A Phororhacoid bird from the Eocene of Africa
- First Online:
- Cite this article as:
- Mourer-Chauviré, C., Tabuce, R., Mahboubi, M. et al. Naturwissenschaften (2011) 98: 815. doi:10.1007/s00114-011-0829-5
The bird fossil record is globally scarce in Africa. The early Tertiary evolution of terrestrial birds is virtually unknown in that continent. Here, we report on a femur of a large terrestrial new genus discovered from the early or early middle Eocene (between ∼52 and 46 Ma) of south-western Algeria. This femur shows all the morphological features of the Phororhacoidea, the so-called Terror Birds. Most of the phororhacoids were indeed large, or even gigantic, flightless predators or scavengers with no close modern analogs. It is likely that this extinct group originated in South America, where they are known from the late Paleocene to the late Pleistocene (∼59 to 0.01 Ma). The presence of a phororhacoid bird in Africa cannot be explained by a vicariant mechanism because these birds first appeared in South America well after the onset of the mid-Cretaceous Gondwana break up (∼100 million years old). Here, we propose two hypotheses to account for this occurrence, either an early dispersal of small members of this group, which were still able of a limited flight, or a transoceanic migration of flightless birds from South America to Africa during the Paleocene or earliest Eocene. Paleogeographic reconstructions of the South Atlantic Ocean suggest the existence of several islands of considerable size between South America and Africa during the early Tertiary, which could have helped a transatlantic dispersal of phororhacoids.
KeywordsAvesEoceneAlgeriaSouth AmericaPaleobiogeographyTransatlantic dispersal
Aves Linnaeus, 1758; Gruiformes Reichenbach, 1852; Cariamae Fürbringer, 1888; Phororhacoidea Patterson 1941; Family incertae sedis; Lavocatavis africana gen. et sp. nov.
Lavocat, for René Lavocat who first reported these fossiliferous outcrops, avis, Latin word for bird, and africana from Africa.
Type locality and age
Hammada Gour Lazib 51. Gour Lazib area, intermediate member of the Glib Zegdou Formation. The charophyte assemblage, discovered throughout this intermediate member, is composed of several species belonging to the genera Raskyella, Stephanochara, Nodosochara, Peckichara, Harrisichara, Nitellopsis, and Maedleriella. This floral assemblage indicates a late early or early middle Eocene age (Mebrouk et al. 1997). Furthermore, mammalian biostratigraphic data also suggests an age earlier than late middle Eocene. Really, several groups (namely, anthracotheriid artiodactyls, anthropoid primates, and hystricognathous rodents), which entered Africa during late middle Eocene faunal exchange(s) between Asia and North Africa, are lacking in the intermediate member of the Glib Zegdou Formation (e.g., Tabuce and Marivaux 2005; Gheerbrant and Rage 2006).
Large-sized femur, straight shaft, caudal face of the shaft flattened, condylus lateralis and trochlea fibularis projecting laterally and distally, great development of the fossa poplitea, of the crista supracondylaris medialis, and of the osseous bar which joins the condylus medialis to the condylus lateralis. This unique combination of characters is found only in the family Phorusrhacidae. Comparatively to the Phorusrhacidae of similar or larger size (i.e., all the Phorusrhacidae apart from the Psilopterinae), Lavocatavis has the smallest distal end in relation to the total length of the femur. In addition, the proximo-distal diameter of the caput femoris, proportionally to the total length, is smaller in Lavocatavis than in the members of the family Phorusrhacidae.
Description and comparisons
The proximal and distal parts of this femur are incompletely preserved on the cranial side. The bone is better preserved on its caudal side, except for a part of the condylus lateralis which is missing. The caput femoris is crushed. The shape is very rectilinear. The shaft is conspicuously flattened on its caudal face and rounded on its cranial face. At the proximal part, the caput femoris is strongly projecting medially, the crista trochanteris is present but incompletely preserved, and on the caudal face there are strong muscular impressions (impresiones obturatoriae). At the distal part, on the caudal face, the condylus medialis is well preserved, the condylus lateralis is partly missing, and the trochlea fibularis is strongly projecting laterally. The laterodistal part of the articulation is laterally elongated and projects farther distally than the mediodistal part. The fossa poplitea is oval shaped, proximo-distally elongated, and very deep. It penetrates about 15 mm deep under an osseous bar which joins the condylus medialis to the condylus lateralis. The crista supracondylaris medialis is situated in the prolongation of the condylus medialis and is extremely developed. It extends on the mediocaudal angle of the shaft, on about 4/5 of the total length of the bone. On the cranial face, both condyles are not preserved and it is not possible to see the shape of the sulcus patellaris. There is no visible linea intermuscularis cranialis.
Considering the least shaft circumference of its femur (79 mm), Lavocatavis had an estimated weight of ca. 32 kg (Campbell and Marcus 1992) or even 45–50 kg (Alvarenga and Höfling 2003). Now, among the recent birds, the maximum weight attained by the flying forms stands around 14 kg, some exceptionally reaching 20 kg (Dunning 1993; Mayr and Rubilar-Rogers 2010). It is therefore possible to assume that Lavocatavis was very probably flightless. The late Mesozoic and Cenozoic bird fossil record offers evidence that several large forms among primitive, palaeognathous, and neognathous birds were also flightless.
Among primitive forms, the genus Gargantuavis, from the late Campanian or early Maastrichtian of France, has been described from a synsacrum and a referred femur, and is placed in the Ornithothoraces (Buffetaut and Le Loeuff 1998, 2011). Mayr (2009, p. 21) indicates that this synsacrum could also be attributed to a very large pterosaur. However, the femur referred to Gargantuavis is very different from the pterosaurs femora which are relatively long and narrow, with a caput femoris strongly projecting proximally (see Wellnhofer 1980, figure 31), and we consider it as avian. This femur is short and stout, strongly incurved mediolaterally, and shows a large crista trochanteris. The shape of this crista trochanteris, strongly projecting laterally and rounded, is due to the strong compression of the bone during the processus of fossilization. If one could make a three-dimensional restoration of its form, it would not be very different from a femur of Gastornithidae. It differs from Lavocatavis, the femur of which is very straight.
The oldest known palaeognath is Diogenornis fragilis, from the late Paleocene of Itaborai, Brazil (Alvarenga 1983), which is placed in the recent family Rheidae, but its femur is unknown. The other palaeognathous birds of the Paleogene are the Palaeotididae, the Remiornithidae, and the Eremopezidae. The femur is known in Palaeotis weigelti, from the middle Eocene of Germany (Houde and Haubold 1987). This femur is elongate and shows a strong curvature, caudally directed, at its distal end. In the Remiornithidae, from the late Paleocene of France (Mayr 2009), and in the Eremopezidae, from the late Eocene or earliest Oligocene of Fayum, Egypt (Rasmussen et al. 2001; Seiffert 2006), the femur is unknown but the other known elements show some similarities with those of other extinct or recent palaeognaths. In the recent large-sized Struthioniformes, as well as in the extinct Dinornithidae and Aepyornithidae, the femur is generally short and stout, and strongly widened at both extremities. The section of the diaphysis is circular in shape and it is not flattened on its caudal surface. The crista trochanteris is generally projecting proximally; it is strongly projecting in the Dinornithidae and Aepyornithidae, and moderately projecting in the recent forms with the exception of Struthio where the facies articularis acetabularis extends slightly proximally beyond the crista trochanteris. In the recent large ratites, the crista trochanteris is also projecting on the cranial side. In the recent ratites and the Aepyornithidae, there is a very deep popliteal fossa. The osseous bar which links the medial condyle and the lateral condyle is very oblique compared to the longitudinal axis of the shaft, while in Lavocatavis this bar is almost perpendicular to this axis. The crista supracondylaris medialis is weakly developed and is not situated along the mediocaudal border of the shaft.
Several forms of large flightless neognathous birds have been described in the Paleogene. In the Gastornithidae, known from the Paleocene to the middle Eocene of Europe, North America, and Asia, with the genera Gastornis and Diatryma, the femur is short and stout, strongly widened at both ends (Andors 1992; Mayr 2009). The Australian Dromornithidae are known mainly from the late Oligocene, with a tentative record of footprints in the early Eocene (Murray and Vickers-Rich 2004). In these birds, the femur is strongly mediolaterally incurved, both ends are much widened, and the crista trochanteris is strongly projecting. In conclusion, the Glib Zegdou femur is very different from that of the other flightless birds, extinct or extant, and cannot be attributed to any of these groups.
By contrast, Lavocatavis presents affinities with Phorusrhacidae. All the authors who have described the femur of the Phorusrhacidae (Andrews 1899; Ameghino 1920; Sinclair and Farr 1932) have insisted on the following characteristics: very straight shape; no curvature in the craniocaudal direction, nor in the mediolateral direction, with only a slight cranial convexity at the distal end; great development of the fossa poplitea and of the crista supracondylaris medialis; and condylus lateralis projecting distally and laterally. All these characteristics are present on the Glib Zegdou femur. The femur of Lavocatavis also presents, at its proximal end, a flattened facies articularis antitrochanterica, and there is no projecting trochanter femoris. The caput femoris is strongly projecting medially and the collum femoris is weakly indicated, but this part is not well preserved.
The suprafamilial name Phororhacoidea Patterson 1941 designates the whole of the fossil birds known as Terror Birds and, according to different authors, includes either several families, for example in Patterson and Kraglievich (1960), or only one family, Phorurhacidae, with five subfamilies, for example in the recent revision of Alvarenga and Höfling (2003). In the present paper, we have considered Lavocatavis as member of a family incertae sedis in the superfamily Phororhacoidea.
Among the Psilopterinae, which are the smallest Phorusrhacidae, the femur is known in the genera Psilopterus and Procariama (Sinclair and Farr 1932; Alvarenga and Höfling 2003). In these two genera, the shaft has a more circular section and is less flattened on its caudal face than in the other Phorusrhacidae, and the distal part is medially incurved. The femur of Mesembriornis (subfamily Mesembriornithinae, middle-sized forms with very elongate tarsometatarsi) is proportionally stouter than the Glib Zegdou one and shows a trochlea fibularis projecting very laterally and distally (Alvarenga and Höfling 2003, figure 33, A and D). In the genera Physornis and Paraphysornis (subfamily Brontornithinae, large-sized forms), the femur is short, stout, and strongly widened at both extremities (Alvarenga and Höfling 2003). In the subfamily Phorusrhacinae are found truly gigantic forms with very elongate hindlimbs (Alvarenga and Höfling 2003; Bertelli et al. 2007). The femur is known in the three genera of the subfamily Patagornithinae, namely Patagornis, Andrewsornis, and Andalgalornis. The femur of Glib Zegdou looks more similar to those of the Patagornithinae than to any other member of the Phorusrhacidae.
In the Patagornithinae, the facies articularis acetabularis has a trilobate appearance. The description of Patagornis marshi by Andrews (1899, p. 78) indicates that the femoral head presents a large blunt-pointed process situated in proximal position, and that, distally, the articular surface is divided by a shallow groove into an anterior (cranial) and a posterior (caudal) lobe, so that when looked at from the inner (medial) side, the head appears to be imperfectly trilobate. This feature is clearly visible on one of the illustrations (Moreno and Mercerat 1891, pl. XIV, figure 7a, right femur, medial view). On the Glib Zegdou femur, this part is not well preserved, but it is possible to see, in cranial view, a flattened cranial lobe, ellipsoidal in shape, and in medial view two lobes, a cranial one and a caudal one, separated by a groove. Distally to these lobes, the collum femoris is slightly excavated. On the cranial face, the crista trochanteris is incompletely preserved. On the caudal face, it is possible to see an oblique ridge, extending from the lateral border of the shaft to the middle of the facies articularis antitrochanterica. This ridge corresponds to the extremely distinct muscular impressions that are present on the femur of Patagornis marshi (Andrews 1899, pl. XVII, figure 8).
Ameghino (1920, p. 248) describes the femur of Patagornis marshi as follows: "In posterior view, the distal part of the bone is flattened and ends into a very deep popliteal fossa, which is separated from the intercondylar fossa by an exceptionally developed bony bar" (our translation). These characteristics are also present on the Glib Zegdou femur.
Measurements (in millimeters) of the femur of the Glib Zegdou Phororhacoidea, Lavocatavis africana gen. et sp. nov., compared with femora of Phorusrhacidae (Alvarenga and Höfling 2003)
Lavocatavis africana UM/HGL 51–55
Patagornis marshi BMNH A 516
Andrewsornis abbotti FMP 14678
Andalgalornis steulleti MACN 69321
220 as preserved
67.0 as preserved
Width of the shaft in the middle
Depth of the shaft in the middle
In conclusion, Lavocatavis is more similar to the members of the Phorusrhacidae than to any other extinct or extant large flightless bird. We place it in family incertae sedis in the superfamily Phororhacoidea because it is known only by a single bone.
Fossil record of Phororhacoidea and paleobiogeographic consequences
The earliest Phorusrhacidae are known from the late Paleocene of Itaborai, Brazil, and the Eocene of Antarctica (Mayr 2009). A fragment of bill and a distal part of tarsometatarsus of a phororhacoid bird have been discovered in the late Eocene, age about 36 Ma, of Seymour Island, Antarctic Peninsula (Case et al. 1987; Tambussi and Acosta Hospitaleche 2007). Footprints of a very large, three-toed, bird have been reported from Fides Peninsula of King George Island (South Shetland Islands, northern tip of the Antarctic Peninsula) in early Eocene deposits (Case et al. 1987). These footprints have been attributed either to a ratite or to a phororhacoid. A femur has been discovered in the early Maastrichtian of Vega Island, Antarctic Peninsula (Case et al. 2006). This femur shows similarities with that of cursorial birds and could correspond to an ancestral form of both cariamids and phororhacoids, or to a basal cariamid, ancestral to phororhacoids. During the Neogene, after the establishment of the Panamanian land bridge, when South America became linked to North America, some taxa arrived in North America where they are known from the Pliocene of Texas, and the late Pliocene/early Pleistocene of Florida (MacFadden et al. 2006). The latest Phorusrhacidae are documented from the late Pleistocene of Uruguay (Alvarenga et al. 2010). Phorusrhacidae had been reported from the European Paleogene, but this attribution has been contested and the concerned taxa are now placed in a distinct family, Ameghinornithidae (Alvarenga and Höfling 2003; Mayr 2005). The Ameghinornithidae are now considered as belonging to the suborder Cariamae but outside a clade including the extinct Idiornithidae and Phorusrhacidae, and the recent Cariamidae (Mayr 2009). As a result, the Phororhacoidea were absent from the Laurasiatic continent during the Paleogene (Alvarenga and Höfling 2003; Mayr 2009), and they were also unknown in Africa up to now. The discovery of a probable flightless phororhacoid at Glib Zegdou as well as the geographic distribution of the group could be explained by either vicariance or oceanic dispersal between South America and Africa. Such a dispersal through the Laurasiatic continent can be excluded because they are unknown as fossils in the rich avifaunas of the Northern Hemisphere.
In this perspective, the paleobiogeographic history of several continental vertebrate groups from the early Paleogene of Africa and South America needs to be discussed here. In addition to Lavocatavis, the Glib Zegdou fauna includes several fishes and mammals, plus a turtle. The latter is a representative of the podocnemidid turtles, a family known since the Cretaceous to Recent in South America, Europe, Caribbean, and Africa. Although discussed, this distribution could be linked to a vicariant event after the mid-Cretaceous opening of the southern Atlantic Ocean (Cadena et al. 2010). The vertebrate fauna from the Glib Zegdou also comprises some freshwater fishes (polypterids and protopterids) which have their closest relatives in South America, indicative of some vicariant event during the Gondwana break-up (Gheerbrant and Rage 2006). After this major tectonic event, and due to tectonic isolation of Africa and South America, the early Cenozoic terrestrial vertebrates in these two landmasses have had a long history of endemic evolution. This endemism gave rise to the emergence of some emblematic clades such as the African Afrotheria and the South American “ungulates”, Xenarthra, and phororhacoids. Interestingly, some of the South American “ungulates” (Notoungulata and Astrapotheria at least) were recently proposed as closely related to the Afrotheria ( Agnolin and Chimento 2011; but see Billet and Martin 2011). Likewise, we must address the Atlantogenata clade, a molecular group that unites the mammalian superorder Afrotheria with Xenarthra (Tabuce et al. 2008). According to this molecular hypothesis, the assumed Cretaceous divergence of Xenarthra and Afrotheria is linked to the Gondwanian break-up (Wildman et al. 2007). Afrotherians are widely distributed during the Eocene in Africa and one of their representatives, the hyraxes, are abundant in the Glib Zegdou fauna. Afrotherian mammals as well as Xenarthra and South American “ungulates” first occur only during the Paleocene (Cifelli 1983; de Muizon and Cifelli 2000; Gheerbrant 2009). So, if the hypothesis of the common ancestry between Afrotheria, South American “ungulates” and Xenarthra is correct, a deep Cretaceous history of vicariant evolution between African and South American mammals is currently unsupported by the fossil record.
A vicariance mechanism can also be ruled out for the phororhacoids because they first appeared in South America well after the onset of the Gondwana break-up, around 100 Ma. Moreover, according to the molecular data, the split which gave rise to the clade including the higher land birds, and among them the Cariamidae, occurred slightly before the K/T boundary (Ericson et al. 2006; Mayr 2009).
It is likely that the Phorusrhacidae appeared in South America because it is there that the oldest forms are known (Late Paleocene), and there that this group became strongly diversified during all the Cenozoic. In addition, it is also in South America that are found the Cariamidae which are the present-day closest relatives of the Phorusrhacidae (Acosta-Hospitaleche and Tambussi 2005; Mayr 2009).
The oldest known forms, Psilopterinae, were relatively small and Tambussi and Noriega (1996) think that they were still capable of limited flight. It is possible to hypothesize that small forms dispersed from South America to Africa by flying and then evolved to give large flightless forms with morphological characteristics very similar to that of Patagornithinae.
If the possibility of a convergent evolution is excluded, the alternative hypothesis is that flightless phororhacoids dispersed from South America to Africa by rafting or by island hopping. Large flightless phororhacoids were also present in the late Eocene of Antarctica (Case et al. 1987; Tambussi and Acosta Hospitaleche 2007), but it is unlikely that they dispersed to Africa because, as early as the late Cretaceous, Africa was already very far from Antarctica, while it was still relatively close to the North-East of South America (Scotese et al. 1988; Smith et al. 1994).
Until now, only six or seven transatlantic dispersal events are known for terrestrial vertebrates; they notably concern the caviomorph rodents, platyrrhine anthropoids, and several groups of reptiles (Flynn and Wyss 1998; Vidal et al. 2010). Platyrrhines and caviomorphs are sister groups of two African taxa (catarrhines and phiomorphs, respectively); they probably arrived in South America around the same time, during the late Eocene (Sallam et al. 2009). Their dispersal follows the prevailing westward paleocurrents and paleowinds suggesting that rafting on floating islands, in addition to island hopping, was their primary transoceanic modes of dispersal (Houle 1999; Sallam et al. 2009). The oldest known South American rodents have been identified in the Tinguiririca fauna in Chile, dated from 31.5 Ma, while the oldest known South American primates are dated from 26.4 Ma (Flynn and Wyss 1998). The age of the Glib Zegdou phororhacoid is much older, which suggests that these birds would have crossed from South America to Africa during a much older period, during the Paleocene or the earliest Eocene, crossing furthered by the fact that the Atlantic Ocean was then narrower.
Moreover, by comparison with caviomorphs and platyrrhines, the phororhacoid dispersal is an eastward dispersal. Considering the dominant westward South Atlantic paleocurrents and paleowinds during the early Tertiary, an eastward rafting on floating islands is improbable, suggesting that phororhacoids have dispersed to Africa only by island hopping. Some possibilities exist for such dispersion. The reconstructions of paleobathymetry made by de Oliveira et al. (2008) indicate that at 50 Ma the shortest distance between Africa and South America was around 1,000 km in a straight line (from present-day Sierra Leone to Paraiba State in Brazil). The ocean was wider further south, “but several islands of considerable size (more than 200 km in length) persisted along the present-day submerged Rio Grande Rise and Walvis Ridge. Between 20 and 30°S, at 50 Ma, a long series of close islands stretched from the African shore, and at least one large island (around 500 km in length) was formed by the emergent top of the Rio Grande Rise” (de Oliveira et al. 2008, p. 59). Another set of islands was present along the Brazilian coast at 20°S. These paleogeographic features, which are underwater today, might have reduced considerably the distance of a possible migration of terrestrial vertebrates. It is not possible for the moment to choose between the two competing hypotheses, (1) either a dispersal of small forms, still able of limited flight, which later would have evolved convergently with the Phorusrhacidae, or (2) a dispersal of flightless forms by island hopping across the Atlantic Ocean. We expect that current and ongoing field research projects in latest Cretaceous and earliest Paleogene horizons in Africa and South America may document (1) other punctual faunal exchanges between the two landmasses, and (2) paleontological evidence in favor of the still poorly supported common ancestry between the so-called endemic African and South American mammals.
We thank the vice-chancellors of Tlemcen and Oran Universities, and the authorities from Bechar and Tindouf districts, who assisted fieldwork in the Gour Lazib area. We thank H. Alvarenga, S. Chapman, and M. Daniels for supplying information, and A.-L. Charruault for technical assistance. We also thank two anonymous reviewers for constructive comments. This research was supported by the French ANR-PALASIAFRICA Program (ANR-08-JCJC-0017).