Arguments Against Apachesaurus: the problem of small metoposaurids

Had to skip last week because of a hectic last week of classes schedule, but I'm back this week with a blog post that's near and dear to my heart and that's been on my mind a lot between doing some collections work with Aaron Kufner (PhD student at the University of Wisconsin-Madison) a few weeks back and going out to Madison this week to give a talk on the subject matter at their student symposium. It's almost the two-year anniversary of my first paper too! For a lot of folks, you already know what I'm going to talk about, but for the uninitiated, I'll be talking about my favourite temnospondyls this week - metoposaurids - more specifically all the problems with a particular taxon, Apachesaurus. I've published several papers on this taxon and probably seem like I have a vendetta against it at this point... Because I didn't even have a website when most of these papers were published, let alone blog, this is also a great chance to read up on some of my existing publications outside of my dissertation work!

Metoposaurids?
Really everyone should know about metoposaurids, but for people who have been unfortunate enough to never cross paths with one, they are an extremely abundant Late Triassic freshwater temnospondyl found across much of North America and western Europe but also in northern Africa, Madagascar, and India. Their name comes from the position of their eyes, which are relatively far forward: 'metoposaurid' = 'front lizard.' They're in the medium-sized range (adults around 2 m long) for Mesozoic temnospondyls, but you still wouldn't want to run into one. They're probably one of the most common fossils in many Late Triassic deposits, but a lot of people sadly ignore them (or deliberately step on them *cough cough* archosaur workers).

Photo of the holotype of Apachesaurus (housed at the UCMP in Berkeley, CA). My lens cap for scale.

Comparative reconstructions of the two metoposaurids from the Chinle Formation: Koskinonodon perfectus and Apachesaurus gregorii. Art by Jeff Martz; one Bill Parker for scale.

So Apachesaurus? 
As I noted above, most metoposaurids got appreciably large. The one apparent exception to this is Apachesaurus gregorii, first described from the Redonda Formation of New Mexico by Hunt (1993) (but mentioned by a number of other workers for over a decade prior). This taxon occurs relatively late in the Triassic and was apparently unique in being substantially smaller than all other metoposaurids, which led to a suggestion that in the last stages of their evolution, metoposaurids got really small in response to changing environmental conditions. I've included a literal screenshot of the page (Hunt, 1993:85) below because a certain someone told me at a conference that size diminution had never been claimed and that Apachesaurus was obviously just another regular ol' metoposaurid. 

The situation
Hunt (1993) justified Apachesaurus as a new taxon, specifically of a small-bodied and definitely not juvenile metoposaurid, through four criteria:

1. Juveniles of other metoposaurids look like their "adult" forms
2. Elongate intercentra (part of the backbone) that are diagnostic for Apachesaurus are also found with non-elongate intercentra of about the same size, so elongation cannot be a juvenile feature
3. There are thousands of vertebrate specimens from the Redonda Formation, and none of them belong to large metoposaurids
4. The pineal foramen is posteriorly situated within the parietal bones

The problem(s)
The above lines of evidence are not fundamentally bad justifications. But there are a lot of potential alternative interpretations that are equally likely. Considering that the sample size of metoposaurids is better than 95% of other extinct tetrapods, you'd think we actually know basically everything there is to know. FALSE. There are perhaps hundreds of skulls of large-bodied metoposaurids, including dozens that are complete. The sample size of small-bodied (either juveniles or small "adults') metoposaurids (#1) is around two dozen, and most of those are so incomplete or beat up that it can't even be determined which genus they belong to without relying on secondary support like which geographic region they occur in. This isn't unique to metoposaurids (in fact almost any extinct tetrapod is plagued by these issues), but it does mean that we still know next to nothing about metoposaurid development - which anatomical features change, when they change, how much variation there can be early in development, etc (#2, #4). Additionally, although the absence of large-bodied metoposaurids (#3) sounds compelling, it's important to remember that the fossil record is always biased. Almost never do we get even a half complete growth series of most extinct tetrapods, and major growth stages are entirely absent from the fossil record in many cases. For example, fossil amphibian larvae (i.e. tadpoles) are exceptionally rare (see Yuan et al., 2004 for an example). This doesn't mean that tadpoles didn't exist, but rather that they were not preserved for a number of reasons (e.g., no bones, not living in environments that get preserved, decay easily, etc.). As scientists like to say, absence of evidence is not evidence of absence. 

The most important question is how do you prove that a taxon represents a diminutive form (implying a major developmental shift to substantially smaller body size)? I do want to point out that although a lot of people (including myself) use the term 'dwarfism' to refer to this purported phenomenon in Apachesaurus, you need a phylogeny to show that it's a trend from large -> small and not from small -> large (which gets all kind of complicated because dwarf taxa can screw up your phylogeny, and there is no good one for metoposaurids anyway...). So I'm just referring to it as size diminution here, as I did with my last paper on the matter (Gee & Parker, 2018b). Morphological observations may clue you in, but ultimately you can only make relative assessments of an animal's age and maturity from its skeleton alone. You can't look at someone's skeleton and know that they were born on August 23, 1978. This is naturally even more poignant for extinct tetrapods for which we know next to nothing about growth patterns. 

I've gone three different routes to try to better tackle this problem using specimens from Petrified Forest National Park (PEFO) in Arizona:

1. ​Assess maturity specifically of small Apachesaurus and Apachesaurus-like specimens through paleohistology (Gee & Parker, 2017; Gee & Parker, 2018b)
2. Assess growth patterns in metoposaurids by comparing large and small specimens through paleohistology (Gee et al., 2017)
3. Look for large metoposaurids in similarly aged rocks to those that Apachesaurus occurs in (Gee & Parker, 2018a)

Photograph and line drawing of a small partial skull of a metoposaurid (figure from Gee & Parker, 2018b).

The intercentra that go with the above skull (figure from Gee & Parker, 2018b).

Approach #1: maturity determination in small metoposaurids
The most immediate way to test whether Apachesaurus is a diminutive metoposaurid would be to section the holotype. Bone histology can give you more definitive data about an animal's development (amongst other things). Too bad no one lets you section holotypes... Also, the holotype is only a skull, which is even less likely to be green-lighted to be partially destroyed. So the next best thing was finding similar small metoposaurid specimens with associated postcranial material that we could histologically section and then use to calibrate morphological features in the skull with a relative age. I more or less rediscovered or salvaged three specimens from PEFO with both postcranial material and partial skulls in association. The first was just sitting in the back of a mystery cabinet, the second came back from being on loan in about 10 pieces that ended up fitting with another 20 pieces that were divided between 10 unmarked and poorly labeled bags that I only recognized fit together because of the relatively unique preservation (that's the one on the left), and the third...uh had to be rescued after the jacket collapsed during a hasty field collection by people who will not be named.

The fact that all three of these skulls have associated intercentra is important for establishing a direct correlation between a skeletal element that is not particularly diagnostic but that can be histologically sampled and thus directly aged (the intercentra) and a skeletal element that is diagnostic but not able to be histologically sectioned (the skull). Without that association, it's impossible to know what size postcranial elements go with what size skull, so that correlation isn't possible. A good example is the holotype of Apachesaurus. It's only the skull and jaws, but it's found in a bonebed with dozens of intercentra. Do any of those go to that specimen? Maybe, but maybe not. You can't say for sure. 

One skull looks like Apachesaurus. One skull sorta looks like Apachesaurus. One skull does not look like Apachesaurus (the largest of the three). All three have histologically immature intercentra. So now we can make some correlations between features that seem to be unique in small metoposaurids and ontogenetic immaturity (e.g., total absence of a tabular horn and an otic notch). The largest of these three specimens has a little bit of a tabular horn and a modest otic notch, which fits into this picture.

Approach #2: growth patterns in metoposaurids
​If Apachesaurus truly were a small-bodied adult metoposaurid, then it should exhibit the same osteological markers of maturity but at a much smaller size when comparing the same skeletal element. But which element? Limbs are the traditional go-to, but we don't really have too many limbs at Petrified Forest. So instead we used intercentra, a large disc-like element in the backbone. These are nice because they're very robust, extremely common, and easy to section. So we went out and grabbed a bunch of differently sized ones, looking for both small and elongate ones (allegedly diagnostic for Apachesaurus) and large and not-so-elongate ones. You can see the sagittal sections (front-to-back cut) on the right from that paper (Gee et al., 2017).

Amongst other things, you may notice that there is a very distinct gradual trend of increasing size and increasing relative height (i.e. shortening). A-C are the smallest and the most elongate, and J is the largest and the most truncated. This is also born out in the bone histology and the microanatomy; the smallest ones show no growth marks, no remodeling, and lots of what we call calcified cartilage (not the pathological kind that apparently can be found in humans). These are all markers of ontogenetic immaturity! In fact, the smallest ones still have a thoroughgoing canal for the notochord, a cartilaginous rod found in early development in vertebrates that is rarely found in adult individuals. Furthermore, as you increase in size in our sample, the relative abundance of all of these features changes in a unidirectional fashion. It's not until the largest one that we see any evidence of growth marks, and even then there are only two, so even that animal was still quite young when it died.

Sagittal sections of variably sized metoposaurid intercentra. All scale bars = 4 mm (figure from Gee et al., 2017).

Comparisons of relative abundance of secondary remodeling (more indicates maturity) and calcified cartilage (more indicates immaturity) across three size bins. A-B is the smallest; C-D is a medium-sized one; E-F is the largest one. Figure from Gee et al. (2017).

Partial mandible of a large metoposaurid (scale bar = 3 cm) from the Petrified Forest Member (Norian) of the Chinle Formation (figure from Gee & Parker, 2018a).

Approach #3: looking for large metoposaurids
​One of my biggest suspicions regarding the apparent absence of large-bodied metoposaurids in the later parts of the Late Triassic is that people just didn't collect them. The sample size for metoposaurids is so atypically large that a lot of people ignore specimens in the field unless they're an immaculate skull. Of course, this misrepresents the fossil record when one looks in collections because their paucity could actually reflect selective choices made by people who often do not work on temnospondyls (no, I'm not bitter at all, why do you ask?). 

So it was quite convenient when Bill Parker walked into the lab and told me to go look at something on the floor in collections, which turned out to be the partial jaw of a distinctly large-bodied metoposaurid that you can see on the left. Only the posterior half of this jaw is preserved, but it already measures 29 cm and could have been twice as long. That's a big metoposaurid! 

Primary conclusions

• Intercentra elongation in small metoposaurids from the Chinle Formation is associated with immaturity. 
• Intercentra show a strong correlation between relative height:width ratios (i.e. elongation) and size.
• There is no way to tell whether a small, elongate intercentrum belongs to a juvenile of a large-bodied taxon or to a purported small adult without histology in the first place, so it should not be a diagnostic feature until we find a clear juvenile that is properly referable to another existing metoposaurid that has articulated intercentra (still looking for that...).

So how do we explain the patterns that we see and reconcile them with previous interpretations?

1. Juveniles of other metoposaurids look like their "adult" forms: There are very few juvenile metoposaurids that can be identified to a particular taxon, so the comparative dataset is limited. Also, this does not rule out the option that Apachesaurus could represent a juvenile of a distinct large-bodied metoposaurid taxon.
2. Elongate intercentra (part of the backbone) that are diagnostic for Apachesaurus are also found with non-elongate intercentra of about the same size, so elongation cannot be a juvenile feature: First of all, intercentra vary in shape and size along the backbone. Caudals (tail) don't look like cervicals (neck). It's also important to point out that there are other stereospondyls in the Late Triassic - they're rare, but they exist. And lastly, individuals can vary in development such that two of the same size may not be the same age. So just because there are small and shorter intercentra doesn't mean that elongation isn't a juvenile feature.
3. There are thousands of vertebrate specimens from the Redonda Formation, and none of them belong to large metoposaurids: I've promoted this idea of niche partitioning between life stages in metoposaurids, which makes a lot of sense when you think about what a lot of living animals do, including modern amphibians. Essentially, adults and juveniles are not always living together, either in the same space or in a social sense. This can happen for lots of reasons. Juveniles may be less mobile or prefer different habitats (e.g., more vegetated ones for protection). Adults may move away from breeding ponds to avoid cannibalism. If some of these habitats are more likely to be preserved, you can end up with major biases in the preservation that remove entire growth stages from the fossil record. We notice at PEFO that a lot of the small metoposaurids are found in pond deposits that also preserve rare animals and in greater articulation (Loughney et al., 2011). So if large metoposaurids don't live in ponds normally, of course they wouldn't be found there. This probably explains the major shift in relative abundances of metoposaurid size bins because higher energy river settings are the more common deposit in earlier parts of the Late Triassic and preserve mostly large individuals.
4. The pineal foramen is posteriorly situated within the parietal bones: The degree and timing of this shift in metoposaurids is unclear. The so-called "adult" condition can occur at different times and in a piecemeal fashion; it's not like one day you wake up and have progressed through the entirety of puberty. Natural development does not occur at the same time or pace across the body, so while certain features can be associated with adults and not found in really young individuals, they may show up well before the animal achieves anything comparable to its maximum size. Because we have such a poor record of juvenile metoposaurids, we don't have good constraints on this feature. 

TLDR: There is no evidence that Apachesaurus is a small-bodied adult. 

P.S. I want to go on record as saying that I did not change the Wikipedia page for Apachesaurus, which now redirects to Koskinonodon (the large-bodied metoposaurid that co-occurs with Apachesaurus​). I think it's funny, but don't believe everything you see on the internet. This misconception probably comes from my first paper in PeerJ where I suggested that Apachesaurus might be a juvenile of an existing metoposaurid taxon, the most likely candidate of which is Koskinonodon. I didn't formally synonymize anything though, didn't reiterate that specific claim in more recent papers, and don't currently think that that's true, and suffice to say, it's essentially someone overextrapolating from my work; frankly I have no idea who edits a lot of these Wikipedia pages...

---

Refs

• Gee BM, Parker WG. 2017. A juvenile Koskinonodon perfectus (Temnospondyli, Metoposauridae) from the Upper Triassic of Arizona and its implications for the taxonomy of North American metoposaurids. Journal of Paleontology 91(5):1047-1059. doi: 10.1017/jpa.2017.18
• Gee BM, Parker WG, Marsh AD. 2017. Microanatomy and paleohistology of the intercentra of North American metoposaurids from the Upper Triassic of Petrified Forest National Park (Arizona, USA) with implications for the taxonomy and ontogeny of the group. PeerJ 5: e3183. doi: 10.7717/peerj.3183
• Gee BM, Parker WG. 2018a. A large-bodied metoposaurid from the Revueltian (late Norian) of Petrified Forest National Park (Arizona, USA). Neues Jahrbuch für Geologie und Paläontologie-Abhandlungen 287(1):61-73. doi: 10.1127/njgpa/2018/0706
• Gee BM, Parker WG. 2018b. Morphological and histological description of small metoposaurids from Petrified Forest National Park, AZ, USA and the taxonomy of Apachesaurus. Historical Biology, pp.1-31. doi: 10.1080/08912963.2018.1480616
• Hunt AP. 1993. Revision of the Metoposauridae (Amphibia: Temnospondyli) and description of a new genus from western North America. In Aspects of Mesozoic Geology and Paleontology of the Colorado Plateau, ​ed. Morales, M. Museum of Northern Arizona Bulletin 59:67-97.
• Loughney KM, Fastovsky DE, Parker WG. 2011. Vertebrate fossil preservation in blue paleosols from the Petrified Forest National Park, Arizona, with implications for vertebrate biostratigraphy in the Chinle Formation. Palaios 26(11):700-719. doi: 10.2110/palo.2011.p11-017r
• Yuan C, Zhang H, Li M, Ji X. 2004. Discovery of a middle Jurassic fossil tadpole from Daohugou region, Ningcheng, Inner Mongolia, China. Acta Geologica Sinica 78(2):145-148. [Link]