Correction EVOLUTION. For the article ‘‘A proboscidean from the late Oli- gocene of Eritrea, a ‘missing link’ between early Elephanti- formes and Elephantimorpha, and biogeographic implications,’’ by Jeheskel Shoshani, Robert C. Walter, Michael Abraha, Seife Berhe, Pascal Tassy, William J. Sanders, Gary H. Marchant, Yosief Libsekal, Tesfalidet Ghirmai, and Dietmar Zinner, which appeared in issue 46, November 14, 2006, of Proc Natl Acad Sci USA (103:17296–17301; first published November 3, 2006; 10.1073͞pnas.0603689103), the authors note that several errors appeared in their article. On page 17296, left column, first paragraph of the Introduction, line 16, ‘‘the only known late Oligocene mammal sites’’ should instead read: ‘‘the only known late Oligocene proboscidean sites.’’ On page 17296, right col- umn, first sentence under Holotype, ‘‘National Museum of Eritrea (NME) field number DOG87.1’’ should instead read: ‘‘National Museum of Eritrea (NME) field number DOG97.1.’’ Last, in Table 2, line 11, ‘‘Superfamily Gomphotherioidea Maglio, 1973 (in part)*’’ should instead read: ‘‘Superfamily Gomphotherioidea Hay, 1922.’’ The corrected Table 2 appears below. These errors do not affect the conclusions of the article. Table 2. A partial, nonranked classification of Proboscidea with emphasis on families discussed in this paper and the taxon from Dogali, Eritrea (modified after ref. 11) Proboscidea Illiger, 1811 Elephantiformes Tassy, 1988 Family Palaeomastodontidae Andrews, 1906* Family Phiomiidae Kalandadze and Rautian, 1992* Elephantimorpha Tassy and Shoshani, 1997 Family incertae sedis Eritreum melakeghebrekristosi gen. et sp. nov.* Mammutida Tassy and Shoshani, 1997* Family Mammutidae Hay, 1922* Elephantida Tassy and Shoshani, 1997 Superfamily Gomphotherioidea Hay, 1922 Family Gomphotheriidae Hay, 1922* Superfamily Elephantoidea Gray, 1821 Family Stegodontidae Osborn, 1918* Family Elephantidae Gray, 1821 Arrangements of taxa correspond to the branching pattern in Fig. 5. The nonranking applies to ranks or categories above the Superfamily level fol- lowing Recommendation 29a and Article 36a of the International Commission of Zoological Nomenclature (11, 33), see text for details. *Extinct taxon. www.pnas.org͞cgi͞doi͞10.1073͞pnas.0610824104 CORRECTION www.pnas.org PNAS ͉ January 16, 2007 ͉ vol. 104 ͉ no. 3 ͉ 1103 Downloaded by guest on October 2, 2021 A proboscidean from the late Oligocene of Eritrea, a ‘‘missing link’’ between early Elephantiformes and Elephantimorpha, and biogeographic implications Jeheskel Shoshani*†‡, Robert C. Walter§, Michael Abraha¶, Seife Berheʈ, Pascal Tassy**, William J. Sanders††, Gary H. Marchant†, Yosief Libsekal‡‡, Tesfalidet Ghirmai¶, and Dietmar Zinner‡§§ *Department of Biology, University of Asmara, P.O. Box 1220, Asmara, Eritrea; †Elephant Research Foundation, 106 East Hickory Grove Road, Bloomfield Hills, MI 48304; §Department of Earth and Environment, Franklin and Marshall College, Lancaster, PA 17604-3003; ¶Eritrean Geological Survey, Department of Mines, Ministry of Mines and Energy, P.O. Box 272, Asmara, Eritrea; ʈGlobal Resources, P.O. Box 4588, Asmara, Eritrea; **USM203͞Unite´ Mixte de Recherche 5143 Centre National de la Recherche Scientifique Pale´ obiodiversite´ , CP 38, Muse´ um National d’Histoire Naturelle, 57 Rue Cuvier, 75231 Paris Cedex 05, France; ††Museum of Paleontology, University of Michigan, Ann Arbor, MI 48109; ‡‡Department of Archeology, National Museum of Eritrea, P.O. Box 1220, Asmara, Eritrea; and §§Department of Cognitive Ethology, Deutsches Primatenzentrum, Kellnerweg 4, D-37077 Go¨ ttingen, Germany Edited by David B. Wake, University of California, Berkeley, CA, and approved September 9, 2006 (received for review May 5, 2006) We report on a late Oligocene proboscidean species from Eritrea, Bridge, Mfwangano, Muruorot, Mwiti, Rusinga, and Songhor, dated to 26.8 ؎ 1.5 Mya. This ‘‘missing link’’ between early Kenya), and Namibia (12–15). Before the discovery of fossil elephantiformes and Elephantimorpha is the oldest known non- proboscideans at Chilga, Ethiopia, and now from Dogali, there gomphothere proboscidean to probably display horizontal tooth was an absence of evidence from the time period between the displacement, typical of elephants [Elephantimorpha consists of earlier (moeritheres and numidotheres) and later radiation Mammutida (mastodons) and Elephantida, and Elephantida in- (Mammutida and Elephantida) of proboscideans. In addition, cludes gomphotheres, stegodons, and elephants]. Together with the phylogenetic connections of elephantimorphs to more ar- the newly discovered late Oligocene gomphotheres from Chilga, chaic taxa were uncertain and subject to considerable debate (2, Ethiopia, the Eritrean taxon points to the importance of East Africa 8, 16–17). The fossil from Dogali (Figs. 2 and 3) is therefore well as a major area for the knowledge of the early evolution of suited temporally as well as morphologically to bridge the gap of Elephantimorpha before the faunal exchange between Eurasia lack of evidence (acting as a ‘‘missing link’’); and it helps us to and Africa. better understand the biogeographical inferences of early pro- boscidean radiation between Africa and Arabia. Furthermore, evolution ͉ NE Africa ͉ palaeontology ͉ horizontal tooth displacement the new data from Dogali enable us to test competing hypotheses about the relationship among elephantimorphs, phiomiids, and roboscideans have been a part of the Afro-Arabian land- Ϸ palaeomastodonts. The two prevailing hypotheses tested are (i) Pscape since at least the early Eocene, 55 Mya (1). Paleo- Phiomia, family Phiomiidae, is a precursor of Elephantida gene proboscidean fossils have been recovered from northern (Gomphotherioidea and Elephantoidea), and Palaeomastodon, Africa (Algeria, Egypt, Libya, Mali, Morocco, Sudan, and family Palaeomastodontidae, is a precursor of Mammutida Tunisia), western Africa (Senegal), central Africa (Angola) (1–6), eastern Africa (Ethiopia) (7, 8), and the Arabian Penin- (mastodontids), that is, Elephantimorpha is paraphyletic, and (ii) sula (Oman) (9). Aside from a small, isolated tusk fragment from Phiomia and Palaeomastodon are a sequence of sister taxa to Baluchistan, Pakistan, dated to the late Oligocene and showing Elephantimorpha (Mammutida and Elephantida) (18–20). ivory structural pattern consistent with attribution to gomphoth- eres, stegodons, or elephants (10), the fossil record of definitive Systematic Paleontology. Class Mammalia Linnaeus, 1758. Paleogene proboscideans and the evidence for the major early Order Proboscidea Illiger, 1811. radiations of proboscideans (moeritheres, numidotheres, and Nonranked category Elephantida Tassy and Shoshani, 1997. palaeomastodonts) appear to be best documented in Afro- Family incertae sedis. Arabia. The fossil site of Dogali (Eritrea, Fig. 1) described here Eritreum melakeghebrekristosi gen. et sp. nov. and the Chilga site in Ethiopia (7, 8) are in close proximity and are the only known late Oligocene mammal sites from the whole Holotype. National Museum of Eritrea (NME) field number of Africa. The paleobiogeographical implications of these sites in DOG87.1, left mandibular ramus with m2 and m3, right man- the context of proboscidean evolution in Africa and across the dibular ramus with m2 and partial first lophid of m3, portions of Arabian Peninsula are discussed below. mandibular symphysis (in two parts) with remnants of tusk The initial radiation of Elephantimorpha that replaced the alveoli (Figs. 2 and 3). archaic Elephantiformes proboscideans and deinotheres was centered in Africa and was primarily a basal Neogene event Etymology. Eritreum from the name of the new nation of Eritrea [Elephantiformes includes Phiomiidae, Palaeomastodontidae, in the Horn of Africa, and melakeghebrekristosi in honor of the and Elephantimorpha; Elephantimorpha consists of Mam- farmer, Melake Ghebrekristos, who found the specimen and mutida (mastodons) and Elephantida (gomphotheres, steg- realized its importance. odons, and elephants) (11)]. Although the earliest gomphothere is from the late Oligocene, Ϸ27 Mya (7, 8), the oldest known mammutid dates to the early Miocene, 22 Mya (12), and the Author contributions: J.S. performed research; J.S., R.C.W., M.A., S.B., P.T., W.J.S., G.H.M., subsequent diversification of these groups occurred during the Y.L., T.G., and D.Z. analyzed data; and J.S., R.C.W., S.B., P.T., and W.J.S., wrote the paper. early Miocene, Ϸ22–20 Mya. Early to early-middle Miocene The authors declare no conflict of interest. African sites dated to 22–16 Mya and yielding mammutid and This article is a PNAS direct submission. gomphothere fossils are known from northern Africa (Kabylie, ‡To whom correspondence may be addressed. E-mail: [email protected] or jshosh@ Algeria; Moghara and Siwa, Egypt; and Gebel Zelten, Libya), sun.science.wayne.edu. eastern Africa (Buluk, Legetet, Karungu, Loperot, Meswa © 2006 by The National Academy of Sciences of the USA 17296–17301 ͉ PNAS ͉ November 14, 2006 ͉ vol. 103 ͉ no. 46 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0603689103 EVOLUTION Fig. 1. Location of Dogali in Eritrea and enlarged map of the discovery site of Eritreum melakeghebrekristosi gen. et sp. nov. (A) Simplified map of Eritrea (Inset, bottom left), and a close-up view of Dogali and the location of the bone-bearing site (map was constructed from data collected by using Total Station). (B) SW–NE section of the Dogali site through the place where Eritreum was found. Geological sections along the heavy lines in A are depicted in B.