Journal of African Earth Sciences 107 (2015) 28–35

Contents lists available at ScienceDirect

Journal of African Earth Sciences

journal homepage: www.elsevier.com/locate/jafrearsci

Pliocene Carnivora (Mammalia) from the Hadar Formation at Dikika,

Lower Awash Valley, Ethiopia
Denis Geraads a,⇑, Zeresenay Alemseged b, René Bobe c, Denné Reed d
a
Sorbonne Universités – CR2P – MNHN, CNRS, UPMC-Paris6, CP 38, Muséum National d’Histoire Naturelle, 8 rue Buffon, F-75231 PARIS Cedex 05, France
b
Department of Anthropology, California Academy of Sciences, 55 Music Concourse Dr., San Francisco, CA 94118, USA
c
Department of Anthropology, Center for the Advanced Study of Hominid Paleobiology, The George Washington University, Washington, DC 20052, USA
d
Department of Anthropology, University of Texas at Austin, Austin, TX 78712, USA

a r t i c l e i n f o a b s t r a c t

Article history: We report here on further study of the Carnivora collected by the Dikika Research Project at Dikika, in the
Received 8 August 2014 Hadar Formation south of the type locality since 2000. The Canidae and the otter Enhydriodon have been
Received in revised form 27 February 2015 described elsewhere, so we focus here on the other Mustelidae and on the Felidae and Hyaenidae. All
Accepted 29 March 2015
Hyaenidae are referred to Crocuta, but differences in size and tooth proportions suggest two species that
Available online 7 April 2015
might belong to distinct lineages. An associated set of upper and lower teeth is made the type of a new
species of Lutra that must be close to the divergence of Lutra palaeindica, Lutra lutra, and Hydrictis mac-
Keywords:
ulicollis. Sample size is still small, but the Dikika assemblage differs from others of similar age in the
Pliocene
Africa
abundance of hyenas relative to felids.
Dikika Ó 2015 Elsevier Ltd. All rights reserved.
Carnivora
Hyaenidae
Mustelidae

1. Introduction Ethiopia, Addis Ababa; RVRME: Rift Valley Research Mission in

Ethiopia.
The Dikika Research Project has been working since 1999 in the
Pliocene Hadar Formation south of the Awash River. From base to 2. Systematic paleontology
top, the Dikika composite sequence includes the same members as
in the type area north of the river, i.e., the Basal, Sidi Hakoma, Order Carnivora.
Denen Dora and Kada Hadar Members, dated to ca. 3.5–2.9 Ma, fol- Family Herpestidae.
lowed by an unconformity that marks the base of the overlying Genus Herpestes Illiger, 1811.
Busidima Formation (Wynn et al., 2008, and Refs. therein). Most Cf. Herpestes sp.
of the sediments at Dikika belong to the oldest two members,
and are highly fossiliferous; they yielded numerous remains of A tooth fragment collected during screening at locality DIK-1
Australopithecus afarensis, of which the juvenile skeleton nick- comprises most of a thick P4 blade, but lacks the protocone; it
named ‘‘Selam’’ is the most famous (Alemseged et al., 2006; has a tall paracone and a well-marked carnassial notch, showing
Wynn et al., 2006), associated with a rich vertebrate fauna. that it is certainly herpestid. An anterior mandibular fragment with
Among the carnivores, new species of a canid and of a gigantic incomplete p1–p3, also from DIK-1 (Fig. 1C), could well be of the
otter have already been described (Geraads et al., 2010, 2011), same species. The generic identification is tentative; the material
and we continue here the study of this group. is too limited for fruitful comparisons with other African forms.
Abbreviations: MNHN: Muséum National d’Histoire Naturelle,
Paris; KNM: National Museums of Kenya, Nairobi; RMCA: Royal Family Hyaenidae.
Museum for Central Africa, Tervuren; NME: National Museums of Genus Crocuta Kaup, 1828.
Crocuta dietrichi Petter and Howell, 1989.

⇑ Corresponding author. Tel.: +33 1 40 79 37 62. DIK-32–1 is a right mandible from the Basal Member at Ilanle
E-mail addresses: geraads@mnhn.fr (D. Geraads), zalemseged@calacademy.org (Fig. 1A and B). It includes most of the corpus, the anterior portion
(Z. Alemseged), bobe@gwu.edu (R. Bobe), reedd@mail.utexas.edu (D. Reed). of the ascending ramus, and well-preserved but heavily worn

http://dx.doi.org/10.1016/j.jafrearsci.2015.03.020
1464-343X/Ó 2015 Elsevier Ltd. All rights reserved.
 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35 29

Fig. 1. A-B; Crocuta dietrichi, mandible with p2–m1 DIK-32-1 in A, occlusal, and B, buccal views. C, Cf. Herpestes sp., mandible with c–p3 from DIK-1 in buccal view. D and E,
Crocuta cf. dietrichi, maxilla DIK-43-14 with P3–P4 in D, occlusal and E; buccal views. F–H, Crocuta cf. dietrichi, DIK-76-4c in F, occlusal, and G and H, buccal views. I–K, Lutra
hearsti n. sp., holotype DIK-50-35; I, occlusal view of C-M1; J, lingual view of p2–m1, K, occlusal view of c-m2. L–N, Homotherium hadarensis, mandible DIK-96-1; L, front view
of the symphysis with the canine; M, lateral view of the same, N, lateral view of right dentary. O and P, Crocuta eturono, maxilla with I3-M1 DIK-73-1; O, occlusal view; P,
lateral view of P3-P4. Scale = 10 cm for Figs. B, G, L–N, 5 cm for Figs. A, D–F, H, O–P, 2.5 cm for Figs. C, I–K.

p2–m1. It is among the smallest known Crocuta, being only slightly
larger than the specimen from Meob, Namibia (Morales et al.,
2011). Compared to other species of Crocuta, the mandibular cor-
pus below the ascending ramus is deep, probably because of the
old age of this individual. The corpus is shallower below the pre-
molars and diastema, but remains deep rostrally, so that it is likely
that the canine was rather upright; there is a large mental foramen
below p2, as in modern C. crocuta.
Due to the advanced wear that differentially affected the distal
portions of the teeth, the crowns of p2 and p3 are not horizontal
but inclined backwards. Additionally, their relative positions may
have been modified but p2 did not insert significantly higher than
p3. Its insertion in the jaw is thus similar to that observed in other
early fossil forms such as C. dietrichi from Laetoli (Laet 75-2953,
Laet 78-5107) and Koobi Fora (KNM-ER 721) and contrasts with
the derived condition found in C. crocuta and most closely related
forms, and in C. dbaa from the late Pliocene of Ahl al Oughlam,
Morocco (Geraads, 1997). Although there are exceptions, in the
derived condition p2 usually inserts much higher than p3, so that
their cervixes are at different levels. The height of p3 of DIK-32-1
cannot be estimated, but its position in the jaw shows that it
was much less tall than in C. crocuta. It is, however, broad with a
rectangular outline, as in other Crocuta. The p4 is heavily worn
and displays no diagnostic features. The carnassial blade is also
well worn, with a small vestigial metaconid and a short and simple
talonid. This tooth is remarkably long relative to the premolars, as
shown in Fig. 2 where the length of m1 is plotted against the sum
of the lengths of the premolars. DIK-32-1 is morphologically simi-
lar to the Meob specimen, and shares with it a long m1 relative to
the premolars. As shown in Fig. 2, such a long carnassial is also
found in C. dietrichi (the length of the missing p2s for the holotype
Laet 75-2953 and for Laet 76-3970 were estimated at 13 and
14 mm, respectively; these are maximum estimates from what
remains of these teeth). Not unexpectedly, among the large sample
30 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35
Fig. 2. Plot of m1 length vs. the sum of the lengths of p2, p3 and p4 in some Crocuta.
Data mainly from Turner (1984), Werdelin and Lewis, (2008), Werdelin and
Dehghani, (2011), Werdelin and Lewis (2013), Petter (1973), and original.
of C. crocuta a few specimens also have relatively long m1s but, by
contrast, all specimens of C. dbaa have longer premolars.
From Hadar, Howell and Petter (1976a) reported some hyaenid
specimens, casts of which are available in the MNHN. Among them,
AL-202-1 from the Sidi Hakoma Member seems to have puzzled
these authors, but we suspect that weathering is the cause for
the small size of its teeth. It most likely belongs to C. dietrichi.
Thus the mandible DIK-32-1 can be assigned confidently to
C. dietrichi, but species identification of other specimens is more
problematic.
Crocuta cf. dietrichi Petter and Howell, 1989
DIK-43–14 is a maxilla with P3 – P4 and an associated but iso-
lated M1, from the base of the Sidi Hakoma Member (Fig. 1D and
E). The P3 is crocutoid in its rectangular rather than rhomboidal
outline, with weak mesio-lingual and distal accessory cusps, but
the mesio-buccal corner is not expanded and this tooth is narrow
relative to P4. The carnassial is remarkably long and slender
(Fig. 3), with a metastyle distinctly longer than the paracone. In
contrast to this hyper-crocutoid feature, the M1 is much less
reduced than in modern C. crocuta.
The shape of P3, narrower and less square than in C. crocuta, and
the large size of M1 suggests DIK represents a primitive form of
Crocuta. Its overall size is consistent with Crocuta, but its narrow
carnassial sets it apart from other forms. However, an outlier speci-
men with narrow carnassials also occurs among Siwalik Crocuta
(L. Werdelin, comm. pers.), suggesting that this may reflect individ-
ual variation. DIK-43-14 is virtually identical in morphology, tooth
Fig. 3. Plot of P4 length vs. P4 width in some Crocuta. Main data as for Fig. 2.
measurements and ratios (especially in its narrow P3 and long P4)
to ER-940, a maxilla with P3–P4 assigned to C. ultra by Werdelin
and Lewis (2013), which comes from Bura Hasuma, a locality
roughly 1 My younger than the base of the Sidi Hakoma Mb.
Although we realize that this is not a sound basis for species iden-
tification, we believe that this difference in age makes assignment
to C. ultra unlikely, and we tentatively call it C. cf. dietrichi.
Specimen DIK-76-4 from the Sidi Hakoma Member consists of a
partial right maxilla with P1 and P2 DIK-76-4a, a left maxilla with
P3 and P4 DIK-76-4b, a mandible with p2 – m1 and DIK-76-4c. The
latter tooth is incomplete, and most of the teeth are very worn,
cracked and suffered ‘‘expanding matrix distortion’’. The P1 is a
small, monoradiculate tooth sandwiched between the C and P2,
with no diastema between them. The P2 is a broad tooth with a
rectangular rather than rhomboidal outline and the P3 is also
broad, especially distally, and squarish; the mesio-lingual crest
bears a small cusplet. The upper carnassial is too damaged to
describe in detail but was certainly thicker than the other speci-
mens. The presence of an M1 alveolus indicates that this tooth
was present and probably not vestigial.
On the lower jaw DIK-76-4c (Fig. 1F–H), the p2 is implanted at
the same level as the p3. The latter is short and relatively broad
compared to that in most modern C. crocuta, and was certainly tall.
The mesial accessory cuspid of p4 is much reduced, as sometimes
occurs in the modern C. crocuta, where its size varies considerably.
The lower carnassial is damaged, but its metaconid was at most
minuscule, and had a short talonid. This tooth is only a trifle longer
than that of DIK-32-1, but the premolars are distinctly longer and
broader.
The lower jaw of DIK-76-4 is distinctly more similar to
C. crocuta than is DIK-32-1 assigned above to C. dietrichi, in spite
of the relatively narrow temporal gap between the two, probably
closer to 0.1 than to 0.5 My. Still, it remains more primitive than
the modern form in the positioning and large size of its p2, as in
C. dbaa and many C. dietrichi specimens, while its short, broad p3
sets it somewhat aside from most other Crocuta (Fig. 4). The latter
peculiarity is the only significant metric difference with modern
C. crocuta, and with several specimens described as C. crocuta ultra
from Omo, Koobi Fora and Olduvai (Howell and Petter, 1976a;
Petter, 1973; Werdelin and Lewis, 2013), although all these speci-
mens are at least 1 My younger. Instead, the small mesial cuspid of
p4 and the broad p3 are reminiscent of C. eturono Werdelin and
Lewis, 2008, whose holotype is slightly older, so the assignment
to this species would be arguable, though the much smaller size
of DIK-76-4 (Fig. 2) would exceedingly expand the size range of
this species. Therefore, we also tentatively assign DIK-76-4 to C.
cf. dietrichi,
Fig. 4. Plot of p3 length vs. p3 width in some Crocuta. Main data as for Fig. 2.
 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35 31

This species likely documents an early stage of a lineage leading

to derived Crocuta, of which C. dbaa documents a slightly more
derived form at a later period. There is no doubt that these species
are closely related, but we believe that lumping them in the same
taxon, as was done by Werdelin and Peigné (2010) and Werdelin
and Lewis (2013) obscures our understanding of the early
evolutionary patterns of the genus.
The only hyaenid post-cranial remains are DIK-53-2 from the
Basal Member, which consist of associated proximal ulna and dis-
tal radius, of small size. The olecranon of the ulna has a weaker
caudal tuberosity than modern specimens, and the distal radius
also has poorly developed grooves and tuberosities. The small size
and early age suggest assignment to C. dietrichi.

Crocuta eturono Werdelin and Lewis, 2008

DIK-73-1 is a partial maxilla with I3-M1 from the Kada Hadar Fig. 5. Plot of p3 length vs. m1 length in some Crocuta. Main data as for Fig. 2.
Member just below the unconformity, at probably just less than
3.0 Ma. This specimen suffered much expanding matrix distortion,
that other Crocuta species reach only much later, and because of
so that the measurements provided in Table 1, which take into
its short p3. This species differs from other Pliocene Crocuta, such
account the resulting enlargement of the teeth, are approximate.
as C. dietrichi, C. dbaa and some other specimens that Werdelin
It is clear, anyway, that these teeth are extremely large, certainly
and Lewis (2013) called C. ultra, in its larger size, and its p3 that
close to the maximum size reached by modern and Pleistocene
is, on average, relatively broader.
Crocuta. The anterior teeth are too distorted to describe, but P3
It is obviously difficult to compare DIK-73-1, a maxilla, with the
was certainly short and squarish in shape. The carnassial has a
holotype of C. eturono, which is a mandible, but the long, slender
long, slender blade in which the parastyle is much shorter than
blade of the upper carnassial and broad M1 of the former would
the paracone, which is shorter than the metastyle. The M1 is rela-
match the long lower carnassial and relatively large m1 talonid
tively large, being short but not considerably narrower than the P2.
of the latter (and of KNM-ER 3748, another specimen assigned to
The large size prompts comparisons with Pachycrocuta, a genus
this species by Werdelin and Lewis, 2013), while the size, much
that has been described from East Africa (Werdelin, 1999).
larger than that of all other contemporaneous Crocuta, is another
Werdelin (1999) was uncertain about Pachycrocuta’s presence at
major similarity. Hence, we assign DIK-73-1 to C. eturono, of which
Hadar, as he stated that this genus is absent, but noted it in his
it represents the first known complete upper dentition, allowing us
chronological chart (Werdelin, 1999, Fig. 3). It is clear, however,
to add the little reduced M1 to its known characters, in sharp con-
that DIK-73-1 does not belong to this genus, whose P4 buccal cusps
trast to C. crocuta.
are sub-equal in length, whereas the metastyle of DIK-73-1 is
The Dikika collection of Pliocene Crocuta, although not very
almost twice as long as the parastyle, more like the proportions
large, significantly increases the known diversity of this group,
observed in DIK-43-14 or other Crocuta, and the broad P3 confirms
while highlighting the difficulty in drawing taxonomic boundaries
this generic identification, but the large size of M1 is a clear differ-
within it. We hope that future, large-scale revisions will allow for a
ence with C. crocuta.
better understanding of the geographic origin of the modern form,
Again, comparison with C. eturono Werdelin and Lewis, 2008, is
which Sheng et al. (2014) considered to be Eurasia.
needed. The type-specimen of this species is the mandible KNM-
WT 40181. According to the authors, its diagnostic features are
Family Felidae Fischer de Waldheim, 1817.
the large size, the great depth of the mandibular ramus and espe-
Genus Homotherium Fabrini, 1890.
cially the great length of the lower carnassial; they did not provide
Homotherium hadarensis Petter and Howell, 1988.
the raw data on tooth measurements but from their Fig. 3 the m1 is
33–33.1 mm long. This tooth is imperfectly preserved, but careful
DIK-96-1includesseveralfragmentsofamachairodontmandible:
comparison with other Crocuta specimens in the KNM shows that
part of the symphysis, a right i2 or i3, the left lower canine, an incom-
this is an over-estimate, and that the actual length was certainly
plete p4, and the posterior half of the dentary (Fig. 1L–N). Enough is
close to 30.5–31 mm. Once this measurement is corrected, the
preserved of the front part of the symphysis to show its square outline
ratio Lp4/Lm1 of KNM-WT 40181 is within the range for modern
and its slightly depressed central part, pierced by a small foramen
C. crocuta (see, e.g., Turner, 1984), while the ratio Lp3/Lm1 remains
nearits ventral border,and a largerone atmid-height.This rostral part
low in part due to the short p3 (Fig. 5). Still, we believe that
of the symphysis is sharply delimited from the lateral part by pillars
C. eturono is a valid species because KNM-WT 40181 is of a size

Table 1
Linear dental measurements of Dikika Carnivora in mm.

Taxon Number L P1 W P1 L P2 W P2 L P3 W P3 L P4 W P4 L M1 W M1 L M2 W M2
cf. Herpestes sp. DIK-1 Lower teeth 3.8 4.7
Crocuta dietrichi DIK-32-1 Lower teeth 11.9 8.5 16.8* 12.1 17.5 11.6 25* 10.8
Crocuta cf. eturono DIK-76-4 UPPER TEETH 5 5.3 16 10.8 22.2 14.9 35 19.6
‘‘ Lower teeth 14.1 9.3 17.6 14.1 20.1 13 25.3
Crocuta cf. eturono DIK-73-1 UPPER TEETH 5.5 6.1 19 16 25 18* 41* 22* 7.5 15
Crocuta cf. eturono DIK-43–14 UPPER TEETH 20.2 13.4 36.4 17.5 6.3 12.5
Lutra hearsti DIK-50–35 UPPER TEETH 5.9 4 11.3* 8.4 8.2* 12.3*
‘‘ Lower teeth 5.9 3.4 8 3.9 12.3 6.9 4.7 5
*
Measurements corrected to account for wear or EMD.
32 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35
that extend ventrally into small parasagittal flanges, so that in ante-
rior view the ventral border of the symphysis is concave on either side
of the midline. A large and very rostrally located mental foramen
opens below the root of the canine. The dorsal part of the symphysis
is missing, but positioning the canine in its alveolus allows an esti-
mate of its depth, which was moderate. The ventral border of the
ramus is slightly concave, and the coronoid process, which is not
extremely reduced, ascends gradually behind m3. None of these teeth
isworn. Bothkeels ofthelower canineand the mesialkeelsof i2 and p4
are crenulated. The canine is much larger than the preserved incisor,
which in turn is much larger than i1, as estimated from its alveolus.
The p4 is thick with a robust anterior cuspid, and this tooth was proba-
bly not very long; its alveolus shows that the carnassial was long, but
the relative proportions of these teeth cannot be estimated.
DIK-83-3 is a felid distal metapodial (articular width 16.5)
whose size would match DIK-96-1, but our inclusion in the same
species is tentative, because the supra-articular tubercles, unless
reduced by weathering, are weak for Homotherium.
Remains of Homotherium have been recovered from several East
African sites, but most of the mandibular remains are fragmentary.
From Omo, Homotherium ethiopicum Arambourg, 1948 is diagnosed
based upon a mandibular fragment with very incomplete p4 and
m1. Too much is missing of these teeth to reliably estimate their
original shape, but nothing demands inclusion in this genus; the
presence of a large foramen below the distal root of p3 shows that
the holotype is felid but we believe, following Howell and Petter
(1976b), that ‘H.’ ethiopicum is more probably a Dinofelis, because
the distal alveolus of p3 is large, the mesial cuspid of p4 is small,
and m1 is much shorter, both absolutely and relative to p4, than
in East African Homotherium. Puzzling features, perhaps due to
the young age of the specimen, are the masseteric fossa which does
not reach the level of the posterior border of m1 and a shallow cor-
pus. In any case the species name should be regarded as a nomen
dubium rather than as a senior synonym of Dinofelis petteri
Werdelin and Lewis (2001).
Petter and Howell (1988) described a cranium from the Denen
Dora Member at Hadar as H. hadarensis. From South Turkwel,
Werdelin and Lewis (2000) reported only postcranial fragments.
From West Turkana, Harris et al., 1988 described a skull as
Homotherium problematicum (Collings, 1972), but the material
includes no mandible. Werdelin and Lewis (2013) described
mandibular remains of Homotherium sp. from Koobi Fora, but nothing
among the few overlapping elements suggests that they differed from
DIK-96-1. Werdelin et al. (2014) reported Homotherium sp. from
Woranso-Mille, slightly older than Hadar, but did not illustrate or
describe mandibular material. Homotherium is encountered in
Africa until the late early/early Middle Pleistocene of Tighenif in
Algeria but again the material includes no comparable parts.
A distinctive feature of DIK-96-1 is the position of the mental
foramen, which is more rostrally located than in other
Homotherium, including NME-KL-232-1 collected by the RVRME
from the base of the Matabaietu Formation (Kalb, 1993).
However, this difference might just be due to variation, and we
tentatively include the Dikika material in the Hadar species
because of their geographic and temporal proximity, although we
recognize that revision of the Hadar material and of African
machairodonts could lead to a re-evaluation of the taxonomy of
African Pliocene forms.
Family Canidae Gray, 1821.
Genus Nyctereutes Temminck, 1838.
Nyctereutes lockwoodi Geraads, Alemseged, Bobe and Reed, 2010.
This species has been described by Geraads et al. (2010) and
nothing can be added concerning this species, known only from
the Sidi Hakoma and Denen Dora Members.
Family Mustelidae Fischer de Waldheim, 1817.
Genus Enhydriodon Falconer, 1868.
Enhydriodon dikikae Geraads, Alemseged, Bobe and Reed, 2011.
The Dikika material of this species has been described by
Geraads et al. (2011). Since then, it has also been described from
Kanapoi (Werdelin and Manthi, 2012) and might also be present
at Koobi Fora, alongside E. afman Werdelin and Lewis, 2013 (pers.
obs. R.B.).
Genus Lutra Brünnich, 1771.
Lutra hearsti sp. nov.
Holotype: DIK-50-35, consisting of a maxilla DIK-50-35a with
C–M1 (P2 missing) and associated mandible DIK-50-35b with c–
m2; from the lower part of the Sidi Hakoma Member at Dikika, c.
3.3 Ma. (Fig. 1I–K).
Etymology: in acknowledgment of the support provided by
Margaret and Will Hearst.
Diagnosis: a species similar in size to L. lutra and Hydrictis mac-
ulicollis. P4 with a distinct parastyle and a small protocone; M1
large but with lingual part only slightly expanded, paracone–meta-
cone axis making a strong angle with P4 blade; p4 with a strong
disto-buccal accessory cuspid; m1 low and broad, virtually no lin-
gual cingulum, a closed trigonid with short paraconid, and a dis-
tinct hypoconulid.
Description: Overall, the dentition resembles those of the mod-
ern palearctic Lutra lutra and sub-Saharan H. maculicollis, but there
are some differences.
On P4, the protocone is much shorter than the blade, so that the
tooth has a concave disto-lingual border; there is a distinct para-
style, stronger than in all L. lutra and slightly larger than in
H. maculicollis. M1 has a poorly expanded hypocone and almost
no mesio-lingual cingulum, so that it is shorter lingually than
bucally; its lingual part is distinctly less expanded than those of
H. maculicollis and especially of L. lutra. A crack has slightly dis-
placed the paracone (raw measurements have been corrected
accordingly in Table 1), but even taking this into account, the para-
cone–metacone axis is more inclined relative to the sagittal plane
than in modern Lutra, so that it forms a less obtuse angle with the
P4 blade.
On the mandible, the masseteric fossa reaches the level of m2;
p4 has a disto-buccal accessory cuspid stronger than in modern
L. lutra. The carnassial has a weaker lingual cingulum than modern
Lutra, but the main difference is in the shape of the trigonid: in
DIK-50-35b the paraconid is short and directed mesio-lingually,
so that the tip of the protoconid is about equidistant from the para-
conid and metaconid; in H. maculicollis and still more so in L. lutra,
instead, the paraconid is long so that the tip of the protoconid is
closer or much closer to the metaconid. Because of this short trigo-
nid, the m1 of DIK-50-35b has a low length to width ratio (Fig. 6).
The tooth being unworn, there are distinct hypoconid and hypoco-
nulid, and the talonid is slightly concave and bordered by a lingual
ridge; these features are identical to those of modern Lutra.
Another resemblance with L. lutra and H. maculicollis is the low
crown.
Among the fossil forms, L. simplicidens from the European
Pleistocene (Willemsen, 1992) resembles L. lutra; differences in
talonid width and hypoconulid development (Willemsen, 1992,
Fig. 7, pl. 1) look subtle, and the post-cranial differences noted by
Willemsen (1992) should probably be checked against a large
modern sample.
Lutra fatimazohrae Geraads, 1997 from the late Pliocene of Ahl al
Oughlam in Morocco differs from DIK-50-35 in its larger size, very
short P4 protocone not extending distally beyond the paracone,
lingually expanded M1 resembling modern Lutra (unpublished
 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35 33

associated P4 and M1 AaO-4819), taller m1 crown, and m1 talonid
short and slightly narrower than the trigonid, but resembles DIK-
50-35 in its weak lingual cingulum and strong p4 distal cuspid.
Lutra palaeindica Falconer, 1868 is mostly known from a skull
M37151 and mandible M37152 in the NHML, from the Upper
Siwaliks. Their tooth measurements are similar to those of DIK-
50-35; the teeth are much worn and abraded but M1 is also similar
in its poorly expanded lingual part, and oblique paracone–meta-
cone axis. Unfortunately, most of the m1 trigonid is now broken
Fig. 6. Plot of m1 length vs. m1 width in some Lutra and ‘‘Hydrictis’’. Data from
Willemsen (1992), Werdelin and Lewis (2013) and original (L. lutra in MNHN,
H. maculicollis in RMCA).
off, but it was certainly longer relative to the talonid than in DIK-
35-1. Another difference is that the masseteric fossa reaches the
level of m1, thus it extends farther mesially than in DIK-35-1 and
most modern Lutra. Interestingly, L. palaeindica resembles
H. maculicollis more than L. lutra in its relatively weak post-orbital
processes and frontals that are about as broad between the orbits
as at the post-orbital constriction.
Torolutra ougandensis was erected by Petter et al. (1991) for
early Pliocene specimens from Uganda, and several fossils have
been assigned to the same genus or species. Those from the
Middle Awash (Haile-Selassie, 2008; there is also a mandible col-
lected by the RVRME, NME-KL-225-1) are of similar age, but the
specimen from Toros-Menalla in Chad (Peigné et al., 2008) is older,
while those from Koobi Fora (Werdelin and Lewis, 2013) are much
younger. Of the generic characters, the position of the masseteric
fossa is of doubtful value, as in the specimens from Koobi Fora it
reaches almost as far mesially as in Lutra; it is likely that the fossa
migrates rostrally with the shortening of the mandible. However,
the larger size, the small size of the m1 talonid, the flattening of
its surface, and the elevation of the cuspids of p4 and of the m1
trigonid, could be distinctive characters of Torolutra, in which case
L. fatimazohrae could also belong here, although it is smaller
(Werdelin and Lewis, 2013, fig. 4.6 — the p4 AaO-3065 was erro-
neously assigned to Lutra by Geraads, 1997 but belongs to
Mellivora).
The poorly known L. libyca Stromer, 1913 also has a small talo-
nid, but the cuspids are low and the masseteric fossa extends far
rostrally; it could be a true Lutra.
Hydrictis gudho Werdelin and Lewis, 2013 is based upon a
poorly preserved partial cranium and posterior part of mandible.
Werdelin and Lewis (2013) stated that the m1 talonid is ‘‘clearly
basined’’, in contrast to Torolutra, but as in Lutra and
H. maculicollis, the spotted-necked otter, and connected it to the
latter species rather than to Lutra on biogeographic grounds.
However, the poor preservation of the type-specimen renders
examination of its morphology difficult; furthermore H. gudho is
significantly larger than both L. lutra and H. maculicollis, and the
talonid of m1 is narrow and tapers distally, unlike in these species,
and we believe that the affinities H. gudho remain unclear. In any
case, it is certainly distinct from L. hearsti.

3. Discussion

Fig. 7. Single shortest tree (L = 18; ci = 77; ri = 60) resulting from the matrix of
Given the very incomplete preservation of the fossil species
Table 2. Numbers refer to characters in Table 2; rectangles are apomorphies, open L. fatimazohrae, L. palaeindica, and L. hearsti, any reconstruction of
ones being homoplasies. a phyletic tree must be considered as very tentative, especially

Table 2
Characters and scores used in the parsimony analysis of otters. The hypothetical outgroup has the likely primitive character states. All characters are treated as additive.

Outgroup Lutra Hydrictis Lutra Lutra Aonyx Lutra

(primitive) lutra maculicollis hearsti palaeindica spp. fatimazohrae
0 1 2
0 Post-orbital Weak Strong 0 1 0 ? 0 0 ?
constriction
1 Post-orbital process Strong Weak 0 0 1 ? 1 0 ?
2 P1 in adult Present Absent 0 0 0 0 0 1 0
3 P4 parastyle Absent or weak Distinct 0 0 1 1 1 1 ?
4 P4 protocone Small Moderate Very large 0 1 1 0 0 2 0
5 M1 lingual part Little Expanded Very 0 2 1 0 1 2 1
expanded expanded
6 M1 parac/metacone Slanting Parasagittal 0 1 1 1 1 1 1
axis
7 p4 distobuccal cuspid Present Weak 0 1 0 0 ? 0 0
8 m1 crown Low Tall 0 0 0 0 ? 0 1
9 m1 proportions Long narrow Short broad 0 0 0 1 0 1 0
10 m1 trigonid Very short Short Long 1 2 1 1 2 0 1
34 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35
because scoring of the characters is often far from straightforward.
On the basis of the few, mostly dental characters listed in Table 2, a
parsimony analysis using TNT (Goloboff et al., 2008) yielded a sin-
gle shortest tree, shown in Fig. 7. All otters considered here with
the exception of L. hearsti are united by some enlargement of the
lingual part of M1, but the difference is subtle, and should not be
given too much significance, even though it is in agreement with
chronology. The same is true of other branches, of which none is
supported by more than one character change, but we may note
that all fossil forms occupy more basal positions than the modern
ones. The cladogram unites Aonyx with Lutra lutra, their synapo-
morphy being the more expanded M1 lingual part. This result is
surprising at first sight, because these taxa do not share many
derived characters, but is in agreement with DNA sequence analy-
sis (Koepfli and Wayne, 2003; Wolsan and Sato, 2010) and with
their late appearances in the fossil record; it implies that the char-
acters of the soft parts shared by L. lutra and H. maculicollis (Pocock,
1921) are primitive. However, if indeed the lineages leading to
these two latter species diverged near the Mio-Pliocene boundary
(Koepfli et al., 2008, Fig. 2), they evolved quite slowly. Indeed,
the African spotted-necked otter had long been included in Lutra,
but Wilson and Reeder (2005) favored generic distinction, as H.
maculicollis (Lichtenstein, 1835). Koepfli et al. (2008) confirmed
this on the basis of their molecular analysis, in which this species
never appears as the sister taxon of Lutra lutra, and this systematic
distinction is now generally accepted, although the precise posi-
tion of the spotted-necked otter varies with the DNA sequence ana-
lyzed (Koepfli and Wayne, 2003), and the methods used (Koepfli
et al., 2008). However, the topology of the tree (Fig. 7) would
require a distinct genus name for each species, and we prefer to
take the conservative view of using for the incompletely known
fossil taxa the earliest name, Lutra, while acknowledging that this
genus is almost certainly paraphyletic.
4. Conclusions
The current small samples from Dikika probably do not repre-
sent the full diversity at the site and future research will likely alter
our assessments of carnivore taxonomic richness. Because very lit-
tle screening has been undertaken there is a clear bias toward lar-
ger forms of carnivores. Here we note that the Hyaenidae is the
most common family of Carnivora. This is in sharp contrast with
the Upper Unit of the Laetolil beds, where herpestids and felids
far outnumber hyenas (Werdelin and Dehghani, 2011); felids are
also more common than hyaenids in the Tulu Bor Member of
Koobi Fora (Werdelin and Lewis, 2013), and in Members A-C of
the Shungura Formation at Omo (unpublished catalogue). Given
the wide range of felid habitats, it is unlikely that their rarity at
Dikika has any environmental significance; a tentative hypothesis
that only further collecting might confirm is that taphonomic fac-
tors and depositional conditions explain the observed hyaenid
abundance.
Acknowledgments
We thank all members of the Dikika Research Project for col-
lecting the fossils, Margaret and Will Hearst for their financial sup-
port, as well as those people who allowed access to collections in
their care: C. Argot and C. Bens (MNHN), E. Gilissen (RMCA),
E. Mbua and F. Kyalo (KNM), and G. Senishaw (NME). Thanks to
the editor and an anonymous reviewer for their comments on
the manuscript, which benefited from the very detailed and
insightful comments of L. Werdelin.
References
Alemseged, Z., Spoor, F., Kimbel, W.H., Bobe, R., Geraads, D., Reed, D., Wynn, J.G.,
2006. A juvenile early hominin skeleton from Dikika, Ethiopia. Nature 443, 296–
301.
Arambourg, C., 1948. Mission Scientifique de l’Omo 1932–1933. T.1: Géologie-
Anthropologie. Fasc. 3. Contribution à l’étude géologique et Paléontologique du
bassin du lac Rodolphe et de la basse vallée de l’Omo. Deuxième partie:
paléontologie. Editions du Muséum, Paris, pp. 231–562.
Collings, G.E., 1972. A new species of machaerodont from Makapansgat. Palaeont.
afr. 14, 87–92.
Falconer, H., 1868. Paleontological memoirs and notes. . .Hardwicke, London.
Geraads, D., 1997. Carnivores du Pliocène terminal de Ahl al Oughlam (Casablanca,
Maroc). Geobios 30, 127–164.
Geraads, D., Alemseged, Z., Bobe, R., Reed, D., 2010. Nyctereutes lockwoodi n.sp., a
new canid from the Middle Pliocene of Dikika, Ethiopia. J. Vertebr. Paleontol. 30,
981–987.
Geraads, D., Alemseged, Z., Bobe, R., Reed, D., 2011. Enhydriodon dikikae n. sp.
(Carnivora: Mammalia), a gigantic otter from the Pliocene of Dikika, Lower
Awash, Ethiopia. J. Vertebr. Paleontol. 31, 447–453.
Goloboff, P.A., Farris, J.S., Nixon, K.C., 2008. TNT, a free program for phylogenetic
analysis. Cladistics 24, 774–786.
Haile-Selassie, Y., 2008. New observations on the Late Miocene -Early Pliocene
Lutrinae (Mustelidae: Carnivora, Mammalia) from the Middle Awash, Afar Rift,
Ethiopia. Comptes Rendus Palevol 7, 557–569.
Harris, J., Brown, F., Leakey, M., 1988. Stratigraphy and paleontology of Pliocene and
Pleistocene localities west of Lake Turkana, Kenya. Contributions in Science,
Natural History Museum of Los Angeles County 399, 1–128.
Howell, F.C., Petter, G., 1976a. Indications d’âge données par les Carnivores de la
formation d’Hadar (Ethiopie orientale). Comptes Rendus de l’Académie des
Sciences II 282, 2063–2066.
Howell, F.C., Petter, G., 1976b. Carnivora from Omo group formations, southern
Ethiopia. In: Coppens, Y., Howell, F.C., Isaac, G.Ll., Leakey, R.E.F. (Eds.), Earliest
Man and Environments in the Lake Rudolf Basin. University of Chicago Press,
Chicago.
Kalb, J., 1993. Refined stratigraphy of the hominid-bearing Awash group, Middle
Awash Valley, Afar depression, Ethiopia. Newslett. Stratigr. 29, 21–62.
Koepfli, K.P., Wayne, R.K., 2003. Type I STS Markers are more informative than
Cytochrome b in phylogenetic reconstruction of the Mustelidae (Mammalia:
Carnivora). Syst. Biol. 52, 571–593.
Koepfli, K.P., Deere, K.A., Slater, G.J., Begg, C., Begg, K., Grassman, L., Lucherini, M.,
Veron, G., Wayne, R.K., 2008. Multigene phylogeny of the Mustelidae: resolving
relationships, tempo and biogeographic history of a mammalian adaptive
radiation. BMC Biol. 6 (10).
Lichtenstein, K.M.H., 1835. Ueber Lutra maculicollis Lichtenst. aus dem Kafferlande.
Archiv für Naturgesch. 1, 89–92.
Morales, J., Senut, B., Pickford, M., 2011. Crocuta dietrichi from Meob, Namibia:
implications for the age of the Tsondab Sandstone in the coastal part of the
Namib Desert. Estudios Geológicos 67, 207–215.
Peigné, S., de Bonis, L., Likius, A., Mackaye, H.T., Vignaud, P., Brunet, M., 2008. Late
Miocene Carnivora from Chad: Lutrinae (Mustelidae). Zool. J. Linnean Soc. 152,
793–846.
Petter, G., 1973. Carnivores pléistocènes du ravin d’Olduvai (Tanzanie). Fossil
Vertebrates of Africa 3, 43–100.
Petter, G., Howell, F.C., 1988. Nouveau félidé machairodonte (Mammalia, Carnivora)
de la faune pliocène de l’Afar (Ethiopie): Homotherium hadarensis n. sp. Comptes
Rendus de l’Académie des Sciences II 306, 731–738.
Petter, G., Howell, F.C., 1989. Une nouvelle espèce du genre Crocuta Kaup
(Mammalia, Carnivora, Hyaenidae) dans la faune pliocène de Laetoli
(Tanzanie): Crocuta dietrichi n.sp.; origine du genre. Comptes Rendus de
l’Académie des Sciences II 308, 1031–1038.
Petter, G., Pickford, M., Howell, F.C., 1991. La loutre piscivore du Pliocène de
Nyaburogo et de Nkondo (Ouganda, Afrique orientale): Torolutra ougandensis
n.g., n.sp. (Mammalia, Carnivora). Comptes Rendus de l’Académie des Sciences II
312, 949–955.
Pocock, R.I., 1921. On the external characters of some species of Lutrinae (otters).
Proc. Zool. Soc. Lond. 1921, 535–546.
Sheng, G.-L., Soubrier, J., Liu, J.-Y., Werdelin, L., Llamas, B., Thomson, V.A., Tuke, J.,
Wu, L.-W., Hou, X.-D., Chen, Q.-J., Lai, X.-L., Cooper, A., 2014. Pleistocene cave
hyenas and the recent Eurasian history of the spotted hyena, Crocuta crocuta.
Mol. Ecol. 23, 522–533.
Stromer, E., 1913. Mitteilungen über Wirbeltierreste aus dem Mittelpliozän des
Natrontales (Ägypten). 1. Affen; 2. Raubtiere. Zeitschrift der Deutschen
geologischen Gesellschaft 65, 350–372.
Turner, A., 1984. The interpretation of variation in fossil specimens of spotted
hyaena (Crocuta crocuta Erxleben, 1777) from Sterkfontein valley sites
(Mammalia:Carnivora). Annals Transvaal Museum 33, 399–418.
Werdelin, L., 1999. Pachycrocuta (hyaenids) from the Pliocene of East Africa.
Paläontologische Zeitschrift 73, 157–165.
Werdelin, L., Dehghani, R., 2011. Carnivora. In: Harrison, T. (Ed.), Paleontology and
Geology of Laetoli: Human Evolution In Context. Springer, Dordrecht, pp. 189–
232.
Werdelin, L., Lewis, M.E., 2000. Carnivora from the South Turkwel hominid site,
Northern Kenya. J. Paleontol. 74, 1173–1180.
 D. Geraads et al. / Journal of African Earth Sciences 107 (2015) 28–35
Werdelin, L., Lewis, M.E., 2001. A revision of the genus Dinofelis (Mammalia,
Felidae). Zool. J. Linn. Soc. 132, 147–258.
Werdelin, L., Lewis, M.E., 2008. New species of Crocuta from the early Pliocene of
Kenya, with an overview of early Pliocene hyenas of East Africa. J. Vertebr.
Paleontol. 28, 1162–1170.
Werdelin, L., Lewis, M.E., 2013. Koobi Fora Research Project. Volume 7: The
Carnivora. California Academy of Sciences, San Francisco.
Werdelin, L., Manthi, F.K., 2012. Carnivora from the Kanapoi hominin site, northern
Kenya. J. Afr. Earth Sci. 64, 1–8.
Werdelin, L., Peigné, S., 2010. Carnivora. In: Werdelin, L., Sanders, W.J. (Eds.),
Cenozoic Mammals of Africa. University of California Press, Berkeley, pp. 603–
658.
Werdelin, L., Lewis, M., Haile-Selassie, Y., 2014. Mid-Pliocene Carnivora from the
Woranso-Mille area. J. Mammal Evol, Afar region, Ethiopia, doi: 10.1007/
s10914-013-9250-5.
35
Willemsen, G.F., 1992. A revision of the Pliocene and Quaternary Lutrinae from
Europe. Scr. Geol. 101, 1–115.
Wilson, D.E., Reeder, D.M., 2005. Mammal Species of the World: A Taxonomic and
Geographic Reference, 3rd ed. Johns Hopkins University Press, Baltimore.
Wolsan, M., Sato, J.J., 2010. Effects of data incompleteness on the relative
performance of parsimony and Bayesian approaches in a supermatrix
phylogenetic reconstruction of Mustelidae and Procyonidae (Carnivora).
Cladistics 26, 168–194.
Wynn, J.G., Alemseged, Z., Bobe, R., Geraads, D., Reed, D., Roman, D.C., 2006.
Geological and palaeontological context of a Pliocene juvenile hominin at
Dikika, Ethiopia. Nature 443, 332–336.
Wynn, J.G., Roman, D.C., Alemseged, Z., Reed, D., Geraads, D., Munro, S., 2008.
Stratigraphy, depositional environments and basin structure of the Hadar and
Busidima formations at Dikika, Ethiopia. Geol. Soc. Am. Special paper 446, 87–
118.