Fossil Friday: Sea Cows and the Abrupt Origin of Sirenia and Desmostylia
Günter Bechly
This Fossil Friday features the reconstructed skeleton of the “walking sea cow” Pezosiren from the Eocene of Jamaica, because today we look into the origins of the placental mammal order Sirenia, commonly known as sea cows. This order of herbivorous aquatic mammals includes only four living species of manatees and dugongs as well as the giant Steller’s sea cow, which was unfortunately exterminated by overhunting just a few decades after its discovery in the mid 18th century. Together with Proboscidea (elephants) and the extinct Desmostylia and Embrithopoda, the order Sirenia belongs to a subgroup of afrotherian mammals that is called Tethytheria (McKenna 1975, Gheerbrant et al. 2005, Nishihara et al. 2005, Berta et al. 2006, Seiffert 2003, 2007, Tabuce et al. 2008, Domning et al. 2010, O’Leary et al. 2013, Self-Sullivan et al. 2014, Domning 2018b), because they are thought to have evolved on the coast of the ancient Tethys Ocean (Heritage & Seiffert 2022).
The fossil record of sirenians is comparatively rich and even includes a lot of more or less complete skeletons. Domning et al. (1982: table 1), Domning (1994), and Self-Sullivan (2014)provided lists of all the ancient fossil sirenians known at that time. Diedrich (2013: fig. 1) and Springer et al. (2015: fig. 5) featured very good charts of the stratigraphic distribution of fossil sirenians, while Heritage & Seiffert (2022) provided the most comprehensive and up-to-date study or their phylogenetic relationships, ages or divergence, and paleobiogegraphic history. Based on molecular data, Sirenia are believed to have originated in the Earliest Paleocene about 65 million years ago (Springer et al. 2015: fig. 4), but when do sirenians first appear in the actual fossil record?
First Fossil Appearance
A putative fossil sirenian, Ishatherium subathuensis, was described by Sahni & Kumar (1980)and Sahni et al. (1980) from the early Eocene (Ypresian, 55.8-48.6 mya) Subhatu Formation in the Himalayas. It might represent the oldest known sirenian (Sereno 1982) but it could as well be an anthracobunid perissodactyl or a moeritheriid proboscidean (Wells & Gingerich 1983, Domning et al. 1986, Zalmout et al. 2003, Cooper et al. 2014, Rose et al. 2019). Domning et al. (1982) questioned the sirenian affinity as well as the dating. Two poorly preserved maxillary fragments from the Early Eocene (48.6-48.0 mya) Casamayoran Formation of Patagonia, which were named Florentinoameghinia mystica, were later attributed to Sirenia by Sereno (1982), but Domning (2001b) considered them as mammal of uncertain affinity.
The oldest unequivocal fossil sea cow is the quadrupedal prorastomid Prorastomus sirenoides from the Chapelton Formation (Yellow Limestone Group) in west-central Jamaica. It was first described by Richard Owen (1855), the famous enemy of Charles Darwin. Savage et al. (1994) re-described the holotype and described a second specimen. Even though Prorastomus is considered to be the most primitive known sirenian, it has some derived traits “that exclude it from the direct ancestry of other sirenians” (Savage et al. 1994). It is often stated to be of early Middle Eocene (48.6-40.4 mya) age, but according to Savage et al. (1994) the layers of the Stettin Member, where both known specimens were found, rather dates to the late Early Eocene (Ypresian) (Robinson 1988, Gold et al. 2018), thus about 49-50 million years ago.
About the same age of 48 million years (Late Ypresian / Early Lutetian), if not even slightly more ancient (Black 2013), is a petrosal bone from Chambi in Tunisia described by Benoit et al. (2013), which seems to be even more primitive than the Eocene prorastomids from Jamaica.
Only slightly younger are the prorastomid Pezosiren portelli from Chapelton Formation in Jamaica (Domning 2001a), which is of early Middle Eocene (Lutetian 48.6-40.4 mya, but rather 47 mya) age (Donovan 2002), and an unnamed prorastomid from the Lutetian of Senegal (Hautier et al. 2012). The skeletal reconstruction of Pezosiren, which is featured in this Fossil Friday article and became quite popular as the “walking sea cow” (Berta 2012, Prothero 2015), actually represents a composite of some hundred isolated bones collected from the same layers and believed to belong to the same species by Domning (2001a). The distal part of the tail and most of the feet bones were not found and added as “educated guesses.” This does not mean that the reconstruction is wrong, but at least some caution may be warranted.
From layers of the same Middle Eocene (Lutetian, 48.6-40.4 mya) age, several protosirenids have been found: Ashokia antiqua from the Harudi Formation in India (Bajpai et al. 2009), Libysiren sickenbergi from the Wadi Thamit Formation in Lybria (Domning et al. 2017), and Protosiren eothene from the early Lutetian Habib Rahi Formation in Pakistan, which is “virtually the same age as the prorastomids” according to Zalmout et al. 2003, early Lutetian). Furthermore, there are Sobrarsiren cardieli from the late middle Lutetian (SBZ15/C19r, 42 mya) Sobrarbe Formation in Spain (Díaz-Berenguer et al. 2018, 2020), and a possible sirenian vertebra from the early Lutetian of Israel (Goodwin et al. 1998). All these protosirenid stem sea cows still had four legs and had an amphibic way of life (actually Sobrarsiren is not a genuine protosirenid but more closely related to the fully aquatic sirenians).
The oldest fully aquatic sea cows, with reduced hind legs and (likely) a fluke, are the Middle Eocene (Lutetian, 48.6-40.4 mya) “dugongids” Anisosiren pannonica from Hungary (Kordos 1979, 2002), Eosiren abeli from Egypt (Sickenberg 1934), Eotheroides sp. (Samonds et al. 2009), and Sirenavus hungaricus from Felsogalla in Hungary (Kretzoi 1941, Kordos 1981, 2002). These advanced stem sirenians resembled modern dugongs, but crown group Sirenia first appear with genuine Dugongidae at the Eocene/Oligocene boundary about 33.9 million years ago (Heritage & Seiffert 2022).
Nevertheless, based on dental synapomorphies, Domning et al. (2010) had considered Eotheroides as a crown-group sirenian closer related to Dugongidae than to Trichechidae, which led to the use of this taxon as calibration data point for the age of crown group Sirenia by Benton et al. (2015, also see Fossil Calibration Database), who erratically also cited “Gheerbrant et al. (2005)” in support of this hypothesis even though these authors did not even mention Eotheroides. Such a crown group position was also suggested for Eotheroidesand Eosiren by the cladistic studies of Savage (1976), Springer et al. (2015), Vélez-Juarbe & Domning (2015), and Balaguer & Alba (2016). Domning et al. (2017) not only recovered Eotheroides as a crown group sirenian in Dugongidae, but even resolved the quadrupedal protosirenids Ashokia, Libysiren, and Protosiren as crown group sirenians closer related to Trichechidae than to Dugongidae (compare Savage 1976). This would make fully aquatic sirenians diphyletic, as already indicated by Diedrich (2013). On the other hand, the studies by Domning (1994), de Buffrénil et al. 2010, Díaz-Berenguer et al. (2018), and especially Heritage & Seiffert (2022) had the protosirenid genera as well as Eotheroides and Eosirenresolved well outside the crown group. I am sorry to say that phylogenetics turns out to be nothing but junk science when you look at the actual studies and their highly incongruent results and not just at the fancy polished text book figures. Darwin’s modern bulldogs like Richard Dawkins and Jerry Coyne are either totally ignorant or deliberately spreading falsehoods when they make their readers believe that there is one well-established tree of life. Nothing could be further from the truth. Phylogenetics is a mess!
Anyway, the sirenian fossil record, just like that of whales, is remarkable in featuring early representatives that still were quadrupedal and amphibic, which arguably supports common descent of sea cows from terrestrial ancestors and thus a secondary adaptation to a fully aquatic way of life (Sickenberg 1931, Heal 1973, Savage 1976, Domning & Gingerich 1994, Domning 1982, 2000, 2001a, 2001b, 2001c, 2018b, Berta et al. 2006, Uhen 2007, de Buffrénil et al. 2010, Berta 2012, 2017, Self-Sullivan et al. 2014, Prothero 2015, Díaz-Berenguer et al. 2020, Heritage & Seiffert 2022, and Wikipedia).
Everything OK So Far?
So, is every thing OK with Darwinism after all? No so fast. Actually, there are some problems that do not square well with a Darwinian scenario:
1.Sirenians appear abruptly in the fossil record at the onset of the Middle Eocene, together with other placental mammal orders, without a long transitional series establishing any kind or gradual development.
2.There is a distinct morphological gap between the quadrupedal forms (Prorastomidae and Protosirenidae) and fully aquatic sea cows (Dugongidae and Trichechidae).
3.Fully aquatic sirenians that looked like modern dugongs appear more or less around the same time as the primitive quadrupedal stem sirenians. Thus, sirenians immediately appear with a large diversity in the Middle Eocene.
4.The origin of Trichechidae is totally in the dark, without any clear connection to the Dugongidae and their stem line.
Even though the fossil evidence in my view indeed supports common descent, it contradicts Darwinian expectations and does not at all support an unguided mechanism of evolution, which would imply slow and gradual transformations with small changes accumulating over long periods of time via numerous transitional species that only slightly differ from each other. The saltational pattern in the fossil record rather suggests very quick and dramatic changes within only a few transitional species, which arguably requires an infusion of new information from outside the system (also known as intelligent design).
Before we move on, I would like to share a little trivia that shows under what strange circumstances some important fossil finds were made: Voss et al. (2019) published one of the oldest fossil sea cows (Prototherium spec.) from Europe, which was found in Middle Eocene limestone from Spain that is dated to an Early Bartonian age of about 40 million years (also see Astibia et al. 2010). More precisely, the fossil was discovered in the paving stones of a shopping mall in the city of Girona in Catalonia, where thousands of people walked over this treasure for several years before its importance was recognized.
The Enigmatic Desmostylia
In the context of the fossil history of sirenians it is also necessary to discuss the extinct mammal order Desmostylia, because these semiaquatic herbivorous marine mammals were somewhat similar to early quadrupedal sirens and possibly closely related. Initially they were even erroneously considered to be sirenians (e.g., Hannibal 1922, Simpson 1945). The degree of their aquatic adaptation is still a matter of scientific debate (Inuzuka et al. 1994, Domning 2002, Clementz et al. 2003, Gingerich 2005, Uhen 2007, Hayashi et al. 2013). Desmostylia are only known from the Early Oligocene to Late Miocene of the North Pacific rim in 2-4 families and 10-12 genera with 13-14 species (Beatty 2009, Domning 2018a, Matsui & Tsuihiji 2019). Some relic species may have survived until the Pliocene (Kimura 1966). The most primitive and among the oldest representatives are the Behemotopsidae (Ray et al. 1994, Beatty & Cockburn 2015, Domning 2018a), while the families Paleoparadoxiidae and Desmostylidae (= Cornwalliiidae) are more derived and usually younger. But if we look more closely into the data, the fossil record tells a somewhat different story:
According to PaleoDB all the earliest desmostylian fossils are of Chattian age (28.4-23.03 mya) and include:
Behemotopsidae:
Behemotops proteus from the Pysht Formation in Washington and the Sooke Formation on Vancouver Island 24.8-24.1 mya.
Behemotops katsuiei from the Moravan Formation in Japan
Seuku emlongi from Yaguina Formation in Oregon
Desmostylidae:
Ashoroa laticosta from the Moravan Formation in Japan (Inuzuka 2000)
Cornwallius sookensis from the Sooke Formation on Vancouver Island, Yaquina Formation in Oregon, Unaslaska Formation in Alaska, and Baja California in Mexico
Many sources (including Wikipedia) still cite an Early Oligocene (Rupelian) age for Behemotops, which would make it to the oldest desmostylian. However, this appears to be based on an obsolete dating of the Pysht Formation in Washington (Domning et al. 1986, Barnes & Goedert 2001), which has been re-dated as Late Oligocene / Chattian (Prothero et al. 2001, Beatty & Cockburn 2015). Actually, the layers where Behemotops proteus was found in the Pysht Formation (Chron C6Cr) have been more precisely dated to 24.8-24.1 million years ago (Prothero et al. 2008). A specimen of Behemotops from Hokkaido in Japan was initially believed to be older, but rather was more or less contemporaneous with the North American congeneric specimens (Saito et al. 1988).
In one of the more recent studies on Desmostylia the authors presented a diagram of the stratigraphic distribution (Matsui & Tsuihiji 2019: fig. 4), in which some of the more derived Desmostylidae not only appear together with the more primitive Behemotopsidae, but even predate them in the fossil record. The desmostylid Ashoroa laticosta is shown around the Rupelian/Chattian boundary about 29-27 mya and Cornwallius sookensis even from the middle Rupelian about 31 mya. The authors do not cite their sources for these more precise stratigraphic ranges, but they are congruent with the dating for all the localities of Cornwallius sookensis by Beatty (2002, 2006a, 2006b, 2009) to a Late Oligocene / Zemorrian (33.5-22 mya) age, which overlaps with the Rupelian and Chattian.
Finally, there is an isolated atlas vertebra from an undetermined putative desmostylian from the Lincoln Creek Formation in Washington, which dates to the Eocene/Oligocene boundary, about 37-36 million years ago (Prothero & Armentrout 1985), and arguably represents the oldest known fossil record of Desmostylia (Barnes & Goedert 2001).
Unlike Any Living Mammals
Desmostylians were quite unlike any living mammals, maybe resembling hippos, but their life reconstruction is still controversial even though complete skeletons are known (Inuzuka 1984, Halstead 1985, Domning 2002). The evolutionary origins of Desmostylia remain totally in the dark, and also their phylogenetic affinities are still hotly debated. Usually, they are considered as relatives of elephants and sea cows within Tethytheria (Reinhart 1953, McKenna 1975, Domning et al. 1986, Novacek & Wyss 1987, McKenna & Bell 1997, Domning 2018a), but is unclear if they are closer related to elephants (Domning et al. 1986, Clementz et al. 2003, Berta et al. 2006, Uhen 2007, Beatty 2009, Asher & Seiffert 2010) or to sea cows (Vélez-Juarbe & Domning 2015).
However, several new studies (Cooper et al. 2014, Rose et al. 2019; also see Beatty & Cockburn 2015 and Heritage & Seiffert 2022) suggested that desmostylians are no tethytheres at all but rather odd-toed ungulates (Perissodactyla), thus not even members of the afrotherian clade. This would align with the fact that demostylians have “no African record whatsoever” (Asher et al. 2003). Gheerbrant et al. (2016) recovered desmostylians either as sister group of Paenungulata or as stem perissodactyls, but also found evidence for long-branch attraction between Desmostylia and Paenungulata (Tethytheria), suggesting possible convergent similarity. Therefore, Matsui (2017) and Matsui & Tsuihiji (2019) considered desmostylian affinities as controversial. If a perissodactyl relationship should be corroborated, then all the morphological similarities with Tethytheria would have to be reinterpreted as independently acquired convergences and thus not based on common ancestry. If you followed my previous articles in this series, this will hardly come as a surprise. I recently wrote a Fossil Friday article about the strange phenomenon of horizontal tooth displacement that independently originated three times within Tethytheria and confused scientists (Bechly 2022). As I said: phylogenetics is a mess and calling it science is misplaced and overselling this kind of fancy storytelling and educated guessing based on highly incongruent data!
Next Fossil Friday we will look into the final member of the Afrotheria, the order Proboscidea, which includes elephants and their fossil relatives.