New publication: A re‐description of the late Carboniferous trematopid Actiobates peabodyi from Garnett, Kansas (Gee & Reisz; Anatomical Record)

Title: A re‐description of the late Carboniferous trematopid Actiobates peabodyi from Garnett, Kansas
Authors: B.M. Gee, R.R. Reisz
Journal: Anatomical Record
​DOI to paper: 10.1002/ar.24381

General summary: Back in 1973, American paleontologist Theodore Eaton described a new dissorophoid from the late Carboniferous of Kansas. Represented by the pancake-iest of the pancaked fossils, Actiobates was confusing because it looked like a cross between a dissorophid and a trematopid, so much so that Eaton suggested that these weren't distinct groups (also bearing in mind that terrestrial amphibamiforms were lumped in with dissorophids at the time). Eaton's taxonomy wasn't widely accepted, but it took arguably decades to fully resolve the dissorophid-trematopid issue, which was compounded by disputed affinities of Ecolsonia, first described in 1969 but subsequently known from much better material described in 1985.

The reconstruction to the right is the only figure of Actiobates in any form (drawing, photograph, interpretative line drawing, interpretative dance) that existed until it was photographed in brief in a paper on other trematopids by Andrew Milner in 2019. To top it off, Eaton alluded to a large portion of the postcranial skeleton that was articulated with the skull and indicated that he was going to describe it, but it never happened before his passing in 1981 (he retired from a professorship at the University of Kansas in 1977). Many trematopids have no postcrania, which has complicated their phylogenetic relationships a bit, but there is obviously a need to describe postcrania when it does exist. Actiobates has also rarely been included in phylogenetic analyses because it's hard to assess the accuracy of the original description without photos to validate the reconstruction. The holotype and only known specimen has also been categorized as a "larval form" and explicitly excluded from these analyses. All of these uncertainties clearly warranted a re-examination of Actiobates, which happened mostly by accident in the first year of my PhD when we were doing a return-specimens/collections trip to Kansas (note: they have metoposaurid material in their collection too).

Eaton's original reconstructions of Actiobates peabodyi.

On the beaches of Kansas

Actiobates comes from a well-known late Carboniferous locality in Kansas near the town of Garnett. The site, located on a private farm, was discovered in 1931 by Norman Newell and then periodically worked over the next six decades by various paleontologists. Kansas is now just a flat grassland, but it's been repeatedly invaded by oceans over the course of North America's evolution. A transgressive-regressive sequence (changing sea levels) produced a number of distinct layers within the locality that preserve various other organisms, such as marine invertebrates typical of shallow-water habitats and terrestrial plants typical of conifer forests. The vertebrates largely come from the middle of the Rock Lake Shale. The model proposed by Reisz et al. (1982) is a freshwater stream channel incising into limestone beds that were exposed as sea levels first dropped and then a flooding of this region in a subsequent sea level rise, which produced estuarine mudflats. The site is definitely better known for its amniote assemblage, which includes Petrolacosaurus kansensis, long regarded as the earliest known diapsid, and a number of pelcyosaur-grade synapsids like Edaphosaurus and Ianthasaurus. Actiobates is in fact the only definitive non-amniote - a second one, Hesperoherpeton (Peabody, 1958), has been regarded as everything from an anthracosaur to a rhipidistian fish but is most likely some kind of terrestrially capable temnospondyl and can't clearly be differentiated from Actiobates (Milner, 1985). The vast majority of specimens from the locality are reposited at either the Kansas Museum of Natural History in Lawrence or at the Royal Ontario Museum up here in Toronto. 

The extremely undiagnostic holotype of Hesperoherpeton garnettense (figure from Milner, 1985).

The perils of pancakes

Photograph and interpretive line drawing of the skull in dorsal profile (figure 2 from the paper).

​As is common for many Carboniferous tetrapods, the holotype of Actiobates is pancake-flat; the entire skull is about the thickness of a normal kettle chip. For this reason, you rarely see these specimens preserved from more than one side, or if they are, one side is prepared and then embedded in a transparent resin to stabilize it for preparation from the other side (an old-school method that doesn't usually age well). It's not entirely clear what happened, but Actiobates ​was damaged at some point, probably post-publication, because it's impossible to get Eaton's reconstruction from the present state of the specimen (below). You only need half the skull to get the whole picture, but the specimen presently lacks details on regions such as the medial edge of the orbit or the tip of the snout that Eaton reconstructed with an apparent semblance of confidence. Probably someone was trying to clean up part of the skull roof or something and cracked through the entire skull on account of it being so thin. In any event, this has really complicated a lot of interpretations because many of the diagnostic features or trematopid synapomorphies that were listed by Eaton aren't preserved anymore (assuming they were to begin with). We took all of these things into account in un-pancaking the specimen to form a new reconstruction (parts C and D below) to try to get more reasonable proportions, and the diagnosis had to be substantially tweaked to account for this.

Comparison of Eaton's reconstructions (A and B) with our new reconstructions (C and D) (figure 1 from the paper).

Young at heart

Actiobates is the oldest trematopid in North America, with the Garnett locality pinpointed to the late Kasimovian in the neighborhood of 305 Ma - this is also the oldest fossil that I've published on to date! Depending on the inferred correlation with the type locality of the enigmatic dissorophid Astreptorhachis from Ohio, Actiobates may be the oldest known olsoniform on the continent. Therefore, it holds some import for understanding the anatomy of early olsoniforms, which is largely lacking in the postcranial region for trematopids and lacking in just about everything for dissorophids. 

The flip side to this is the relative maturity of the individual represented by the holotype of Actiobates. Berman et al. (1985) argued that it was a larval trematopid, in part based on another larval trematopid described by Olson (1985). Well if you pay attention to my publications, Olson's specimen turned out to be a mature amphibamiform, not a larval trematopid. At the time, we had a really poor understanding of ontogeny in trematopids (arguably it is still not great), and so a number of the arguments from back then for a larval status haven't aged particularly well. Eaton described elements of the hyobranchial apparatus, but we didn't clearly identify anything that looked like it would be a part of a structure supporting internal gills (or external gills), and these structures weren't part of Eaton's reconstruction, so we think he mis-identified some ribs or something else from the anterior trunk region. The holotype is small for a trematopid, but it's hardly the smallest - there are three other taxa with smaller type specimens (Mattauschia laticeps, Mordex calliprepes, Rotaryus gothae), and none of those are considered to be larval forms. The postcranial skeleton (this is why postcrania are important) is appreciably well-ossified - the wrist and ankle bones are there, all of the limbs have distinct features, etc. It's not fully-grown - the coracoids weren't ossified, for example, but just about everything that we expect is there. 

The postcranial skeleton of Actiobates peabodyi (figure 4 from the paper). [by the way, the typeset figure is unusually small for some reason; if you want a larger copy, email me]

The humble beginnings of trematopids

There's a lot we don't know about trematopids' early evolution, and that of olsoniforms more broadly (the crappy dissorophid record in the Carboniferous doesn't help). The Carboniferous trematopids are widely dispersed across a range from New Mexico to the Czech Republic, and almost all of them are represented only by one relatively complete specimen. What we can see from the skull of Actiobates is a mixture of features between the oldest trematopids at Nýřany and the more "prototypical" morphology of Permian taxa - for example, the frontal contributes to the orbital margin, but the nasal lacks a prominent ventrolateral expansion that would subdivide the nostril from above. Eaton reported caniniform marginal dentition, historically a hallmark of trematopids (but not in the Nýřany taxa), but we couldn't verify that in our re-examination; the predicted region doesn't have good preservation of the teeth. We did identify a lateral exposure of the palatine (LEP), which is a feature increasingly identified in dissorophoids in the post-1970s studies that often was missed because it wasn't know that this exposure was a thing. In Actiobates, the position is the same as in most other dissorophoids: sandwiched between the lacrimal and the jugal and above the maxilla. The postcrania are remarkably similar to other trematopids. Not that many differences have been previously identified across the whole range of trematopids other than possible dorsal closure of the intercentra in large Acheloma, but aspects of the skeleton and the degree of development are quite similar to those of Permian forms (e.g., a stout scapula; short, longitudinally expanded iliac crest). Not clear whether there was a supinator process, a historical trematopid synapomorphy that may only appear in adults (per Milner, 2019). No osteoderms, as in most trematopids.

Collectively, this indicates that most of the trematopid bauplan showed up early in their evolutionary history and then persisted largely unchanged, somewhat contrary to dissorophids, which see an explosion in osteoderm morphology and some distinct differences in limb proportions and rib morphology. In essence, trematopids probably had what it took to be pretty successful and widely dispersed (they extend from New Mexico to Germany in the Permian) but were restricted spatially by wetter climates in the Carboniferous that would have favoured other taxa or at least not favoured terrestrial dissorophoids. As things got hotter and drier, the dissorophoids probably had a greater advantage over taxa like eryopoids and edopoids that allowed them to become extremely numerous in terrestrial paleoenvironments. Need more dissorophids to examine this further!

---

Refs

• Berman DS, Reisz R, Eberth DA. 1985. Ecolsonia cutlerensis, an Early Permian dissorophid amphibian from the Cutler Formation of north-central New Mexico. New Mexico Bureau of Mines and Mineral Resources Circular 191: 1-31. [PDF]
• Eaton TH. 1973. A Pennsylvanian dissorophid amphibian from Kansas. Occasional Papers of the Museum of Natural History, The University of Kansas, Lawrence, Kansas 14: 1-8. [link]
• Milner AR. 1985. On the identity of the amphibian Hesperoherpeton garnettense from the Upper Pennsylvanian of Kansas. Palaeontology 28(4): 767-776. [link]
• Milner AR. 2019. Two primitive trematopid amphibians (Temnospondyli, Dissorophoidea) from the upper Carboniferous of the Czech Republic. Earth and Environmental Science Transactions of the Royal Society of Edinburgh 109(1-2): 201-223.
• Olson EC. 1985. A larval specimen of a trematopsid (Amphibia: Temnospondyli). Journal of Paleontology, 59(5):1173-1180. [link]
• Peabody FE. 1958. An embolomerous amphibian in the Garnett fauna (Pennsylvanian) of Kansas. Journal of Paleontology 32(3): 571-573.
• Reisz RR, Heaton MJ, Pynn BR. 1982. Vertebrate fauna of late Pennsylvanian rock lake shale near Garnett, Kansas: Pelycosauria. Journal of Paleontology ​56(3): 741-750.