Introduction

In the early 1980s, some spectacular mammoth fossils were recovered in northern China, but they have received little attention from the broader scientific community.

In May 1980, two partial skeletons of huge mammoths were discovered in the Lingquan Strip Mine of Zhalainuoer, Inner Mongolian Autonomous Region, China, at 49°20â€ČN, 117°35â€ČE. The radiocarbon date of coprolites found with the skeletons is 33,760 ± 1,700 years BP (PV-170) (Li et al. 1982, 1984; Li 1984; Liu and Li 1984). Today these skeletons are stored in the Inner Mongolian Museum collection, labelled Nos 1 and 2. A composite skeleton cast is exhibited at the same museum and at the Ibaraki Nature Museum in Japan.

Four years later, in March 1984, a third very large and near-complete skeleton, No. 3, was found in the same mine and horizon, at the bottom of the alluvium of an old river bed (Fig. 1). It is currently exhibited at the Zhalainuoer National Mine Museum, but no rigorous study has been made of it.

Fig. 1
Fig. 1
Full size image

The Zhalainuoer III mammoth location, Inner Mongolian Autonomous Region (asterisk)

Today these skeletons are attributed to the species Mammuthus sungari by most Chinese scholars (Wei et al. 2010). Nevertheless, according to Wei et al. (2010) remains previously assigned to M. sungari can be divided in two species: M. trogontherii and M. primigenius.

The Zhalainuoer mammoth fossils are among the richest M. trogontherii finds recovered from northern latitudes (Fig. 2). This study mainly focuses on the description of the whole skeleton found in 1984.

Fig. 2
Fig. 2
Full size image

Map of the Eurasian region showing the geographical distribution of complete/partial skeletons, and a selection of isolated remains. The Zhalainuoer mammoths are among the northernmost sites known. Data according to Kosintsev et al. (2004), Kubiak (2001), Palombo and Ferretii (2004), Van der Made (2001), Athanassiou (2011), Lister and Bahn (2007), Shpansky et al. (2008)

The first impression of the mounted skeleton is its enormous size, very well preserved and complete. Only a few parts are restored in plaster, for example the left fibula and fragments of other bones. The only large bone missing is the sternum. All long-bone epiphyses are fused; this is indicative of a fully grown mature individual. The original skull and tusks are preserved in a glass case. The tusks are clearly of Mammuthus morphology, and more detailed study indicates the skeleton belongs to the species M. trogontherii (Pohlig 1888) (see below) (Fig. 3).

Fig. 3
Fig. 3
Full size image

The Zhalainuoer III mammoth skeleton at the Zhalainuoer National Mine Museum. The original skull and tusks are preserved in the case behind

It is very rare to find such a complete and well-preserved steppe mammoth skeleton; in fact, the Zhalainuoer III (No. 3), is the best-preserved M. trogontherii skeleton ever found and surpasses the almost complete Kostolac mammoth skeleton discovered in 2009 in Pleistocene deposits adjacent to the Drmno opencast lignite mine in the Serbian Danube Basin (Lister et al. 2012). If the Zhalainuoer find was dated correctly, it is also the only near-complete skeleton of this species from the Late Pleistocene. The completeness of the Zhalainuoer III skeleton provides new information about the evolutionary status of some anatomical features such as caudal vertebrae count and the animal’s overall morphology.

The Late Pleistocene absolute age of the Zhalainuoer mammoths should be treated with caution. First, the radiocarbon dates were based on coprolites, but it is not possible to know if these came from the mammoths. Second, for dates obtained before the mid-1980s cleaning of collagen was usually insufficient, so they should be treated with caution, and re-dated where possible (Jacobi et al. 2006; Lister and Stuart 2013). However, a recent sedimentological study (Bin Xue et al. 2003) of the northern part of the Hulun Lake suggests that the sediment within which the Zhalainuoer mammoths fossils were found was formed during the Last Glaciation.

An isolated M3 of M. trogontherii SMD:8 found in Xi’an, was recently dated by the thermal ionization mass spectrometric U-series technique in the Radiogenic Isotope Laboratory, Centre for Microscopy and Microanalysis, University of Queensland, Australia; the result was 33.858–24.857 ka BP (Wei et al. 2010). This date can be regarded as reliable, so it seems that M. trogontherii survived into the Late Pleistocene in China. Other M. trogontherii material from Yuxian was dated to approximately 45 ka BP (Tong 2010) but no laboratory information was given.

Materials and methods

The Zhalainuoer III skeleton, exhibited at the Zhalainuoer National Mine Museum, was found in the 1984 excavations. The skull is badly damaged, the lower jaw is in very good condition, M3 and m3 are preserved, a large portion of the right tusk and a small portion of the mid part of the left tusk are preserved, but no measurements could be taken because they are stored in a glass case. The axial skeleton is represented by seven cervical, 19 thoracic, four lumbar, five sacral, and 23 caudal vertebrae and all 38 ribs. The appendicular skeleton is represented by partial scapulae, both humeri, both ulnae, the left complete radius, the proximal and distal ends of the right radius, all carpals, all metacarpals, most of the phalanges (15), except the distal phalanges, one sesamoid bone, the complete pelvis, both femora, both patellae, the complete right tibia, the proximal and distal end of the left tibia, the right fibula, all tarsals, all metatarsals except the first one of the right pes and the fourth one of the left pes, and practically all phalanges (16) except the distal ones, only two of which are preserved. The bones are brown, and natural preservation is good (Fig. 4).

Table 1 Measurements (mm) of mammoth teeth from Zhalainuoer
Full size table
Fig. 4
Fig. 4
Full size image

Damaged skull with perfectly preserved lower jaw of the Zhalainuoer III mammoth, in left lateral view

Measurements are given in millimetres and were taken with sliding callipers or flexible tape. Measurements taken on a cast of the lower jaws were after Beden (1980). The postcranial measurements were taken in accordance with Göhlich (1998), with additions from Lister (1996) and Mol et al. (1999). Individual age was determined by the state of wear and eruption of the molars, in accordance with Laws (1966) and Jachmann (1988), averaged across the preserved molars, and the state of fusion of the long-bone epiphyses in relation to those of the recent elephants Loxodonta africana and Elephas maximus, and the woolly mammoth M. primigenius (Lister 1999).

Table 2 Measurements (mm) of cervical vertebrae of the Zhalainuoer III mammoth
Full size table

To calculate the mammoth body mass, two methods were used:

  • Water displacement: on the basis of a rigorous multi-view skeletal restoration (Fig. 10) an accurate model, 1:15 scale, was sculpted by a paleo-life artist. Its volume was measured twice in a water-filled acrylic glass case with an aperture to a measuring cylinder with an accuracy of ±2 %.

    Table 3 Measurements (mm) of thoracic vertebrae of the Zhalainuoer III mammoth
    Full size table

  • Graphic double integration: GDI is a method proposed by Jerison (1973) for estimation of the volumes of endocasts from dorsal and lateral views. Hurlburt (1999) applied it to whole animals for the first time. The body or body part under research is modelled as an elliptical cylinder to determine the volume (Jerison 1973). The body volume of the Zhalainuoer III mammoth was calculated on the basis of GDI of the body parts: head, body trunk, forelimbs, hindlimbs, tusks, trunk, and tail. The volume of the two ears was estimated to be approximately 10 l, slightly less than that of Asian elephants (approx 13 l), each ear of which weighs 6.5 kg (Shoshani and Foley 2000).

    Table 4 Measurements (mm) of lumbar vertebrae of the Zhalainuoer III mammoth
    Full size table
    Fig. 5
    Fig. 5
    Full size image

    Right tusk of the Zhalainuoer III mammoth, in lateral view

A specific gravity of 0.99 was applied to the animal’s total volume to determine the mass (see below).

M. trogontherii skeletons used for comparison were: West Runton specimen (Lister and Stuart 2010), Kostolac specimen (Lister et al. 2012), Azov II and Azov I specimens (Bajgusheva and Titov 2010; Baigusheva et al. 2011), Pyatiryzhsk specimen (Shpansky et al. 2008a), LoussikĂĄ specimen (Athanassiou 2011), and Odessa specimen (Yatsko 1948; Shpansky et al. 2008b).

Table 5 Measurements (mm) of ribs of the Zhalainuoer III mammoth
Full size table

Description of the skeleton

Cranium

The cranium is quite damaged, although some parts are well preserved, including the occipital condyles and the right zygomatic arch. There is so much plaster over the whole cranium it is impossible to know how much real bone is underneath, but some parts can be clearly seen, in particular the orbital process of the frontal, the temporal line, the nuchal line, and the occipital. A large amount of the frontals and parietals is missing and the external nasal area is extensively reconstructed.

Table 6 Measurements (mm) of caudal vertebrae of the Zhalainuoer III mammoth
Full size table

The overall size is relatively small in relation to the postcranial material. On the basis of measurements taken from the mandible cast (see below) and by use of photographs, the length from the top of the cranium to the tip of the premaxilla was estimated to be 140 cm, which is comparable with the size of the Steinheim mammoth skull (Dietrich 1912). The cranium found in Azov (SW Russia) in 1964 (Baigusheva and Titov 2010) is much larger. Although the Azov cranium is complete, it is crushed and the shape cannot be accurately compared with that of the Zhalainuoer specimen. The other apparently complete cranium, from Gelsenkirchen (Germany), was found in 1926 (Siegfried 1956) but is extensively reconstructed. Further partial crania include the Nolhac specimen (Mol and Lacombat 2010), the superior part of which is entirely preserved. Its general form is similar to that of Zhalainuoer: high-domed, short, and with the occipital vertically developed.

Table 7 Measurements (mm) of scapulae of the Zhalainuoer III mammoth
Full size table

The premaxillary bones are very long and massively built, although they are restored at an angle too open relative to the maxilla and separated too much from each other.

Dentes

The skull and the two hemimandibles contain the hindmost molars, M3 and m3, indicative of an old individual. No measurements could be taken, but the total number of lamellae present on the complete third lower molars was counted as 19 (excluding talon and platelet), the anterior talon has signs of erosion and approximately 12 or 13 plates are worn. The plate count for m3 is the average for M. trogontherii (Lister and Joysey 1992; Lister and Sher 2001). An isolated upper third molar ZLNE001, recovered from the same location and horizon, has x19x plates, a lamellar frequency of 6.6 and 2.4 mm of enamel thickness (Wei et al. 2010), which also falls within the range for M. trogontherii. The other two skeletons from Zhalainuoer each have m2 and m3 present. One of the specimens has the m3 with the complete number of plates, 18, 3 mm of enamel thickness, and a lamellar frequency of seven, the other has +17 plates and a lamellar frequency of 6.5 (Cheng et al. 1982; Wei et al. 2010), which is likewise consistent with M. trogontherii (Table 1). Another m3 from Yuxian, dated to the Late Pleistocene, is assigned to M. trogontherii and has x18 plates (Tong 2010).

Incisivi

The right tusk is entirely preserved, but only a small portion of the central part of the left tusk was found. The right tusk is not very long but very robust and thick, and is curved and spirally twisted. It emerges approximately vertically from the alveolus before curving outward and then inward toward the tip. The overall shape and size are very similar to those of the West Runton right tusk (Lister and Stuart 2010, fig. 5), although it looks slightly more massive (Fig. 5).

Mandibula

The mandible is complete and in very good condition. Its morphology clearly indicates an advanced form of Mammuthus that can hardly be referred to a more primitive species such us M. meridionalis, because of the very short symphysis and mandibular rostrum of this specimen. The mandibular body is very short and robust, which is typical for M. trogontherii. The symphyseal process is nearly absent. The ramus is wide and barely dorso-ventrally rounded, the coronoid process is upright and lower than the condyle. No medial mental foramen (MMF) is observed, which is unlike the West Runton and Kostolac specimens. Medial foramina are present in 74 % of M. trogontherii mandibles (n = 19) (Lister et al. 2012; Ferretti and Debruyne 2011). The only measurements possible were taken from the cast, as follows: greatest height: 525 mm, greatest length (horizontal length of mandible taken parallel to the sagittal plane): 610 mm, greatest breadth: 625 mm. The size and general shape are very similar to those of the M. trogontherii mandibles from Pyatiryzhsk and West Runton (Shpansky et al. 2008a; Lister and Stuart 2010).

Columna vertebralis

The vertebral column is entirely preserved and includes all caudal vertebrae. This is a significant finding because it is the first tail known for M. trogontherii.

All the cervical vertebrae are in very good condition, and the anterior and posterior plates are fused to the bodies. The atlas is the widest, but the rest, including the axis, are similar in breadth. A spinous process is present from C3 to through C7; on C3 it is very short, but on the other neck vertebrae the spinous processes are increasingly taller (Table 2).

The number of thoracic vertebrae, 19, is the same as for the European M. trogontherii specimens with complete thoracic series, Kostolac (Serbia) and Novogeorgievsk (Ukraine). Most of the vertebrae are well preserved, but some of the vertebral processes are partially broken and have been restored in plaster. Several thoracic and lumbar vertebrae have bodies with unfused anterior and posterior plates. As is usual for mammoths, the spines of the first few thoracic vertebrae are very long. I was able to measure the height of the second vertebra in the anatomical position and found it to be 800 mm, substantially larger than that of the West Runton mammoth T3, which measures 680 mm (Lister and Stuart 2010). The longest are T3 and T4, behind which the lengths of the spines decrease dramatically, so that they bring about a sloping outline of the back (Table 3). The lumbar vertebrae (four) are thicker than the thoracic vertebrae. They are very similar in height and breadth, and the anterior and posterior plates are not totally fused with the bodies. The last has the right transverse process broken (Table 4).

The number of sacral vertebrae is five, as for the Kostolac mammoth (Lister et al. 2012). The West Runton M. trogontherii, in contrast, has six sacral vertebrae, and the Novogeorgievsk skeleton has four (Lister and Stuart 2010). In the Zhalainuoer individual the sacrum is quite small: the maximum length is approximately 450 mm and the breadth 270 mm, slightly smaller than that of the Kostolac specimen, the length of which is 500 mm (Lister et al. 2012).

All caudals are preserved. Some are damaged but restored with plaster. All vertebrae were inspected individually with a small but powerful light to ensure that all, including the restored ones, had the original fossil preserved. Number 3 is the first M. trogontherii skeleton found with all the tail vertebrae. This is very valuable because it enables us to trace the evolution of this feature in the mammoth lineage. The number of caudal vertebrae is 23, two more than for AMNH 14559, Borna and Beresovka M. primigenius specimens (Osborn 1942; Zalensky 1903) and generally less than for extant elephantids, which have 24–33 caudal vertebrae (Osborn 1942; Mol and van Essen 1992). Since in living species there is much variation of caudal number, some variation is expected in mammoths also, so the difference from M. primigenius is not very significant. All tail bones are compact and thick, which led to a very short tail, just one metre in the flesh (Fig. 10; Table 6). It has been suggested the short tail of M. primigenius is an environmental adaptation that prevents heat loss and frostbite because of very cold temperatures (Lister and Bahn 2007). The evidence from the Zhalainuoer individual suggests this adaptation had already been achieved by M. trogontherii (Tables 2, 3, 4, 5, 6).

Costae

All ribs are preserved, 38 in total. A few are slightly damaged but were accurately restored. The first is the most robust. It is wide, and, with its length of only 730 mm, very short compared with the 810 mm of the M. trogontherii individual from Pyatiryzhsk (Shpansky et al. 2008a). The second rib is of similar structure but longer. The others are narrower and elongated. The necks of the ribs are slim, become thicker along the body, and finish wider at their distal ends. The longest are found within the series R5–R12, R9 being the largest, it measures 1,400 mm along the outer curve and 1,180 mm in a straight line, nearly identical in size with the largest rib of the Pyatiryzhsk individual: 1,420 mm × 1,140 mm (Shpansky et al. 2008a). From R9 to R19 the ribs gradually shorten (Table 5).

Scapulae

The scapulae are incomplete, with apparently only the distal parts preserved. It is unclear whether any original bone is preserved in the proximal (dorsal) regions, as these are extensively reconstructed, so the maximum measured lengths should be regarded as estimates. The glenoid cavities are relatively small compared with those in other Eurasian individuals (Table 7).

Humeri

Both humeri are well preserved. They are similar in size to those of other male individuals of M. trogontherii and sightly stronger than the allegedly female humerus from Azov (Table 8). The right one is broken in the middle of the shaft, so the length of the left one is more credible. The tuberculum majus protrudes proximally beyond the caput. The capitulum and lateral epicondyle of the left humerus are broken but restored in plaster. The overall length is similar to analogous data obtained for other Mammuthus males from the Middle Pleistocene (Table 8).

Table 8 Measurements (mm) of humeri of the Zhalainuoer III mammoth
Full size table

Ulnae

The left ulna is entirely preserved and the right one nearly so, with only a small part of the distal shaft being damaged. The tuber olecrani is only slightly curved. Both ulnae have the epiphyses fused, which indicates a fully grown individual, because the last bones to fuse in elephants and mammoths are the distal ulna and radius (Roth 1984; Lister 1999). However, the fusion line is easily visible, which indicates that fusion occurred shortly before the death of the individual (Table 9).

Table 9 Measurements (mm) of ulnae of the Zhalainuoer III mammoth
Full size table

Radii

The left radius is complete, but a large part of the mid-shaft of the right radius is extensively restored. These limb bones are very long and robust compared with those of other M. trogontherii individuals (Table 10), especially the female one from Azov. The ratio of the minimum width of the diaphysis to the bone’s length (width/length × 100) is 7.71 for the Zhalainuoer III specimen and 5.98 for the female from Azov. As for the ulnae, the epiphyses are fused. In this case too, there are indications that fusion occurred shortly before the death of the individual (Table 11).

Table 10 Measurements (mm) of radii of the Zhalainuoer III mammoth
Full size table
Table 11 Measurements (mm) of carpal bones of the Zhalainuoer III mammoth
Full size table

Manus

The manus are very large and remarkably complete (Fig. 6). All the right carpals, metacarpals, and nearly all the phalanges are preserved. Only the distal phalanges are missing and one sesamoid bone is mounted. The left manus, also, is almost complete, although some digits are missing (Tables 12, 13).

Fig. 6
Fig. 6
Full size image

Schematic representation of the left and right manus of an elephant in dorsal view. The bones missing from the Zhalainuoer skeleton are shaded in grey

Table 12 Measurements (mm) of metacarpal bones of the Zhalainuoer III mammoth
Full size table
Table 13 Measurements (mm) of manus phalanges bones of the Zhalainuoer III mammoth
Full size table

Pelvis

Both halves are well preserved, especially the right half. The end of the pubis and the epiphysis of the left half are quite damaged. The size is enormous, and in this respect surpasses all other male individuals in this study (Table 14). The maximum breadth of the pelvic girdle in the anatomical position is 1,866 mm, but as the left ilium is slightly deformed and the size of the restored pubis is too large, the correct maximum pelvic width would be approximately 1,800 mm, which indicates the animal had a huge body (Fig. 10).

Table 14 Measurements (mm) of pelvic bones of the Zhalainuoer III mammoth
Full size table

The pelvic morphology of mammoths provides information enabling determination of sex (Lister 1996), because the pelvic aperture is relatively smaller in males.

The pelvic aperture is very wide (650 mm), substantially wider than the apertures of the male West Runton and Kostolac individuals’ pelves (Table 14). The ratio of the breadth of the pelvic aperture to the maximum breadth of the pelvis is 0.361 for the Zhalainuoer III specimen, 0.278 for West Runton, 0.323 for Kostolac, and 0.320 for Azov II, supposedly a female individual. The ratio for the Zhalainuoer mammoth is the highest observed in this study, which suggests a female individual. However, some male M. primigenius pelves from Ahlen, Gewande, Taymir, Gydan, and Liakhov (Lister 1996) have similar ratios. If these pelves are scaled to the pelvis breadth of Zhalainuoer III, the horizontal widths of the pelvic apertures of the said individuals would range from 620 to 690 mm.

The ratio of the left ilium width (from the tuber coxae to the point nearest to the pelvic aperture; 670 mm) to the maximum width of the pelvic aperture (650 mm) is 0.97, a value similar to that for several Mammuthus primigenius males from the study by Lister (1996). This ratio may, however, not be sufficiently accurate for determination of gender, because the ratio is similar for the Azov II female (0.955).

Unfortunately no measurements of the minimum width of the of ilium were taken, because the ratio of the minimum width of the ilium shaft to the maximum horizontal width of the pelvic girdle seems to be the best indicator of the gender of mammoths (Lister 1996). This ratio was found to be 0.120 for the Azov II individual, a value much smaller than those published for the Kostolac (0.146) and West Runton (0.139) male mammoths, so it is likely the Russian specimen was a female individual.

Male individuals also have a proportionally longer ilium and a more developed tuber coxae, so they have a relatively larger ilium. The length of the right ilium of the Azov II female is 1,000 mm, which is small compared with the 1,177 mm of Zhalainuoer III, because both have a similar maximum horizontal width of the pelvic girdle.

Femora

The right femur is complete whereas the left is quite damaged distally. The epiphyses are clearly fused. The proportions are nearly identical with those of the West Runton femur (Table 15).

Table 15 Measurements (mm) of femora of the Zhalainuoer III mammoth
Full size table

Patellae

Both patellae are preserved. Their size and proportions are very similar to those of patellae of other large individuals of M. trogontherii (Table 16).

Table 16 Measurements (mm) of patella of the Zhalainuoer III mammoth
Full size table

Tibiae

The right tibia is in very good condition, but of the left one only the proximal and distal parts are preserved. The proximal and distal epiphyses are completely fused. The size and proportions of the bone are very similar to those of other tibiae of large male individuals of the Middle Pleistocene Mammuthus (Table 17).

Table 17 Measurements (mm) of tibiae of the Zhalainuoer III mammoth
Full size table
Table 18 Measurements (mm) of fibulae of the Zhalainuoer III mammoth
Full size table

Fibulae

The right fibula is complete and the left was restored on the basis of the former. Their length is sightly smaller than that of the tibiae. The proximal and distal epiphyses are fused (Table 18).

Pedes

The pedes are much smaller than the manus. They also are almost complete with all tarsal bones preserved and only a few metatarsals and digits missing (Fig. 7; Tables 19, 20, 21).

Fig. 7
Fig. 7
Full size image

Schematic representation of the left and right pes of an elephant in dorsal view. The bones missing from the Zhalainuoer skeleton are shaded in grey

Taxonomic assignment

Three different candidates are plausible for identification: M. primigenius, M. sungari, and M. trogontherii.

M. primigenius must be taken into account because of the age of the deposits and the geographic coordinates of the site (Northeastern China), where fossils of this species have been identified (Dong et al. 1999). However, because of the scarcity of materials and the inadequate amount of study, it would be necessary to re-study the material found in this area for better species assignment. On the other hand, the size of the skeleton is too large for a woolly mammoth, because the skeletal shoulder height of this species rarely exceeds 300 cm. Nevertheless, a very large skeleton from Steinheim, originally identified as M. primigenius fraasi, is of a size similar to that of Zhalainuoer III. To judge by molar characteristics and size, the Steinheim skeleton seems to be an intermediate between M. trogontherii and M. primigenius, and, according to some authors (Lister and Stuart 2010), this individual should be referred to M. trogontherii. On the basis of molar morphology, M. primigenius may be excluded because the number of enamel plates, 19, is quite low for this species, although 20 plates can be found in some M. primigenius individuals (Lister and Sher 2001). The reported absence of an MMF in the mandible, could eliminate M. primigenius, because this foramen is always present in woolly mammoth mandibles (Ferretti and Debruyne 2011). Finally the combination of the large size of the skeleton and the available data on the molars of the Zhalainuoer I and II specimens (Cheng et al. 1982) and on other M3s recovered from the same location and horizon (Table 1), definitely suggest that M. primigenius can be excluded.

Table 19 Measurements (mm) of tarsal bones of the Zhalainuoer III mammoth
Full size table
Table 20 Measurements (mm) of metatarsal bones of the Zhalainuoer III mammoth
Full size table
Table 21 Measurements (mm) of pes phalanges of the Zhalainuoer III mammoth
Full size table

Mammuthus sungari is a species described by Chow et Chang (1959) (Mammoth skeletons unearthed from Zhalainuoer, Inner Mongolia Autonomous Region), and the name is widely used by Chinese scholars. The species is characterized by large size and a morphology more primitive than that of the woolly mammoth. In fact the Zhalainuoer III skeleton was, until recently, labelled as M. sungari, and the other two large individuals stored at the Inner Mongolian Museum were also attributed to this species. Wei et al. (2010) suggested that M. sungari is a junior synonym of M. trogontherii, and that the three skeletons found in Zhalainuoer during the 1980s should be referred to this species. According to observations made by this author, the diagnosis by Wei et al. is correct: all characteristics of the skeleton, including bone lengths and ratios (Fig. 8), point to the species M. trogontherii. As was suggested above, this means, contrary to the widely held view that it became extinct much earlier, M. trogontherii was able to survive into the Late Pleistocene. The last European record of M. trogontherii dates from approximately 200 ka BP (Lister and Stuart 2010). An M. trogontherii mandible with two m3, found in Yuxian, was recently dated to 45 ka BP (Tong 2010), and an M3 from Gaoling county was dated to 34–25 ka BP (Wei et al. 2010) (Tables 19, 20, 21).

Fig. 8
Fig. 8
Full size image

Ratios between limb bones and pelves of some M. trogontherii skeletons from Eurasia (lengths in mm)

Vital statistics

Shoulder height

Many authors (Lister and Stuart 2010; Kosintsev et al. 2004; Christiansen 2004; Baigusheva et al. 2011), define or calculate the shoulder height as the distance from the ground up to and including the projection of the neural spines of the thoracic vertebrae above the scapula. This concept was probably developed because nearly all mounted skeletons have the backbones above the scapula. However, if one observes walking living elephants of both extant genera, the scapulae can be seen rising and dropping several centimetres above the spines as they walk. Even in a standing position, both scapulae extend above the neural spines, more noticeably in the African elephant, which is high-shouldered as mammoths. Shoulder height should therefore be measured or calculated to the top of the scapula.

Many authors estimate shoulder height incorrectly because, on the basis of skeletons, they calculate shoulder height by regression of published shoulder heights from lengths of long bones, or they apply an incorrect percentage of the bones related to the shoulder height (Lister and Stuart 2010; Shpansky 2008a; Baigusheva et al. 2011; Athanassiou 2011). One problem is that published shoulder heights of different mounted skeletons often misrepresent the living animal because, as mentioned above, they are sometimes mounted with the scapulae in a very low position relative to the chest, which makes the skeleton too high. Another reason is that the forelimb is often mounted in an incorrect anatomical position. There is, also, sometimes too much space between the bones, which makes the skeleton taller. Osborn (1942) estimated skeletal shoulder height to the top of the scapula by adding the articular lengths of the thoracic limb bones, but the results of this method amount to slight over-representation because, although the forelimbs of derived proboscideans are columnar, the bones are not arranged in a totally straight line.

A direct measurement of shoulder height was obtained from the mounted Zhalainuoer III skeleton. The result was 361 cm, but this measurement is not accurate because the forelimb is not in an anatomical position and, as mounted, the manus are too flat. To calculate the shoulder height correctly, the forelimb was digitized on the basis of measurements and photographs, and rigorously restored in anatomical position, which gave 369 cm as the result (Fig. 9).

Fig. 9
Fig. 9
Full size image

Reconstruction of the forelimb of the Zhalainuoer III mammoth in the anatomical position. The reconstruction was based on measurements taken and on photographs

A widely quoted case is the attempt to calculate the shoulder height of the large steppe mammoth male from Azov (the Azov I skeleton). Garutt and Nikolskaya (1988) claimed 450 cm, a value which was based on the inaccurately mounted skeleton. Shpansky et al. (2008b) estimated 420 cm for this individual, Baigusheva et al. (2011) suggested 416 cm, and Kosintsev et al. (2004) proposed 400 cm as the correct height. On the basis of the preserved material, all these height estimates are exaggerated: the correct skeletal shoulder height should be approximately 375 cm (Larramendi, manuscript in preparation). When based on limb bone lengths, all published skeletal shoulder heights of steppe mammoth skeletons were overestimated, so they have been recalculated (Larramendi, manuscript in preparation) (Table 22).

Table 22 Vital statistics of the Zhalainuoer III mammoth
Full size table

In life, fossil proboscideans would have been taller than their mounted skeletons. To calculate how much taller, several aspects must be taken into account. First, the thickness of the skin: of extant elephants, the skin along the back of an E. maximus cow was measured at 3.2 cm (Shoshani et al. 1982); the skin on the back of the larger L. africana can range from 3.0 to 4.0 cm (Skinner et al. 2005). The much larger M. trogontherii is predicted to have skin at least 4.0 cm thick in the shoulder region. Roth (1990) calculated 10 cm of total skin thickness for the back and sole of the foot of an Asian elephant cow with a shoulder height of 227 cm, i.e. more than 4 % of the skeletal height. Not only must skin be added, but soft tissues, muscles, and a thin layer of cartilage in the joints of the bones must also be taken into account. The Zhalainuoer III skeleton was restored on the basis of all these, which resulted in a shoulder height of 389 cm in the flesh (Fig. 10; Table 22), which means an extra 5.5 % of the total skeletal shoulder height, nearly the same as the 6.33 % added by Osborn (1942).

Body mass and appearance

Body mass predictions for extinct proboscideans have been made by many authors, who applied different methods (Roth 1990; Fariña et al. 1998; Lister and Stuart 2010; Ferretti 2007, 2010; Palombo and Giovinazzo 2005, 2010; Christiansen 2004; Fortelius and Kappelman 1993; Athanassiou 2011; Shipman 1992). A commonly used technique is to compile data on body mass and shoulder height of living elephant populations, and to compare these with those available for extinct taxa (Roth 1990; Fortelius and Kappelman 1993; Ferretti 2007, 2010; Palombo 2010; Lister and Stuart 2010; Athanassiou 2011). Another method, based on bone dimensions, is to derive an allometric scaling formula from a large number of living examples and to apply it to the fossil form (Shipman 1992; Christiansen 2004).

These methods could be problematic, especially the first, because extinct forms were often larger than extant elephants and may have had different morphological proportions and significant differences in body mass/bone dimension relationships (Haynes 1991; Paul 1997). The best way to estimate the body mass of extinct forms is a volumetric method based on rigorous reconstruction of skeleton and muscles (Haynes 1991; Paul 1997). On the basis of bone measurements and photographs the M. trogontherii individual from Zhalainuoer was restored by use of image-editing computer software. The damaged skull was completely restored on the basis of the Nolhac specimen, although the preserved material suggests a sightly less developed parietal dome than the French specimen. The whole body reconstruction was made bone by bone, and by adding flesh, so that comparative anatomy and data on extant proboscideans could be applied (Larramendi, manuscript in preparation) (Fig. 10). Two methods were used to calculate the volume of the mammoth. The first method consisted in production of a physical model on an accurate scale to determine its water displacement; the second, graphic double integration (Jerison 1973; Hurlburt 1999), was applied directly to the digital restoration (Fig. 10). Both methods gave very similar results: a volume of approximately 10,500 l (WD = 10,650 l, and GDI = 10,450 l). If one assumes the animal had the same density as water, the volume implies a mass of 10.5 tons; however, although many extant animals have an overall density equal to that of water (Alexander 1985), some land mammals sink and others float. If one observes swimming elephants in a relaxed position, only about 1 % of the whole animal is seen to rise above the water, which indicates a density nearly equal to that of fresh water. The assumption of a specific gravity (SG) of 0.99 is therefore valid and results in a value of 10.4 tons for the body mass.

Fig. 10
Fig. 10
Full size image

Rigorous multiview skeletal restoration of Mammuthus trogontherii from Zhalainuoer

However, the restored physical model and the digital reconstruction were made without hair. The nature of the hair coat of M. trogontherii is unknown, but given the very short tail of the Zhalainuoer skeleton, this animal was evidently cold-adapted and may have carried a thick coat of hair. To estimate how much the hair could have weighed, partial pelts of several muskoxen, Ovibos moschatus, were weighed at the Quagga Associats studio at Barcelona. Muskox hair is good for comparison because its pelt has a structure which is similar to that of the woolly mammoth: very dense, short woollen underfur, and very long, straight guard hairs (Haynes 1991; Sokolov et al. 1982). The hair distribution around the body of M. primigenius is as follows: long hairs on the tail, decreasing lengths on foreleg, thigh, and belly, and the shortest hair on the tip of the trunk (Sokolov et al. 1982; Stewart 1979; Haynes 1991). The longest hairs on northern M. primigenius are up to 100 cm (Haynes 1991), approximately twice as long as those measured on the pelts of muskoxen, and probably longer than the hairs of the less cold-adapted M. trogontherii. The average weight of 1 m2 of muskox pelt is 2.5 kg. These pelts, with 2 mm thickness of attached skin, had been tanned and reduced, so their live weight would have differed. The skin alone would weigh approximately 2.2 kg if its specific gravity were the same as that of human skin (SG = 1.1) (Leider and Buncke 1954). However, because the skin was dried its SG should be considerably lower. If one assumes an SG of 0.9, the skin alone would weigh 1.8 kg and the hair alone 0.7 kg. Because the pelts were reduced, the real weight with all the hairs probably exceeded 1 kg. Finally, the hair was much heavier when fresh, and probably reached 2 kg/m2. The coat of the extinct species M. primigenius, with measured hairs approximately twice as long as those of Ovibos moschatus hairs measured in this study, would have weighed approximately 4 kg/m2. M. trogontherii may have had less and shorter hair, which weighed 3 kg/m2 on average, as suggested in this study. To determine the total skin surface of the Zhalainuoer III mammoth, the GDI formula was adapted to calculate the curved surface area as an elliptical cylinder, which resulted in a total skin surface of 37 m2. Approximately 110 kg of hair is therefore predicted for the Zhalainuoer III mammoth. This amounts to approximately 1 % of the total body mass.

The total body mass estimate for the Zhalainuoer mammoth, with the weight of the hair added, is 10.5 tons. Extant elephant body masses can vary up to 10 % within a year, depending on daily cycles of feeding, drinking, and defecating, and even more on seasonal changes and other factors (Lister and Stuart 2010). So estimated body mass should be taken as approximate only.

The body masses of the other steppe mammoths mentioned in this study were determined on the basis of volumetric calculations (Table 22). The body size of Zhalainuoer III is comparable with that other large steppe mammoths from the Middle Pleistocene, although the enormous specimen from Irtysh would have been 1 ton heavier (Table 22). After a review of the Inner Mongolian Museum collection, where the Zhalainuoer I and II skeletons are stored, it may be stated that Zhalainuoer III is probably the smallest of the three specimens. In the flesh the other two individuals would have reached and exceeded four metres at the shoulder. The bones of the two specimens were mixed, but can easily be differentiated because of the gigantic size of one of them. The largest specimen is colossal; the bones are at least 10 % larger than those of Zhalainuoer III, which indicates a shoulder height in the flesh of approximately 430 cm and body mass over 13.5 tons. It is worth noting that these individuals were considerably younger than Zhalainuoer III, because both still have M2/m2, so they would have continued growing to an even more gigantic size if they had lived longer.

Gender and age determination

The gender of the Zhalainuoer III mammoth is somewhat problematic for the following reasons:

  • Pelvic form. The best way to determine the gender is to analyse the pelvis (Lister 1996), but, as pointed out above, the pelvic girdle gives contradictory information. On the one hand, the proportions of the pelvic aperture may be indicative of a female individual, but, on the other, the shape of the iliac crest suggests a male individual, as do the ilium wing length proportions.

  • Body size. Both extant elephants are sexually dimorphic in body size. This feature is observed for woolly mammoths, M. primigenius, also (Lister and Stuart, 2010). The overall size of the Zhalainuoer skeleton is very large, comparable with that of the largest male specimens of M. trogontherii from Europe, Russia, and Kazakhstan (Table 22). The Azov II female individual is itself very large, but the Zhalainuoer III specimen is considerably larger, with a difference of 20 cm in shoulder height and approximately 1 ton in mass (Table 22). Limb bones are also proportionally more robust than those of the Azov II female. The degree of slenderness, defined as the ratio of the minimum diaphyseal width to the bone length, is 12.2 for the humerus of Zhalainuoer III, and 11.7 for that of Azov II. For the ulna the analogous values are 13.2 and 11.3, for the radius 7.7 and 6.0, for the femur 13.7 and 12.4, and for the tibia 14.9 and 15.2, respectively. The tibia ratio is higher for Azov II, but the proximal and distal width and depth are substantially larger for Zhalainuoer (Table 17). The body proportions are strongly indicative of the male gender.

  • Tusk size. Strong sexual dimorphism in tusk size (especially with regard to diameter) is observed among different species of mammoth (Vereshchagin and Tikhonov 1986). It was not possible to take measurements from the tusks, because they are stored in a glass case. They are very thick, however (personal observation): the maximum diameter must easily exceed 200 mm, so it surpasses the relatively thick tusk of Azov II, which measures 185 mm. This makes male gender likely.

  • Postcranial growth. Is known for extant elephants that complete fusion of long bone epiphyses occurs late among males, around the age of 40, and this is also observed in woolly mammoths. However, in females this process is complete at an approximate age of 25 years (Roth 1984; Haynes 1991; Lister 1999; Lister and Stuart 2010). The ontogenetic age of the Zhalainuoer III specimen is estimated to be approximately 53 years. The distal epiphyses of the radius and ulna of this specimen were fusing or had recently fused, which indicates male gender. Nevertheless, the Azov II female individual, which died at an estimated age of 50, does not have all limb bone epiphyses fused. This is probably because of its gigantic size and its estimated body mass of over 9.5 tons. This female steppe mammoth was three times heavier than females of both extant elephant taxa, so she would have needed more time to reach her huge size. This means that among very large female proboscideans the epiphyseal fusion process took as long as in male individuals, so postcranial growth may not be definitively indicative of gender.

Although the pelvic aperture proportions might be indicative of a female individual, the other characteristics, including the shape of the pelvis, render it likely the gender of the Zhalainuoer III mammoth was male.

The best way to estimate a mammoth’s ontogenetic age is to compare the stage of tooth eruption and wear with analogous data Laws (1966) compiled for a set of mandibles of African elephants (L. africana). His results were revised by Jachmann (1988). Of a total of 19 plates (excluding talon and platelet) of the m3s of the Zhalainuoer III mammoth, 13 were worn, equivalent to 8 or 9 worn plates out of 12 of a typical L. africana. This stage of wear corresponds to the group XXV of Laws’ scheme, which indicates an age at death of ~47 ± 2 African equivalent years. The large body size of the Zhalainuoer III mammoth and other steppe mammoths implies a longer lifespan than that of African elephants. The average body mass of a fully grown steppe mammoth is estimated to be approximately 11 tons (Table 22). If one uses the formula given by Blueweiss et al. (1978) (longevity in days proportional to body mass in grams0.17) and assumes a longevity of 60 years and a body mass of 5.5 tons for extant male L. africana, and a body mass of 11 tons for the species M. trogontherii, as suggested above, a longevity of 68 years is obtained for the latter. So the approximate age at death for the Zhalainuoer III mammoth is 47 × 68/60 = 53 years. This corresponds to the advanced state of epiphyseal fusion of the bones. The ages of the other two M. trogontherii individuals mentioned in this study were recalculated on the basis of the wear stages of the teeth and the lifespan obtained for this species (Table 22).

Conclusions

The discovery of three partial to nearly complete, probably Late Pleistocene, skeletons of M. trogontherii in Zhalainuoer is very important to our understanding of steppe mammoth distribution and evolution. The earliest remains of M. trogontherii recovered from North China (Wei and Lister 2005), and the material described here and by Wei et al. (2010), suggest that the steppe mammoth originated and disappeared in the same area, after having lived for more than 1.6 Ma. The large-scale cooling event between 40 and 30 ka BP pushed M. primigenius to southern latitudes, into the Jin’an area (approx. 36°N), Shandong Province (Takahashi et al. 2007). The time interval corresponds with the latest record of Chinese M. trogontherii. Therefore it was probably during this cooling event that the latest M. trogontherii made way for M. primigenius, which migrated into the area from NE Siberia (Wei et al. 2010). Hence, it is hard to envisage coexistence of the two species during the Late Pleistocene in northern China on the basis of the current data. On the other hand, it is important to remark that the plate number of 18 or 19 found for Late Pleistocene Chinese specimens, is the same as the average for western Eurasian early Middle Pleistocene populations; this might imply a stasis in molar plate number during more than 500,000 years throughout the entire Eurasian region, possibly because of ecological environment characteristics. The examined third skeleton from Zhalainuoer, the most complete skeleton of M. trogontherii ever found, has all the caudal vertebrae preserved, 23 in total. The find provides new information about the evolutionary history of the mammoth linage and about the morphology of M. trogontherii. The very short tail suggests that the Zhalainuoer mammoths lived in cold conditions, which in turn suggests the animal was covered with a thick coat of hair. There is a possibility of shortening of the tail of the last M. trogontherii populations in northern China compared with Middle Pleistocene specimens. The length of the eight caudal vertebrae preserved in the early Middle Pleistonece Kostolac skeleton, is approximately 80 cm (Lister et al. 2012), whereas the total length of the first eight proximal caudal vertebrae of the Zhalainuoer III individual is approximately 70 cm and the latter individual is larger in body size. The short tail is a derived trait of Zhalainuoer mammoths, but the large size, and probably the relatively lower skull dome and primitive molars, might be indicative of introgression which occurred after interbreeding of a more advanced immigrant steppe mammoth population with a more primitive resident one, and which led to the advanced immigrants becoming slightly more primitive in some respects. This might explain the archaic characteristics of the Zhalainouer mammoths. Of course the scarcity of the material means this hypothesis should be regarded with caution. The completeness of the Zhalainuoer III skeleton gives the opportunity to accurately restore the animal and to investigate the overall shape, shoulder height, and body mass in life. Finally, no significant variation of appendicular skeleton proportions is observed among steppe mammoth skeletons.