hum
ba,
e Un
Paleoecology
are
ding
als
. The
pos
. Th
wev
Forma
003) h
l but o
a is us
tandin
rly Ho
members results from periodic episodes of high rainfall in the Ethi-
opian plateau that resulted in northward expansion of Lake Turkana
(Butzer, 1971; Brown and Fuller, 2008; McDougall et al., 2008). The
consistencies of the fauna among these threemembers re?ect similar
habitats. The depositional environments characterize thewet phases
of the interglacials, as they are deltaic and littoral deposits.
basin has an arid to semiarid tropical climate, characterized by high
temperatures (Butzer, 1971) and low precipitation, with mean
annual rainfall ranging from 320 to 380 mm and a maximum mean
temperature of 40 C (Carr, 1976), based on records of nearby sta-
tions such as Ileret, Todenyang, and Lokomarinyang (Butzer, 1971).
Although all fossil-bearing localities from the Kibish Formation
come from deltaic and littoral beds on both sides of the Omo River,
major sedimentary environments of the Omo Kibish area today also
include crystalline uplands, pediments, and piedmont alluvia. The
Contents lists availab
Journal of Hum
.e l
Journal of Human Evolution 55 (2008) 501?512* Corresponding author.environment, including the community in which they lived. The
mammalian fauna from Kibish represents a poorly known but
critical phase in the east African fossil record. According to the
newly revised dates (McDougall et al., 2005, 2008), the three
members of the Kibish Formation represented by mammalian
fossils sample the interglacial phases of three different isotopic
stages. Member I accords with isotopic stage 7 (240?200 ka),
Member III with isotopic stage 5 (130?75 ka), and Member IV with
the early stages of the Holocene (<10 ka). Speci?cally, each of the
Background
Vegetation
Today the Omo Kibish area supports both wet and dry vegeta-
tion types following the course of the meandering Omo River.
Closed to open woodlands dominate the inner bends of the river.
The ?oodplains in the extended area away from the riparian
woodland support open, dry habitats such as shrubland, grassland,
and bushland thickets. At a larger scale, the lower Omo regionalIntroduction
Recent expeditions to the Kibish
Valley of southern Ethiopia (1999?2
mammal remains that comes from al
of the formation. Mammalian faun
vegetation structure and for unders
ations of homininsdin this case, eaE-mail address: zassefa@gmail.com (Z. Assefa).
0047-2484/$ ? see front matter  2008 Elsevier Ltd.
doi:10.1016/j.jhevol.2008.05.015vegetation types. These vegetation forms are similar to current habitats surrounding the Omo River.
 2008 Elsevier Ltd. All rights reserved.
tion in the Omo River
ave yielded a sample of
ne of the four members
eful for reconstructing
g the ecological associ-
mo sapiensdwith their
In this paper, we discuss the vegetation and fauna that exist in
the Omo Valley today. This is followed by a description of the fauna
recovered from the members of the Kibish Formation. We then
reconstruct the paleoenvironment of the Kibish Formation based
on these faunal remains and discuss the mammalian communities
within which Homo sapiens fossils were recovered.Kibish
Middle Stone Ageand both indicate more closed habitats. Comparative analyses of the Kibish faunal remains using the
ecological-diversity approach document close associations with edaphic grassland and woodlandHylochoerus Hylochoerus meinertzhageni (giant forest hog), taxa that are extremely rare in the African fossil record,The large-mammal fauna from the Kibis
Zelalem Assefa a,*, Solomon Yirga b, Kaye E. Reed c
aArchaeobiology and Human Origins Programs, Smithsonian Institution, National Muse
bBiology Department, Addis Ababa University, Science Faculty, P.O. Box 1176, Addis Aba
c Institute of Human Origins, School of Human Evolution & Social Change, Arizona Stat
a r t i c l e i n f o
Article history:
Received 24 April 2007
Accepted 15 May 2008
Keywords:
East Africa
Ethiopia
Extant fauna
a b s t r a c t
The Kibish faunal remains
sapiens and for understan
assemblage of large mamm
from the Kibish Formation
the overall taxonomic com
living near the Omo River
nivore species are rare. Ho
journal homepage: wwwAll rights reserved.useful for reconstructing the habitat of the earliest documented Homo
the community within which early modern humans existed. A diverse
, including many species of bovids, suids, and equids, has been recovered
re are no extinct large mammals represented in the fossil assemblage, and
ition of the fossil fauna is similar to the modern-day wildlife community
e fossil faunal assemblage shows a paucity of arboreal primates, and car-
er, the faunal sample includes possible Cephalophus (duiker) remains andFormation
of Natural History, 4210 Silver Hill Road, Suitland, MD 20746-2863, USA
Ethiopia
iversity, P.O. Box 874101, Tempe, AZ 85287, USA
le at ScienceDirect
an Evolution
sevier .com/locate/ jhevolinteraction of various soil types, topographic changes, and
availability of moisture generally provide a mosaic of plant com-
munities in the region, which were assigned by Carr (1976) to the
broad habitat types of plains environments (grasslands), riverine-
associated environments (riparian forest and woodland), and
anomalous habitats (e.g., sporadic uplands and outcrops, streams,
hot springs). Figure 1 shows a schematic pro?le of vegetation types
across grasslands and riparian woodland.
Both grasslands and riverine-associated vegetation structures
are dependent on soil types, rainfall, and water-drainage patterns.
Higher elevations support higher plant diversity, while in the
mud?ats and ?oodplains, trees are essentially nonexistent (Carr,
1976, 1998). There are diverse mesic depositional environments,
such as deltas, silt berms, levees, back swamps, and adjacent ?ats
that also support high ?oral diversity. Vegetation types in the
modern delta and silt berms share many similarities. The riverine
woodlands occupy the natural levee crest of the river and the
immediate backslope. Carr (1976, 1998) noted progressive shifts in
the following sequences along south-to-north meanders of the
river: grassland and thicket, open woodland, closed woodland, and
forest. Such sequences particularly characterize levee vegetation
types along inner bends of the lower 50?60 km of the river. Farther
upstream, including the area where the Kibish Formation is
exposed, the forest-cover declines and is replaced by intermediate
variable grazers, browser-grazer intermediates, generalists, and
browsers. Thus, in its modern ecological setup, the region supports
diverse species preferring both open and closed habitats (Gagnon
and Chew, 2000; Cerling et al., 2003). However, grazers of both
obligate and variable types appear to have higher representation
among the extant bovids of the Omo Kibish area (Yalden et al.,
1984). The majority of grazing species are found mainly within the
different areas of grassland habitats. In the extant Omo region,
there are no documented forest frugivorous/browsing bovids (i.e.,
cephalophines), probably indicating that themodern riverine forest
is lacking in the type of fruiting plant species required to support
these bovids.
The Omo Kibish area is known for one of the few sympatric
distributions of Burchell?s (Equus burchellii) and Grevy?s zebras
(Equus grevyi). Their joint presence signi?es diverse precipitation
and temperature ranges and a wider range of accumulated vege-
tation indices, distinct from the predominantly milder or harsher
climates in the nonoverlapping ranges of these two equid species
elsewhere (Bauer et al., 1994). The representation of two species of
suids, the common warthog (Phacochoerus aethiopicus) and the
bushpig (Potamochoerus porcus), is also consistent with the mosaic
environments of Omo Kibish. The Omo Valley also supported the
last known survivors in Ethiopia of the once widespread black
Z. Assefa et al. / Journal of Human Evolution 55 (2008) 501?512502vegetation between woodland and forest. The riverine woodlands
include large trees, such as ?g (Ficus sycomorus), and two major
species of shrubsdgundi (Cordia sinensis) and confetti tree (May-
tenus senegalensis), which are often found in associationwith dense
thickets. The riverine forest, on the other hand, is composed of
woody plants that grow to a height of 25?30 m, and also maintain
dense canopy covers. Herbs and shrubs are also found associated
with the riverine forest. For a list of plant species, see Carr (1976,
1998).
Fauna
The extant and historically documented faunal composition of
the Omo Kibish area is characterized by diverse taxa of large
mammals (Table 1). Following Gagnon and Chew?s (2000) dietary-
preference assessment of extant African bovids, the species of
bovids reported from the Omo area include obligate grazers,Fig. 1. Pro?les of soil and vegetation types at the Lrhinoceros (Diceros bicornis), which preferentially inhabits
ecotones between forests and grasslands. Of the diverse species of
primates documented in the Lower Omo Valley, the forest and
woodland support guereza (Colobus guereza), blue monkey (Cer-
copithecus mitis), and De Brazza?s monkey (Cercopithecus neglectus),
whereas other transitional zones, as well as plains and adjacent
?ats, support Senegal galagos (Galago senegalensis), vervet mon-
keys (Chlorocebus aethiops), and anubis baboons (Papio anubis). The
number of primate species inhabiting the forest today indicates the
richness and lateral extent of the forest along the Omo River. Most
riparian forests in semiarid regions only have Chlorocebus aethiops
and occasionally Cercopithecus mitis or Colobus guereza.
Materials and methods
In the four ?eld seasons conducted between the years 1999 and
2003, fossil fauna was collected from 122 localities sampling allower Omo Valley (modi?ed after Carr, 1976).
manTable 1
Extant large mammals of the Omo Kibish region.
Z. Assefa et al. / Journal of Humembers of the Kibish Formation. Consistent with its short
depositional history, Member II is devoid of faunal samples, and is
missing from the stratigraphic sequence in many places across the
landscape, so faunal-bearing localities derive from Members I, III,
Family Subfamily/tribe Species and common names
Bovidae Alcelaphini Alcelaphus buselaphus (hartebeest)
Damaliscus korrigum (topi)
Cephalophini Sylvicapra grimmia (bush duiker)
Neotragini Madoqua guentheri
smithi (Gunther?s dik-dik)
Oreotragus oreotragus (klipspringer)
Ouerbia ourebi (oribi)
Reduncini Kobus ellipsiprymnus
defassa (waterbuck)
Redunca redunca (bohar reedbuck)
Redunca fulvorufula
(mountain reedbuck)
Antilopini Gazella granti (Grant?s gazelle)
Gazella thomsonii (Thomson?s gazelle)
Litocranius walleri (gerenuk)
Hippotragini Oryx beisa (beisa oryx)
Tragelaphini Tragelaphus strepsiceros (greater kudu)
Tragelaphus scriptus
decula (bushbuck)
Tragelaphus imberbis (lesser kudu)
Taurotragus oryx (common eland)
Bovini Syncerus caffer (African buffalo)
Hippopotamidae Hippopotamus amphibius
(river hippopotamus)
Suidae Phacochoerus aethiopicus
(desert warthog)
Potamochoerus porcus (river hog)
Giraf?dae Giraffa camelopardalis
reticula (giraffe)
Equidae Equus grevyi (Grevy?s zebra)
Equus burchellii
boehmi (Burchell?s zebra)
Rhinocerotidae Diceros bicornis (black rhinoceros)
Galagidae Galago senegalensis (Senegal galago)
Cercopithecidae Colobinae Colobus guereza (guereza)
Cercopithecinae Cercopithecus mitis (blue monkey)
Chlorocebus aethiops (vervet monkey)
Cercopithecus neglectus
(De Brazza?s monkey)
Papio Anubis
doguera (Anubis baboon)
Elephantidae Loxodonta africana (African elephant)
Procaviidae Heterohyrax brucei
(yellow spotted hyrax)
Procavia capensis (rock hyrax)
Leporidae Lepus habessinicus (Abyssinian hare)
Orycteropodidae Orycteropus afer (aardvark)
Canidae Lycaon pictus
(African wild dog)
Otocyon megalotis (bat-eared fox)
Canis adustus (side-striped jackal)
Canis aureus (golden jackal)
Canis mesomelas (black-backed jackal)
Viverridae Genetta genetta (common genet)
Viverra civettina (civet)
Herpestidae Helogale hirtula
(desert dwarf mongoose)
Helogale parvula (dwarf mongoose)
Herpestes ichneumon
(Egyptian mongoose)
Galerella sanguinea (slender mongoose)
Hyaenidae Hyaena hyaena (striped hyena)
Crocuta crocuta (spotted hyena)
Felidae Felinae Acinonyx jubatus (cheetah)
Felis silvestris (wild cat)
Leptailurus serval (serval)
Caracal caracal (caracal)
Pantherinae Panthera pardus (leopard)
Panthera leo (lion)and IV. Most of the fossils were collected from the surface, and
because of topographic relief in the area, there are many places
where several members overlie one another, creating ambiguity
about the assignment of some specimens to a speci?c member. This
is particularly problematic for specimens found at localities where
Member III unconformably overlies Member I. In such situations,
many of these fossils appear to have come from extensive aprons of
dark sediments at the base of Member III. Generally, we counted
specimens found in such contexts as deriving from Member III.
Sampling of the different members of the formation was not
uniform. Efforts concentrated on Member I because of the focus of
the project on hominins from that member. In addition, exposures
of Member I are of relatively low relief and are thus readily
accessible.
The collection procedure followed different strategies, yielding
differences in the completeness, identi?ablity, and taxonomic
category of each specimen. With bovids and equids, the focus was
mainly on crania, horn cores, mandibles, and isolated teeth. Post-
cranial bones were collected, but tarsals such as astragali and cal-
canei were especially targeted. Faunal remains of all non-ungulate
taxa were always collected due to their rarity. Only the diagnostic
dental remains and horn cores that were identi?ed to family or
more speci?c taxonomic ranks were included in the analyses
presented here. Illustrations and other information about the
Kibish mammalian fauna analyzed in this paper may be found at:
http://turkanabasin.org/research/omo-kibish.
Taxonomic identi?cation of most of the specimens, in particular
the isolated teeth, bene?ted from the use of a digital archive on
extant east African large mammals, developed by Assefa (http://
ripley.si.edu/hop/dentition/dentition.htm). Information calculated
from the fauna of the Kibish localities included number of identi-
?able specimens (NISP) andminimumnumber of individuals (MNI).
In quantifying the latter, we took specimens from each locality as
separate individuals. The sums of MNIs from localities associated
with each member provide the MNI value of speci?c taxa within
a given member. For example, we have a total of 13 NISP of Syncerus
caffer that were recovered from localities Kib-53, Kib-69, Kib-93,
Kib-141, Kib-149, Kib-150, Kib-152, and Kib-106. Except for locality
Kib-106, which has yielded six nonidentical specimens identi?ed to
Syncerus caffer, each of the other seven localities is represented by
single specimens. Lumping the former six specimens from Kib-106
as one, Syncerus caffer?s MNI for Member I is eight. This same
procedure was applied to all levels of taxonomic identi?cation,
which comprises species, uncertain species (cf.), genus, and tribal
levels.
The estimates of MNIs at the species level included only speci-
mens that were securely identi?ed. Specimens that were tenta-
tively identi?ed to a species (cf.) were added to the MNI only if
there were no specimens of more than a single species of the same
genus securely identi?ed to species. In conditions where we have
the representation of both securely and tentatively identi?ed
specimens, such as Kobus cf. K. ellipsiprymnus and Kobus ellipsi-
prymnus, only the latter counts was included in the MNI calcula-
tions. This procedure of calculating MNI follows mainly Bobe and
Eck?s (2001) approach, which was designed to control for the
possibility of counting one individual more than once.
For comparative analysis of the Kibish fauna with extant and
fossil specimens, it was necessary to modify the level of taxonomic
abundance to the level of genus or tribe. The MNIs for genus and
tribe were derived following the same approach cited above. The
sum of MNI counts for each species of the same genus provides the
MNI for that genus. Similarly, the tribal MNI is the sum of MNIs of
all genera from the same tribe.
The faunal data were placed in a data matrix to evaluate
Evolution 55 (2008) 501?512 503representation and abundance of major taxonomic groups within
and among the three members. Diversity indices were employed
to assess the diversity of fauna among members. Differences
and similarities in the patterns of faunal representations among
There is not a single fossil specimen unambiguously identi?ed to an
Z. Assefa et al. / Journal of Human504extinct species. Because of the small sample sizes, we doubt that
the faunal assemblages from any member are suf?ciently large to
sample all taxa that might have been present, especially rarer
species such as carnivores and tubulidentates. While the absence of
extinct taxa, in particular from Member I, suggests exclusive
representation of extant taxa in the later-middle Pleistocene
deposits of the Kibish Formation, we cannot be sure at this juncture
if this is the case. Representations of only extant fauna, however,
are documented from the middle Pleistocene fossil collections at
Lainyamok, Kenya (Potts and Deino, 1995), and thus this pattern
may indeed be the case at Kibish as well.
Fauna from the Kibish Formation includes taxa not often
recovered from east African fossil localities, such as Cephalophus
and Hylochoerus meinertzhageni. Both taxa are found in forest
habitats with dense vegetation cover, which is likely why they have
limited representation in the fossil record. The H. meinertzhageni
specimens (Fig. 2), in particular, are rare and listed in the Paleobi-
ology Database1 as occurring only at Kaiso (Uganda), Matupi Cave
(Democratic Republic of Congo), and Lamto Station (Cote d?ivore).
Uncertain Hylochoerus remains have also been reported from
Kanjera, Kenya (Plummer et al., 1994). All of the con?rmed fossil
Hylochoerus remains come from sites located within the tropical
rainforest belt or its fringes. In its modern-day distribution, thethe different members were further examined using chi-square
goodness-of-?t tests. Similarly, univariate, bivariate, and corre-
spondence analyses were employed for reconstructing past habitats
of the Omo Kibish area using extant faunal data gathered from
several modern game parks throughout Africa. We used corre-
spondence analysis to evaluate integrated patterning (in terms of
abundance) of various adaptations amonghabitats, and to determine
into which type of habitat(s) the faunal data from Kibish best ?t.
Correspondence analysis (CA) is an exploratory tool for comparing
Kibish faunal assemblages with modern faunal communities from
a variety of habitats. As with any multivariate technique, the algo-
rithm reduces large numbers of variables to two or three
dimensional vectors so that they can be graphically displayed to
better understand underlying patterns in the input data. Here, we
used the generic distribution of taxa in habitats for the ?rst CA; for
the second CA, we used the adaptations exhibited by those taxa. As
much of the variation in the data is placed on the ?rst axis, this
variation is expressed as a deviation from an expected chi-square
value and described as inertia away from this expectation. The
localities that plot close together in two-dimensional space are most
similar to each other in the pattern represented by the genera or
adaptations, whereas the genera or adaptations plot together based
on the habitats from which they derive.
Results
Fossil fauna
In general, fauna collected from all members of the Kibish
Formation are the same diverse species of large mammals, with
most taxa known throughout the later-middle and late Pleistocene
(Table 2). There are also some large mammalian taxa that show
sporadic representation among the three Kibish members. The
majority of large mammals represented in the Omo Kibish
assemblage are the same as the extant fauna in the study area.1 The data were downloaded from the Evolution of Terrestrial Ecosystems (ETE)
database on 15 June 2002 using the group name ??mammal?? and the following
parameters: middle and late Pleistocene, Africa.species has a preference for forests and forest mosaics. Occurrence
of this suid species with faunal remains from Kibish is consistent
with a closed, wet habitat.
Reliable identi?cation of two different equids based exclusively
on the distinct morphology of their dentitions is often problematic.
This is particularly so for isolated teeth, which constitute the
majority of the equid sample from Kibish. As noted above, the Omo
area is one of the few regions where Equus burchellii and E. grevyi
are sympatric, a distributional pattern that may extend deep into
the Quaternary. In addition, to morphologically distinguish the
Kibish equid species from one another, we measured the occlusal
length and breadth of 52 (20 Equus grevyi and 32 E. burchellii)
maxillary and mandibular dentitions housed in the National
Museum of Natural History (Smithsonian Institution) for compar-
ison. Differences in the occlusal length of most of the cheek teeth of
these equids were noted and these were used in conjunction with
distinctive morphological features to differentiate E. grevyi from E.
burchellii.
In general, the faunal samples from Kibish document consider-
able diversity of large mammals, with ungulates predominant
throughout the sequence. Primates and carnivores are poorly rep-
resented, and we do not believe that this is due to taphonomic
factors because of the recovery of even smaller mammals such as
Thryonomys, Hystrix, and Heterocephalus in the assemblages. Extant
and fossil faunas from the Kibish show considerable overlap, par-
ticularly among bovids. The exceptions are the lack of some bovid
taxa, such as Taurotragus, Litocranius, and Oreotragus in the fossil
record. Among nonbovid ungulates, there are a few remains of
Ceratotherium simum, the white rhino. Damaliscus cf. D. niro re-
mains from an unrecorded locality represent the only extinct large
mammal in the assemblage. Because of contextual uncertainty, this
specimen can neither be incorporated into the analysis nor taken as
evidence for the presence of an extinct taxon in the Kibish
assemblage.
Cephalophus and Hylochoerus meinertzhageni are not found in
the Kibish area today. The only primate specimen, a complete skull
of Papio cf. P. anubis, comes from the uppermost member of the
formation. The 1967 expedition to Kibish by the National Museums
of Kenya recovered two additional primate specimens, including
a partial skull identi?ed as Colobus sp. (S. Frost, personal commu-
nication). However, we are unsure of the exact provenience of this
specimen within the formation and so have not included it in
further analyses.
Table 2 summarizes the numbers of identi?ed specimens (NISP)
in each member. The sample from Member I (118) is considerably
larger than the samples from Members III (39) and IV (22). Overall,
the limited sample sizes from all members, particularlyMembers III
and IV, do not allow de?nitive conclusions regarding the absence of
taxa in the faunal assemblage. Despite noticeable differences in the
occurrence and abundance of fossil taxa among the threemembers,
the concentration of efforts on surveying Member I may partly
account for observed sample-size differences.
Throughout all members, bovids constitute more than 60% of
the faunal sample at each stratigraphic level. Among nonbovids,
equids and suids are well represented, whereas primates and car-
nivores make up only a small portion of the assemblage; these taxa
are therefore low in both diversity and abundance. Small sample
sizes, particularly in the upper twomembers, do not allow a reliable
estimate of differences or similarities in faunal abundances among
the three members.
In reviewing the presence and absence of different mammalian
taxa from each member, no conspicuous difference in overall pat-
tern of faunal representations was noted. Chi-square tests for
goodness-of-?t conducted on three species, selected for their large
Evolution 55 (2008) 501?512sample sizes, also show a lack of signi?cant differences among
members. Results from these tests are: Phacochoerus africanus, c2
manTable 2
List and abundance of faunal remains from the Kibish Formation
Family Subfamily/tribe Genus and species
Bovidae Alcelaphini Alcelaphini indet.
Alcelaphus sp.
cf. Damaliscus
Damaliscus cf. D. niro*
Cephalophini Cephalophini indet.
Cephalophus cf. C. weynsi
Neotragini Madoqua sp.
Madoqua cf. M. guentheri
Madoqua cf. M. kirkii
Reduncini Reduncini indet.
cf. Kobus
Kobus sp.
Kobus cf. K. ellipsiprymnus
Kobus ellipsiprymnus
Redunca sp.
Redunca cf. R. redunca
Z. Assefa et al. / Journal of Hu(2, n? 19)? 2, p> 0.05; Syncerus caffer, c2 (2, n? 20)? 1.3,
p> 0.05; andHippopotamus amphibius, c2 (2, n? 15)? 4.8, p> 0.05.
Notwithstanding the possible effects of small sample size and
a biased sampling strategy, the fauna from Member IV re?ects
a paucity of some bovid tribes (Fig. 3), such as Tragelaphini,
Cephalophini, Neotragini, and Antilopini, almost all of which
appear to be more abundant in Member I and, to some degree,
Member III. On the other hand, the abundances of Oryx cf. O. gazella
and Kobus ellipsiprymnus show moderate increases in Member IV.
Sample size is greater for Member III than for Member IV, and
the faunal diversity is also slightly higher. Despite differences in
abundance, bovid species representation in Member III is similar to
that in Member I. Member III lacks equids, but three species of suids
(Phacochoerus africanus, Hylochoerus meinertzhageni, and Pota-
mochoerus porcus) were documented only from this member.
Phacochoerus is the best-represented species of suid throughout the
sequence at Kibish.
For reasons cited earlier, the composition and abundance of
fauna from Member I is considerably higher than that of the two
overlying members. Cephalophus and Giraffa are known exclusively
Redunca redunca
Antilopini Antilopini indet.
Gazella sp.
Hippotragini Hippotragini indet.
Oryx cf. O. gazelle
Tragelaphini Tragelaphini indet.
Tragelaphus strepsiceros
Tragelaphus cf. T. strepsiceros
Tragelaphus cf. T. scriptus
Tragelaphus scriptus
Bovini Bovini indet.
Syncerus caffer
Syncerus cf. S. caffer
Aepycerotini cf. Aepyceros*
Hippopotamidae Hippopotamus amphibius
Suidae Phacochoerus africanus
Hylochoerus meinertzhageni
Potamochoerus porcus
Giraf?dae Giraffa camelopardalis
Equidae Equus sp.
Equus cf. E. grevyi
Equus cf. E. burchellii
Elephantidae Loxodonta africanus
Rhinoceratidae Ceratotherium simum*
Thryonomyidae Thyonomys swinderianus
Hystericidae Hystrix cristata
Bathyergidae Heterocephalus
Herpestidae Herpestinae Herpestes
Hyaenidae Hyaena hyaena
Cercopithecidae Cercopithecinae Papio cf. P. anubis
* Faunal samples from locations that could not be identi?ed to a speci?c member.Members
I III IV Unspeci?ed* (NISP)
5 (7) 1 (3)
1
1
1
1 (2)
1 (3)
1 1 7
1 1
3 (10)
1
1 (2) 1 1 2
1
2 1 (3) 3 (4) 3
3 3 2 5
2
1 1
Evolution 55 (2008) 501?512 505from Member I. In addition to Syncerus caffer and Phacochoerus
africanus, the faunal sample from Member I includes abundant
remains of Hippopotamus amphibius. In Members III and IV, S. caffer
and P. africanus show higher representations, respectively, in
comparison to Member I.
Paleoecological reconstruction
Researchers employ different methods of paleofauna-based
environmental reconstructions. These include the taxon-based
(phylogenetic) approach (Vrba, 1980; Dodd and Stanton, 1990), the
functional morphological (ecomorphological) approach (Plummer
and Bishop, 1994; Kappelman et al., 1997; DeGusta and Vrba, 2003,
2005), and amethod based on adaptation and community structure
(ecological diversity) (Andrews et al., 1979; Reed, 1997, 1998;
Mendoza et al., 2005). If we followed a taxonomic uniformitarian
principle to reconstructing the environments at Kibish, then the
presence of Cephalophus and Hylochoerus meinertzhageni would
suggest closed-habitat environments in Members I and III. Rather
than taxonomic uniformitarianism, however, we prefer an
1 1
1 1
1 1
1 1
1 1 3
1
1 1
1
6 (7) 1
5 2 (3)
7 3 2 5
9 (14) 5 (10) 6 (7) 13
1 1
2
9 (10) 3 3 (5) 9
9 (22) 6 4 (8) 6
2
1
3 1 3
4 (8)
1 (2) 1 4
2
1 1
4
3 1 1
1
1
1
1
1
ecological-diversity method because of its comprehensive com-
munity-based approach, which combines analysis of morphology
and ecological adaptation of each taxon to suggest overall
vegetation.
The ecological-diversity approach, as originally outlined by
Andrews et al. (1979), assigns to each mammalian taxon trophic,
locomotor (or substrate), and body-size attributes. Subsequent
work by Reed (1997, 1998) expanded this approach, with a focus on
the proportion of each adaptation within each habitat type (e.g.,
forests, closed woodlands, openwoodlands, bushlands, shrublands,
substrate adaptations, and was limited to mammals greater than
500 g in size. The locomotor and trophic categories include ?ve and
twelve types, respectively (see Reed, 1997, 1998). Species in extant
African mammal communities were assigned to substrate and
trophic categories based on observations or stomach contents, and
the percentage of species in each category of adaptation were
calculated for each habitat. Univariate and bivariate methods of
analysis were employed to differentiate vegetation types by the
proportion of trophic and substrate adaptations. This method
identi?es ranges of adaptation percentages that predict habitats
due to signi?cant differences among habitats in those particular
adaptations. These signi?cantly different adaptations can then be
implemented in reconstructing broad habitat types for fossil
assemblages. Taking substrate adaptations as an example, high
proportions of terrestrial mammals usually indicate more grass-
land-dominated vegetation, whereas high percentages of arboreal
mammals are consistent with forest-dominated vegetation.
We classi?ed the Kibish fossil fauna identi?ed either to species
or genus level with speci?c substrate and trophic adaptations
assigned to each of the extant large mammals. We incorporated
genera in our classi?cations only when there were no species of
mammal identi?ed in the member, and when the trophic and
substrate adaptations were identical across all extant members of
the genus (e.g., Giraffa). Table 3 shows the taxonomic percentages
of large mammals from all members of the Kibish Formation sorted
by their respective adaptations.
Figure 4 compares the faunal data from Kibish, divided by
member, to the rangeof trophic and substrate adaptations across the
sixmajor vegetation types of Reed (1997). Some adaptations, such as
arboreality and other locomotor and trophic variables, are excluded
from the analysis because there were not any taxa from Kibish
Fig. 2. Hylochoerus meinertzhageni (giant forest hog) mandible from Kibish.
Z. Assefa et al. / Journal of Human Evolution 55 (2008) 501?512506grasslands, and deserts). Reed?s method used only trophic andFig. 3. Relative representation of largeassigned to these speci?c categories. Adaptations showing discretemammals by stratigraphic layer.
manTable 3
Abundance (percentage) of large mammals categorized by locomotor and trophic
ecovariables
Taxon Member Substrate Trophic
I III IV
Alcelaphus 1 d d T G
Damaliscus d 1 d T G
Cephalophus cf. C. weynsi 1 d d T FL
Madoqua cf. M. guentheri d 1 d T B
Madoqua cf. M. kirkii 3 d d T FL
Kobus ellipsiprymnus 3 3 2 T FG
Redunca redunca 1 d d T FG
Oryx cf. O. gazelle 1 1 3 T G
Tragelaphus strepsiceros 1 1 d T B
Tragelaphus scriptus 5 2 d T B
Syncerus caffer 9 5 6 T MF
Z. Assefa et al. / Journal of Huseparation among habitats, as pointed out by Reed (1997, 1998), are
restricted to arboreal substrate and total frugivory (a trophic
strategy that combines the eating of fruit, fruit and insects, and fruit
and leaves). The box plot towards the center of the ?gure shows
discrete proportions of total frugivory in different vegetation zones
compared with analogous data sets from Kibish. There are discrete
differences in the abundance of total frugivory in different vegeta-
tion zones, with forests, different varieties of woodlands and bush-
lands, and grasslands showing high, medium, and low abundances,
respectively. The data from Kibish approach values from extant
grasslands and shrublands.
Other substrate and trophic adaptations also provide useful,
though less speci?c, indications of habitat. Fossorial mammals tend
to have higher representation in open rather than closed environ-
ments, and the fauna from Kibish lacks fossorial large mammals.
The exception is Member I, which has a proportion close to what is
expected for more closed habitats. On the other hand, most
Hippopotamus amphibius 7 2 1 AQ FG
Phacochoerus africanus 9 6 4 T G
Potamochoerus d 1 d T OM
Hylochoerus meinertzhageni d 2 d T MF
Giraffa camelopardalis 3 d 1 T B
Equus cf. E. grevyi 1 d 1 T G
Equus cf. E. burchellii 2 d d T G
Loxodonta africanus d 1 1 T MF
Thryonomys swinderianus 3 1 d T FG
Hystrix 1 d d F R
Herpestes d 1 d T CI
Hyaena d 1 d T CB
Papio cf. P. anubis d d 1 TA FL
Adaptations Member
I III IV
SUBSTRATE
A 0.0 0.0 0.0
AQ 13.7 6.9 5.0
F 2.0 0.0 0.0
T 84.3 93.1 90.0
TA 0.0 0.0 5.0
TROPHIC
B 17.6 13.8 5.0
CB 0.0 3.4 0.0
CI 0.0 3.4 0.0
FG 27.5 20.7 15.0
FL 7.84 0.0 5.0
G 27.45 27.6 40.0
MF 17.64 27.6 35.0
OM 0.0 3.4 0.0
R 2.0 0.0 0.0
TC 0.0 6.8 0.0
TF 7.84 0.0 5.0
Abbreviations are as follows: A? arboreal, AQ? aquatic, F? fossorial, T? terrestrial,
TA? terrestrial/arboreal, B? browser, CB?meat/bone, CI?meat/invertebrates,
FG? fresh grass, FL? fruit/leaves, G? grass, MF? leaves/grass, OM? omnivorous,
R? roots/bulbs, TC? total carnivory, TF? total frugivory.fossorial mammals are small- to medium-sized and therefore may
be missing from our collection due to taphonomic bias. The other
important substrate adaptation, aquatic, is obviously found wher-
ever there is a river or lake, irrespective of speci?c terrestrial
habitat. Therefore, the proportion of aquatic mammals in varying
types of habitat shows considerable variation and overlap. Fauna
from different members of the Kibish Formation include aquatic
mammals, albeit in variable proportions, suggesting a type of
habitat close to water. Likewise, the high proportion of fresh-grass
grazers is closely associated with edaphic grasslands in lake mar-
gins, ?oodplains, river banks, and deltas. The proportion of fresh-
grass grazers from all members of the Kibish Formation is high,
suggesting that edaphic grasslands associated with overbank ?ood
deposits predominated.
Forest habitats are separated from other habitats because of the
complete absence of grazers andmeat/bone-feedingmammals, and
relatively small abundances of terrestrial mammals. Excluding
Member III, the Kibish faunal data represent a virtual lack of meat/
bone-feeding carnivores, such as hyaenids, showing a pattern
consistent with forest-dominated habitat. Conversely, the repre-
sentation of grazers and terrestrial mammals at Kibish is more
closely aligned with what is documented from more open settings
of grassland environments. As all meat/bone consumers are rare
taxa in the fossil record, and the Kibish sample sizes are small, it is
possible that these animals existed but are not represented in our
fossil sample.
An interesting picture of Kibish habitats is evident in bivariate
plots of two adaptations, such as total frugivory or fresh-grass
grazing against total terrestrial substrate use (all substrate adap-
tations except arboreality) and fresh-grass grazing against aquatic
substrate use. The percentages of fresh-grass grazing and aquatic
locomotion (Fig. 5), as outlined by Reed (1998), separate edaphic
grasslands from other vegetation types. Areas of abundant edaphic
grasslands, ?oodplains, wetlands, and swamps are positioned in
areas outside of the dotted box in Fig. 5. All data points derived
from the Kibish fauna fall in areas well outside the dotted box. As
stated above, there are no arboreal mammals documented securely
from any speci?c member of the Kibish Formation. Therefore, total
percentage of terrestrial mammals is 100% throughout the
succession (although we know that Colobus sp. from Kibish was
likely arboreal, we do not know its provenance). Despite high
percentages of terrestrial mammals in the Kibish fauna, the degree
of frugivore representation from all members, particularly Member
I (7.84%) and Member IV (5 %), suggests a grassland-dominated
habitat (Fig. 6). The percentage of fresh-grass grazers from all
members of Kibish is also high (ranging from 27.5% in Member I to
15.0% in Member IV), whereas the percentage of aquatic mammals
is generally lower (Member I 13.7%; Member III 6.9%; Member IV
5%). All members of the Kibish Formation feature habitats that in-
clude rivers, wetlands, and edaphic grasslands (Fig. 7). This is
consistent with the earlier observation in the plot of percentage of
fresh-grass grazing vs. aquatic locomotion.
In general, the bivariate plots comparing the fossil faunal
communities from Kibish to extant faunal communities indicate
mosaic habitats such as riparian forests and closed woodlands in
combination with grasslands of mostly edaphic origin. Based on
Reed?s (1997) analysis of numerous Plio-Pleistocene faunal locali-
ties from southern and eastern Africa, the faunal communities
from Kibish show close resemblance to communities from Krom-
draai A (South Africa), the Okote and KBS Members of the Koobi
Fora Formation (Kenya), and Swartkrans (Members 2 and 3; South
Africa). In all of these localities, total percentage of terrestrial
mammals is considerably high, while the percentages of frugivores
and fresh-grass grazers are moderate. The percentages of fresh-
Evolution 55 (2008) 501?512 507grass grazers from Kibish are often higher than those observed
at the other Plio-Pleistocene localities, perhaps indicating more
0
2
4
6
8
10
12
14
0
10
20
30
40
0
10
20
30
40
50
0
4
8
12
16
20
0
4
8
12
16
20
Fo
re
st
C
W
 /T
ra
ns
.
Bu
sh
la
nd
s
Sh
ru
bl
an
d
G
ra
ss
la
nd
Ki
b 
I
Ki
b 
III
Ki
b 
IV
0
20
40
60
80
100
T
0
4
8
12
16
20
AQ
0
5
10
15
20
25
30
35
F
0
5
10
15
20
25
30
FG
0
4
8
12
16
20
24
28
 TA
CI
0
10
20
30
40
50
G TC
R
0
20
40
60
Ki
b 
I
Ki
b 
III
Ki
b 
IV
TF CB
O
pe
n 
w
oo
d
Fo
re
st
C
W
 /T
ra
ns
.
Ki
b 
IV
Sh
ru
bl
an
d
Bu
sh
la
nd
s
O
pe
n 
w
oo
d
G
ra
ss
la
nd
Ki
b 
I
Ki
b 
III
Fo
re
st
C
W
 /T
ra
ns
.
Bu
sh
la
nd
s
O
pe
n 
w
oo
d
Sh
ru
bl
an
d
G
ra
ss
la
nd
Ki
b 
I
Fo
re
st
C
W
 /T
ra
ns
.
Bu
sh
la
nd
s
O
pe
n 
w
oo
d
Sh
ru
bl
an
d
G
ra
ss
la
nd
0
4
8
12
16
20
B
0
10
20
30
40
MF
0
4
8
12
16
OM
FL
Ki
b 
III
Ki
b 
IV
Substrate Trophic
Fig. 4. Box plot comparing percentage of mammals exhibiting varying locomotor and trophic adaptations classi?ed by vegetation types. On right side of each plot, the relative
percentages of mammals from Kibish classi?ed according to each adaptation are provided. Abbreviations are as in Table 3.
Kibish I
Kibish III
Kibish IV
0 2 4 6 8 10 12 14 16 18
Percentage of aquatic locomotion
0
5
10
15
20
25
30
P
e
r
c
e
n
t
a
g
e
 
o
f
 
f
r
e
s
h
 
g
r
a
s
s
 
g
r
a
z
i
n
g
Forest
Closed woodland/transit
Bushlands
Open woodland
Shrubland
grassland
Kibish
Fig. 5. A bivariate plot showing the relation between percentage of aquatic locomotion and fresh-grass grazers from modern communities sampling varying vegetation types,
including faunal samples from Kibish. According to Reed (1998), localities outside the dotted line (10% on both axes) represent edaphic grasslands, wetlands, or swamps. All
members from Kibish fall outside of the dotted range.
Z. Assefa et al. / Journal of Human Evolution 55 (2008) 501?512508
60 65 70 75 80 85 90 95 100
Percentage of total terrestrial locomotion
5
15
25
35
45
55
P
e
r
c
e
n
t
a
g
e
 
o
f
 
t
o
t
a
l
 
f
r
u
g
i
v
o
r
y
Forest
Closed woodland/ Transit
Bushlands
Open Woodland
Shrubland
Grassland
Kibish
Kib IV
KibI 
Kib III
Fig. 6. A bivariate plot showing the relation between percentage of total terrestrial mammals and total frugivorous mammals from modern communities sampling varying veg-
Z. Assefa et al. / Journal of Human Evolution 55 (2008) 501?512 509widespread edaphic grassland. The small percentage of aquatic
mammals recovered from all members is consistent with what is
observed from Okote, which has been characterized as having
edaphic grasslands of ??shallow wetlands or marshy swamps??
(Reed, 1997).
Further assessment of the Kibish faunal data in reference to
extant communities provides a clearer picture of the above argu-
ment. Results of the correspondence analyses are illustrated in
Fig. 8 (CA with all fauna in taxonomic groups) and Fig. 9 (CA only
with fauna exhibiting diagnostic adaptations). The correspondence
analysis illustrated in Figure 8 includes extant faunal communities
(as row data) contrasted with the percentages of different taxo-
nomic groups of mammals (as column data). Extant faunal data are
from Reed (1998: Appendix 3). Axis 1 and axis 2 represent only
etation types, including faunal samples from Kibish.28.21% and 25.72% of the inertia, respectively. Such low axis values
suggest that the space de?ned by the genera is not signi?cant in the
60 65 70 75 80
Percentage of total 
0
5
10
15
20
25
30
35
40
P
e
r
c
e
n
t
a
g
e
 
o
f
 
f
r
e
s
h
 
g
r
a
s
s
 
g
r
a
z
i
n
g
Forest
Closed woodland/ Transit
Bushlands
Open Woodland
Shrubland
Grassland
Kibish
Fig. 7. A bivariate plot showing the relation between percentage of total terrestrial mammals
including faunal samples from Kibish.distribution of habitats, despite the moderate clustering by habitat
of the modern localities.
None of the Kibish faunal localities show close association with
extant localities of a particular habitat type. This is likely the result
of the limited number of taxa recovered from thesemembers, and it
does not suggest a unique habitat for the ancient Kibish.
Conversely, in Fig. 9, the taxa are represented as community
percentages of signi?cantly different trophic and substrate adap-
tations, which accurately separate extant communities into habi-
tats (Reed 1997, 1998). These adaptations include terrestrial,
arboreal, aquatic, frugivorous, and fresh-grass-grazing mammals.
The faunal data from Kibish were included to compare with the
extant communities.
The total inertia represented on these two axes is 95.06%
(80.54% and 14.52%, respectively) and thus accounts for almost all
of the variation among the extant communities. The Kibish faunal
85 90 95 100
terrestrial locomotion
Kib IV
KibI 
Kib III
and fresh-grass grazers from modern communities sampling varying vegetation types,
36
Dimension 1; Eigenvalue: .29155
(28.21% of Inertia)
D
i
m
e
n
s
i
o
n
 
2
;
 
E
i
g
e
n
v
a
l
u
e
:
 
.
2
6
5
8
0
(
2
5
.
7
2
%
 
o
f
 
I
n
e
r
t
i
a
)
12
3
4
5
6 7
8
910
11
12
13
14
15
16
1718 19
20
21
22
23
24
25 26
27
28
29
30
31
32
33
34
35
37
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
Kibish I
Kibish III
Kibish IV
Fauna
Forest
CW/bushland Transit
Bushland
Woodland
Shrubland
Grassland
Kibish
n ty
atio
Z. Assefa et al. / Journal of Human Evolution 55 (2008) 501?512510assemblages are now distributed with extant communities of
grassland and shrubland, particularly those habitats possessing
abundant water and edaphic grasslands, such as the Kafue Flats
(central Zambia) and the Linyanti Swamp, with surrounding riv-
erine woodland located just north of Okavango Delta in Botswana.
Aquatic mammals and fresh-grass grazers are more abundant in
these edaphic grassland and swampy regions. Therefore, the Kibish
Fig. 8. Correspondence analysis plotting extant communities sampling varying vegetatio
The Kibish faunal assemblages are included to show closely associated type(s) of vegetfaunal assemblages also possess mammals exhibiting aquatic sub-
strate use and fresh-grass grazing. These results support the view
presented above that the Kibish faunal remains were deposited in
Adaptation
Forest
CW/bushland transit
Bushland
Woodland
Shrubland
Grassland
Kibish
Lainyamok
Dimension 1; Eigenvalue: .18283 (80.54% of In
Di
m
e
n
s
i
o
n
 
2; 
Ei
g
e
n
va
l
u
e
: 
.03297 
(14.52% 
o
f
 
I
n
e
r
t
i
a
)
Arboreal (S)
F1
F2
F3
F4
F7
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
-1.4 -1.2 -1.0 -0.8 -0.6 -0.4
Total frugivore (T)
F5 F6
Fig. 9. Correspondence analysis plotting extant communities sampling varying vegetation
following diagnostic locomotor and trophic adaptive strategies (column): arboreal, aquatic,
faunal assemblages are included to show the closely associated type(s) of vegetation identan environment where edaphic grassland, shrubland, and riparian
forests were dominant on a mosaic landscape. While there is no
clear evidence of forest, as arboreal mammals are poorly repre-
sented in the assemblage, some of the rarer mammals (i.e., Ceph-
alophus and Hylochoerus) recovered from the assemblage indicate
a riparian forest.
As noted above, the Kibish fauna resembles that of the middle
pes (row) and percentage of different genera of large mammals they support (column).
n and the related taxa of large mammals. Refer to Appendix 1 for abbreviations.Pleistocene faunal assemblage from Lainyamok, Kenya, which
also is composed entirely of extant species of large mammals, and
suggests a close association with more open habitats such as
ertia)
Fresh grazer (T)
Aquatic (S)
CW8 CW9
B12
B13
B14
W16
W18
W19
W20
S23
S26
G28
G29
G30
Kibish I
Kibish II
-0.2 0.0 0.2 0.4 0.6 0.8
Total terrestrial (S)
CW11
W17
S25
S21
S24
Kibish IV
CW10 B15 S22
S27
Lainyamo k
types (row) and percentages of the different genera of large mammals exhibiting the
and total terrestrial locomotion, and total frugivory and fresh-grass grazing. The Kibish
i?ed by the diagnostic adaptations. Refer to Appendix 1 for abbreviations.
over brief periods during wet phases of the late-middle and late
Pleistocene (McDougall et al., 2005, 2008). Accordingly, faunal
References
Andrews, P., Lord, J.M., Nesbitevans, E.M., 1979. Patterns of ecological diversity in
fossil and modern mammalian faunas. Biol. J. Linn. Soc. 11, 177?203.
Bauer, I.E., McMorrow, J., Yalden, D.W., 1994. The historic ranges of three equid
species in north-east Africa: a quantitative comparison of environmental
tolerances. J. Biogeogr. 21, 169?182.
Bobe, R., Eck, G.G., 2001. Responses of African bovids to Pliocene climatic change.
Paleobiology 27 (Suppl. 2), 1?47.
Brown, F.H., Fuller, C.R., 2008. Stratigraphy and tephra of the Kibish Formation,
southwestern Ethiopia. J. Hum. Evol. 55, 363?403.
Butzer, K.W., 1971. Recent History of an Ethiopian Delta. The University of Chicago,
Chicago.
Carr, C.J., 1976. Plant ecological variation and pattern in the Lower Omo Basin. In:
Coppens, Y., Howell, F.C., Isaac, G.L., Leakey, R.E.F. (Eds.), Earliest Man and
Environments in the Lake Rudolf Basin. University of Chicago Press, Chicago, pp.
433?467.
Carr, C.J., 1998. Patterns of vegetation along the Omo River in southwest Ethiopia.
Plant Ecol. 135, 135?163.
11 Felinae CW11 Lake Merwu
12 Giraf?dae B12 West Lunga
National Park
13 Herpestidae B13 Serengeti National Park
14 Hippotragini B14 Serengeti Bush
15 Hippopotamidae B15 Rukwa Valley
16 Hyaenidae W16 Kafue National Park
17 Hyracoidea W17 Kruger National Park
18 Hystricidae W18 Linyanti Swamp
19 Insectivora W19 SS Woodland
20 Lagomorpha W20 Sudan
21 Lorisidae S21 Southwest Arid
22 Manidae S22 Kalahari TV
23 Cricetinae S23 Sahel
24 Mustelidae S24 Chobe
25 Neotragini S25 Amboseli National Park
26 Papionini S26 Tarangire National Park
27 Pedetidae S27 Okavango
28 Peleinae G28 Kafue Flats
29 Pongidae G29 Serengeti Plains
30 Reduncini G30 Southern savanna grassland
31 Rhinocerotidae
32 Suidae
33 Thryonomidae
34 Tragelaphini
35 Tragulidae
36 Tubulidentata
37 Viveridae
mansamples from different layers of the Kibish Formation represent
only episodic records sampling the wet interglacial phases of the
late-middle and late Pleistocene. Predominance of moist grass-
lands may also explain the lack of the large-bodied and hyp-
sodont extinct taxa, such as Pelorovis, in the Kibish faunal
assemblage. Dental and other morphological (locomotor, shape
and size of horns) characteristics, paleoenvironmental associa-
tions, as well as the distribution of both large- and small-bodied
extinct African late Pleistocene ungulates are consistent with dry-
forage specialization and adaptation to open environments
(Klein, 1984; Marean, 1992). Such groups of ungulates must have
been extremely rare or totally absent within the deltaic
environments of the Kibish area, where moist vegetation types
such as edaphic grassland were predominant.
Conclusion
The paleoecological reconstruction of the Omo Kibish deposits
suggests a habitat that is very similar to that seen in the Omo Valley
today, although moister and probably less arid further away from
the river. Cephalophus and Hylochoerus live in forest environments
where the vegetation cover is often dense and there is a regular
supply of water. This conclusion is also indicated by the abundance
of fossil ?sh (Trapani, 2008) and water birds (Louchart, 2008)
recovered from the Kibish deposits. Mammalian taxa such as
Thryonomys, diverse species of tragelaphines, and reduncines also
suggest a more closed habitat, one with tall grasses and abundant
standing water. There was also likely a variety of dense, light, and
intermediate covers depending on soils and distance from the an-
cient Omo River. The Homo sapiens fossils from Kibish were also
recovered from this mosaic habitat. It is evident that these homi-
nins existed in landscapes extremely similar to extant habitats and
with a fauna that still exists today in Africa. The fact that the Kibish
fauna lacks carnivores and nonhuman primates is not an indication
that these taxa (and other rarer and smaller mammals) were not
present, but indicates the lack of mammals that were rare on the
landscape in the Kibish sediment. Perhaps further excavation in
additional areas of the Kibish Formation would provide a larger
faunal community with which to analyze hominin interactions.
Acknowledgements
We are grateful to the Authority for Research and Conservation
of Cultural Heritage (ARCCH) and the Ethiopian National Museum
for granting us permits to study the deposits of the Omo Kibish
area. Regional of?ces of the Ministry of Tourism and Culture
provided all necessary assistance for the project. John Fleagle is
responsible for the collection of the bulk of the faunal samples, and
Denne Reed assisted with identi?cation of some of the small
mammals. Essayas G/Mariam assisted with the curation. Special
thanks go to Curtis Marean for inspiring the revisit of the Kibishshrublands and grasslands. In addition, the faunal remains
from Lainyamok and Kibish share more similarities in their
open-habitat-dominated faunal composition. However, differ-
ences between the two assemblages are also apparent, with the
Kenyan assemblage showing more association with open-dry
vegetation zones, such as Chobe and Serengeti plains, as opposed
to the open-wet habitats that the Kibish faunal remains
document. This is consistent with the predominantly deltaic en-
vironments that predominated in the Kibish area during the later
Pleistocene. Sediments of the Kibish Formation were deposited
Z. Assefa et al. / Journal of HuFormation 30 years after the initial exploration by Richard Leakey.
Assefa thanks the Smithsonian?s Human Origins and ArchaeobiologyCerling, T.E., Harris, J.M., Passey, B.H., 2003. Diets of East African Bovidae based on
stable isotope analysis. J. Mammal. 84, 456?470.
DeGusta, D., Vrba, E., 2003. A method for inferring paleohabitats from the func-
tional morphology of bovid astragali. J. Archaeol. Sci. 30, 1009?1022.
DeGusta, D., Vrba, E., 2005. Methods for inferring paleohabitats from discrete traitsPrograms for providing support during the analysis and writing of
this paper. This paper was improved greatly by the comments and
suggestions of Rene? Bobe and Frank H. Brown.
Appendix 1. Keys to abbreviations in Figs. 8 and 9
Figure 8 Figure 9
1 Aepycerotini F1 East of Niger
2 Alcelaphini F2 Congo Basin
3 Antilopini F3 Knysna
4 Bovini F4 Kilimanjaro
5 Canidae F5 West of Niger
6 Cephalophini F6 East of Cross
7 Cercopithecini F7 Makakou
8 Colobinae CW8 Rwenzori National Park
9 Elephantidae CW9 Guinea Woodland
10 Equidae CW10 Natal Woodland
Evolution 55 (2008) 501?512 511of the bovid postcranial skeleton. J. Archaeol. Sci. 32, 1115?1123.
Dodd, J.R., Stanton, R.J., 1990. Paleoecology: Concepts and Applications. Wiley, New
York.
Gagnon, M., Chew, A.E., 2000. Dietary preference in extant African Bovidae. J.
Mammal. 81, 490?511.
Kappelman, J., Plummer, T., Bishop, L., Duncan, A., Appleton, S., 1997. Bovids as
indicators of Plio-Pleistocene paleoenvironments in East Africa. J. Hum. Evol. 32,
229?256.
Klein, R.G., 1984. Mammalian extinctions and Stone Age people in Africa. In:
Martin, P.S., Klein, R.G. (Eds.), Quaternary Extinctions: A Prehistoric Revolution.
University of Arizona Press, Tucson, pp. 553?573.
Louchart, A., 2008. Fossil birds of Kibish Formation. J. Hum. Evol. 55, 513?520.
Marean, C.W., 1992. Implications of late Quaternary mammalian fauna from Luken-
ya Hill (south-central Kenya) for paleoenvironmental change and faunal
extinctions. Quatern. Res. 37, 239?255.
McDougall, I., Brown, F.H., Fleagle, J.G., 2005. Stratigraphic placement and age of
modern humans from Kibish, Ethiopia. Nature 433, 733?736.
McDougall, I., Brown, F.H., Fleagle, J.G., 2008. Sapropels and the age of the hominins
Omo I and II, Kibish Formation, Ethiopia. J. Hum. Evol. 55, 409?420.
Mendoza, M., Janis, C.M., Palmqvist, P., 2005. Ecological patterns in the trophic size
structure of large mammal communities: a ??taxon-free?? characterization. Evol.
Ecol. Res. 7, 505?530.
Plummer, T.W., Bishop, L.C., 1994. Hominid paleoecology at Olduvai Gorge, Tanzania
as indicated by antelope remains. J. Hum. Evol. 27, 47?75.
Plummer, T.W., Kinyua, A.M., Potts, R., 1994. Provenancing of hominid and mam-
malian fossils from Kanjera, Kenya, using EDXRF. J. Archaeol. Sci. 21, 553?563.
Potts, R., Deino, A., 1995. Mid-Pleistocene change in large mammal faunas of East
Africa. Quatern. Res. 43, 106?113.
Reed, K.E., 1997. Early hominid evolution and ecological change through the African
Plio-Pleistocene. J. Hum. Evol. 32, 289?322.
Reed, K.E., 1998. Using large mammal communities to examine ecological and
taxonomic structure and predict vegetation in extant and extinct assemblages.
Paleobiology 24, 384?408.
Trapani, J., 2008. Quaternary fossil ?sh from the Kibish Formation, Omo Valley,
Ethiopia. J. Hum. Evol. 55, 521?530.
Vrba, E.S., 1980. The signi?cance of bovid remains as indicators of environment and
predation patterns. In: Behrensmeyer, A.K., Hill, A.P. (Eds.), Fossils in the Mak-
ing: Vertebrate Taphonomy and Paleoecology. The University of Chicago Press,
Chicago, pp. 247?271.
Yalden, D.W., Largen, M.J., Kock, D., 1984. Catalogue of the mammals of Ethiopia. 5.
Artiodactyla. Monit. Zool. Ital. 19 (Suppl. 4), 67?221.
Z. Assefa et al. / Journal of Human Evolution 55 (2008) 501?512512