Color profile: Disabled

Composite Default screen

Systematics and Phylogeny

of the Horse

C.P.
C.P. Groves
1Systematics Groves
and
and 1 and O.A. Ryder2
Phylogeny
O.A. Ryder
1Department of Biological Anthropology, Australian National

University, Canberra, ACT 0200, Australia; 2Center for Reproduction

of Endangered Species, Zoological Society of San Diego, PO Box 551,
San Diego, CA 92112–0551, USA

The Order Perissodactyla 1

The Family Equidae 2
Genetics of the Equidae 4
Generic Limits 5
Early History of the Genus Equus 6
The species question 7
The subspecies question 8
Nomenclature of Domestic Animals 9
A Taxonomy of the Genus Equus 9
Subgenus Equus: horses 11
Subgenus Asinus: ass, onager and kiang 13
Subgenus Hippotigris: zebras 18
References 22

The Order Perissodactyla

The order Perissodactyla, known as odd-toed hoofed mammals, are character-
ized by the relative enlargement of digit III on each extremity. Other charac-
teristics include, in the skull, the persistence of the tuber maxillare, and, in the
dentition, the basic π-shape of the occlusal pattern of the maxillary molars
(modified in living Equidae, but extremely useful for recognizing primitive
fossil perissodactyls).
The order contains three living groups, the horses, tapirs and rhinos,
and two major extinct groups, the brontotheres (or titanotheres) and the
ancylopods (or chalicotheres). The brontotheres arose in the Early Eocene and
survived into the Oligocene but no longer; the ancyclopods arose in the late
Eocene and survived into the Pleistocene. Tapirs and rhinos differentiated in
the Late Eocene.
©CAB International 2000. The Genetics of the Horse (eds A.T. Bowling and A. Ruvinsky) 1

1
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:22 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

2 C.P. Groves and O.A. Ryder

The following outline classification of the Perissodactyla is based on

Prothero and Schoch (1989), but excludes the hyraxes which (contrary to the
classification of Prothero and Schoch) are not perissodactyls but are related to
the elephants and sirenians.
Order Perissodactyla
Suborder Titanotheriomorpha – brontotheres
Suborder Hippomorpha
Superfamily Pachynolophoidea
Superfamily Equioidea
Family Palaeotheriidae
Family Equidae
Suborder Moropomorpha
Parvorder Ancylopoda – chalicotheres
Parvorder Ceratomorpha – tapirs and rhinos
The Pachynolophoidea, restricted to the Eocene, were small perissodactyls
with simple teeth, but with some development of the complexity of the cheek
teeth which characterizes equids; in particular, the premolars were somewhat
molarized (meaning that they had come to resemble the molars to some
degree). The Palaeotheriidae, from the Late Eocene and Early Oligocene, were
larger (some of them tapir-sized) and longer necked, but still extremely primi-
tive; they include Palaeotherium, which was described by Georges Cuvier in
1804, and was one of the very first fossil mammals to be discovered.

The Family Equidae

The Equidae are known from the earliest Eocene. The famous Hyracotherium,
described by Richard Owen in 1841, is very primitive, and the various species
which have been included within it, and which differ only very slightly from
each other, are now regarded as probable ancestors to the Pachynolophoidea,
the Palaeotheriidae and probably the Moropomorpha as well as the Equidae,
so many authors now place them on cladistic grounds into different genera.
The species described by Owen, Hipparion leporinum, is thought by Hooker
(1984) to be the ancestor of the Palaeotheriidae. The ancestor of the Equidae,
according to Hooker, is the species formerly known as Hyracotherium
cuniculum, which he places in a new genus Cymbalophus.
The general outline of the evolution of the Equidae is summarized by
Evander (1989). After Cymbalophus of the earliest Eocene (54 million years
ago (mya)) of Europe, the line moves to North America, and goes more or less
straight through a series of genera which (except for the last two, which do
overlap) simply mark points on the tree which are represented by good
fossil material: Orohippus (Early Eocene, 50–47 mya), Epihippus (Middle and
Late Eocene, 47–40 mya), Mesohippus (Late Eocene and Early Oligocene,
40–30 mya), and Miohippus (Latest Eocene and Oligocene, 37–25 mya). The
line leading to modern horses had thus gone through more than half of its

2
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:23 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 3

evolutionary history with hardly any branching until the latest stages. The
narrow basicranium so characteristic of modern horses was absent in the
primitive stages but had been developed by Mesohippus, and the dentition of
each is advanced over its predecessor(s): more complex, with more cusp
development, but still without any uniting of the cusps into ridges.
The line then, at the beginning of the Oligocene some 25 mya branched
into the Anchitheriinae and the Equinae. Both these lines were advanced
in having molar ridges, unlike their predecessors; the Anchitheriinae, which
lasted until about 12 mya in North America and 7 mya in Asia, lost some
features of the molars and the foot skeleton which the Equinae retained, while
the Equinae developed a close-packed foot skeleton suitable for running in
open grasslands.
The line of the Equinae continued through to the Middle Miocene (15
mya) via Kalobatippus, Archaeohippus and Parahippus, with increasingly
complex molars. From Archaeohippus on, the skull developed a post-orbital
bar, a complete strut of bone behind the orbit, separating it from the temporal
fossa. From the later species of Parahippus on, the crowns of the molar and
premolars had become high, covered with cementum, and suitable for shear-
ing silica-rich grasses, and the radius and ulna had become fused. The line
then, about 15 mya, split into three branches: the Protohippini, Hipparionini
and Equini, though Hulbert (1989) considers that the Protohippini are actually
a composite, made up of stem forms of the other two tribes and of their
common ancestors, and includes Protohippus itself in the Equini; all these
late groups are difficult to work out, and there was a good deal of parallel
evolution in such features as large size, development of a pre-orbital fossa and
retraction of the nasal notch. However, the Equini were the only horses to
reduce their toes to one on each foot (with the laterals, represented by
metapodials II and IV, retained as ‘splint bones’), whereas the Hipparionini
developed their lateral toes into support digits, perhaps for marshy country.
The Hipparionini, the last of the three-toed horses, lived in North America until
the beginning of the Pliocene, 5 mya, but survived in the Old World until
about 1 mya, disappearing last from Africa.
Evander’s (1989) classification of the family is (abbreviated and slightly
modified) as follows:
Family Equidae
Cymbalophus
Orohippus
Epihippus
Mesohippus
Miohippus
Subfamily Anchitheriinae
Subfamily Equinae
Kalobatippus
Archaeohippus
Parahippus

3
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:23 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

4 C.P. Groves and O.A. Ryder

Tribe Protohippini
Tribe Hipparionini
Tribe Equini
Dinohippus
Hippidion
Onohippidion
Astrohippus
Pliohippus
Equus
In this classification, the five unranked genera coordinate with the two
subfamilies, and the three coordinate with the three tribes of Equinae, are
given the status of Plesion: a category which means that they are of limited
diversity (one or a few species in each), and primitive for that group (and very
likely ancestral to the remainder).
The genus Dinohippus, which lived in North America between about 8
and 5 mya, is the stem genus of Equus, and Hulbert (1989) even includes it in
Equus. It emerged gradually from Pliohippus, which lived in North America
from about 14.5 to 6 mya; Astrohippus lived in the same region from about 6 to
4.5 mya. Hippidion and Onohippidion were large single-toed horses that lived
in South America from the Early Pliocene until the end of the Pleistocene; their
earliest species lived in North America about 5 mya.

Genetics of the Equidae

Studies of chromosomes and DNA have provided a rich source of information
for interpretation of morphological and palaeontological data (George and
Ryder, 1986; Ryder et al., 1978; Oakenfull et al., unpublished observations).
The phylogenetic analysis of DNA sequence data sheds new light on the
systematics and taxonomy of Equus and is highly relevant for conservation
action plans for equid taxa (Oakenfull, equid action plan).
The first DNA-based studies of Equus involved the analysis of mitochon-
drial DNA restriction maps and required purification of mitochondrial DNA
(George and Ryder, 1986). With the advent of the polymerase chain reaction
(PCR), DNA sequence data have become the accepted standard in studies of
molecular evolution; PCR also allows a wide variety of samples previously
unusable for genetic analysis to be utilized in genetic studies. The complete
16,660 nucleotide sequence of a domestic horse mitochondrial DNA has been
published (Xu and Arnason, 1994).
The first extinct organism to have its DNA cloned was the quagga, Equus
(burchelli) quagga, an event of technical wizardry made even more remark-
able for its accomplishment before PCR was invented (Higuchi et al.,
1984). Samples of dried tissue, a residue of a roughshod taxidermy, provided
sufficient DNA for the construction of a library that included quagga DNA
sequences.

4
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:24 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 5

Today, bone and tissue fragments, dried blood and dung samples provide
routine sources of DNA for comparative genetic studies. Despite these
advances, it is still the case that DNA sequence data are lacking for some equid
populations and named taxa that confound the assembly of a clear picture of
the systematic relationships of extant and recently extinct taxa. The Somali
wild ass, Syrian, Indian and Mongolian wild asses, and the kiang have yet to be
the subject of reports utilizing molecular methods on wild populations.
Although most equid taxa have been the subject of at least preliminary DNA
sequencing studies, the amount of data is still rather small, may not include
nuclear loci and may incorporate limited sampling of the potential variability
within extant wild populations. The investigation of molecular genetics of
captive and wild equids is an area of great current interest. Thus, we can
anticipate that new findings will soon become available (Oakenfull et al.,
unpublished observations) and that as more intensive analysis incorporating
additional loci and more extensive sampling of extant populations is carried
out, a more refined picture of evolutionary relationships and the resultant
taxonomy will emerge.
The most definitive genetic data pertinent to equid evolution collected
to date involve sequence analysis of mitochondrial DNA. The region of the
circular mitochondrial DNA at which strand displacement for the initiation of
DNA replication takes place (the d-loop) is thought to be the most rapidly
evolving portion of the approximately 16,500 bp molecule. Accordingly, this
region can identify recent divergences due to mutations. Other portions of the
mitochondrial DNA, such as the 12S region and the cytochrome b gene,
accumulate mutations at a slower rate and there is a proportionally smaller
chance that the same sequence of nucleotides is present as a result of two
mutations (a ‘forward’ mutation and a ‘back’ mutation). For this reason,
evidence of the divergence of lineages at the base of the phylogenetic tree
may be derived less ambiguously from more slowly evolving sequences.

Generic Limits
While most specialists have been content to include all living equids in a single
genus, Equus, from time to time different authors have proposed to set aside
one or more species into separate genera, on the general grounds that they
were ‘different enough’. Part of the philosophy was no doubt that horses,
asses, onagers and zebras are all the living species that we have in the family
Equidae, and there is sufficient ‘taxonomic room’ for several genera. The wish
to divide up the genus in this way persists in the modern era: Trumler (1961),
Groves and Mazák (1967) and Bennett (1980) are examples of this. A different
philosophy is behind Quinn’s (1957) multigeneric scheme: that author – incor-
rectly, as most specialists now concur – saw the different modern groups as
the end points of lineages which could be traced back deep into the Miocene,
and had even achieved monodactyly independently.

5
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:24 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

6 C.P. Groves and O.A. Ryder

Although proposals have been made to link taxonomic ranking to time

depth, these have not achieved general acceptance, and the only widely
acknowledged criterion for taxonomic categories (above the species level, at
least) is monophyly. The single versus multigeneric schemes are essentially a
matter of taste. The fossil record of the Equidae is one of the best known
among mammals, and is replete at every level with genera, leaving much less
‘room’ for multiple genera among the living fauna. There is also some doubt
what these genera would actually be: will Equus, Asinus and Hippotigris
suffice, or must we have Hemionus, Dolichohippus and Quagga as well? What,
really, are the interrelationships of the living species? We are of the opinion, in
accord with most other equid specialists, that the less formal category of sub-
genus will have to do; but the question remains, what are they? The question
in general narrows down to asking, what are the monophyletic groups and
how well-supported are they? Below, we argue that three subgenera, Equus
(horses), Asinus (asses and hemiones) and Hippotigris (zebras), adequately
represent the major monophyletic groupings among living species, although
the diversity of species makes it necessary to recognize species-groups within
the latter two.

Early History of the Genus Equus

Before examining the living members of Equus, a brief survey of fossil forms is
in order. Like the family as a whole, the genus Equus evolved in North Amer-
ica, and spread from there into the Old World. It is likely that horses, asses and
zebras all evolved in North America, and separately spread into the Old World
when the Bering land bridge was open, from the mid-Pliocene onward.
Azzaroli (1998) has reviewed the North American species. The earliest, Equus
simplicidens, which lived from 4 to 2.5 mya, was very primitive, with many
Dinohippus-like features, and was probably ancestral to all subsequent
species; it is conventionally placed in a separate subgenus, Allohippus. The
subgenus Equus was represented by an indigenous species, E. niobrarensis, in
the Middle Pleistocene, and the immigrant E. ferus in the Late Pleistocene,
surviving into the Holocene in Alaska. The subgenus Asinus had emerged by 2
mya, and was well known by fossils by 1.2 mya (E. semiplicatus) and survived
into the Middle Pleistocene (E. francisci). Possible representatives of the
subgenus Hippotigris were E. idahoensis (Late Pliocene to Early Pleistocene,
2.6–1.0 mya), E. excelsus and E. occidentalis (Middle and Late Pleistocene,
respectively). A further subgenus, Amerhippus, is represented by early remains
at 2 mya, by a Middle Pleistocene species (E. fraternus) and one or two Late
Pleistocene species (E. conversidens and, possibly, E. mexicanus). Why equids
eventually became extinct in North America is unknown, but the last of them,
E. occidentalis (known from many skeletons from the tar pits at Rancho La
Brea, Los Angeles, California), was perhaps exterminated by the first human
inhabitants in the Early Holocene.

6
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:25 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 7

It is worth noting, too, that one branch of Equus entered South America in
the Late Pliocene. This is the zebra-like subgenus Amerhippus, which diversi-
fied into a variety of pampas and even Andean species (seven in number,
according to Azzaroli, 1992a), which survived alongside Hippidion and
Onohippidion until the Late Pleistocene, and finally died out for reasons that
are unclear; however, Azzaroli (1992b) has argued that it was a case of ‘pre-
historic overkill’, i.e. overhunting by early humans, perhaps abetted by climatic
change.
In both Europe and Asia, the earliest certain appearance of monodactyl
equids is at about 2.5 mya, apparently part of a major faunal turnover as forest
mammals disappeared and open-country fauna took their place (Azzaroli,
1990); they entered Africa around 2 mya (Azzaroli, 1998). The earliest species
(E. stenonis and its relatives) were nearly identical to North American E.
simplicidens and are placed in the subgenus Allohippus. After 1 mya, they
began to be replaced by species related to modern forms: by the Middle
Pleistocene, both the subgenus Equus (E. mosbachensis and later species)
and the subgenus Asinus (E. altidens, a member of the E. hemionus group)
had entered Eurasia; and in the Late Pleistocene, E. graziosii, a member of the
E. asinus group known only from Italy, and the widespread but still very
poorly known E. hydruntinus, which may be either a hemione or a zebra.
The African fossil representatives of Equus are almost entirely zebras.
From 2 mya on, one or several large species related to E. grevyi were wide-
spread throughout east and south Africa (E. capensis, E. oldowayensis,
E. koobiforensis). Remains of other zebra groups are rare, becoming common
only in the Late Pleistocene. A fossil ass is known from Olduvai Bed II
(~2 mya) and from the Middle (E. tabeti) and Late (E. melkiensis) Pleistocene
of the Maghreb. Fragmentary horse remains (E. algericus) also occur in the
Maghreb in the Late Pleistocene.

The species question

Most non-taxonomists still operate on a misunderstanding of the biological

species concept (BSC) of Mayr (1942). The popular misapprehension is
that different species cannot interbreed; some go a step further, believing that
species are sometimes able to interbreed, but hybrids between them are
sterile. The six universally recognized ‘species’ (here, species-groups) of Equus
therefore qualify under any rubric as valid species: hybrids between horses
and asses, for example, are (almost always) sterile. Many commentators would
go on to say that if hybrids between Indian and Persian wild asses prove to be
fertile, this would indicate that they, by contrast, are not valid species.
This is not in fact what Mayr said: he proposed that species ‘do not inter-
breed under natural conditions’, and emphasized that this reproductive isola-
tion might be the result of either pre- or post-mating isolating mechanisms.
The post-mating mechanisms are those which cause hybrids to be sterile; the
pre-mating ones include such things as ethological mechanisms (different

7
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:25 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

8 C.P. Groves and O.A. Ryder

courtship displays, for example), which can be broken down under unnatural
conditions, such as captivity. If the pre-mating mechanisms break down,
we may well discover that post-mating mechanisms are not in place – and
perfectly fertile hybrids will result.
Clearly, sympatry between two taxa is prima facie evidence for the
existence of reproductive isolating mechanisms between them; but, under the
BSC, allopatric taxa are simply not amenable to being assessed objectively,
unless we are able to conduct breeding experiments in captivity. Even then, if
crossing them is unsuccessful, or hybrids between them are sterile, we can say
that they are reproductively isolated and so are distinct species; but if they
interbreed to give fertile hybrids, we are not at liberty to claim that they are
members of the same species. All too often, of course, breeding experiments
are simply not feasible, which means in practice that the majority of sexually
reproducing species cannot even be tested. This is the case with living equids:
hybrids are known in captivity between (for example) Indian and Persian wild
asses, but their fertility has not been tested; and, today, the concern has always
been to preserve the pure gene pool, and multigeneration crossing is of low
priority.
Several variants of the BSC have been proposed, but when all is said and
done the biologist needs an operational definition of this concept, the species,
which we all agree is so basic. Cracraft (1983) gave voice to such thoughts
when he pointed out what an arbitrary decision it could be to assess whether
two taxa might or might not interbreed were their ranges to meet, and pro-
posed the phylogenetic species concept (PSC), whose operational criterion is
simply the diagnosability (the consistency) of the candidates for species status.
This is the attitude we take here: where two taxa are consistently different,
given the available evidence, we rank them as separate species. This is why
we recognize four species of the E. hemionus group, and two of the E. zebra
group, and why we do not recognize more than one species in the E. quagga
group, whose subspecies merge insensibly into each other.
The potential importance of the reproductive factor is unquestioned, but it
can never in itself be a criterion. We usually do not know whether there is
interbreeding between two species, or we may even know that there actually
is; the significant factor is that, if there is interbreeding, any resulting gene flow
does not affect the character discontinuity between them.

The subspecies question

Conceptually, there is no disagreement that subspecies are geographic

segments of a species which are morphologically differentiated to some
extent. They are geographic: by definition, they can never be sympatric. They
are morphologically differentiated, but not consistently, not 100%: this would,
under the PSC, make them different species. Mayr’s 75% rule, i.e. that
three-quarters of individuals of one subspecies must be distinguishable from
all individuals of all the others, is usually upheld, but this is a rule-of-thumb.

8
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:26 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 9

The subspecies is just a point on the continuum of degrees of differentiation at

which it becomes helpful (or desirable, or simply convenient) to recognize that
differentiation with a trinomial.
Subspecies are often the steps on a stepped cline. The subspecies of
E. quagga are an example here. On a stepped cline (as within E. africanus),
one can say, within limits, where subspecies A ends and subspecies B begins,
and between the two points is a more rapid change in gene frequencies, metri-
cal averages, whatever one is talking about. If the clinal gradient is insensibly
smooth, obviously to dignify its two ends as subspecies is rather arbitrary: and
this is why we recognize a single living subspecies of E. quagga south of the
Zambezi.
Note that either a species has at least two subspecies, or it has none.
Subspecies exist in relation to one another: there is no such thing as a species
having one single subspecies. One of the subspecies is called the nomino-
typical subspecies, and its subspecific name repeats its specific name; this will
be the subspecies that exists at the species’ own type locality. There is no
assumption that one subspecies is more ‘typical’ of the species than another.

Nomenclature of Domestic Animals

Groves (1995) has argued that there is, and can be, no definitive answer to the
question of whether domestic animals are the same species as their wild
relatives, or different species. They are not subspecies because they are
sympatric; they are not, or generally not, different species because except in a
few outstanding cases they are not diagnosably different. Groves (1995) has
called them paraspecies. Corbet and Clutton-Brock (1984) argued that, in
most cases, it is convenient to maintain the fiction that, for nomenclatural
purposes at least, they are different species.
The wild species may have subspecies; the domestic species does not.
Domestic breeds, however distinct they are, are sympatric; they arise and
merge within very short periods of time.

A Taxonomy of the Genus Equus

C.P.G. collected a number of characters from the literature (mainly Bennett,
1980; Groves and Willoughby, 1981; Eisenmann, 1986), but found that most
of the unique states are clearly autapomorphic: the elongated metapodials
of Hemionus, and so on. As in Bennett (1980), Dinohippus is taken as the
outgroup. In the end, 26 characters seem to yield phylogenetically informative
information (Table 1.1). The data matrix is given in Table 1.2.
The resulting cladogram (Fig. 1.1) separates the living equids into three
clades: horses, zebras, and asses plus hemiones. These same three clades are
found by studies of mitochondrial DNA (George and Ryder, 1986), whereas a
repetitive DNA sequence placed ass, hemione and the zebra group as three

9
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:26 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

10 C.P. Groves and O.A. Ryder

Table 1.1. Morphological interrelationships within living Equus.

1. Presence and development of isthmus on lower molars (B, no. 6)

2. Presence and development of infundibulum on third lower incisor (B, no. 11)
3. Presence of ‘cranial broadening complex’ (B, no. 12)
4. Degree of cranial flexion (B, no. 13)
5. Relationships of mastoid, paramastoid and mastoid temporal (B, no. 14)
6. Orientation of post-orbital bar relative to horizontal plane (B, no. 15)
7. Presence of frontal doming (B, no. 18)
8. Relative skull breadth (B, no. 19)
9. Scapula length > 102% of radius, humerus length < 86% of scapula (GW) – polarity
unknown
10. Forefoot terminal phalanx breadth > 125% of length (GW)
11. Forefoot terminal phalanx length > 55% of proximal phalanx, > 14% of radius (GW)
12. Forefoot terminal phalanx plantar length > 125% of height (GW)
13. Proximal humerus bicipital grooves deep (GW) – polarity unknown
14. Skull vertex length > 560 where basal length standardized = 500 mm (GW)
15. Palate–vomer length < vomer–basion (GW, E)
16. Tooth row length > 170 where basal length standardized = 500 mm (GW)
17. Occipital height > 298 (state 1), > 320 (state 2) where basal length
standardized = 500 mm (GW)
18. Nasal end of pre-maxilla narrowed, insinuated between nasal and maxilla (GW)
19. Choanal opening horseshoe-shaped (GW) – polarity unknown
20. Hypostylid present on third lower deciduous molar (GW) – polarity unknown
21. Shape of wolf tooth (GW) – polarity unknown
22. Ectoloph styles angle into walls (GW)
23. Palate length < 52% of basal length (E)
24. Muzzle length > 50% of palate length (E)
25. Choanal length > 15% of basal length (E)
26. Choanal breadth > 8.5% of basal length (E)
Morphological characters used in the cladistic analysis of species-groups of living equids
(B = Bennett, 1980; GW = Groves and Willoughby, 1981; E = Eisenmann, 1986; in general,
the derived states are described).

Table 1.2. Matrix of character states.

Horse 12111100111110110010100001
Hemione 21000001011001010010001001
Ass 20000000010001001100111001
Mountain zebra 00111100000001002001011110
Quagga 00111111011001102111100100
Grévy’s zebra 00111111111111000001111110
The characters are listed in rows 1–26 as per Table 1.1. State
0 denotes the primitive state, state 1 the derived, state 2
(where relevant) most derived.

equal branches of a non-horse clade (Sakagami et al., 1999). A striking point of

agreement is that the zebras form a monophyletic group; the common ancestor
of the three extant species was probably a striped animal and each taxon of

10
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:26 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 11

Fig. 1.1. Cladogram of living species-groups of Equus, based on the matrix of Table 1.2. The
single most parsimonious tree has length 48, consistency index 0.604, retention index 0.486. The
numbers above the lines are bootstrap values based on 1000 replicates. The horse/zebra clade is
defined by derived states of characters 3, 4, 5 and 6; the hemione/ass clade by the derived state
of character 1; the zebra clade by derived states of characters 7, 20 and 24; and the moun-
tain/Grévy’s clade by derived states of characters 22, 23, 25 and 26.

zebra is more closely related to other zebras than it is to any other equid. A
minor difference is that the ass/hemione clade separates first in the morpho-
logical tree, the horse clade in the genetic tree. Another possible area of
disagreement is that asses and hemiones are associated unequivocally in the
morphological tree, but less definitely in the genetic tree: whether hemiones
and African (‘true’) wild asses are more closely related to horses or whether
they are most closely related to each other and form a separate lineage has yet
to be resolved unequivocally by the DNA data, and note that in the morpho-
logical data they are united by only a single derived condition (Fig. 1.1).
In what follows, we have adopted the cladistic results (above), and
lumped all zebras into one subgenus, all ‘asses’ into another, with the horses in
a third. We briefly survey the evidence for species within each, giving less
attention to subspecies.

Subgenus Equus: horses

Externally, horses are recognizable by the long-haired tail; the mane that is
both long and thick and tends, at least in winter and/or with increasing age, to
fall to one side; the rounded croup; the usual presence of chestnuts on
hindlimbs as well as forelimbs; the broad, rounded hooves; and the poor
countershading, weak dorsal stripe and dark lower limbs with general traces of
stripes on carpus and tarsus. Cranially, they have a relatively small skull,
reduced cranial flexion, long diastema (the gap between the incisors and the
cheek teeth), large pterygopalatine fossa (behind the hard palate) and a long,
rounded occipital crest. The nasal end of the pre-maxilla is truncated or
rounded, so that the nasal bone forms the angle of the narial notch. Post-
cranially, they have a long scapula and stout metapodials and phalanges. The
metacarpus is short compared with the metatarsus, so that their hindlegs are

11
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:38 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

12 C.P. Groves and O.A. Ryder

longer than their forelegs. The pelvis is broad and splayed compared with
other subgenera; the biischial breadth is high compared with the biacetabular
(i.e. it is flared at the haunch compared with the hip joint); the height of the
pelvic inlet is strongly sexually dimorphic, increasing the width of the female’s
birth canal.
Groves (1986, 1993) argued that all horses that survived into historic times
belonged to one species, Equus ferus Boddaert, 1785, with three subspecies:
E. f. ferus (Steppe tarpan), E. f. sylvestris (Forest tarpan) and E. f. przewalskii
(Przewalski horse). The evidence that the Przewalski horse is the eastern rep-
resentative of the species that contained the European tarpan is admittedly
inferential: an apparent gradation of colours from west to east, and apparently
the occasional appearance of Przewalski-like colours in Europe (including, as
many people have noticed, on the walls of terminal Pleistocene caves such as
Lascaux). This needs to be tested further (if possible!), but it does seem at the
moment as if there was a continuum across Eurasia.
A large number of species of wild horse have been described from the
Late Pleistocene and Early Holocene of Europe, Siberia and China (see
Forsten, 1988, and, for a recent survey, see Azzaroli, 1998). Eisenmann (1996)
examined characters that have commonly been used to differentiate wild horse
species; very few actually seem to characterize credible taxonomic units.
Przewalski horses have a short vomer compared with other horses, both wild
and domestic, including tarpans, and differ in the metacarpal proportions. On
the other hand Przewalski horses and tarpans have shorter proximal phalanges
compared with the metapodials (i.e. their feet are short compared with their
lower limbs) than do the Late Pleistocene wild horses of France, Ukraine and
Russia; however, the Portugese wild horses and the European Bronze Age
(presumably domestic) horses resemble Przewalski horse and the tarpan.
Przewalski’s horses have 2n = 66 chromosomes, while normal domestic
horses possess 2n = 64 chromosomes.
There is no evidence that there were differentiated subspecies within what
conventionally has been considered the range of E. f. przewalskii. Analysis of
the stud book of the Przewalski’s horses identifies that just four mitochondrial
DNA haplotypes may descend from the 13 founders. However, DNA sequence
analysis of the control region of these four matrilines suggests that only two
distinct haplotypes survive (Oakenfull and Ryder, 1998). Previous studies of a
single mitochondrial control region haplotype of Przewalski’s horses in com-
parison with domestic horses resulted in the suggestion that Przewalski’s
horses may have been derived from domestic horses (Ishida et al., 1995), but
more recent studies involving all surviving mitochondrial DNA control region
haplotypes suggest that insufficient information exists to identify the phylo-
genetic relationships of Przewalski’s horses based on mitochondrial DNA.
Erect mane, lack of forelock and moulting of the hair at the base of the tail
are all primitive characteristics noted in wild, but not domestic, equids that
are typical features for Przewalski’s horses. Studies of nuclear genetic
variation in domestic and Przewalski’s horses may shed additional light on
to the phylogeny of caballine horses. Of particular interest will be studies of

12
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:39 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 13

domestic horses in Mongolia whose mitochondrial and nuclear diversity may

provide otherwise unavailable insights into the relationship of Przewalski’s
and domestic horses.

Subgenus Asinus: ass, onager and kiang

These are characterized externally by having a tufted tail, chestnuts on

forelimbs only and at least some degree of countershading. Cranially they are
characterized by the squared, upturned occipital crest and wide external audi-
tory meatus. Post-cranially, they have a short scapula; and, like horses, the
metacarpus is short compared with the metatarsus, the biischial breadth is high
compared with the biacetabular, and the height of the pelvic inlet is strongly
sexually dimorphic.
We can divide this subgenus into two certainly monophyletic species-
groups, the E. hemionus group and the E. asinus group. This is necessary
because of the existence of more than one living species of ‘hemiones’, and of
multiple fossil species of ‘true asses’.

Equus hemionus group

The E. hemionus group (onagers or hemiones or Asian wild asses) are charac-
terized externally by the short, clipped-appearing mane, short erect hair form-
ing a broad dorsal stripe 50–100 mm broad, and white underparts and lower
limbs. Cranially, they are characterized by the very narrow squared upturned
occipital crest, shortened brain case, long vomer, short diastema and short
pterygopalatine fossa. The nasal end of the pre-maxilla is truncated or
rounded, so that nasal bone forms the angle of the narial notch. The tuber
maxillae extend back, hiding the pterygopalatine fossa in the ventral view,
like quaggas and unlike other equids. Post-cranially, they are distinguished by
the short humerus and femur, the elongated, slender metapodials (so that the
lower limb segments are long and fine) and the elongated distal phalanx.
Groves and Mazák (1967) argued that the kiang, of the Tibetan plateau, is
a distinct species (E. kiang) from the onagers (E. hemionus) of the more
low-lying Asian deserts. The differences are very striking, and this separation
has been widely followed. The subspecies onager, kulan, castaneus and
hemionus form an intergrading series, overlapping (at the extremes) in their
traits but each strongly distinct as a unit. The two other taxa usually placed as
subspecies of E. hemionus, i.e. hemippus and khur, are in fact diagnosable
taxa (in the PSC sense) and are here regarded as distinct species.
The number of chromosomes in Asiatic wild asses varies from 2n = 56 to
2n = 50. This is a large amount of numerical and structural chromosomal poly-
morphism to be segregating within a single species. Tibetan kiangs, Equus
kiang holdereri, apparently have fixed chromosomal differences from other
subspecies of Asiatic wild asses, based on the available chromosomal data
(Ryder et al., 1978; Ryder, 1986; Ryder and Chemnick, 1990). Mitochondrial
DNA studies, including restriction fragment length polymorphism (RFLP)

13
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:39 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

14 C.P. Groves and O.A. Ryder

analysis (Ryder and Chemnick, 1990) and control region and 12S RNA
sequence data (Oakenfull et al., unpublished observations), find that named E.
hemionus subspecies are paraphyletic. Phylogenetic analysis identifies mito-
chondrial lineages encompassing the examined diversity of Asiatic wild asses
that are present in multiple subspecies. Based on available data, extant
mitochondrial DNA variation appears to reside largely within the named
subspecies and not to be partitioned between onagers, kulans and kiangs.
Subspecies of E. hemionus may hybridize, and be fertile, in spite of
the chromosomal polymorphism (Pohle, stud book), providing evidence that
reproductive isolation has not yet developed. The implication from mito-
chondrial DNA analyses is that the extant subspecies share a common pool of
mitochondrial variation; divergence and genetic isolation are minimal, imply-
ing that these processes have begun only recently. The most divergent of the
Asiatic wild asses based upon genetic analysis is the kiang, supporting their
status as a separate species.
Significant gaps exist, though, in the genetic studies of the hemione group.
In particular, the khur of the Little Rann of Kutch has not been the subject of
published molecular studies. Mongolian wild asses and the western and
southern kiang similarly are absent from the literature on molecular evolution
in Equus. Accordingly, there are opportunities for additional studies and a
need to provide a more detailed view of nuclear and mitochondrial variation in
Asiatic wild asses for conservation assessments, monitoring and management.

Equus kiang Moorcroft, 1841: Kiang The kiang or Tibetan wild ass is of large
size; it has a large head and thick muzzle, a relatively long mane and long hairs
not restricted to the tail tuft but extending some way up either side of the tail.
The pattern of contrasting dark (reddish) body blocks and white underside is
characteristic: the demarcation between them on the flank is oblique from stifle
to croup, and the white rump patch is infused with the reddish tone of the
haunch. The dorsal stripe is thin and never bordered with white; it extends to the
tail tuft. There is a dark ring round the hoof. The ear is 165–178 mm long.
The skull resembles that of E. hemionus, except that the incisors tend to sit
more vertically in the jaws (except in aged individuals, in which alveolar reces-
sion tends to reveal the oblique roots), and the highest point on the cranial
profile is often directly above the posterior rim of the orbit, instead of behind
it.
There are three subspecies, which differ (as a whole, but not absolutely)
as follows:
1. Equus kiang kiang Moorcroft, 1841. Western kiang. Dark red-brown colour
in summer, dark brown in winter; the legs have brown tones. Size large; nasal
bones short; tooth row relatively short. The southwestern corner of Tibet, into
India (Ladakh) and Pakistan.
2. Equus kiang holdereri Matschie, 1911. Eastern kiang. Colour not so dark:
strong red in summer, darker red-brown in winter; the legs are pure white. Size
large; nasals very long; tooth row long. Eastern Tibet (Lhasa district) to Chinghai,
Ganssu and the Seshu region of Sichuan.

14
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:40 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 15

3. Equus kiang polyodon (Hodgson, 1847). Southern kiang. Colour dark as in

E.k. kiang; size very small; nasals and tooth row long as in E. k. holdereri. The
Gayam Tso region of Tibet, into Sikkim.

Equus hemionus Pallas, 1775: Onager The onager varies from kiang-sized to
khur-sized. The mane is very short and ‘clipped’; the dorsal stripe is thick, often
bordered with a white line on either side; the white of the rump is not infused;
the demarcation between the reddish flank block and whitish underside runs
parallel to the body outline, before turning sharply up towards the dorsal stripe.
The dorsal stripe extends to the tail tuft. There is a dark ring round the hoof. The
nasal bones are relatively straight, the skull resembles that of E. kiang.
The number and definition of the subspecies are disputed, but the follow-
ing seem recognizable:
1. Equus hemionus hemionus Pallas, 1775. Mongolian wild ass. This is the least
endangered subspecies, still found throughout the Mongolian desert regions,
and formerly into Transbaikalia. Height 110–130 cm; poor demarcation of dark
and light (off-white) areas; the only subspecies usually lacking a white border to
the dorsal stripe. The concept that Groves and Mazák (1967) and Groves (1986)
used was: (i) a disruptively coloured subspecies, occupying a northerly distribu-
tion extending from Transbaikalia to Kazakhstan, and (ii) a grading-toned one,
called E. h. luteus, restricted to the Gobi desert. However, Denzau and Denzau
(1999) have shown that the type illustration of Pallas’s E. hemionus, which was
from Transbaikalia, in fact corresponds to the grading subspecies. The disrup-
tively patterned ‘northern’ subspecies therefore never extended to Transbaikalia
but was purely western in range, and must be called:
Equus hemionus castaneus (Lydekker, 1905). Probably now extinct:
formerly from Dzungaria through Kazakhstan to Uzbekistan. Also large; clear
demarcation of coloured and white areas; much white on body and head.
2. Equus hemionus kulan (Groves and Mazák, 1967). Turkmenian wild ass or
kulan. Badkhyz Reserve, Turkmenia. Smaller: height 108–120 cm. Coloured and
white areas strongly demarcated like castaneus. Relatively larger teeth than the
large Mongolian/Kazakhstan forms.
3. Equus hemionus onager Boddaert, 1785. Persian wild ass. Iran, east of the
Zagros Range. Also small; demarcation between coloured and white areas less
strong than in kulan, from which it differs in pattern details. Broader occipital
crest; even larger teeth. This still occurs in the Bahram-e-Gur and Touran
Reserves in Iran; according to Denzau and Denzau (1999), these two remnant
populations differ in size.
4. Equus hemionus blanfordi Pocock, 1947. Probably now extinct; formerly,
known only from Sham Plains (Pakistan) and Kandahar (Afghanistan). Also
small, and large-toothed; with relatively narrow occipital crest and long nasals;
extensive dark areas on flanks; the dorsal stripe fades out halfway down the tail.

Equus khur Lesson, 1827; Khur (Indian wild ass) The Indian wild ass or
khur is sharply distinct from E. hemionus, and because it can always be
distinguished from other forms we recognize it as a full species. The coloured

15
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:40 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

16 C.P. Groves and O.A. Ryder

blocks on the flank and haunch are very small, so the predominant colour is
white, and the lower 45% or more of flank is whitish; the demarcation on the
lower haunch slants upward from the front (stifle) to the back. There is a dorsal
stripe with a clear white border on either side. The white wedge between the
haunch and flank blocks nearly or fully reaches the spine. The legs are pure
white. The dorsal stripe fades out halfway down the tail. There is no dark ring
round the hoof.
The facial profile is concave; the nasal bones are raised (making the whole
facial profile strongly concave), and comparatively short (Groves, 1986, Figure
1); and the teeth are small. The skull is noticeably high crowned. The choanae
are small and the orbits are high. Height at withers is 110–130 cm. The
metapodials are less elongated than in E. hemionus. The ear is very long:
187–210 mm.
The ass survives in the Rann of Kutch, Gujarat, India; formerly known
from the Sind deserts, Pakistan.

Equus hemippus I. Geoffroy St Hilaire, 1855: Syrian wild ass The extinct
Syrian wild ass or achdari was likewise diagnostically different from other
hemiones. It was very small in size; the evidence suggests that this difference has
come about since the end of the Pleistocene (Turnbull, 1986). The height at the
withers was about 1 m. Colour was very grading; a sandy-brown flank patch
extended well down, grading into off-white on the underside; only the lower
20% or less of flank was whitish. There was a dorsal stripe with a clear
white border on either side; this became obfuscated with age, and eventually
disappeared. The white wedge between the haunch and flank blocks was
vague, strongly infused with body tone. The legs were strongly infused with
body tone. The dorsal stripe faded out halfway down the tail. There was no dark
ring round the hoof.
The nasal bones were raised, and relatively longer than in other onagers
(Groves, 1986, Fig. 1.1); the teeth were relatively large. Otherwise the skull,
with its concave profile, high-placed orbit, and high crown, resembled a small
E. khur. The metapodials were more elongated than those of other species; the
terminal phalanges were shortened.
It was known from Syria and northern Iraq.

Equus asinus group

The Equus asinus group (true asses or donkeys) are characterized externally
by a long, thin ‘scruffy’ mane, thin dorsal stripe, usually traces of stripes on the
legs (at least fetlocks) and a restricted white (or pale) region on the underside.
Cranially they have a very long cranium, short palate, long diastema, large
pterygopalatine fossa and a squared, upturned occipital crest. The nasal end of
the pre-maxilla is narrow, insinuated into a corner of the narial notch. The
orbit is high, rounded and not anteroposteriorly elongated like other equids
except mountain zebras. Post-cranially, there are very short, narrow distal
phalanges, translating externally to small, narrow hooves.

16
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:41 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 17

Eisenmann (1995) reviewed recently the palaeontological evidence for the

evolution of this species group, and the evidence for extinct Early Holocene
populations.
There is one wild species:

Equus africanus Heuglin and Fitzinger, 1866: African wild ass All African
wild asses tend to be more reddish in summer, greyer in winter, with
contrastingly white legs and a less contrastingly whitish underside; the white
wedges behind the shoulder and in front of the haunch, so prominent in
hemiones, are evanescent.
Groves (1986) showed that, to some degree, there is clinal change from
the Atbara population via the Red Sea coastal populations to Somalia. There is
quite a marked change, a distinct step in the cline, between northern and
southern Eritrea, perhaps representing a bottleneck to gene flow in the
Massawa region, where the Highlands approach the sea, and this justifies the
recognition of at least two subspecies. A supposed wild ass was described as
Asinus taeniopus Heuglin (type locality: Little Dobar, south of Berbera), and
even today this is sometimes recognized as a subspecies, but no such wild
ass actually exists: it was based on a cross between a Somali wild ass and a
domestic ass.
A north African wild ass, with strong, often doubled, shoulder-cross and
well-marked leg stripes was depicted in both rock art and Roman era mosaics,
and was stated to survive at Siwa, on the Libyan–Egyptian border, by Hufnagel
(1965). It is often called Equus africanus atlanticus, but should not be, as that
name was first given to a fossil north African zebra. Eisenmann (1995) dis-
cusses whether the name Equus melkiensis (described from a genuine fossil
ass, from the Late Pleistocene of Allobroges, Algeria) might apply to this form,
though one ought to be rather cautious in associating Pleistocene fossils with
Holocene rock depictions.
In the earliest Holocene, wild asses were also present in northern Arabia
(Ducos, 1986; Groves, 1986); a subspecies Equus africanus mureybeti Ducos,
1986 has been described from pre-pottery levels in Iraq, but Eisenmann (1995)
is not convinced that the remains are ass rather than onager.
Only the Somali wild ass has been studied genetically, and there are signif-
icant doubts that any other forms of African wild ass survive. Chromosomal
polymorphisms in Somali wild asses have been described (Houck et al., 1998).
Mitochondrial DNA analyses place Somali wild ass as a sister group to domes-
tic donkey but, without additional study, the phylogenetics and systematics of
African wild asses remain tentative.
There are several subspecies:
1. Equus africanus africanus Heuglin and Fitzinger, 1866. Nubian wild ass.
From Atbara River to Red Sea coast and northernmost Eritrea. May be extinct.
Shoulder height 115–121 cm. A dorsal stripe is always present, and nearly
always complete from mane to tail tuft; crossed by an usually thin, fairly short
shoulder stripe. Leg stripes, where present are restricted to a few bands at the
fetlocks. The diastema is relatively short, and the post-orbital constriction well

17
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:41 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

18 C.P. Groves and O.A. Ryder

marked. The nasal process of the pre-maxilla ends bluntly. There is never a
‘bridge’ between metaconid and metastylid in the lower premolars and molars.
Groves (1986) found that specimens from the Atbara differ on average from
those from the Red Sea Hills (Sudan) and Eritrea, and they could be
subspecifically distinct. The Nubian wild ass is probably not, contrary to
‘received opinion’, the ancestor of the domestic donkey.
2. Equus africanus somaliensis Noack, 1884. Somali wild ass. Ears shorter,
187–200 mm; shoulder height 120–125 cm; the dorsal stripe is often absent, and
when present is often incomplete, broken at some point along the dorsum.
Shoulder-cross often absent. Leg stripes are present from hooves to above the
carpus and tarsus. Diastema is relatively long, and the post-orbital constriction
less marked. There is a thickened bar of bone behind the orbits, marking the
highest point on the profile. The nasal process of the pre-maxilla is thin and
pointed. There is always at least a trace of a ‘bridge’ between metaconid and
metastylid in the lower premolars and molars. Somali asses seem to be longer
legged, and shorter bodied than Nubian ones. Differences between Somalian
and Danakil plus Djibouti populations exist, but are less marked than those
between the two populations of E. a. africanus. Somali wild asses still exist in
two small population nuclei: in the Afar (Danakil) country of southern Eritrea,
and in the Nogal valey, northern Somalia.
3. Equus africanus subsp. Saharan wild ass. Still more a rumour than a fact, but
best authenticated for Ahaggar, Tibesti and Fezzan; the appearance of the
(apparently indigenous) wild ass of the Sahara was reconstructed by Groves
(1986) as closest to E. a. africanus but smaller and greyer, and with a long, thin
shoulder-cross.

Subgenus Hippotigris : zebras

These are characterized externally by the long, rather thick upright mane,
tufted tail, chestnuts on forelimbs only, and striking black and white stripes.
Cranially, the occiput is high and raised; the post-orbital constriction is
deep; the muzzle long; and the vomer long. Post-cranially, the metacarpus
is long compared with the metatarsus, so that the forelimbs are longer than
the hindlimbs. The biischial breadth is low compared with the biacetabular, so
that the hip joints are further apart, giving zebras a fat-rumped appearance; the
height of the pelvic inlet is not strongly sexually dimorphic, so that males, as
well as females, have a large, high space corresponding to the birth canal.
Chromosomal and molecular data, as well as the morphological analysis
presented here, all support the position of monophyly for zebras – that all
extant zebras share a recent common ancestor with each other. Mitochondrial
DNA RFLP study suggested that plains zebra and Grévy’s zebra are sister taxa,
as did study of a repeated sequence (Sakagami et al., 1999), but in another
study involving control region and 12S gene sequencing, the mountain zebra
and plains zebra were identified as sister taxa (Oakenfull and Ryder, 1998).
The morphological tree, however, places mountain and Grévy’s zebras closer.

18
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:41 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 19

Equus zebra group

The E. zebra group (mountain zebras) is characterized externally by the long,
rather thick upright mane, small dewlap, and stripes absent from venter, form-
ing what has been called a ‘gridiron’ pattern (short transverse stripes meeting
the uppermost of a long, thick, oblique/longitudinal series) on the croup.
Cranially, the nasal end of the pre-maxilla is truncated or rounded, so that
the nasal bone forms the angle of the narial notch. The external auditory
meatus is wide, and (uniquely) directed horizontally instead of upward and
backward, probably connected with the large external ears. The ventral bar
of the orbit is slender and the orbit is high, rounded like that of asses. Post-
cranially, the scapula is short; the distal phalanges are short and narrow, so
that the hoof is small and fine like that of a donkey. The skull is relatively large
compared with the rest of the skeleton.
The two forms usually ranked as subspecies of a single species, E. zebra,
should certainly be regarded as distinct species. They are strikingly different
externally, and some of the skull measurements do not overlap, although note
that sample sizes are small. For example, occipital crest breadth is 63–71 mm
in male zebra, 74–78 mm in male hartmannae (n = 5 of each); in females,
63–68 mm in zebra, 70–86 mm in hartmannae (n = 3 and 6, respectively). In
the present sample, there is a sexual size difference in zebra (mean skull
length 516.0 mm in males, 530.8 mm in females), but not in hartmannae
(548.0 and 549.9 mm, respectively).

Equus zebra Linnaeus, 1758. Cape mountain zebra. Mountains of the south-
ern Cape (Skead, 1980). It was nearly extinct in the 1950s but has been very
successfully preserved and reintroduced over much of its former range.
Size smaller; black stripes broader than white interspaces. Occipital crest
narrow; females apparently larger than males.

Equus hartmannae Matschie, 1898. Hartmann’s mountain zebra Arid, hilly

regions from southern Angola formerly to northwestern Cape. It still occurs in
Angola and northern Namibia, but is very much less numerous than formerly.
Size much larger; black stripes narrower than white interspaces. Occipital
crest broad. Sexes equal in size.

Equus quagga group

The E. quagga group (plains zebras) have a mane varying from long, thick,
neat to shorter, thinner, and even absent altogether; striping varies from dark
brown and white on head and neck only to striking black and white over the
whole body, including the venter, and a simple oblique/longitudinal pattern
on the croup and haunch.
The post-orbital constriction is relatively narrow, the vomer long, the
diastema long, the teeth relatively small, and the occiput extremely raised. The
narial notch is less deep than that of other zebras. The interorbital part of
the frontal bone is strongly convex, so that these zebras have a forehead that is
rounded and convex from side to side. The bizygomatic width (cheekbones

19
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:42 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

20 C.P. Groves and O.A. Ryder

behind eyes) generally exceeds the biorbital width, so that the eyes do not
protrude like those of other equids. The nasal end of the pre-maxilla is narrow,
insinuated into the corner of the narial notch. Pre-maxillae, unlike those
of other zebras, are curved downward below the level of the alveolar line of
the cheek teeth, so that the upper lip is unusually deep. The tuber maxillae
extends back, hiding the pterygopalatine fossa in the ventral view, like
Hemionus and unlike other equines. The foramen magnum (the hole through
which the spinal cord enters the brain case) is an unusual, uniquely rectangu-
lar shape. The metapodials are somewhat lengthened, so that the lower legs
are relatively long. Distal phalanges are less reduced than in the E. zebra
group, translating to larger, broader hooves.
The cranial differences between plains and mountain zebras are given by
Eisenmann and de Giuli (1974) and Smuts and Penzhorn (1988), who agree in
most respects, though the latter add a few characters, notably the important
difference in the foramen magnum, in which this species group is unlike any
other subgenus.
Although in this species group two species, E. quagga and E. burchelli, are
often recognized, they grade insensibly into each other; some of the skins
described and illustrated by Rau (1974), especially the Mainz female ‘true
quagga’ and another specimen in Mainz, the type of paucistriatus Hilzheimer,
are difficult to allocate to one or the other. There seems no prospect of
breaking the cline from the Cape to Sudan and Somalia, into species; though,
the cline being stepped, the subspecies seem clear enough. The quagga was in
fact the first extinct organism from which DNA was extracted (Higuchi et al.,
1984); the result showed clearly its close affiliation to living plains zebras.
The plains zebra is the most abundant wild equid and, though it can be
divided into several fairly clear-cut subspecies, there is no clear picture to
present from the genetic perspective until nuclear and mitochondrial DNA
studies of populations across the range of the species has been undertaken.

Equus quagga quagga Gmelin, 1788 The extinct true quagga lived west of
the Drakensberg and south of the Vaal-Orange system (Skead, 1980). It had
head and neck stripes, but on the body the stripes were incomplete at best;
body colour was fawn, but legs were white. The available museum material was
fully discussed by Rau (1974, 1978), who showed that there are some specimens
that are so intermediate between ‘true quagga’ and burchelli that we cannot
recognize any species distinction. The physical geographic barriers between the
two are, however, sufficient to make a strong step in the cline, and so confirm
their subspecific distinction. I have measured only three adult male skulls from
the wild (Leiden, Berlin and Frankfurt). They range from 485 to 528 mm in
length, and so average smaller than other subspecies apart from boehmi.

Equus quagga burchellii Gray, 1824 Known in different regions as

Burchell’s, Chapman’s, Wahlberg’s and Damara zebra. From Kwazulu-Natal and
the Vaal-Orange system north to the Zambezi and Namibia. They are large in
size. Three or four stripes (very rarely two or five) meet (sometimes they do not

20
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:42 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 21

quite meet) the median ventral line between the elbow and the stripe that bends
back to form the ‘saddle’ of the lumbar region. Colour is ochrey or off-white,
never pure white; shadow stripes are usually well marked; leg stripes are absent
or poor, almost never complete to the hooves. The infundibulum (‘cup’) on the
lower incisors is better expressed than in other subspecies. The mane is well
developed. Northerly populations (Zimbabwe, northern Mozambique) are
paler, with less strikingly marked shadow stripes and more complete leg stripes
than other populations. Those from Kwazulu-Natal and also northern Namibia
(Etosha and Kaokoveld) are more ochrey with stronger shadow stripes and
fewer leg stripes. There is simply no prospect of breaking up this subspecies as
has usually been done. The least striped phenotype, which even lacks stripes on
the lower haunches, occurs in both Etosha and Kwazulu-Natal and formerly
occurred in the Free State; it has sometimes been called ‘true Burchell’s zebra’,
but there was never a population even predominantly characterized by this type,
and ‘true Burchell’s zebra’ has never existed as a discrete population.

Equus quagga crawshayi de Winton, 1896 From north of the Zambezi and
east of the Luangwa: so, in easternmost Zambia, Malawi, northern Mozambique
and southern Tanzania. They are of large size. Stripes are numerous and narrow;
there are always at least five stripes (often six to eight) meeting the median ven-
tral line between the elbow and ‘saddle’ stripe; body tone is white or off-white;
almost never even traces of shadow stripes; leg stripes complete to hooves.

Equus quagga zambeziensis Prazak, 1898 From Angola and Zambia east to
the Luangwa, and perhaps into Shaba. Large in size. Stripes are broad; only three
or four meet the median ventral line between the elbow and ‘saddle’ stripe;
colour varies from ochrey through off-white to white; shadow stripes vary from
fairly prominent to absent; leg stripes are usually complete, or nearly so.

Equus quagga boehmi Matschie, 1892: Grant’s or Boehm’s zebra Small in

size. Externally it resembles the previous form closely, but is separated from it by
E. q. crawshayi and by Lake Tanganyika. The infundibulum is generally absent.
It intergrades with crawshayi in southern Tanzania, and then ranges north
through Tanzania and Kenya into southern Somalia and Ethiopia, Karamoja
(Uganda) and southeast Sudan, east of the Nile. Most northerly populations of
this subspecies have high frequencies of manelessness, and the ears are either
very vaguely striped or completely white; they could perhaps be separated as a
different subspecies but there are differences among them: (i) manelessness
appears to be fixed in the northeastern ones (northeast of Mount Kenya, into
Somalia), but merely high frequency (especially frequent in males) in the north-
western ones (Baringo, Karamoja and Sudan), and apparently does not occur at
all in Ethiopia or around Lake Turkana, and (ii) whereas northeastern zebras are
as small as other boehmi or even slightly smaller, northwestern ones are larger,
as big as the more southerly subspecies such as zambeziensis. If it is felt that they
should be recognized as different subspecies: the name for the northeastern one
would be granti, for the northwestern, borensis.

21
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:43 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

22 C.P. Groves and O.A. Ryder

Equus grevyi group

The E. grevyi group, Grévy’s zebra, has a long thick upright mane, very small
chestnuts, short erect black hairs along the dorsal stripe and narrow, almost
dazzling, black and white stripes on the whole body except for the venter and
croup, forming a complex triradial pattern on the haunch. The skull is very
elongated, exceeding the cervical spine in length. The nasal end of the
pre-maxilla is rounded, wedged into the nasal bone. Post-cranially, the scapula
is lengthened, the metapodials somewhat lengthened, and distal phalanges
small.
There is only one living species, i.e. Equus grevyi Oustalet, 1882. Grévy’s
zebra. Provisionally, two subspecies can be recognized, based entirely on size,
although they need to be tested on larger samples: (i) Equus grevyi grevyi
Oustalet, 1882. Greatest skull length 529–615 mm. From the Somali Arid zone
proper: Harar region to northernmost Kenya; and (ii) Equus grevyi subsp.
Larger: greatest skull length 608–639 mm. From the Guaso Nyiro district,
Kenya.

References
Azzaroli, A. (1990) The genus Equus in Europe. In Lindsay, E.H. (ed.), European
Neogene Mammal Chronology. Plenum Press, New York, pp. 339–356.
Azzaroli, A. (1992a) The present state of knowledge on the Ecuatorian species of the
genus Equus. Bollettino della Società Paleontologica Italiana 31, 133–139.
Azzaroli, A. (1992b) Ascent and decline of monodactyl equids: a case for prehistoric
overkill. Annali Zoologici Fennici 28, 151–163.
Azzaroli, A. (1998) The genus Equus in North America – the Pleistocene species.
Palaeontographia Italica 85, 1–60.
Bennett, D.K. (1980) Stripes do not a zebra make, part I: a cladistic analysis. Systematic
Zoology 47, 272–287.
Corbet, G.B. and Clutton-Brock, J. (1984) Appendix: taxonomy and nomenclature.
In: Mason, I.L. (ed.), Evolution of Domesticated Animals. Longman, London,
pp. 434–438.
Cracraft, J. (1983) Species concepts and speciation analysis. In: Johnston, R.F. (ed.)
Current Ornithology 1, 159–187.
Denzau, G.and Denzau, H. (1999) Wildesel. Jan Thorbecke Verlag, Stuttgart.
Ducos, P. (1986) The equid of Tell Muraibit, Syria. In: Meadow, R.H. and Uerpmann,
H.-P. (eds), Equids in the Ancient World. Dr Ludwig Reichert Verlag, Wiesbaden,
pp. 237–245.
Eisenmann, V. (1986) Comparative osteology of modern and fossil horses, half-asses,
and asses. In: Meadow, R.H. and Uerpmann, H.-P. (eds), Equids in the Ancient
World. Dr Ludwig Reichert Verlag, Wiesbaden, pp. 67–116.
Eisenmann, V. (1995) L’origine des ânes: questions et réponses paléontologiques.
Ethnozootechnie 56, 5–26.
Eisenmann, V. (1996) Quaternary horses: possible candidates to domestication. Pro-
ceedings of the XIII Congress, International Union of Prehistoric and Protohistoric
Sciences 6, 27–36.

22
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:43 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

Systematics and Phylogeny 23

Eisenmann, V. and de Giuli, C. (1974) Caractères distinctifs entre vrais zèbres (Equus
zebra) et zèbres de Chapman (Equus burchelli antiquorum) d’après l’étude de 60
têtes osseuses. Mammalia 38, 509–543.
Eisenmann, V. and Shah, N. (1996) Some craniological observations on the Iranian,
Transcaspian, Mongolian and Indian hemiones. EEP Yearbook, 1995/6, 396–399.
Evander, R.L. (1989) Phylogeny of the family Equidae. In: Prothero, D.R. and Schoch,
R.M. (eds), The Evolution of Periisodactyls. Oxford University Press, Oxford,
pp. 109–127.
Forsten, A. (1988) The small caballoid horse of the upper Pleistocene and Holocene.
Journal of Animal Breeding and Genetics 105, 161–176.
George, M. and Ryder, O.A. (1986) Mitochondrial DNA evolution in the genus Equus.
Molecular Biology and Evolution 3, 535–546.
Groves, C.P. (1986) The taxonomy, distribution, and adaptations of recent equids. In:
Meadow, R.H. and Uerpmann, H.-P. (eds), Equids in the Ancient World. Dr Ludwig
Reichert Verlag, Wiesbaden, pp. 11–65.
Groves, C.P. (1993) Morphology, habitat and taxonomy. In: Boyd, L. and Houpt, K.A.
(eds), Przewalski’s Horse: the History and Biology of an Endangered Species. State
University of New York Press, Albany, pp. 39–59.
Groves, C.P. (1995) On the nomenclature of domestic animals. Bulletin of Zoological
Nomenclature 52, 137–141.
Groves, C.P. and Mazák, V. (1967) On some taxonomic problems of Asiatic wild asses
with a description of a new subspecies (Perissodactyla, Equidae). Zeitschrift für
Säugetierkunde 32, 321–355.
Groves, C.P. and Willoughby, D.P. (1981) Studies on the taxonomy and phylogeny of
the genus Equus. 1. Subgeneric clasification of the recent species. Mammalia 45,
321–354.
Higuchi, R., Bowman, B., Freiberger, M., Ryder, O.A. and Wilson, A.C. (1984) DNA
sequences from the quagga, an extinct member of the horse family. Nature 312,
282–284.
Hooker, J.J. (1984) A primitive ceratomorph (Perissodactyla, Mammalia) from the early
Tertiary of Europe. Zoological Journal of the Linnaean Society 82, 229–244.
Houck, M.H., Kumamoto, A.T., Cabrera, R.M. and Benirschke, K. (1998) Chromosomal
rearrangements in a Somali wild ass pedigree, Equus africanus somaliensis
(Perissodactyla, Equidae). Cytogenetics and Cell Genetics 80, 117–122.
Hufnagel, E. (1965) Libyan Mammals. Oleander, Harrow.
Hulbert, R.C. (1989) Phylogenetic interrelationships and evolution of North American
late Neogene Equinae. In: Prothero, D.R. and Schoch, R.M. (eds), The Evolution of
Periisodactyls. Oxford University Press, Oxford, pp. 176–196.
Ishida, N., Oyunseren, T., Mashima, S., Mukoyama, H. and Saitou, N. (1995) Mitochon-
drial DNA sequences of various species of the genus Equus with special reference
to the phylogenetic relationship between Przewalski’s wild horse and domestic
horse. Journal of Molecular Evolution 41, 180–188.
Lydekker, R. (1905) Notes on the specimens of wild asses in English collections.
Novitates Zoologicae 11, 583–596.
Mayr, E. (1942) Systematics and the Origin of Species. Columbia University Press, New
York.
Oakenfull, E.A. and Ryder, O.A. (1998) Control region and 12S rRNA variation in all the
extant mitochondrial lineages of Przewalski’s horse (Equus przewalskii). Animal
Genetics 29, 456–459.

23
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:44 A3759:AMA:First Revise: 20 January 1999 CHAP-1
Color profile: Disabled
Composite Default screen

24 C.P. Groves and O.A. Ryder

Prothero, D.R. and Schoch, R.M. (1989) Classification of the Perissodactyla: summary
and synthesis. In: Prothero, D.R. and Schoch, R.M. (eds), The Evolution of
Periisodactyls. Oxford University Press, Oxford, pp. 530–537.
Quinn, J. (1957) Pleistocene Equidae of Texas. Bureau of Economic Geology, University
of Texas, Austin, Report no. 33, pp. 1–51.
Rau, R.E. (1974) Revised list of the preserved material of the extinct Cape Colony
Quagga, Equus quagga quagga (Gmelin). Annals of the South African Museum 65,
41–86.
Rau, R.E. (1978) Additions to the revised list of preserved material of the extinct Cape
Colony Quagga and notes on the relationship and distribution of southern Plains
Zebras. Annals of the South African Museum 77, 27–45.
Ryder, O.A. (1986) Genetic investigations: tools for supporting breeding programme
goals. International Zoo Yearbook 24/25, 157–162.
Ryder, O.A. and Chemnick, L.G. (1990) Chromosomal and molecular evolution in
Asiatic wild Asses. Genetica 83, 67–72.
Ryder, O.A., Epel, N.C. and Benirschke, K. (1978) Chromosome banding studies of the
Equidae. Cytogenetics and Cell Genetics 20, 323–350.
Sakagami, M., Hiromura, K., Chemnick, L.G. and Ryder, O.A. (1999) Distribution of the
ERE-1 family in Perissodactyla. Mammalian Genome 10, 930–933.
Skead, C.J. (1980) Historical Mammal Incidence in the Cape Province. 1. The Western
and Northern Cape. Department of Nature and Environment Conservation, Provin-
cial Administration of Cape of Good Hope, Cape Town.
Smuts, M.M.S. and Penzhorn, B.L. (1988) Descriptions of anatomical differences
between skulls and mandibles of Equus zebra and Equus burchelli from Southern
Africa. South African Journal of Zoology 23, 328–336.
Trumler, E. (1961) Entwurf einer Systematik der rezenten Equiden und ihrer fossilen
Verwandten. Säugetierkundliche Mitteilungen 24, 206–218.
Turnbull, P.F. 1986. Measurements of Equus hemionus from Palegawra Cave (Zarzian,
Iraq). In: Meadow, R.H. and Uerpmann, H.-P. (eds), Equids in the Ancient World.
Dr Ludwig Reichert Verlag, Wiesbaden, pp. 319–365.
Xu, X. and Arnason, U. (1994) The complete mitochondrial DNA sequence of the horse,
Equus caballus: extensive heteroplasmy of the control region. Gene 148, 357–362.

24
Z:\Customer\CABI\A3688 - Bowling - The Genetics of the Horse\A3688 - Bowling - The Genetics of the Horse - Chps 1-10 #C.vp
03 April 2000 10:35:44 A3759:AMA:First Revise: 20 January 1999 CHAP-1