The two living elephants are the sole survivors of a diverse order of mammals originating in the Paleocene. The extant elephants represent two different genera, Loxodonta and Elephas, which share their origins in the late Miocene of Africa. Although similar in overall body plan, there are striking differences in cranial and dental morphology. General osteological descriptions of the extant elephants can be found in a variety of sources (de Blainville, 1845; Gregory, 1903; Weber, 1928; Osborn, 1942; Deraniyagala, 1951, 1955; Piveteau, 1958; Sikes, 1971; Maglio, 1972, 1973; Beden, 1979, Shoshani, 1996; and others), but most of these studies concentrate on either the African elephant, Loxodonta africana (Blumenbach, 1797), or the Asian elephant, Elephas maximus (Linnaeus, 1758). There has been no recent, detailed osteological analysis of the elephants in recent years and no comparative analysis of the two genera. As both species represent end-points of their respective evolutionary lineages, morphological characters that distinguish the two species are potentially useful for phylogenetic analysis of extinct members of each genus and other species within the Family Elephantidae.
There are many sources of morphological variation within-species that cannot be examined on fragmentary remains and minimal numbers of fossil species and specimens. This variation includes differences due to sexual dimorphism and subspecies differences, as well as ontogenetic differences, all of which can confound identification of phylogenetically useful characters. Thus, with large sample sizes, the extant elephants provide an excellent model for examining variation in morphology and the identification of a character data set for comparing fossil specimens.
The general differences between forest and savannah populations of Loxodonta have been noted for some time. The presence of pygmy elephants (Loxodonta pumilio) in African forests and in North American zoos (Noack, 1906; Hornaday, 1923) has also been mentioned in the literature, but many of these are now considered to have been immature specimens of the forest elephant, L. a. cyclotis (genetic analysis by Debruyne et al. (2003). Some of the recent genetic studies on Loxodonta have indicated significant differences which warrant division at the species level [i.e., L. africana (savanna or bush elephant) and L. cyclotis (forest elephant)] (Barriel et al., 1999; Roca et al., 2001; Comstock et al., 2002; Vogel, 2002), while others do not (Eggert et al., 2002; Nyakaana et al., 2002; Debruyne, 2005a, b). There is also some controversy about the results that are obtained from nuclear DNA versus mtDNA further complicating the issue (Roca et al., 2005).
The habitat fragmentation and geographic separation of populations of Elephas in southern Asia have obviously had an impact on the species. Barriel et al. (1999) found evidence in the cytochrome c region that E. maximus maximus from Sri Lanka and E. maximus indicus from continental India are closer genetically to each other than to other island populations of Elephas. Significant genetic differences were also found by Fernando et al. (2000) between specimens from Sri Lanka and the Asian mainland, as well as significant differences within regions in Sri Lanka itself. Vidya et al. (2005) found genetic differences between populations of E. maximus in southern India. There is also very new evidence of pygmy elephants in Borneo that are genetically distinct from E. maximus, but very little is known of these elephants as yet (Fernando et al., 2005).
The discrepancies in the genetic data suggest that other lines of evidence need to be examined. While general descriptions of the differences between forest and savanna elephants have been noted (such as differences in overall size, ear shape, tusks, etc.) (Frade, 1931, 1933; Allen, 1936; White, 1993), no detailed morphological analysis exists. There are also few descriptions of osteological differences between subspecies of Elephas.
Both species of living elephants are sexually dimorphic, with most of the differences due to size, as males are considerably larger. However, there are no detailed descriptions of the morphological differences due to sex in the modern elephants,
The purpose of this study is to provide a detailed comparison of the cranial–dental morphology of the two living elephants, with additional comparisons of the variation due to sex and to subspecies differences. This will also establish a character data set for comparing variation and relationships among fossil specimens of elephants, useful for identification and phylogenetic analysis, which will be presented in a future publication.
MATERIALS AND METHODS
The study sample includes 142 crania, 140 mandibles, and 504 molars of the modern elephants. Subspecies designations as listed on museum catalogues were retained. For L. africana, 99 crania, 103 mandibles, and 344 molars were examined. Within this collection, 5 crania, 5 mandibles, and 22 molars were attributed to L. africana cyclotis (Forest elephant). For E. maximus, 43 crania, 37 mandibles, and 160 molars were examined, most of which had no subspecies designation. There were 4 crania, 4 mandibles, and 18 molars attributed to E. maximus maximus, 3 crania, 3 mandibles, and 22 molars attributed to E. maximus indicus, and 1 cranium and mandible with teeth associated with E. maximus sumatranus.
Specimens were assigned a relative age category according to the molar in wear in the jaw, i.e., adult with M3 or M2 in wear, subadult with M1 in wear, juvenile with dM4-dM3, or infant with dM2/dM3. Broad age categories were used so that comparison with fossil species (where ontogenetic information is difficult to evaluate) is possible. Assignment to sex was done whenever possible. Although juvenile species were examined during the course of this study, the character descriptions presented here are based only on adult specimens. The characters described in this study are descriptive, although a few require estimates of angles, which were measured using a standard protractor.
In both elephants, the skull is extensively pneumatized, forming compartments for air spaces to lighten the skull, as well as a system of bony interior buttresses to provide strength and support (Shoshani, 1996). The overall shape of the cranium is very different in the two genera. The Loxodonta cranium is much more rounded in profile than Elephas, which is compressed anterior/posteriorly. This compression is accompanied by elongation or “heightening” of the cranium. Increased crown height of molars in the Asian elephant is accommodated by ventral depression of the palate and elevation of the occipital region. The only accommodation to increased crown height in L. africana is the ventral depression of the palate. Less elevation of the occipital region is needed because crown height in teeth of the African elephant is much lower than Asian elephant teeth.
The shape of the premaxillary tusk sheaths is quite different between the two genera as are the tusks themselves. In Loxodonta, the premaxillaries are massive and flaring laterally with the tusks directed down, outward, and slightly curved. Tusks in Elephas are directed downward and are much straighter before they begin to curve, resulting in premaxillaries that are closer together, parallel, and smaller overall. Specific differences in the crania are illustrated in Figs. 1–4.
The mandible in both elephants is composed of dense bone. Because of the unique way in which the mandible moves during mastication, its morphology is very important to the functional anatomy of the skull as a whole (Maglio, 1972). Differences in the mandible of the two living genera are quite dramatic and are directly related to the structure of the cranium. The mandible of the Asian elephant has a very wide corpus, almost “swollen” in appearance, compared with the African elephant, which has a relatively gracile corpus. Based on the position of the condylar process relative to the mandibular corpus, the center of gravity is anterior to the coronoid, and the ramus forms an acute angle to the corpus. In the African elephant, the center of gravity is more posterior, under or behind the coronoid, and the ramus forms a right angle to the corpus. If the mandible is placed on a horizontal surface, the mandible will balance itself at the center of gravity. In superior view, the mandible is V-shaped in Loxodonta and U-shaped in Elephas. The condyles are very different in shape, position, and orientation. In the African elephant, the condyles are rounded and the articular surface is level whereas in the Asian elephant, the condyles are elongated medio-laterally and the articular surface slopes down medially. Specific differences between the two extant elephants are illustrated in Figs. 5–7.
Molars have long been the principle element for identification of fossil elephants. They are quite diagnostic of genus and species and are easy to collect and curate. Elephants have a unique system of tooth eruption and replacement among the mammals. The milk molars are replaced by permanent molars, as in other mammals, but this replacement is continuous throughout their lifetime. There are only six sets of teeth, and the final wear of the last of these sets heralds the death of an individual (Cooke, 1947). Dental formula for both living species is incisors 1/0, canines 0/0, premolars 3/3, and molars 3/3 (Shoshani, 1996). All incisors have been lost except I2 which has elongated to form the tusks in the upper jaw. A number of extinct proboscideans had lower tusks, but these have been lost in more recent species. The canines have also been lost in all but the earliest species. Apart from the Sirenidae, elephants have the most derived tooth replacement pattern in mammals. The first three teeth, dM2, dM3, dM4, are referred to as “milk molars.” These teeth are molariform, but developmentally homologous to dp2, dp3, and dp4 in other mammals, but as they resemble permanent molars, they are commonly designated dM2, dM3, and dM4. There are no permanent premolars or deciduous M1. The last three molars (M1–M3) correspond to the permanent molars of other mammals (Shoshani, 1996).
The living elephants have very distinctive teeth, and many of these features can also be used to identify fossil specimens. The most striking difference between Loxodonta and Elephas is the shape of the enamel figure. Loxodonta is characterized by a “lozenge-shape” of these enamel figures, whereas enamel figures of Elephas have roughly parallel edges. African elephant teeth have much thicker enamel and there is usually no folding or crinkling of the enamel figure, although variation does occur. A number of specimens do exhibit enamel folding, but this feature is not consistent throughout the genus as it is in Elephas. Asian elephants also have much higher numbers of molar plates per tooth. Specific differences are illustrated in Figs. 8, 9.
Differentiating between the sexes has always been important in studies of modern elephant behavior and social structure, but there are many dimorphic features that can be described in morphology as well. Difference in molar size (particularly width) has been noted by Laws (1966) and Sikes (1967, 1971) in the African elephant. In a comparison of variation due to sexual dimorphism of the Asian elephant, Roth and Shoshani (1988) and Roth (1992) found that average lengths of teeth for males exceeded those for females (maximum of 19 mm difference). Males also had consistently wider teeth, with the greatest difference being 5.5 mm, but numbers of lamellae did not differ between the sexes. The differences in length and breadth are primarily due to the differing body size between the sexes, but did not significantly inflate variation when both sexes were in the same sample. Roth (1992) concluded that these differences were not large enough to affect variation, and that samples containing both males and females could be pooled. This is very important for fossil specimens for which there are no accurate information to use to separate the sexes other than overall size (Roth, 1992).
Sexual dimorphism in Loxodonta cranial morphology has been previously noted by Sikes (1971) and Beden (1979), but there are more differences which can be discussed in greater detail. Male crania are much larger overall and have much more robust muscle attachments and markings. The tusk sheaths are massive and flaring, and the inferior border is characterized by the presence of massive osteophytes or bony growths. In females, the sheaths are much smaller, related to smaller tusk size, and the inferior border is smooth (Figs. 10, 11). Differences in the mandible are not very pronounced, apart from overall size and robusticity of muscle markings (Fig. 12).
The cranium of the Asian elephant is generally smaller than the African elephant, but sexual dimorphism is just as pronounced. Apart from overall size, the most distinguishing character is the robusticity of the muscle markings and attachments, and the greater size of the parietal–occipital bosses in males. Females have very broad foreheads and the borders of the incisive fossa are much more rounded than in males. The nuchal crest in males is quite robust compared with females (Figs. 13, 14). Differences in the mandible of the Asian elephant are more difficult to identify, with overall size and robustness being the only obvious difference.
The use of subspecies designations is present in descriptions of both living elephants. While researchers such as Corse (1799), Lydekker (1907) and Deraniyagala (1955) discussed the different “races” of elephants, Shoshani et al. (1991) and others have discussed more concrete defining features, mostly limited to soft anatomy. Very little research has been done comparing the osteology and dentition of these subgroups with the exception of Eltringham (1991).
L. africana cyclotis is the forest elephant and is smaller than the savanna or bush elephant, L. africana africana. There has been much controversy over whether this subspecies should actually be a separate species from the savanna elephant (Matschie, 1900; Greenwell, 1992). The most striking difference is in overall body size, with males averaging 2.3 m at the shoulder and females averaging 2 m (Morrison-Scott, 1947).
In terms of specific morphological differences, the cranium is much smaller, has more gracile muscle markings, and a very rounded shape. Additional characteristics have been included in Figs. 15–18. The coronoid process is much shorter compared with corpus length than in the savannah elephant, and the muscle attachments are not as robust (Fig. 19). The forest elephant has fewer numbers of plates and relatively thicker enamel (Fig. 20). Tusks are much smaller than in L. africana africana and are straight and directly downward rather than curved up and out as in the bush elephant.
E. maximus has been divided into three subspecies based on geographic region. E. maximus maximus is found on the island of Sri Lanka, E. maximus indicus is found on mainland Asia, and E. maximus sumatranus is found on the island of Sumatra (Eltringham, 1991; Sukamar and Santapillai, 1996). Apart from very minor soft anatomical differences, E. maximus maximus and E. maximus indicus are difficult to separate based on their morphology, unlike the African elephant. E. maximus sumatranus does have a very different shape to the cranium, however, compared with the other two subspecies. The forehead and tusk sheaths are quite dished in lateral view and the occipital is flat. The entire skull is elongated and the forehead bosses are strongly divided and enlarged, giving the two-dome appearance on the forehead (Figs. 21, 22). Dental differences are not as pronounced between E. maximus maximus and E. maximus indicus, but there are distinctions from E. maximus sumatranus which include thickness of enamel, amplitude of enamel folding, and spacing of plates (Fig. 23).
Previous phylogenetic analyses of the Proboscidea have included characters that are too plesiomorphic for distinguishing the most recent elephant species (Tassy and Darlu, 1986, 1987; Tassy, 1988, 1990, 1996; Kalb and Mebrate, 1993; Kalb et al., 1996; Shoshani, 1996). Comparison of the cranium, dentition, and postcrania of the extant elephants reveals many diagnostic features which can be used to distinguish the two living genera and can also be applied to fossil species.
Elephants have long been cited as prime examples of anagenetic evolutionary trends, particularly in body size and dental characters. A robust character data set for the modern elephants is critical for testing the presence and direction of these trends. For example, the premaxillary tusk sheaths of the Asian elephant show a significant size decrease from their Pleistocene relatives, but the downward direction and roughly parallel shape has remained the same. Similarly, the tusk sheaths in the African elephant have reduced in size since the Pliocene, but are directed outward and show a tendency to flare away from each other, whereas the forest elephant's tusks are much reduced in size and are directed downward. This may be an adaptation to moving in dense forest cover as outward flaring tusks might hinder movement among trees.
In addition, analysis of evolutionary trends and change requires a basis for delimiting morphological variation at the species and subspecies level. Previous studies by Todd (1997, 2001, 2005, 2006) have discussed the high degree of diversity and variation in species of the Elephantidae. Many fossil species, such as the African Elephas recki for example, exhibit high degrees of variation, but much of this variation is due to lumping of specimens into taxa based on time period of their locality rather than on strict morphological analysis and comparison. The lack of a comparative suite of characters to base species comparisons on (although Maglio, 1973 and Beden, 1979 are excellent initial attempts to create a character set) has resulted in a confusing mixture of species without clear phylogenetic signals. This is further hampered by parallel evolution in many characters at different evolutionary rates in three separate lineages, Elephas, Loxodonta, and Mammuthus and by the fragmentary nature of fossil species collections (the emphasis on collecting teeth only rather than crania, mandibles, and postcrania) (Todd and Roth, 1996).
Thus, the modern elephants, with large and complete sample sizes and good age and sex classes, provide a basis for comparative study and character description which can then be applied to study variation in fossil species. In addition, variability due to sexual dimorphism can also be examined using the extant elephants and as a basis for analyzing fossil specimens.
Not only is variation due to sexual dimorphism important to consider when evaluating species level taxonomy (fossil and extant), variation due to geographic differences is also important to quantify. The morphological differences between the two African elephants are quite distinctive. Differences among the various subspecies of Asian elephant are not as pronounced, except in the Sumatran elephant, which is interesting given the distribution of Elephas on islands and in pockets of forest throughout southeast Asia. Defining species, however, is problematic under the Biological Species Concept, in which species are defined as potentially interbreeding populations that produce fertile offspring. There is overlap in range and territory in the two African species, and between mainland populations of the Asian elephant. Although they may be morphologically distinct, interbreeding may still be occurring. According to Debruyne (2005a), although there are two divergent molecular clades of African elephants, they are not isolated from each other reproductively and there has been recurrent interbreeding. For fossil organisms, however, genetic data are lacking and morphological distinctions are the only parameters for defining species. It is vital to define ranges of variation that are acceptable for a single species, thus data collected from the living species will assist in determining the extent of variation that could be expected in fossil population.
This suite of characters is currently in use to examine variation in fossil African Elephantidae. Characters which separate Loxodonta from Elephas can be identified in fossil representatives of each genus as far back as 4 ma and, in addition, have shed some light on the origin of the Mammuthus lineage in Africa (Todd, 2009). Revision of fossil species and genera based on identifiable and recognizable morphology are now possible, allowing for a new interpretation of the evolution of the family (Todd, 2009).
In summary, comparative analysis of the extant elephants provides a classic method for osteological studies and serves as the basis for a dataset to use in distinguishing between species and subspecies in both living and extinct elephant species. There are a suite of descriptive characters that have utility for identifying living and fossil representatives of each genus, species and subspecies, as well as characters which are useful for examining sexual dimorphism. Significant, identifiable differences were found between the forest and savannah African elephant, and between E. maximus maximus/indicus and E. maximus sumatranus, lending support to possible division at the species level under the morphological species concept. More study is needed, however, to quantify limits of variation for each of these species before elevating subspecies to species rank. In addition, as both genera of living elephants are currently listed as endangered species, increased numbers of species may also have implications for their conservation status.
The author is grateful to the following institutions for allowing access to their collections: the National Museum of Natural History in Washington, D.C.; the American Museum of Natural History in New York; the Peabody Museum in New Haven, CT; the Royal Ontario Museum in Toronto; the Office of the President of Kenya, and the staff in the Osteology Department at the National Museums of Kenya. Many thanks also to G. P. Aronsen and E. Sargis for providing comments on initial drafts of this manuscript. The author would like to dedicate this article to Dr. Jeheskel Shoshani, who was killed by a bus bomb in Addis Ababa on May 20, 2008. He was my cherished mentor, teacher, good friend, and champion of elephants, and the world is a smaller place without him.