvic scars (the raised areas on the shell bones present where the pelvic bones contact the shell). If you need the details see Wood (1976) and Cadena et al. (2008).

Incidentally, C. lewisi makes a guest appearance in one of Jorge González’s excellent illustrations from Sánchez-Villagra et al.’s Urumaco & Venezuelan Paleontology. Part of the picture concerned features on the cover (shown above); you should be able to see the matamata down at bottom left (obscured by the names of the editorial team). If you’re interested in South American Cenozoic vertebrates, Urumaco & Venezuelan Paleontology is a must-have: check it out.

Big – but how big?

What makes these fossil matamatas all the more interesting is that they are somewhat larger than the living ones, perhaps much larger. As we’ve seen in previous articles, modern matamatas are fairly sizeable turtles. Carapace lengths of 40-50 cm have been reliably reported (the record is probably 53 cm). Because the animal’s neck is typically about similar in length to the carapace, a big modern matamata can reach 1 m in total (a matamata this big weighs between 10 and 12 kg). There are rumoured, unconfirmed references to modern matamatas with carapace lengths of over 120 cm (this would mean a total length – with neck outstretched – of around 2 m) but, alas, such accounts are almost certainty exaggerations.

I think exaggerations have also affected the fossil species. The type specimen of C. lewisi has a carapace length (CL) of 45.5 cm and there are specimens with a CL of 50 cm (Wood 1976). This is smaller than the biggest modern matamata. Yet authors have said that C. lewisi is about 10-20% longer in carapace length than the largest C. fimbriata specimen (Wood 1976, Sánchez-Villagra & Scheyer 2010). So far as I can tell, a big C. lewisi would be a large turtle, but not one any bigger than a big modern matamata.

C. colombiana was definitely bigger and both Wood (1976) and Sánchez-Villagra & Scheyer (2010) stated that it could be 50-100% larger than the largest authenticated specimens of C. fimbriata. Yes, as much as twice the size. Assuming this to be correct, it would give a big C. colombiana a CL of about 1 m and hence a total length (again, with outstretched neck) of about 2 m. That’s a very impressive matamata (though note that it’s not bigger than those unconfirmed giant modern specimens). Again, however, I’m not sure that this is precisely correct. The holotype of C. colombiana has a CL of 54.8 cm and the largest reported specimen has an estimated CL of 63 cm (Wood 1976). That’s certainly not 100% bigger than the biggest modern matamata, and it’s not 50% bigger either! Nevertheless, it’s big, and I’m not unhappy with the concept of a matamata that (with neck outstretched) would have reached about 1.2 m in total.


Yet there are claims of much, much larger individuals. Within the last year or two a news items on a Brazilian TV show reported the discovery of a fossil matamata (I assume a specimen of C. colombiana) in which the carapace length alone was 2.46 m. If you don’t believe me, the image above is a screenshot (the relevant segment is online here: not working for me at the moment, alas). Let’s pretend for a moment that this is correct. Assuming that a matamata this big would be isometrically similar to a normal-sized modern matamata, we would have to imagine a neck more than 2 m long, and hence a total length of over (or about) 4 m. A 4 m long matamata. That’s such an awesome idea that I couldn’t help but produce the image you see here. Do take it with a substantial pinch of NaCl.


Of course, that 2.46 m CL is definitely wrong. I think. No matter what it says in that screenshot, there’s no way that the carapace concerned is really that long – it just doesn’t look big enough compared to the other objects that can be observed in the film (like the adjacent table, the floor tiles, the other fossils etc.). So, how did this mistake come about? Perhaps someone said that the whole animal might be 2.46 m long, but even this seems just about unbelievable when the maximum authenticated CL for any matamata is substantially less than 1 m. Or have I missed something?

Whatever the truth, these big, extinct matamatas would undoubtedly have been fascinating creatures, and they were made all the more so by the fact that they lived alongside other giant pleurodires (like Stupendemys), an assortment of giant crocodilians (Purussaurus, Mourasuchus, Charactosuchus and so on) and a diversity of sloths, toxodonts, litopterns and large rodents.

Is it with thoughts of giant fossil matamatas fresh in our minds that we bring the Tet Zoo matamata series to a close, at last. I haven’t done much on other chelids, nor on pleurodires as a whole but, hey, there’s still the future. I owe several people kind thanks for their assistance with the matamata series: thanks to Markus Bühler, Mike Habib, Mark Hollowell and Marcelo Sánchez-Villagra. For the previous articles in the series, see…

And for previous Tet Zoo articles on turtles see…

And for more on neat South American Cenozoic tetrapods, see…

Wow, that’s actually quite a lot. Much more to come.

Refs – -

Broin, F. de & de la Fuente, M. S. 1993. Les tortues fossils d’Argentine: synthèse. Annales de Paléontologie 79, 169-232.

Cadena, E., Jaramillo, C. & Paramo, M. E. 2008. New material of Chelus colombiana (Testudines; Pleurodira) from the Lower Miocene of Colombia. Journal of Vertebrate Paleontology 28, 1206-1212.

Lapparent de Broin, F. de, Bocquentin, J. & Negri, F. R. 1993. Gigantic turtles (Pleurodira, Podocnemididae) from the Late Miocene-Early Pliocene of south western Amazon. Bulletin de l’Institut Francais d’Études Andines 22, 657-670.

- & de la Fuente, M. S. 2001. Oldest world Chelidae (Chelonii, Pleurodira), from the Cretaceous of Patagonia. Comptes Rendus de l’Academie des Sciences de Paris 333, 463-470.

Sánchez-Villagra, M. R., Pritchard, P. C., Paolillo, A. & Linares, O. J. 1995. Geographic variation in the matamata turtle, Chelus fimbriatus, with observations on its shell morphology and morphometry. Chelonian Conservation and Biology 1, 293-300.

- . & Scheyer, T. M. 2010. Fossil turtles from the northern neotropics: the Urumaco sequence fauna and finds from other localities in Venezuela and Colombia. In Sánchez-Villagra, M. R., Aguilera, O. A. & Carlini, A. A. (eds) Urumaco & Venezuelan Paleontology. Indiana University Press, Indianapolis, pp. 173-191.

Wood, R. C. (1976). Two new species of Chelus (Testudines: Pleurodira) from the Late Tertiary of northern South America. Breviora, 435, 1-26


  1. #1 Sordes
    April 29, 2011

    The fossil giant carapace is huge without doubt, but is it really 2,46? In the video where it is shown together with a modern matamata carapace and and with a man as comparison, it looks much more like 1,5 m or so.

  2. #2 Sven DiMilo
    April 29, 2011

    more like it.

  3. #3 Jamie
    April 29, 2011

    Matas are some of my favorite turtles. They are also on my wish list to own some day. Thank you so much for this series on matas. Its really eye opening on the species.

  4. #4 David Marjanović
    April 29, 2011

    What is Chelus etymologically? An idiosyncratic transcription of what is usually spelled “chelys”? If so, yes, feminine.

    and large rodents.

    Rodents Of Unusual Size like Phoberomys, right?

  5. #5 J. S. Lopes
    April 29, 2011

    Greek word was khelys, and it’s feminine. Usual latinization of khelys is chelys, but Greek really sounded , while Athenian Greek sounded < Ü>, like French u. Greek loanwords in Latin usually turned into , but there’s some exceptions, like lynx > Latin luncia, later misunderstood as l’uncia, so uncia (Portuguese onc,a and Spanish onza; onc,a without zedilla in Panthera onca).
    I dont know if Chelus was an attempt to latinize khelys, or just a kind of typo.

  6. #6 Mokele
    April 29, 2011

    One thing I’ve always wondered is whether captive animals can give us insight into the true maximal size of a species. On one hand, they lack the genetic variation of a wild population, thus may lack the 1-in-a-million flukes that produce truly huge snakes/lizards/turtles/whatever. But on the other, they’re raised in as close to optimum conditions as possible, with access to modern vet care and a nearly-unlimited amount of food (possibly too much in many ectotherms, IMHO).

    It’ll be interesting to see what the captive market in matamatas turns up in a few decades as far as maximal size.

  7. #7 stripey_cat
    April 29, 2011

    Early borrowings from Greek into Latin transcribed upsilon as u; y is a pretty late (2nd-1st century BC, five hundred years or more after the basic alphabet was borrowed) addition to the Roman alphabet, to distinguish between Greek and Latin u as Greek became culturally more important following Roman expansion in the eastern Med. So you can find yourself with very closely related words spelled with u or y, or alternating at the whim of the author depending on whether it was perceived as a Latin word that happened to have Greek roots, or a recent loan-word.

  8. #8 Zach Miller
    April 29, 2011

    Matamatas are so cool it hurts.

    Query: I don’t know crap about turtles, but I do know that snapping turtles often have ridged shells and seem to be “too big” for their shells, especially around the legs. Couldn’t help but notice that matamatas also have ridged shells and, from that ventral picture of the two populations of modern matamata, appear to be bulging out of their shells. The plastrons look quite small as well.

    Both also live in brackish swamps. Is there a phylogenetic connection between the two (aside from “they’re both turtles”)?

  9. #9 DMA
    April 29, 2011

    Did you know that wandering albatrosses have been recorded as having a 17.5 ft wingspan. I think someone also reported a marabou stork with a 16 ft wingspan, though I have to check the source. Measuring animals can be difficult from a distance, probably why there are megalania sightings.

  10. #10 220mya
    April 30, 2011

    The other thing that makes me suspicious about the size of the specimen in the Brazilian TV clip is that it appears that a substantial part of the carapace is reconstructed. Note that a good portion of it is smooth, brown and tan, which looks like plaster (or other) reconstruction, with only the irregular grey being the original bone.

    Zach: Chelus is a pleurodire, whereas snapping turtles (chelydrids) are eucryptodires. Basically, they’re on opposite sides of the phylogeny of crown turtles. Unfortunately the Wikipedia page for turtles is awful; just focusing on Linnean classification rather than phylogeny.

  11. #11 220mya
    April 30, 2011
  12. #12 Cameron
    April 30, 2011

    [from Darren: sorry, delayed by spam filter]

    DMA -

    The reference for the wandering albatross with a 17’6″ (5.33 m) wingspan is:

    Parkinson, C. (1900). The great birds of the Southern Seas. Cornhill Magazine (May). London.

    It also claims that the specimen is in the Sydney Museum and that the average wingspan of the species is 11’6″ (3.5 m)!

    I’m not certain about any reference that gives a Marabou a freakish 16 foot (4.87 m) span – which is certainly more than twice the average size – although Animal Facts and Feats seems to think the Meinertzhagen specimen could have been reliably recorded, despite having an incredible ~13 foot/4 m span.


    The Americas have yet another example of a turtle clade with a reduced plastron and three ridges on the shell – the Giant Musk Turtles. Unlike the phylogenetically distant Matamata, molecular phylogeny seems to consistently place Musk Turtles and Snapping Turtles as sister clades. Considering that other Musk Turtles don’t look so snapping turtle-y, it seems that Staurotypus and snappers are to some degree convergent… despite (I think) overlapping in range.

  13. #13 David Marjanović
    April 30, 2011

    to distinguish between Greek and Latin u

    Well, no, to distinguish between the sound [u], which occurred in both languages in Classical times (it is Greek ου and also occurred in αυ and ευ) and was used transcribed as u, and the sound [y], which was missing from Latin.

  14. #14 David Marjanović
    April 30, 2011

    Er, let me try again. In Classical Latin, Greek ου, αυ, ευ were transcribed as u, au, eu, while υ was kept as the specially imported letter y.

    A few hundred years earlier, υ had been pronounced [u], and for at least that long, education in Greek wasn’t widespread in Rome. Therefore, υ shows up as u in sufficiently old Greek loans, like Pollux instead of Polydeuces.

    Late Vulgar Latin inscriptions often or always represent υ as i. People who natively speak languages without [y] usually consider [i] to be the closest sound their native language has to offer.

    In modern Greek, υ is pronounced [i], and there is no more [y] in the language. However, this sound change is thought to have happened only about 1000 years ago.

  15. #15 Neil
    April 30, 2011

    A great series on these weird turtles. A 2m freshwater turtle would be quite a surprise on a pond dip!

    On a related subject – is there a real distinction between turtle and terrapin? i always thought it was terrapin if it lived in freshwater and turtle in marine habitats but I see the term freshwater turtle used more these days.

  16. #16 Cameron
    April 30, 2011


    Present-day American English uses ‘terrapin’ exclusively for Malaclemys, ‘tortoise’ for Testudinidae, and ‘turtle’ for everything else – including semi-terrestrial species (Terrapene, Cuora) which are ‘box turtles’.

    I have seen older American English sources use ‘terrapin’ more freely, and of course British English uses ‘terrapin’ for freshwater/brackish species, ‘turtle’ for marine species, and ‘tortoise’ for terrestrial species. Why the word ‘terrapin’, and Algonquin loanword is far more common on the other side of the Atlantic is rather strange.

  17. #17 Dino
    April 30, 2011

    Fossil matamatas are definitely bigger! This shell from the Miocene of Peru whose picture I posted on my blog some time ago is about 80cm long. I think it looks like your Orinoco turtle despite being from Peru!


  18. #18 Nick Gardner
    May 1, 2011

    Zach: “Query: I don’t know crap about turtles, but I do know that snapping turtles often have ridged shells and seem to be “too big” for their shells, especially around the legs. Couldn’t help but notice that matamatas also have ridged shells and, from that ventral picture of the two populations of modern matamata, appear to be bulging out of their shells. The plastrons look quite small as well.”

    regarding the function of the reduced plastron in snapping turtles, W4TP

  19. #19 Paul White
    May 1, 2011

    I don’t know about maximum size, but I’ve seen a few mata-mata in zoos that are rather large; I’d hazard 18-20″ carapace length although I obviously couldn’t measure them (zoos get so picky about people reaching into exhibits ;) ) Very strange and impressive. Them and our snappers and alligator snappers are probably my favorite turtles but they get too big for me to want to care for. Lots and lots of water changes would be required, even with good filters.

  20. #20 Dartian
    May 2, 2011


    Animal Facts and Feats seems to think the Meinertzhagen specimen could have been reliably recorded

    If the source is Richard Meinertzhagen, I would treat the claim with more than usual scepticism…

  21. #21 johannes
    May 2, 2011

    Rodents Of Unusual Size

    BTW, in spite of their name, the sprawling, polyprotodont ROUS from The Princess Bride can’t be rodents, or even therians.


  22. #22 Cameron
    May 2, 2011


    Yes it was Richard Meinertzhagen – I see that Wood made his statement prior to Meinertzhagen’s rampant fraud being uncovered. Considering the apparent lack of a preserved specimen, the extraordinarily freakish size, and improprieties of the reporter, I suppose the story is most likely a hoax. What a shame there weren’t Leptoptilos falconeri-sized monstrosities flying around in recent times.

  23. #23 Peter Johnston
    May 6, 2011

    Yesterday morning I was reading your excellent articles on the matamata, and by chance in the evening I was looking at Georges Cuvier 1805 (Leçons d’Anatomie Comparée vol 3) for historical background on reptile hyoids and found that the Baron described the hyoid apparatus of the matamata – “très-solide”.

  24. #24 herky stubby
    May 12, 2011

    I’m actually new in these blogs. Its the first time I saw this turtle? Is it?
    It has a snorkel! Very interesting! I’m not a scientist, nor do i have a brilliant mind such as those regular followers of this blog. But I am very interested in this kinds of topics. Specially when it comes to history…. The beginning of all things

    [from Darren: I'm not sure, but suspect this is a spammer, so am deleting their url. Let me know if you know otherwise, thanks]

  25. #25 steven
    July 4, 2011

    if i was a prehistoric turtle i’d attack my prey from far underneath it, using my, odd looking beak. or crawl in and eat my way out.