Quaternary International 255 (2012) 196e205 Contents lists available at ScienceDirect Quaternary International journal homepage: www.elsevier.com/locate/quaint Early tooth development, gestation, and season of birth in mammoths Adam N. Rountrey a, *, Daniel C. Fisher b, Alexei N. Tikhonov c, Pavel A. Kosintsev d, Pyotr A. Lazarev e, Gennady Boeskorov f, Bernard Buigues g a Museum of Paleontology, University of Michigan, 1109 Geddes Ave., Ann Arbor, MI 48109-1079, USA Department of Geological Sciences and Museum of Paleontology, University of Michigan, 1109 Geddes Ave., Ann Arbor, MI 48109-1079, USA c Zoological Institute, Russian Academy of Sciences, Universitetskaya nab. 1, Saint-Petersburg 199034, Russian Federation d Institute of Plant and Animal Ecology, Russian Academy of Sciences, 202 8 Marta St., 620144 Ekaterinburg, Russian Federation e Institute of Applied Ecology of the North, Kalandarashvili str. 5, 677891 Yakutsk, Russian Federation f Diamond and Precious Metals Geology Institute, Siberian Branch of the Russian Academy of Sciences, Lenina prospect 39, 677891 Yakutsk, Russian Federation g International Mammoth Committee, 2 av. de la Pelouse, 94160 Saint Mandè, France b a r t i c l e i n f o a b s t r a c t Article history: Available online 17 June 2011 The Yamal mammoth calf “Lyuba” and the Oimyakon mammoth calf are two recently collected, mummified woolly mammoth calves both dating to around 41 ka. They offer opportunities to study the earliest parts of mammoth life history through microstructural and compositional analyses of tooth dentin. Understanding mammoth paleobiology is critical for assessing how populations might have responded to ecological changes in the late Pleistocene, and such assessment will aid in determining the likely cause(s) of extinction. The state of tooth eruption and wear in Lyuba suggests that she died quite young. Prominent constrictions in the roots of the dP2s and dP3 correspond to neonatal lines in the dentin of each tooth, marking the time of birth. The neonatal line in dP2 allowed us to estimate the duration of prenatal development, determine Lyuba’s age at death, and place birth in the context of variation in isotopic and elemental composition of the dentin. Growth increments and seasonal variation in d15N indicate that Lyuba’s dP2 records over a year of prenatal development, suggesting duration of gestation was similar to that observed in elephants. Death occurred at about 30e35 days, and birth appears to have occurred in spring. The Oimyakon specimen represents an individual older (biologically) than Lyuba. The dP2s have been lost, and most plates of the dP3s are in wear. An unerupted permanent tusk is present, but no deciduous tusk is present in the single recovered premaxilla. The tusk exhibits a circumferential ridge at mid-length that represents the surface expression of a prominent growth line in the dentin. Based on its prominence and elevated dentin Zn/Ca in its vicinity, this is interpreted as a neonatal line. Its occurrence in the tusk is of interest because this implies initiation of permanent tusk mineralization months before birthd yet permanent tusks are not present in Lyuba. Differences in timing of tusk development may result from individual, population, or sexual variation. Growth increments in the tusk indicate death at an age of about 7 months with both d15N and d13C generally declining postnatally. The increment-based age at death is less than an estimate based on the state of cheek tooth eruption and wear (using extant elephants as references), implying that ages based on comparisons to elephants may be inaccurate. Counts of dentin growth increments combined with identification of seasonally varying aspects of dentin composition could be used to calibrate the cheek tooth aging system for Mammuthus primigenius. Ó 2011 Elsevier Ltd and INQUA. All rights reserved. 1. Introduction * Corresponding author. Fax: þ1 734 936 1380. E-mail addresses: arountre@umich.edu (A.N. Rountrey), dcfisher@umich.edu (D.C. Fisher), atikh@mail.ru (A.N. Tikhonov), kpa@ipae.uran.ru (P.A. Kosintsev), plazarev@yandex.ru (P.A. Lazarev), gboeskorov@mail.ru (G. Boeskorov), b.buigues@gmail.com (B. Buigues). 1040-6182/$ e see front matter Ó 2011 Elsevier Ltd and INQUA. All rights reserved. doi:10.1016/j.quaint.2011.06.006 Current views on the life history and autecology of arctic mammoths are largely based on analogy with extant elephants. However, arctic conditions would have required that these animals lived differently. Understanding the paleobiology of this species is critical for assessing how populations might have responded to A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 ecological changes in the late Pleistocene, and such assessment is needed to clarify the likely cause(s) of extinction. The mummified remains of mammoths preserved in permafrost offer unique opportunities for paleobiological study. Such specimens provide access not only to extremely well preserved bones and teeth, but often to skin, hair, muscle, fat, and gut contents. This study investigates the paleobiology of mammoths through analyses of the teeth of two mummified mammoth calves from Siberia. With samples from these young individuals, we seek to address questions regarding length of gestation, season of birth, and timing of tooth development. 1.1. Expected gestation length and season of birth We can make initial postulates of gestation length and season of birth using data from extant elephants and large arctic herbivores. For wild African elephants, mean gestation length is around 656 days (Moss, 1983), or 21.6 months. Data from captive African elephants are almost identical (e.g., 657  6 days, Meyer et al., 2004). Gestation length in captive Asian elephants is similar at 664  11 days (Meyer et al., 2004), or 21.8 months. While it is reasonable to assume that mammoths would have had similar gestation lengths (e.g., Guthrie, 2001), no direct evidence supporting or refuting this hypothesis has been presented. Season of birth depends on gestation length and season of conception. Conceptions occur year-round in African elephants, but there are seasonal peaks in conception rates that occur during rainy seasons when primary productivity is higher (Hanks, 1969; Laws, 1969; Williamson, 1976; Wittemyer et al., 2007). Wittemyer et al. (2007) found that most conceptions carried to term occurred following seasonal peaks in primary productivity (as measured by Normalized Difference Vegetation Index, NDVI), and given the 656day (21.6 month) gestation, births occurred close to the onsets of seasonal increases in primary productivity. They suggest that elephants exhibit condition-dependent estrus, and that conceptions tend to occur after elephants have consumed most of the resources of the high-productivity season. Arctic mammoths would have been subject to similar, although possibly more extreme seasonal variations in nutrient availability and might have timed the birth of calves to coincide with the beginning of the growing season. We look to caribou and muskoxen for comparative information on season of birth for arctic herbivores. In both species gestation is around 7.5 months, and births occur in spring (Rowell et al., 1993; Ropstad, 2000; Barboza and Parker, 2008). In muskoxen, calving can occur up to two months before new plant growth is available, making mothers dependent on body stores of nutrients for milk production (Adamczewski et al., 1997). Similarly, in reindeer and caribou, maternal body protein (as opposed to maternal dietary protein) is the source for around 91% of protein deposited in the calf in the first month after birth (Barboza and Parker, 2008). Therefore, although early growing season births in the arctic require the use of maternal fat and protein stores to support initial milk production, much of the burden of early lactation is on mothers during the summer, when more food is available. If we assume that mammoths, like elephants, exhibited condition-dependent estrus, the optimal time of year for breeding might have been from August to October as is the case in caribou and muskoxen (Adamczewski et al., 1997; Barboza and Parker, 2008). If we also assume that gestation length would have been at least roughly similar to that of extant elephants (i.e., between one and two years), and that births would have been timed to just before the start of the growing season (wMay) we would expect gestation length to have been around 20 months. 197 1.2. Timing of tooth development The state of tooth eruption and wear is often used to estimate the ages of elephants (Laws, 1966; Roth and Shoshani, 1988), and the aging systems derived from elephant data are also used to estimate age for mammoths. However, these systems have not been calibrated for mammoths, and Haynes (1991) has noted differences in apparent wear rates for mammoths compared to African elephants. We are interested not only in calibrating a system for use on mammoths, but also in determining the timing of formation of the teeth. Information on the timing of tooth development is particularly important for analyses in which the composition of teeth (e.g., stable isotopes, trace elements) is studied. Time of initiation of tooth mineralization has not been well studied in African elephant fetuses. Eales (1926) described the deciduous tusk (dI2) and upper and lower dP2, dP3, and dP4 (partial) of a fetal African elephant. The fetus had a body length (forehead to root of tail) of 21 cm. Using data from Hildebrandt et al. (2007), we can estimate the fetal age as 192 days. Eales (1931) also depicted dP2, dP3, and dP4 in an illustration of a mandible from a 9.5-cm African elephant fetus, which corresponds to a fetal age of about 140 days (Hildebrandt et al., 2007). Degree of mineralization is not mentioned. No teeth are present in the mandible of a 6.8-cm (125-day, Hildebrandt et al., 2007) Asian elephant fetus (Eales, 1931), but it is unknown whether the timing of early tooth development is sufficiently similar in Asian elephants to make this observation relevant. We can conservatively state that by 192 days of gestation in the African elephant, mineralized tissue is present in the upper and lower dP2, dP3, and dP4, and in the deciduous tusk. Crown formation in the deciduous tusk is completed after 16 months gestation in African elephants, and root formation is completed in the first three months of postnatal development (Raubenheimer et al., 1995). Mineralization of the permanent tusk begins before 17 months gestational age (Raubenheimer, 2000) and continues throughout life. 1.3. Using teeth to investigate gestation length and season of birth Dentin grows by accretion, and periodic growth lines occur at several temporal scales. In mammoths, first-order growth increments are annual in periodicity, second-order increments have a period of around seven days, and third-order increments are daily (Fisher, 2001). Using these growth increments, one can determine the period of time over which a tooth formed. Furthermore, in a tooth that was accreting dentin and enamel at birth, there are prominent growth lines known as neonatal lines, which mark the timing of birth in the dentin and enamel (Schour, 1936). Analysis of teeth containing neonatal lines allows determination of when the earliest prenatal mineralization occurred, and if the teeth were still forming at the time of death, it also allows determination of age at death. Teeth do not begin to mineralize at conception, so the period of prenatal dentin accretion in a mammoth tooth does not directly give us the gestation period of mammoths. By itself, the period of prenatal dentin accretion gives an underestimate of gestation length. In combination with comparative data from elephants, it permits estimation of relative gestation length. For example, the dP2s in African elephants have a prenatal mineralization period of about 15e17 months (for initiation between 140 and 192 days of gestation). If we observe a similar prenatal mineralization period in mammoth dP2s, we could infer that their gestation lengths were probably similar. Season of birth (and death) can be determined from teeth by taking advantage of the fact that dentin composition often varies predictably with season. Koch et al. (1989) showed that seasonal 198 A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 variation in d18O of precipitation leads to variation in the d18O of carbonate in proboscidean tusk dentin deposited at different times in the year. However, determining seasons using d18Ocarb records from Siberian mammoth tusks can be difficult because the patterns are sometimes reversed relative to expectations, or the data can exhibit no clear seasonal variation (Fox et al., 2007). We can look to other compositional measures for predictable seasonal variation. For example, Rountrey et al. (2007) observed annual oscillations in d15N and d13C in a dentin series from a juvenile mammoth tusk. Oscillations in d15N have also been observed in other juvenile and adult tusks (Rountrey, 2009; Gohmann et al., 2010) as well as in mammoth hairs (Iacumin et al., 2005). However, Iacumin et al. (2005) point out that it is difficult to assign particular seasons to the peaks or troughs in the d15N series. The d15N values of arctic plants cover a large range (w10&) due to differences in N source and mycorrhizal associations (Nadelhoffer et al., 1996), and seasonal preferences for certain plants could produce annual oscillations in tusk collagen d15N. Data for muskoxen from Banks Island, Canada (Larter and Nagy, 1995) provide some reference for seasonal variation in diet. While ruminant muskoxen are not an ideal analog for non-ruminant mammoths, it is still relevant to examine these patterns. The data indicate that muskoxen mostly feed on sedges, but in areas of high population density, willows can become an important component of diet in late spring. Willows tend to be depleted in 15N relative to sedges (Nadelhoffer et al., 1996). Thus, we might expect either little seasonal change in muskox tissue d15N if diet composition remains similar throughout the year, or perhaps a decrease in d15N in the late spring/early summer following the period when willow is consumed to greater degree. Intestinal contents from arctic mammoths indicate that they mostly fed on grasses and sedges, with woody plants usually making up a minor component (van Geel et al., 2008, van Geel et al., 2010). However, the intestinal contents of the Yukagir mammoth, which died in the early growing season, contained a substantial quantity of willow, suggesting that it may have been preferentially consumed in this season (van Geel et al., 2008) as is the case in the above-mentioned muskoxen (Larter and Nagy, 1995). Fisher et al. (2010) analyzed a dentin collagen series from the tusk of the Yukagir Mammoth for d13C and d15N and showed that death occurred during the rising phase of what appears to be an annual oscillation in d15N. This suggests that rising collagen d15N occurs in the winter and spring (Note that there are several possible ways of defining seasonal boundaries. We use the system most commonly used in the United States. In this system, Northern Hemisphere winter begins at the December solstice, and spring begins at the March equinox. Summer begins at the June solstice and ends at the September equinox. Using these definitions, arctic “spring” is ecologically much like winter. Actual green-up may not begin until near the start of summer in June.) In addition to the potential d15N effects from seasonal variation in diet, nutritional stress and amino acid routing during pregnancy could be important factors in producing d15N patterns. Enrichment in 15N due to recycling of body proteins has been observed in cases of nutritional stress (Hobson et al., 1993; Mekota et al., 2006). Mammoths may have experienced nutritional stress in the late winter and spring when diet quality would have been poor, and this could have led to elevated d15N values in dentin collagen produced during this time (e.g., Fisher et al., 2010). Late winter and spring may have also been a time when pregnant female mammoths would have relied more upon body protein than dietary protein for development of fetuses. In reindeer, 96% of fetal growth, which largely occurs in the winter and spring, is supported by maternal body protein (Barboza and Parker, 2008). As d15N of body protein is enriched in 15N over dietary protein (e.g., Schoeninger and DeNiro, 1984), dependence on maternal body protein would lead to 15N enrichment of the calf’s tissues during this time. Our expectation for a tooth that begins to form prenatally would be annual oscillations in d15N with the rising portion of each cycle representing late winter and spring. During this time, maternal nutritional stress and reliance on maternal protein stores to support calf growth might contribute to higher d15N values in the calf’s tissues. This is a preliminary determination based on the presently available data, but work on the d15N of individual amino acids in dentin collagen series might clarify the degree to which the annual oscillations in d15N reflect changes in diet or changes in nutritional stress. 2. Materials and methods 2.1. Mammoth calves In 2004, a partial carcass of a mummified mammoth calf was recovered from the Ol’chan Mine in the Oimyakonskii region, Yakutia, Russian Federation (Boeskorov et al., 2007). The carcass, which has been dated to 41,300  900 years before present (Boeskorov et al., 2007), consists of the head and the anterior portion of the thorax (Fig. 1B). At the time of discovery, the carcass was damaged by excavation equipment with one result being the separation of the premaxillae from the rest of the skull. Only the right premaxilla was recovered, and it contained a small, unerupted permanent tusk (Fig. 2). No deciduous tusk was present. The darkly stained permanent tusk is 7.6 cm in length (Rountrey et al., 2010), with a maximum circumference of 5.9 cm at the apical margin (i.e., the proximal margin; there is some splaying of the tusk near this margin, and the true maximum circumference may have been closer to 5.5 cm). The circumference 1 cm from the tip is 3.2 cm, and the length of the pulp cavity is 5.5 cm measured along the tusk axis. Enamel is present on the tusk, but much of it has spalled off (Fig. 2A). The margin of the area originally covered by enamel extends further ventrally than dorsally and is marked by a change in dentin surface texture (Fig. 2A; Rountrey et al., 2010). A prominent, circumferential ridge on the exposed dentin surface and a change in color occur 4.2 cm from the tip (Fig. 2A; Rountrey et al., 2010). The ridge is the surface expression of a well-defined growth line in the dentin, which we interpret as a neonatal line (Fig. 2B). Boeskorov et al. (2007) estimate the age at death to be between 1 and 1.5 years based on body size, and tusk/tooth development. With the exception of the tusk, no teeth were extracted, so their assessment was carried out using data from a CT scan. In the maxillae, dP3e4 are visible in CT images, as are their counterparts in the mandible (Boeskorov et al., 2007). The dP2s have been shed. The dP3s show wear extending back to the penultimate plate (the dP3s have eight or nine plates), and no wear is apparent on the developing dP4s (Boeskorov et al., 2007). Interplate cementum deposition in dP4s is complete only between anterior plates. In 2007, another mammoth calf carcass was discovered on a bank of the Yuribei River, Yamal Peninsula, Yamalo-Nenets Autonomous Okrug, Russian Federation. The find is the most complete specimen of a mammoth ever discovered (Fig. 1A). The female mammoth calf nicknamed “Lyuba” has been dated (AMS, on bone collagen) to 41,910 þ 550/ 450 radiocarbon years before present (Kosintsev et al., 2010). In 2008, the carcass was partially thawed to allow collection of samples for histological, molecular, palynological, microbiological, and biogeochemical analysis. Teeth were extracted from the left premaxilla, maxilla, and hemimandible at this time. The state of dental progression in Lyuba is not as advanced as in the Oimyakon calf (Fig. 3). No permanent tusks are present, but the left deciduous tusk (dI2) is completely formed (i.e., the root is A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 199 Fig. 1. External views of the two mammoth calves studied. A) “Lyuba”- carcass recovered in 2007 on the Yamal Peninsula, Yamalo-Nenets Autonomous Okrug, Russia; B) carcass recovered in 2004 in the Oimyakonskii region, Yakutia, Russia. Due to the angle at which photograph B was taken, the scale bar is only accurate close to Oimyakon’s head. closed) and was beginning to undergo resorption prior to death (Fig. 4). The dP2s are present and show little occlusal wear (Fig. 4). Some enamel/dentin has been lost due to spallation (during drying), and some areas of enamel exhibit dissolution pits or incomplete mineralization (Fig. 4). The crown of dP2 consists of five lophs, yet cementum does not fill the space between these. The crown of dP2 consists of four lophs increasing in width distally and a distal set of minor cuspules. Cementum does not fill gaps between lophs. The bifurcate roots of the dP2s are well developed, but they remain open, indicating that root formation was ongoing at the time of death. The roots of dP2 and dP2 also show accentuated constrictions 1.5 and 3 mm from the apical margins respectively (Fig. 4). The original distance from the apical margin in dP2 was probably greater, but some of the apical material appears to have been lost to decay. These constrictions are surface expressions of prominent growth lines in the dentin, which we interpret as neonatal lines. The general form of the dP3 crown is complete, but enamel formation is incomplete. The crown consists of nine lophs (with Fig. 2. Right permanent tusk (I2) from the Oimyakon calf. A) Lateral aspect of tusk showing retained enamel, cementum, and a prominent ridge in the dentin at midlength; B) lateral surface of medial portion of tusk after longitudinal sectioning. Note presence of the neonatal line (NnL) and its correspondence to the ridge on the surface. a small set of mesial cuspules) with no cementum filling the gaps between lophs (Fig. 4). There is no occlusal wear. Root formation is in an early stage with no furcation, and there is a constriction near the apical margin similar to that observed in the dP2s (Fig. 4). Crown formation in dP3 is incomplete. The crown consists of eight lophs (with small sets of mesial and distal cuspules), but the distal three lophs are incompletely formed and are not joined to the rest of the tooth (Fig. 4). Cementum does not fill gaps between lophs, and there are no signs of wear on this tooth. dP4s are present, but crown formation is incomplete, and at least in the dP4s, lophs are not yet fused (Fig. 3). There are constrictions near the apical ends of lophs, and prominent lines in the latest-formed dentin. We interpret these as indicating the presence of neonatal lines in the lophs of dP4. Using the African elephant tooth eruption/wear aging system of Laws (1966), Lyuba appears to have died at an age of between 0 and 0.5 years. Data from Asian elephants indicates she was less than 1.5 years of age (Roth and Shoshani, 1988). Fig. 3. State of tooth development and eruption in the right maxilla and hemimandible of Lyuba. Anterior is to the right. A deciduous tusk is present in the premaxilla, but it is not visible in this slice. The dP4 plates are unfused and have shifted within the crypt. The round, high-density structures within and around bones are likely vivianite nodules, which were found in abundance during the dissection (Fisher et al., this issue). 9.4-mm thick slice from CT data. CT data courtesy of GE Healthcare Institute. 200 A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 Fig. 4. Teeth extracted from Lyuba’s left premaxilla, maxilla, and hemimandible. Fragments of dP4s not shown. Anterior is to the left. Arrows mark constrictions in the roots that correspond to neonatal lines in the dentin. Inset: Occlusal aspect of dP2s showing limited occlusal wear. Regions of exposed dentin are probably the result of spalling, dissolution pitting, or incomplete mineralization, rather than wear. b-buccal, d-distal, l-lingual, m-mesial. 2.2. Thin section production The dP2 from Lyuba and the I2 from the Oimyakon calf were chosen for this analysis. The I2 was first cut longitudinally (Fig. 2B) using an Isomet low-speed diamond wafering saw (Buehler, Lake Bluff, Illinois). Subsequent transverse cuts on one half of the tusk were also made using this saw. Thin sections from transverse and longitudinal cut surfaces were made from these segments such that each growth increment in the tusk (with the exception of those in the tip of the tusk) was visible in at least two thin sections. This allowed correlation of growth increments among sections and compilation of a complete series of growth increment thicknesses. Photomicrographs of thin sections were taken with crossed polars on a petrographic microscope. Measurements of growth increments were taken from the photomicrographs using ImageJ (Rasband, 1997e2009) and a custom plugin. The dP2 was bisected along a vertical mesialedistal plane using the Isomet saw. A thin section was made from one of the resulting cut surfaces. Imaging and measurement were carried out as described in the preceding paragraph. 2.3. Serial sampling of dentin The small size of the Oimyakon tusk required that a combination of techniques be used to obtain dentin samples for compositional analyses. Particularly near the tip of the tusk, there is insufficient material to permit the alternating milling pattern described in Rountrey et al. (2007). However, the dentin of this tusk contains many concentric fractures that formed along prominent growth lines, and these fractures often extend for some length along the tusk, allowing correlation of fractures in different transverse segments. These fractures isolate blocks of dentin representing fixed intervals of time. Near the tip, blocks of dentin bordered by fractures were extracted using a scalpel and forceps. The specific growth increments contained in these blocks could be determined by noting the bounding fractures and correlating them to fractures visible in the thin sections. Powder samples were also obtained by milling with a fixed dental drill and 0.5-mm carbide burr. In addition, some alternating block/powder samples were taken farther away from the tip of the tusk. The small size and complex geometry of dentin increments in Lyuba’s dP2 also necessitated an unusual sampling strategy involving peeling off layers of dentin. This type of sampling would not have been possible if not for the incipient demineralization resulting from Lyuba’s diagenetic history (Fisher et al., this issue). Following the initial vertical cut, two additional cuts were made to isolate a portion of the distal wall of the crown, from the pulp cavity to the enamel. The purpose of isolating this part of the tooth was to obtain as complete as possible a dentin sequence in which the growth laminae curved mostly in one direction, as it was expected that attempting to separate layers of dentin curving strongly in two directions might lead to additional fractures or tears. Natural fractures in the dentin were accentuated by repeatedly submerging the fragment in w70% ethanol, then allowing much of the water and A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 ethanol to evaporate prior to submerging again. This led to propagation of existing fractures. Layers of dentin were then peeled from the tooth using a spatula and forceps. The tooth fragment was scanned on a high-resolution flatbed scanner after removal of each sample, effectively recording the location of each sample. Use of this method produced relatively large (w7.5 mg) yet thin dentin sheets, permitting analysis of this tooth at a higher temporal resolution than would have been possible using milling methods. 2.4. Preparation of samples Dentin blocks from the Oimyakon tusk were demineralized in 10 mL of 0.5 M HCl at 4  C for 36 h. The collagenous blocks remaining after demineralization were rinsed five times with 10 mL of ultrapure water. To remove lipids from the blocks (oil-based cutting lubricant was used for Isomet cuts), 10 mL of 2:1 chloroformemethanol was added to each vial, and the vials were placed in an ultrasonicator for 30 min. The chloroformemethanol was discarded, and samples were rinsed five times with 10 mL of ultrapure water. Samples were then freeze-dried and weighed. The d13C and d15N analyses of 1.5-mg (0.1 mg) subsamples were conducted at the Isotope Ratio Mass Spectrometry Laboratory, University of California, Santa Cruz. For elemental analysis, powder samples from the Oimyakon tusk were used. 1.0-mg subsamples were weighed into metal-free 1.5-mL microcentrifuge tubes. 0.5 mL of ultrapure water was added to each tube. Tubes were agitated and left at room temperature for w30 min. Following centrifugation and removal of the supernatant, samples were freeze-dried overnight and re-weighed. 1.0 mL of 0.5 M TraceMetal grade HCl was added to each tube, and samples were refrigerated at 5  C for 24 h. Tubes were then centrifuged, and the supernatant was removed for analysis. The remaining collagen was rinsed, centrifuged, freeze-dried, and weighed. Subsamples of solutions were diluted with 1% HNO3 (1 ppb indium internal standard) to target Ca concentrations between 10 and 14 ppm. Concentrations of Ca and Zn were determined using a Finnigan Element ICP-MS at the University of Michigan, Ann Arbor. For the peeled dentin sheets from Lyuba’s dP2, a different procedure was followed. Subsamples were freeze-dried and weighed into 1.5-mL metal-free microcentrifuge tubes. 1.0 mL of 0.5 M TraceMetal grade HCl was added to each sample. After 24 h at 4  C, the samples were centrifuged, and the supernatant was pipetted off and retained for elemental analysis (ICP-MS). Demineralized dentin sheets were rinsed five times in ultrapure water and then freeze-dried overnight. Analyses for d13C and d15N were performed at the Isotope Ratio Mass Spectrometry Laboratory at the University of California, Santa Cruz. A chloroform/methanol step, as is usually included in similar preparations to remove lipids, was not conducted on these samples. Powdered dentin samples from Lyuba’s dI2 were also analyzed (results not reported here), and the powdered nature of the samples required that they remain in centrifuge tubes, which were not appropriate for use with chloroform. To make results from the powdered and sheet samples comparable, neither set was treated with chloroform/methanol. The lipid content of dentin is low (around 1.7 wt% of demineralized dentin (Wuthier, 1984), and 70% ethanol rather than oil was used as the cutting lubricant for this tooth. Therefore, lipid contamination should be minimal. 3. Results and discussion 3.1. Identification of neonatal lines Much of our interpretation of data from these teeth depends on the correct identification of neonatal lines in the dentin. The lines we selected are the most prominent growth lines in these teeth, 201 and in both individuals, ridges and/or constrictions in the roots occur where the increments reach the external surfaces. Furthermore, the root constrictions clearly represent systemic rather than local disturbances in Lyuba’s teeth as they appear in all teeth with sufficient root development to express the constrictions (Fig. 4). We cannot make the same claim for the Oimyakon tusk, because the tusk is the only tooth that has been extracted at this time. Our identifications are supported by the dentin Zn/Ca data. In series from both individuals, dentin samples taken near the proposed neonatal lines have elevated Zn/Ca (Figs. 5A and 6A). Fetal (cord blood) serum Zn concentrations increase through gestation in humans (Jeswani and Vani, 1991), and Zn concentrations in human milk are highest in the early stages of lactation (Vuori and Kuitunen, 1979). These factors could lead to elevated dentin and/ or enamel Zn/Ca around the time of birth. In a study of human enamel, Ca-normalized Zn concentration was found to be elevated near the neonatal line (Kang et al., 2004). However, additional studies on dentin Zn/Ca are needed to clarify whether or not elevated Zn/Ca consistently occurs in neonatal dentin. 3.2. Age at death and initiation of mineralization In principle, one could count the number of daily growth increments in dentin from the neonatal line to the pulp cavity surface and precisely determine the age at death. However, the clarity of daily growth lines is often variable in different areas of the tooth, making this difficult. In cases where it is not possible to count all daily growth lines, we use counts of second-order (wweekly) growth increments. We also determine mean dentin accretion rate in areas where daily lines are visible and then use the rates to calculate the durations of prenatal and postnatal dentin formation. For Oimyakon, second-order increments are visible in most areas of the tusk, but daily increments are difficult to observe. Second-order increments were also difficult to identify near the tip of the tusk in the early-forming dentin. In one area where secondorder increments and daily increments can be seen together, the second-order increments appear to occur every six or seven days, and we use a seven-day periodicity for our calculations. We counted 75 second-order increments in the dentin of the tusk, and 32 of these occur from the neonatal line to the pulp cavity. The second-order increments, thus, indicate an age at death of about 7.4 months and an onset of mineralization at about 9.9 months before birth. The total second-order increment count differs from that reported by Rountrey et al. (2010; 58 second-order increments) due to inclusion of additional increments from a longitudinal section of the tusk tip. Rountrey et al. (2010) utilized only transverse sections, and the earliest growth increments in the tip of the tusk were not visible in the distal-most transverse section. The number of postnatal second-order increments is one less than reported by Rountrey et al. (2010) because of a change in counting convention (i.e., which of two bounding increments actually contains the neonatal line). The accretion rate based on measurements of 16 clear, daily increments in postnatal dentin is 16  4 mm/day. If this rate is used to estimate age at death based on postnatal dentin thickness, the result is about 7.4 (þ2.4, 1.5) months, just as we estimated from second-order increments. Applying the daily-rate method to prenatal dentin, we would estimate onset of mineralization at about 6.7 (þ2.2, 1.3) months before birth rather than the 9.9 months estimated from second-order increments. This difference could result from a difference in pre- and postnatal accretion rates, difficulty identifying second-order increments in the earliestformed dentin, or a change in second-order increment periodicity. 202 A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 Fig. 5. Elemental and isotopic series from I2 of the Oimyakon calf. Sample locations are tied to second-order (wweekly) growth increments in the dentin and are arranged chronologically from left to right. Horizontal error bars show the increments included in each sample. Vertical error bars are 1s based on standards run with the samples. Vertical dashed line represents position of the neonatal line (time of birth). A) Zn/Ca; B) collagen d15N; C) collagen d13C. B and C include data from Rountrey et al. (2010). While little information is available on the timing of elephant tusk development, Raubenheimer (2000) shows that a mineralized permanent tusk is visible in a radiograph of a 53-kg (17-month) fetus, and Anthony (1933) notes the presence of permanent tusks around the time of birth. Based on Raubenheimer’s (2000) Fig. 6. Elemental and isotopic series from dP2 of “Lyuba”. Sample locations are arranged chronologically from left to right. Vertical error bars are 1s based on standards run with the samples. Vertical dashed line represents position of the neonatal line (time of birth). A) Zn/Ca; B) collagen d15N; C) collagen d13C. observation, 5 months of prenatally produced dentin should be present in an elephant tusk. This is similar to our accretion-rate estimate of 6.7 months for the Oimyakon calf. The state of cheek tooth eruption and wear at death (w7.4 months of age) in the Oimyakon calf appears advanced relative to expectations based on extant elephants. Oimyakon’s dentition approximates Laws (1966) age group IV with an age of around 2 years. Using the system of Roth and Shoshani (1988), we would A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 estimate an age between 1.5 and 2.5 years. This suggests that cheek tooth wear and loss in Oimyakon was considerably faster than in extant elephants. Alternatively, the prominent growth line and constriction in the tusk might not represent a neonatal line. If permanent tusk mineralization normally begins after birth in mammoths (at 4e6 months), as suggested by Maschenko (2002), then Oimyakon might have died at around 19 months of age (5 months for age of onset of mineralization, plus all of the time in Oimyakon’s tusk, 6.7 þ 7.4 months). This would be more in agreement with the age based on tooth wear. However, the unique, prominent nature of the inferred neonatal line occurring 7.4 months before death and its association with elevated dentin Zn/Ca represent strong evidence that it does mark the time of birth. These two hypotheses on age at death could be evaluated through analysis of a dP3, which begins to mineralize before birth in mammoths and extant elephants. If death occurred at 19 months of age, there should be two prominent growth lines in the dentin of dP3 e one representing birth around 19 months before death and one corresponding to the prominent growth line in the tusk around 7.4 months before death. If death occurred at 7.4 months of age, only a single prominent growth line produced 7.4 months before death should be present. In Lyuba’s dP2, daily growth increments are visible in many areas, but second-order increments are not clear. Direct counts of postnatal daily increments by multiple observers gave ages at death of 30e32 days. In addition, we measured the total thickness of postnatal dentin and divided this by the mean daily accretion rate (based on a clear series of 9 daily growth increments in the same region). The mean daily accretion rate for this area was 8.9 mm/day, and the total thickness of postnatal dentin was 312 mm, yielding an age at death of 35 days (Fig. 7). We can approximate the onset of mineralization in this tooth in a similar manner. The mean accretion rate based on 123 prenatal daily increments is 8.2 mm/day, and the total thickness of prenatal dentin in the posterior loph and root is 3273 mm. This suggests that mineralization in this loph began approximately 400 days (just over 13 months) before birth. However, the standard deviation of mean accretion rate from five sampled areas is 1.5 mm/day, so our estimate of initiation is between 11 and 16 months before birth. Fig. 7. Cross-polarized light photomicrograph of dentin near the pulp cavity surface of Lyuba’s dP2. Dentin between the neonatal line (birth) and pulp cavity surface (death) represents postnatal tissue. Nine relatively clear daily growth increments occur over a thickness of 80 mm, giving a postnatal accretion rate of about 8.9 mm/day. The total thickness of postnatal dentin is 312 mm, giving an age at death of w35 days. Direct counts of daily increments give ages at death ranging from 30 to 32 days. The dark line below the neonatal line is a fracture in the dentin. 203 In African elephants, mineralization of the dP2s begins by 192 days of gestation (6.3 months; see Section 1.2), so one would expect to see at least 15 months represented in the prenatal dentin of this tooth. Results from Lyuba indicate a similar prenatal development period for dP2. Age at death from growth increments is in agreement with that determined using the Laws (1966) African elephant system (0e0.5 years). This suggests roughly similar developmental timing, and it supports a hypothesis that gestation length in mammoths did not differ greatly from that of extant elephants. The absence of permanent tusks in Lyuba shows that the timing of permanent tusk development differed from the timing in the Oimyakon calf and extant African elephants. While it is possible that Lyuba would never have developed tusks, the incidence of tusklessness is low (4.16% for females) in a population of African elephants that has not been subjected to hunting pressures (Raubenheimer, 2000), and we are unaware of a bilaterally tuskless juvenile or adult mammoth specimen. Thus, it is likely that Lyuba’s permanent tusks would have developed later. The natural variation in onset of tusk mineralization in extant elephants is unknown, and it is possible that some individuals do not begin to develop tusks until after birth. The differences between Lyuba and the Oimyakon calf may reflect such natural variation among individuals or populations, or between sexes. Our analysis of Lyuba is consistent with Maschenko’s (2002) proposition that most mammoths did not begin to develop permanent tusks until a few months after birth. Why the timing of tusk development would differ between mammoths and extant elephants is an interesting question. Our data from Lyuba’s dP2 do not preclude the possibility that mammoth gestation period was a few months shorter than that of extant elephants, and if the timing of initial tusk mineralization relative to conception were similar in mammoths and elephants, we might expect tusks to begin to mineralize after birth in mammoths. Alternatively, gestation might be essentially the same, but with mineralization of the permanent tusk occurring later (even relative to conception) in mammoths than in elephants. 3.3. d15N, d13C, and season of birth/death Eleven dentin collagen samples represent the time recorded in the Oimyakon tusk (values for nine of the samples were reported in Rountrey et al., 2010). The sequence has five prenatal samples and six postnatal samples (two of these are pulp surface samples and represent the same interval). Analytical errors for d13C and d15N are conservatively <0.1& (1s) based on drift- and samplemass-corrected standards run with the samples. The C/N (atomic ratio) for the samples ranged from 3.2 to 3.3. This is within the range (2.9e3.6) expected for collagen extracted from modern bones (DeNiro, 1985). The mean value obtained for collagen extracted from modern elephant tusk dentin was also 3.2 (Rountrey, 2009). d15N rises in the early part of the series and reaches a peak in the sample representing the first postnatal dentin (Fig. 5B). Values then decline until death. As mentioned in Section 1.3, we expect to see rising d15N in a fetus during late winter and spring assuming maternal nutritional stress and use of maternal body protein for development of the fetus when dietary protein is limited. In this case, birth occurred following a rising trend in d15N suggesting that it occurred in spring. The high values after birth might be due to continued reliance of the mother on her protein stores for early milk production and/or trophic level effects due to nursing. The subsequent decrease in d15N could result from decreased use of maternal body protein for milk production during the summer when higher quality plant foods would be available. The same declining trend could also result if the calf was beginning to ingest plant material in the summer and fall. As death occurred following 204 A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 a declining trend in d15N (Fig. 5B), the season of death was probably autumn or early winter. d13C is relatively high until the first postnatal sample, and it then declines until increasing again in the final sample (Fig. 5C). The declining postnatal d13C values might be associated with either a high relative proportion of isotopically light lipids (from milk) in the diet, a decreasing trophic level effect in C associated with increasing plant food intake, or a change in the diet of the mother. The higher d13C prior to birth could be due to high maternal d13C caused by eating plants in high, dry areas (with less snow cover) in the winter, or due to use of maternal protein stores for the production of fetal proteins (similar to 13C- and 15N-enriched fetal blood cells observed in caribou (Barboza and Parker, 2006). There are 13 dentin collagen samples representing the time recorded in Lyuba’s dP2. The C/N (atomic ratio) for the samples ranged from 3.2 to 3.3. All samples were formed prenatally with the exception of the final sample (Fig. 6). Because second-order increments were difficult to see in the thin section from the dP2, we are unable to tie compositional samples to specific time intervals. However, the samples do represent an ordered sequence of nonoverlapping time intervals, and the dentin thickness and accretion rate suggest that the total time represented in the sample sequence is between 12 and 17 months. The d15N series shows seasonal-scale oscillations (Fig. 6B) similar to those observed in other tusk and hair series (Section 1.3). However, the cycle has a peak-to-peak amplitude of only around 0.5&. In Siberian mammoth tusks and hair, peak-to-peak amplitude in d15N oscillation is typically greater (w1ew3&; e.g., Iacumin et al., 2005; Rountrey, 2009; Gohmann et al., 2010). Mammoth hairs from the Gydanskiy Peninsula (near the Yamal Peninsula) dating to around 10,000 years BP do not show seasonal-scale oscillations in d15N (Iacumin et al., 2006), but we are unaware of any studies of d15N variation in tusks or hair dating to around 40,000 years BP from the Yamal region. Thus, we are unable to determine whether the low amplitude is typical of the region and time or is related to this dentin having been produced prenatally. The complete cycle from sample 4 to sample 10 indicates that dentin formation in this tooth probably took place over more than one year (Fig. 6B). The pattern would not be consistent with dentin formation occurring over more than about two years, but it is difficult to make a precise estimate of the total time represented based only on the oscillatory pattern, because the samples are not uniform in thickness (and thus, the durations represented by points vary). A period of prenatal dP2 mineralization of between one and two years is in agreement with our estimates based on growth increments, and is similar to what is observed in African elephants. As mentioned above, segments of the series in which d15N increases may represent winter and spring. Birth occurs during a period of rising d15N (Fig. 6B) indicating that Lyuba was likely born in spring (i.e., before the June solstice), possibly before her mother had access to new plant growth. The postnatal sample has the highest d15N in the series (Fig. 6B) as is the case for the Oimyakon calf (the first postnatal sample has the highest value; Fig. 5B). The d13C series from Lyuba’s dP2 (Fig. 6C) lacks the oscillation observed in other tusk and hair series. Values tend to increase with time, but the pattern is difficult to interpret. As is the case with the d15N series, data from additional dP2s are needed to determine whether this pattern is typical of arctic mammoths. 4. Conclusions We have shown that accurate and precise age estimates for young mammoths are possible using dentin growth increments. The deciduous tusks, dP2s, dP3s, and dP4s begin mineralization before birth, and the timing of birth is marked by neonatal lines in the dentin of at least the dP2s, dP3s, and dP4. Dentin mineralized around the time of birth shows elevated Zn/Ca, and this may provide a means of testing identifications of neonatal lines, but analyses of additional teeth are necessary to confirm this association. In both calves, birth occurred following increases in d15N, which we interpret as representing winter/spring periods of maternal nutritional stress and/or reliance on maternal body protein for production of fetal tissues. This suggests that births took place in spring. Growth increments in the dentin of Lyuba’s dP2 show that mineralization of the tooth began about 13 months before birth, which is comparable to the timing observed in African elephants (w15 months). This supports the hypothesis that the duration of gestation in mammoths was similar to that in extant elephants. The timing of initial permanent tusk mineralization differs in Lyuba and the Oimyakon calf. In the Oimyakon calf, which appears to have died at an age of about 7.4 months, the permanent tusk began mineralization months before birth, as has been observed in an African elephant. We determined that Lyuba died at an age of about 30e35 days, but no mineralized permanent tusk was observed. This suggests that the timing of permanent tusk formation differed among individuals, sexes, or populations. Further work on a dP3 from the Oimyakon calf is warranted to confirm the age at death and timing of tusk mineralization. This work shows the potential value of detailed individual life history studies using teeth. With work on additional specimens, these techniques could lead to construction of an accurate tooth development schedule for mammoths, and this information is critical for studies of mammoth life history that will help elucidate ecological factors associated with extinction. Acknowledgments We are thankful for thoughtful reviews of this manuscript by J. Codron and an anonymous reviewer. This work was funded in part by grants to D.C. Fisher from the U.S. National Science Foundation (EAR-0545095) and the National Geographic Society (EC-03494-08 and CRE-8503-08). Scott Beld and Ted Huston provided technical expertise. We gratefully acknowledge the GE Healthcare Institute, Waukesha, WI for conducting the CT scans of Lyuba. Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.quaint.2011.06.006. References Adamczewski, J.Z., Flood, P.F., Gunn, A., 1997. Seasonal patterns in body composition and reproduction of female muskoxen (Ovibos moschatus). Journal of the Zoological Society of London 241, 245e269. Anthony, R., 1933. Recherches sur les incisives supérieures des Elephantidae actuels et fossils (Éléphants et Mastodontes): morphologie, structure, evolution ontogénique, interpretation. Archives du Muséum National d’Histoire Naturelle, 6th series 10, 65e124. with 2 plates. Barboza, P.S., Parker, K.L., 2006. Body protein stores and isotopic indicators of N balance in female reindeer (Rangifer tarandus) during winter. Physiological and Biochemical Zoology 79 (3), 628e644. Barboza, P.S., Parker, K.L., 2008. Allocating protein to reproduction in arctic reindeer and caribou. Physiological and Biochemical Zoology 81 (6), 835e855. Boeskorov, G.G., Tikhonov, A.N., Lazarev, P.A., 2007. A new find of a mammoth calf. Doklady Biological Sciences 417, 480e483. DeNiro, M.J., 1985. Postmortem preservation and alteration of in vivo bone collagen isotope ratios in relation to palaeodietary reconstruction. Nature 317, 806e809. Eales, N.B., 1926. The anatomy of the head of a foetal African elephant, Elephas africanus (Loxodonta africana). Transactions of the Royal Society of Edinburgh 54, 491e551. þ12 plates. Eales, N.B., 1931. The development of the mandible in the elephant. Proceedings of the Zoological Society of London 101, 115e127. A.N. Rountrey et al. / Quaternary International 255 (2012) 196e205 Fisher, D.C., 2001. Season of death, growth rates, and life history of North American Mammoths. In: West, D. (Ed.), Proceedings of the International Conference on Mammoth Site Studies. Publications in Anthropology 22. University of Kansas, Lawrence, pp. 121e135. Fisher, D.C., Rountrey, A.N., Beld, S.G., Fox, D.L., Gohman, S., Tikhonov, A.N., Mol, D., Buigues, B., Boeskorov, G.G., Lazarev, P.A., 2010. Life history of the Yukagir mammoth. In: Lazarev, P., Boeskorov, G., Maschenko, E. (Eds.), Proceedings of the IV International Mammoth Conference. Institute of Applied Ecology of the North, Yakutsk, Yakutia, Russian Federation, pp. 54e63. Fisher, D.C., Tikhonov, A.N., Kosintsev, P.A., Rountrey, A.N., Buigues, B., van der Plicht, J., this issue. Anatomy, death, and preservation of a woolly mammoth calf, Yamal Peninsula, Northwest Siberia. Quaternary International, xxexx. Fox, D.L., Fisher, D.C., Vartanyan, S., Tikhonov, A.N., Mol, D., Buigues, B., 2007. Paleoclimatic implications of oxygen isotopic variation in late Pleistocene and Holocene tusks of Mammuthus primigenius from northern Eurasia. Quaternary International 169e170, 154e165. Gohmann, S.C.L., Fox, D.L., Fisher, D.C., Vartanyan, S., Tikhonov, A.N., Mol, D., Buigues, B., 2010. Paleodietary and paleoenvironmental implications of carbon and nitrogen isotopic variations in late Pleistocene and Holocene tusks of Mammuthus primigenius from northern Eurasia. In: Lazarev, P., Boeskorov, G., Maschenko, E. (Eds.), Proceedings of the IV International Mammoth Conference, Institute of Applied Ecology of the North, pp. 177e187. Yakutsk, Yakutia, Russian Federation. Guthrie, R.D., 2001. Reconstruction of woolly mammoth life history. In: Cavarretta, G., Gioia, P., Mussi, M., Palombo, M.R. (Eds.), The World of Elephants. Proceedings of the First International Congress. Consiglio Nazionale delle Ricerche, Rome, pp. 276e279. Hanks, J., 1969. Seasonal breeding of the African elephant in Zambia. East African Wildlife Journal 7, 167. Haynes, G., 1991. Mammoths, Mastodonts, and Elephants: Biology, Behavior, and the Fossil Record. Cambridge University Press, Cambridge. Hildebrandt, T., Drews, B., Gaeth, A.P., Goeritz, F., Hermes, R., Schmitt, D., Gray, C., Rich, P., Streich, W.J., Short, R.V., Renfree, M.B., 2007. Foetal age determination and development in elephants. Proceedings of the Royal Society B 274, 323e331. Hobson, K.A., Alisauskas, R.T., Clark, R.G., 1993. Stable-nitrogen isotope enrichment in avian tissues due to fasting and nutritional stress: implications for isotopic analyses of diet. The Condor 95 (2), 388e394. Iacumin, P., Davanzo, S., Nikolaev, V., 2005. Short-term climatic changes recorded by mammoth hair in the Arctic environment. Palaeogeography, Palaeoclimatology, Palaeoecology 218, 317e324. Iacumin, P., Davanzo, S., Nikolaev, V., 2006. Spatial and temporal variations in the 13 12 C/ C and 15N/14N ratios of mammoth hairs: palaeodiet and palaeoclimatic implications. Chemical Geology 231, 16e25. Jeswani, R.M., Vani, S.N., 1991. A study of serum zinc levels in cord blood of neonates and their mothers. Indian Journal of Pediatrics 58 (5), 683e686. Kang, D., Amarasiriwardena, D., Goodman, A.H., 2004. Application of laser ablationinductively coupled plasma-mass spectrometry (LA-ICP-MS) to investigate trace metal spatial distributions in human tooth enamel and dentine growth layers and pulp. Analytical and Bioanalytical Chemistry 378, 1608e1615. Koch, P.L., Fisher, D.C., Dettman, D., 1989. Oxygen isotope variation in the tusks of extinct proboscideans: a measure of season of death and seasonality. Geology 17, 515e519. Kosintsev, P.A., Lapteva, E.G., Trofimova, S.S., Zanina, O.G., Tikhonov, A.N., van der Plicht, J., 2010. The intestinal contents of a baby woolly mammoth (Mammuthus primigenius Blumenbach, 1799) from the Yuribey River (Yamal Peninsula). Doklady Biological Sciences 432, 209e211. Larter, N.C., Nagy, J.A., 1995. Peary caribou, muskoxen and Banks Island forage: assessing seasonal diet similarities. Rangifer 17 (1), 9e16. Laws, R.M., 1966. Age criteria for the African elephant. East African Wildlife Journal 4, 1e37. Laws, R.M., 1969. Aspects of reproduction in the African elephant, Loxodonta africana. Journal of Reproduction and Fertility Supplement 6, 193e217. 205 Maschenko, E.N., 2002. Individual development, biology and evolution of the woolly mammoth. Cranium 19, 4e120. Mekota, A.-M., Grupe, G., Ufer, S., Cuntz, U., 2006. Serial analysis of stable nitrogen and carbon isotopes in hair: monitoring starvation and recovery phases of patients suffering from anorexia nervosa. Rapid Communications in Mass Spectrometry 20, 1604e1610. Meyer, J.M., Walker, S.L., Freeman, E.W., Steinetz, B.G., Brown, J.L., 2004. Species and fetal gender effects on the endocrinology of pregnancy in elephants. General and Comparative Endocrinology 138, 263e270. Moss, C.J., 1983. Oestrus behavior and female choice in the African elephant. Behaviour 86, 167e196. Nadelhoffer, K., Shaver, G., Fry, B., Giblin, A., Johnson, L., McKane, R., 1996. 15N natural abundances and N use by tundra plants. Oecologia 107, 386e394. Rasband, W.S., 1997e2009. ImageJ, U.S. National Institutes of Health, Bethesda, Maryland, USA. http://rsb.info.nih.gov/ij/. Raubenheimer, E.J., 2000. Development of the tush and tusk and tusklessness in the African elephant (Loxodonta africana). Koedoe 43 (2), 57e64. Raubenheimer, E.J., Van Heerden, W.F.P., Van Nierkerk, P.J., De Vos, V., Turner, M.J., 1995. Morphology of the deciduous tusk (tush) of the African elephant (Loxodonta africana). Archives of Oral Biology 40 (6), 571e576. Ropstad, E., 2000. Reproduction in female reindeer. Animal Reproduction Science 60e61, 561e570. Roth, V.L., Shoshani, J., 1988. Dental identification and age determination in Elephas maximus. Journal of the Zoological Society of London 214, 567e588. Rountrey, A.N., 2009. Life Histories of Juvenile Woolly Mammoths from Siberia: Stable Isotope and Elemental Analyses of Tooth Dentin. Unpublished dissertation, University of Michigan. Rountrey, A.N., Fisher, D.C., Mol, D., Buigues, B., Tikhonov, A.N., Lazarev, P.A., Boeskorov, G.G., 2010. Stable isotope time-series in juvenile mammoth tusks. In: Lazarev, P., Boeskorov, G., Maschenko, E. (Eds.), Proceedings of the IV International Mammoth Conference. Institute of Applied Ecology of the North, Yakutsk, Yakutia, Russian Federation, pp. 235e243. Rountrey, A.N., Fisher, D.C., Vartanyan, S., Fox, S.L., 2007. Carbon and nitrogen isotope analyses of a juvenile woolly mammoth tusk: evidence of weaning. Quaternary International 169e170, 166e173. Rowell, J.E., Pierson, R.A., Flood, P.F., 1993. Endocrine changes and luteal morphology during pregnancy in muskoxen (Ovibos moschatus). Journal of Reproduction and Fertility 99, 7e13. Schoeninger, M.J., DeNiro, M.J., 1984. Nitrogen and carbon isotopic composition of bone collagen from marine and terrestrial animals. Geochimica et Cosmochimica Acta 48, 625e639. Schour, I., 1936. The neonatal line in the enamel and dentin of the human deciduous teeth and first permanent molar. Journal of the American Dental Association 23, 1946e1955. van Geel, B., Aptroot, A., Baittinger, C., Birks, H.H., Bull, I.D., Cross, H.B., Evershed, R.P., Gravendeel, B., Kompanje, E.J.O., Kuperus, P., Mol, D., Nierop, K.G.J., Pals, J.P., Tikhonov, A.N., van Reenen, G., van Tienderen, P.H., 2008. The ecological implications of a Yakutian mammoth’s last meal. Quaternary Research 69, 361e376. van Geel, B., Guthrie, R.D., Fisher, D.C., Altmann, J.G., Broekens, P., Bull, I.D., Gill, F.L., Gravendeel, B., Jansen, B., Nieman, A.M., Rountrey, A.N., van Reenen, G.B.A., 2010. Paleoecological studies of mammoth and mastodon feces and the evidence for coprophagy. Quaternaire, Hors série 3, 98e99. Abstracts of the Vth International Conference on Mammoths and Their Relatives. Vuori, E., Kuitunen, P., 1979. The concentrations of copper and zinc in human milk: a longitudinal study. Acta Paediatrica Scandinavica 68, 33e37. Williamson, B.R., 1976. Reproduction in female African elephants in the Wankie National Park, Rhodesia. South African Journal of Wildlife Research 6, 89e93. Wittemyer, G., Rasmussen, H.B., Douglas-Hamilton, I., 2007. Breeding phenology in relation to NDVI variability in free-ranging African elephant. Ecography 30, 42e50. Wuthier, R.E., 1984. Lipids in dentinogenesis. In: Linde, A. (Ed.), Dentin and Dentinogenesis. CRC Press, Boca Raton, pp. 93e106.