History of side-necked turtle diversification revealed

Brazilian paleontologists have elucidated the evolutionary history of today’s Yellow-spotted Amazon River turtle and of the oldest turtle in Brazil, which inhabited the Northeast 125 million years ago.

By Peter Moon  |  Agência FAPESP – What does the oldest turtle in Brazil, which inhabited the Northeast region 125 million years ago, have in common with the largest turtle that ever existed, a monster with a 3.5 m shell that swam in the fresh waters of a megaswamp 10 million years ago? And what do these two long-extinct reptiles have to do with the Yellow-spotted Amazon River turtle, Podocnemis unifilis?

The answer is that all three belong to the Pleurodira suborder of side-necked turtles, found today only in the southern hemisphere. They are called side-necked because they fold the neck horizontally under their shell, tucking it into the space in front of one of their front legs. In this respect, they differ from Cryptodira, the other living group of turtles, which lower the neck vertically to pull the head straight back into the shell. Brazil’s Yellow-footed tortoise and Red-footed tortoise belong to this suborder, as do the giant tortoises of the Galapagos Islands and all sea turtles. 

Side-necked turtles are found in South America, sub-Saharan Africa, Indonesia, Australia, and New Guinea. There is an isolated species in Yemen on the Arabian Peninsula and another on the island of Madagascar. 

Except for part of the Indonesian archipelago, all these regions were once part of Gondwana, the ancient supercontinent that existed between 250 million and 150 million years ago, until tectonic forces tore it apart. Africa, India and Madagascar separated first. South America sheared off from Antarctica 40 million years ago, followed by Australia 30 million years ago. 

Although side-necked turtles are now confined to regions that were formerly part of Gondwana, their fossil record is present on all continents. “Pleurodira were clearly distributed far more widely in the past,” said paleontologist Max Langer, a professor in the University of São Paulo’s Ribeirão Preto Biology Department. 

The oldest fossils of pleurodire turtles in North America, Europe and North Africa have been dated to between 105 million and 70 million years ago. In North America and North Africa, they survived until at least 35 million years ago, and in Southeast Asia, until at least 10 million years ago.

The biogeographical explanation for this discrepancy between the past and present distribution of side-necked turtles is far from straightforward. Did Pleurodira emerge first on Gondwana and then spread elsewhere? Or are their fossils dispersed across all five continents because they originated in some other place as yet unknown? 

“The phylogenies of pleurodires constructed to date have been partial. Our purpose in undertaking the large phylogenetic analysis now published was to understand the lineage’s evolution during the Mesozoic and Cenozoic Eras,” said Langer, who is the principal investigator of the Thematic Project “The origin and rise of dinosaurs in Gondwana (late Triassic-early Jurassic)”, funded by FAPESP. 

The study to which Langer referred has been published in Royal Society Open Science and is the most comprehensive phylogeny of side-necked turtles yet produced. It assembles phylogenetic, biogeographical and morphological data in search of an explanation for the biogeographical history of Pleurodira, especially the discrepancy between their distribution in the fossil record and in the world today.

In addition to Langer, the authors of the study are paleontologists Gabriel Ferreira and Mario Bronzati, also affiliated with the University of São Paulo (USP), and Argentinian paleontologist Juliana Sterli. Ferreira is the lead author: the study is part of his PhD research, which is supervised by Langer and supported by a scholarship from FAPESP. Bronzati earned his PhD from the Paleontological Museum belonging to Ludwig Maximilian University in Munich, Germany. Sterli is a researcher at Egidio Feruglio Paleontological Museum in Argentinian Patagonia.

The oldest known turtles lived in China 220 million years ago. Curiously, their shell consisted only of the lower half, called the plastron. The carapace, as the upper half is called, is the most familiar feature of the turtles we know but was then evolving and appeared fully formed much later. 

According to Ferreira, “estimates of molecular divergence events suggest that the evolutionary history of Pleurodira began in the Upper Jurassic, between 165 and 150 million years ago, and the oldest pleurodire fossil, Atolchelys, was found in 2009 in a limestone quarry in Alagoas State, Brazil”. 

Atolchelys lived 125 million years ago when Africa was starting to separate from South America. It belonged to the extinct Bothremydidae, one of several groups of pleurodires, of which only three families survive: Pelomedusidae (Africa and Yemen), Chelidae (South America and Australia), and Podocnemididae (South America and Madagascar). The latter include P. unifilis and P. expansa, the Arrau or Giant South American turtle.

By 110 million years ago, the pleurodires had diversified considerably and were geographically more dispersed. In the estuaries that occupied what is now the Araripe Plateau in Northeast Brazil, there were four pleurodires: Brasilemys (an ancestor of Podocnemididae), Cearachelys (like Atolchelys, a member of Bothremydidae), Euraxemys, and Araripemys. The latter two belonged to families that became extinct in the Cretaceous and were kin to Laganemys in what is now Niger, Africa. At approximately the same time, three ancestors of Chelidae lived in Patagonia.

During the next 20 million years, the pleurodires spread to Peru, Bolivia and Southeast Brazil, to North Africa and Madagascar, and to Europe, North America, the Middle East and India. This expansion occurred as Africa and South America were separating completely from the rest of Gondwana, between 105 million and 100 million years ago.

The evidence of continental drift was the basis for the vicariance hypothesis traditionally proposed by paleontologists to explain the broader distribution of Pleurodira in the past. Vicariance is an evolutionary mechanism whereby the distribution of an ancestral species is fragmented into two or more areas by the appearance of a natural barrier, in this case, the opening of the South Atlantic. 

“I’ve always loved turtles. They were my very first research interest,” said Langer, who eventually specialized in the study of dinosaurs. 

“Pleurodira are a very important group,” he added. “Ninety percent of all living pleurodires are Chelidae or Podocnemididae. Why are these animals in South America but not in Africa? What lies behind the group’s geographical distribution? How did this distribution arise?” 

The dozens of living pleurodire species are restricted to terrestrial and freshwater environments; they do not tolerate contact with salt water. If this restriction applied 100 million years ago, the South Atlantic opening must indeed have created an unsurpassable barrier against dispersal of the pleurodires across continents. Continental drift would have separated populations and forced them to adapt to different conditions, leading over time to the emergence of new genera and species.

The vicariance hypothesis made sense while the number of pleurodire fossils was relatively small and their distribution very sparse. The discovery of many extinct genera in the past decade has revealed gaps in the biogeographical narrative that the traditional theory has proved unable to fill.

Take the example of Chelidae, found today in South America and Australia. Until 40 million years ago, these continents were interconnected by Antarctica. The global climate was warmer, and Earth’s southernmost continent was not a sterile desert covered with ice but home to forests full of life. Turtles must have swum in the rivers and lakes of ancient Antarctica. No fossils of Chelidae have ever been found there, however. 

Now observe the Podocnemididae, which inhabit tropical forests in the Amazon and Orinoco basins in South America and lowland rainforests in Madagascar. Why are none of these turtles to be found in the equatorial forests of Central Africa? Again, this absence does not mean that Podocnemididae never lived in African freshwater environments. They undoubtedly did, but no trace of them has yet been detected.

Lastly, there is the case of sea turtles. Today, the oceans are home only to Cryptodira, such as the Green sea turtle (Chelonia mydas) and the Leatherback sea turtle (Dermochelys coriacea). There are no sea turtles in Pleurodira. The situation was different in the past. In the Upper Cretaceous, the Paleogene and continuing into the Cenozoic, there were pleurodires that could tolerate salt water. They belonged to the extinct Bothremydidae, as did Atolchelys and Cearachelys, which inhabited the coast of Northeast Brazil 125 million and 110 million years ago, respectively.

At that time, the South Atlantic was not yet fully open. This development occurred later, between 80 million and 66 million years ago, when Bothremydidae inhabited both sides of the Atlantic. Inaechelys lived on the coast of Pernambuco, and on the other side of the still young (and for this very reason narrow), Atlantic Ocean lived the Portuguese Rosasia, as well as Foxemys and Polysternon, found in Spain and France. Another marine genus of Bothremydidae, Bothremys, was more widely distributed, as shown by fossils found in four US states as well as Morocco and Jordan.

Perhaps none of these pleurodires were good ocean swimmers capable of surviving for long periods on the high seas, as are the migratory Green and Leatherback sea turtles. Even so, for Inaechelys, Rosasia, Foxemys and Polysternon, the young South Atlantic may not have been a sufficiently formidable barrier to prevent their dispersal to other continents – at least not while the distance between South America and Africa was relatively small, perhaps a few hundred kilometers, a fraction of the 3,300 km that currently separate Northeast Brazil from West Africa in the narrowest part of the South Atlantic. 

A new phylogeny

“If the traditional hypothesis was that the current distribution of pleurodires derives from vicariant events linked to continental drift, there was a second hypothesis according to which the group was broadly distributed, and successive extinctions eventually confined its lineages to the areas where they’re found today,” Ferreira said. 

“We imagined a third hypothesis: a complex pattern of dispersal from Gondwanan areas explains the broad distribution found in the past.” 

To test each of these hypotheses, the researchers decided to construct a new phylogeny of Pleurodira, tracing the group’s evolutionary history as broadly as possible in order to reveal unknown patterns of past biogeographical distribution.

The first step in building the phylogeny was a matrix analysis of 245 morphological characters in 101 species. “This matrix of morphological data for Pleurodira included both living and extinct species. The matrix was analyzed using parsimony, and from the analysis, we obtained a new phylogenetic tree for Pleurodira,” Ferreira said. 

“We performed other analyses with the phylogenetic tree, focusing on diversification and biogeography and using the temporal and geographical data for the species in the tree. The diversification analysis took into account both the number of known species in a given geological period and the relative quantity of species in two sibling groups. By comparing these datasets, the analysis identified which groups were more diverse than other groups in the same lineage.” 

According to Ferreira, biogeographical analysis is probabilistic. It takes into consideration the temporal and geographical distribution of the species in a group and reconstructs the ancestral areas of the groups in the tree. 

“We identified the geographical distribution of the groups of interest, finding, for example, that the ancestor of Podocnemididae was in area X, the ancestor of Pleurodira in area Y and so on, and that they left that area and went to area Z,” he explained. 

Saga of endemisms and dispersals

The new phylogeny leads to the conclusion that Araripemys and Euraxemys were relatives of Pelomedusoides, the ancestral group that gave rise to the families Bothremydidae, Podocnemididae and Pelomedusidae.

Moreover, during the Lower Cretaceous, when Araripemys and Euraxemys were extant, both of the main pleurodire lineages already existed. They were the Pan-Chelidae (the group comprising all Chelidae) and Pan-Pelomedusoides (Bothremydidae, Podocnemididae, Pelomedusidae, and the other extinct families).

The new tree suggests that Atolchelys, the oldest known pleurodire (and the oldest member of Bothremydidae, alive in the Lower Cretaceous 125 million years ago in Alagoas) shares a common ancestor with Araripemys and Euraxemys (alive 110 million years ago in what is now Ceará). 

Despite the scarcity of the fossil record for the Lower Cretaceous (half a dozen species are known), the new phylogenetic tree suggests that a large number of lineages of Chelidae and ancestors of Pelomedusoides were already established in the period. 

The mass extinction that wiped out the dinosaurs at the end of the Cretaceous apparently did not correspond to a critical period for the pleurodires, involving neither extinction nor diversification. This finding makes sense, as the turtles were the animals that suffered least from the mass extinction among terrestrial vertebrates. 

The fossil record shows that at least during the Cretaceous, Pan-Chelidae and Pan- Pelomedusoides were restricted to Gondwana. Interestingly, however, the researchers’ reconstruction of the ancestral areas based on the new phylogenetic tree supports an Australian origin for Pan-Chelidae, followed by a spread to South America during the Lower Cretaceous. In other words, the presence of Chelidae in Australia is no surprise if Australia is the origin of their lineage.

The biogeographical history of Pan-Pelomedusoides, in contrast, is dominated by the occurrence of areas of endemism for each group, with several dispersal events to other areas. The exception is Pelomedusidae, which was always endemic to continental Africa.

Some pelomedusids are currently found in Madagascar, the Arabian Peninsula, the Seychelles and other small islands, but the absence of fossil records other than very scarce and fragmentary remains in continental Africa precludes a more detailed account of the biogeographical history of Pelomedusidae. Given the current data, the researchers believe that Pan-Pelomedusidae were always restricted to the African continent and only recently dispersed transoceanically. 

The results also show that the ancestors of Araripemys, Euraxemys and Pan-Podocnemididae originally inhabited Africa, dispersing to South America during the Lower Cretaceous. The ancestors of Podocnemidoidae remained in South America, whereas the ancestors of Bothremydidae returned to Africa.

Bothremydidae diversified significantly in Africa, but several representatives dispersed independently to other areas: at least once to Europe, India, Madagascar and back to South America, and at least three times to North America. 

The results highlight the strong dispersal capability of Bothremydidae due to their marine habits. They were the most widespread group of side-necked turtles during the Cretaceous and Paleocene, when they started to decline in diversity until their complete extinction approximately 50 million years ago.

Third hypothesis

The new phylogenetic tree for pleurodires enabled the researchers to detect and differentiate vicariant events, dispersal events and founding events that occurred in the past 125 million years. The previous hypotheses did not satisfactorily explain the distribution of pleurodires over time. 

“Our third hypothesis, which assumes a complex pattern of dispersals to North America, Europe and Asia from Gondwanan areas (South America and Africa), is the best explanation of the past and present distribution patterns,” Ferreira said.

“Furthermore, we noted that the pleurodire groups with above-normal diversity were precisely those that diversified in different environments – in other words, those that became sea turtles.” 

The Bothremydidae are the most obvious but not the only example, ranging from Pernambuco’s Inaechelys to Portugal’s Rosasia and France’s and Spain’s Foxemys and Polysternon.

The 125-million-year saga of the pleurodire lineages, with their expansion to various magnificent paleobiomes that vanished long ago, includes memorable passages. 

For example, São Paulo State and Minas Gerais were inhabited by dinosaurs at the end of the Cretaceous. They were also home to a curious and highly diversified lineage of terrestrial crocodiles, which did not crawl but ran swiftly on legs as long as wolves’. Podocnemididae such as São Paulo’s Roxochelys and Bauruemys and Minas Gerais’s Cambaremys lived in small lakes amid these huge creatures.

Much further north, in an area ranging from the west of Amazonas State and Acre in Brazil to parts of Peru, Colombia and Venezuela, between 18 million and 10 million years ago, there was a megaswamp of 2 million square kilometers, four times the size of today’s Pantanal in Mato Grosso. 

This megaswamp was a land of giants, where the apex predator was Purussaurus, a 12 m caiman weighing 15 tons. To sate its enormous appetite, Purussaurus devoured buffalo-sized pacaranas (Dinomys branickii, a cousin of the Capybara) and hunted the largest turtle ever, a colossus called Stupendemys, a podocnemidid with a 3.5 m diameter shell. 

We know Purussaurus attacked Stupendemys because a complete shell with a large piece missing has been found in Venezuela. The missing piece is the exact shape of Purussaurus’s mouth.

Langer and Ferreira are right to be captivated by pleurodires, scientifically speaking. They are indeed a fascinating lineage.

The article “Phylogeny, biogeography and diversification patterns of side-necked turtles (Testudines: Pleurodira)” (doi: http://dx.doi.org/10.1098/rsos.171773) by Gabriel S. Ferreira, Max Langer, Mario Bronzati and Juliana Sterli can be read at: rsos.royalsocietypublishing.org/content/royopensci/5/3/171773.full.pdf.