Introduction
Mammuthus is one of several iconic extinct Ice Age genera that define both scientific and popular reconstructions of Late Pleistocene North American landscapes. Our understanding of the paleontological record of species of Mammuthus is largely a reflection of our understanding of dental morphology, and the relationship of that morphology to taxonomy. Variation and morphological patterns in mammoth teeth are mostly considered at continental scales (e.g., Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017), but regional studies highlight the significance of evaluating geographically restricted samples in order test broad hypotheses of change in proboscidean populations in North America (e.g., Saunders et al., Reference Saunders, Grimm, Widga, Campbell, Curry, Grimley, Hanson, McCullum, Oliver and Treworgy2010). Here, we summarize morphometric features of isolated mammoth teeth from Alberta, Canada, a geographic region with documented biogeographic fluidity thought to preserve the remains of both woolly and Columbian mammoths (Harington and Shackleton, Reference Harington and Shackleton1978; Burns and Young, Reference Burns and Young1994; Burns et al., Reference Burns, Baker and Mol2003; Hills and Harington, Reference Hills and Harington2003; Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017). We use those morphometric data, morphometric data from the literature (Lister and Sher, Reference Lister and Sher2015; Widga et al., Reference Widga, Saunders and Enk2017), and both new and previously published radiocarbon data to establish an understanding of and context for quantitative morphological variation in mammoth teeth from Alberta. Finally, we discuss the significance of our observations for existing taxonomic hypotheses of Mammuthus from Alberta.
Taxonomic context
The taxonomy of non-insular North American species of Mammuthus has changed considerably since Osborn's (Reference Osborn1942) seminal work, where 15 species of Mammuthus were recognized. Subsequent work vastly reduced the number of recognized species, with most studies identifying between two and four valid species occurring in the Late Pleistocene (Mammuthus columbi, Mammuthus exilis, Mammuthus jeffersonii, and Mammuthus primigenius; e.g., Maglio, Reference Maglio1973; Kurtén and Anderson, Reference Kurtén and Anderson1980; Madden, Reference Madden1981; Agenbroad, Reference Agenbroad, Martin and Klein1984, Reference Agenbroad, Agenbroad and Mead1994, Reference Agenbroad2005; Graham, Reference Graham, Frison and Todd1986; Saunders et al., Reference Saunders, Grimm, Widga, Campbell, Curry, Grimley, Hanson, McCullum, Oliver and Treworgy2010; Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017; Lucas et al., Reference Lucas, Morgan, Love and Connell2017; Widga et al., Reference Widga, Saunders and Enk2017). For Late Pleistocene faunas of western North America, most authors now restrict morphological identifications to M. primigenius (woolly mammoth) and M. columbi (Columbian mammoth), with the former distributed in “northern” latitudes and the latter in “southern” latitudes.
Dental remains have always played a prominent role in the taxonomy of mammoths. Not only are teeth durable, facilitating their preservation and recovery, but they possess characters used by researchers since the 1800s in order to establish and recognize different mammoth species (Osborn, Reference Osborn1942; Maglio, Reference Maglio1973; Madden, Reference Madden1981; Lister, Reference Lister2017). The total number of lamellae (enamel plates), relative spacing of lamellae, and relative enamel thickness of the last permanent molars are some of the characters used to distinguish mammoth species (e.g., Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017; Widga et al., Reference Widga, Saunders and Enk2017). In recent years, genomic (Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016; Chang et al., Reference Chang, Knapp, Enk, Lippold, Kircher, Lister and MacPhee2017; Palkopoulou et al., Reference Palkopoulou, Lipson, Mallick, Nielsen, Rohland, Baleka and Karpinski2018; van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021) and morphological (Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017; Widga et al., Reference Widga, Saunders and Enk2017) studies documented evidence of hybridization and introgression between M. primigenius and M. columbi, suggesting possible further complexity for our understanding of morphological variation in mammoth dentitions. Minimally, those studies illustrate the need for comprehensive descriptions of morphometric and statistical data for teeth at a regional level in order to understand morphological variability across time and space.
Among the most widely utilized dental characters in mammoth taxonomy are: (1) the total number of lamellae in an unworn molar; (2) the relative crown height (i.e., degree of hypsodonty) in an unworn molar; (3) the relative spacing between lamellae; and (4) the relative thickness of occlusal enamel bands (Osborn, Reference Osborn1942; Maglio, Reference Maglio1973; Madden, Reference Madden1981; Lister and Sher, Reference Lister and Sher2015). These traits can be studied in any of the upper (M) or lower (m) molars produced within the lifespan of an individual, but most researchers place special emphasis on the upper and lower M6—the last molar to form and erupt into the oral cavity (= M3 of some authors). Of particular relevance to our work is the distinction between M. primigenius and M. columbi. Although relative crown height is not particularly different between the two taxa, molars of M. primigenius possess on average a greater number of lamellae, narrower relative spacing between lamellae, and narrower enamel thickness than molars of M. columbi (Lister and Sher, Reference Lister and Sher2015). We note that molars of M. trogontherii, an Early to Middle Pleistocene Eurasian and Beringian species which is considered by some researchers to have given rise to both M. columbi and M. primigenius (Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017; van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021), has dental characters that closely resemble those observed in M. columbi (Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017). Pending further morphological and molecular analyses, the retention of M. columbi and M. trogontherii as “separate species names [is] largely a pragmatic decision based on long historical usage in North America and Eurasia, respectively.” (Lister, Reference Lister2017, p. 27). Morphological and molecular analyses of Early to Middle Pleistocene mammoth specimens will likely continue to clarify that interpretation (e.g., van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021).
The published late Quaternary record of mammoths in Alberta was recently summarized along with tentative species identifications for many previously unreported isolated mammoth teeth from Alberta (Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017). Those identifications were based on preliminary observations of enamel thickness and lamellar frequency (Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017). Teeth retaining characters deemed consistent with M. columbi and M. primigenius were both reported, along with specimens that appeared to show intermediate morphologies, with the caveat that future analyses would permit further evaluation of those taxonomic hypotheses.
Mammoths, megafauna, and a dynamic landscape
Most records of mammoth in Alberta represent isolated data points consisting of single specimens, including isolated teeth (Burns et al., Reference Burns, Baker and Mol2003; Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017). Those records, and other records of megafauna, come mostly from uncertain stratigraphic positions in alluvium, in both primary and secondary depositional settings that may include Quaternary vertebrate remains representing a range of Pleistocene ages (e.g., Hills and Wilson, Reference Hills and Wilson2003). As a result, records of Quaternary megafauna in Alberta are sometimes viewed through the lens of broad time bins (e.g., Jass et al., Reference Jass, Burns and Milot2011). As increasing numbers of individual specimens are directly dated or evaluated in other ways (e.g., aDNA), more specific biological patterns may emerge for individual taxa.
Recent work examining aDNA of separate series of the isolated records of Bison and Mammut has shown that for at least some megafauna, the Quaternary vertebrate record preserved in Alberta is biogeographically complex (Heintzman et al., Reference Heintzman, Froese, Ives, Soares, Zazula, Letts and Andrews2016; Karpinski et al., Reference Karpinski, Hackenberger, Zazula, Widga, Duggan, Golding and Kuch2020). For mastodons, in particular, genetic data indicate the presence of different clades at discrete time intervals through the Pleistocene, reflecting possibly distinct temporal dispersals (Karpinski et al., Reference Karpinski, Hackenberger, Zazula, Widga, Duggan, Golding and Kuch2020). Such patterns likely correspond with the significant environmental and geologic fluctuations documented for the Pleistocene in western Canada (e.g., Dyke et al., Reference Dyke, Moore and Robertson2003; Dyke, Reference Dyke, Ehlers and Gibbard2004, Reference Dyke2005; Dalton et al., Reference Dalton, Margold, Stokes, Tarasov, Dyke, Adams and Allard2020). However, our understanding of the interrelationship of environmental and geologic fluctuations with biogeographic patterns of individual taxa, and the relationship between those biogeographic patterns and morphological patterns preserved in the regional fossil record, remains limited for most taxa (e.g., mammoths).
Some researchers suggested that M. columbi and M. primigenius were allopatric (e.g., Maglio, Reference Maglio1973), but more recent work suggests time-averaged sympatry in the distributions, at least at the margins of the ranges at mid-latitudes (Agenbroad et al., Reference Agenbroad, Lister, Mol, Roth, Agenbroad and Mead1994; Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016; Smith and Graham, Reference Smith and Graham2017; Widga et al., Reference Widga, Saunders and Enk2017). Modern elephants can range over vast geographic spaces within a lifetime, with estimated ranges from hundreds to thousands of square kilometers (e.g., Lindeque and Lindeque, Reference Lindeque and Lindeque1991; Douglas-Hamilton et al., Reference Douglas-Hamilton, Krink and Vollrath2005; Ngene et al., Reference Ngene, Okello, Mukeka, Muya, Njumbi and Isiche2017), and recent work on M. primigenius suggests that similar scales of mobility occurred in Mammuthus (Wooler et al., Reference Wooler, Bataille, Druckenmiller, Erickson, Groves, Haubenstock and Howe2021). Given sympatric records elsewhere, the mobility of elephantids, and evidence for introgression among mammoths (e.g., Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016), the presence of sympatric populations seems a reasonable inference for Alberta, where mammals with northern and southern evolutionary origins are inferred as co-occurring over broad time scales (e.g., Burns, Reference Burns2010; Jass et al., Reference Jass, Burns and Milot2011).
Given that context and the tentative taxonomic identifications proposed by Jass and Barrón-Ortiz (Reference Jass and Barrón-Ortiz2017), we predicted that mammoth teeth recovered from Upper Pleistocene deposits in Alberta would show a particularly large range of quantitative variation, ranging from “typical” M. primigenius to “typical” M. columbi morphologies, and including specimens with intermediate morphologies (see “Materials and Methods” for delimitation of “typical” and intermediate morphologies). Given the currently documented biogeographic distributions of those taxa and rare documentation of M. columbi in northern North America (see Smith and Graham, Reference Smith and Graham2017), we also predicted that M. primigenius would be more prevalent in our samples.
Materials and methods
Following the dental nomenclature of Agenbroad (Reference Agenbroad, Agenbroad and Mead1994), the sixth upper molars (M6) and sixth lower molars (m6) are the last molars to form, erupt, and come into occlusion in Mammuthus; they are referred to by some researchers as the permanent M3/m3 molars (e.g., Lister and Sher, Reference Lister and Sher2015; Widga et al., Reference Widga, Saunders and Enk2017). We studied the sixth molars (M6/m6) of Mammuthus recovered from Pleistocene deposits in Alberta, Canada (Tables 1a, b and 2a, b, Fig. 1). Our study sample (n = 17) was limited to teeth that did not show advanced stages of wear as this can confound the taxonomic identification of mammoth molars (Smith and Graham, Reference Smith and Graham2017). Additionally, teeth in our sample were complete enough and in a state of preservation that permitted measurement of at least three variables: tooth width, lamellar frequency, and enamel thickness (Tables 1a, b and 2a, b). We directly measured 14 specimens housed at the Royal Alberta Museum (RAM) in Edmonton, Alberta, Canada (Figs. 2 and 3), and added published measurements of specimens housed at the Canadian Museum of Nature (CMN; Ottawa, Ontario, Canada) and Royal Ontario Museum (ROM; Toronto, Ontario, Canada). We note that RAM P94.4.3 was reported as a lower left m6 and P97.10.144 as an upper left M6 (Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017), but we here identify them as an upper left M6 and a lower left m6, respectively (Tables 1a, b). We also revise the identifications of RAM P81.3.5 and P97.8.1, which were previously questionably identified as sixth molars (Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017) (Tables 1a, b).
Table 1a. Data for M6 of Mammuthus recovered from Alberta, Canada. All measurements in millimeters.
Key: PF, Plate Formula; ∞, anterior loss through wear; -, anterior loss through breakage; x, anterior talon; p, posterior platelet; i, “accessory” lamellae; P, observed or reconstructed complete lamellar number excluding x and p; L, length of the preserved tooth crown perpendicular to the average orientation of lamellae; W, width of the tooth crown at the widest preserved lamella including cementum; W′, width of the tooth crown at the widest preserved lamella excluding cementum; l and l′, lamella (or lamellae) where measurements were taken (the lamellae are numbered in an anterior–posterior direction [excluding the talon]; l indicates lamella(e) of teeth in which the anterior end of the crown is preserved or reliably estimated; l′ indicates that the anterior portion of the crown is broken or worn and the lamellae are numbered from the anterior end of the preserved crown); C, cementum thickness (estimated based on measurements of cementum preserved on medial and/or lateral sides); H, tooth crown height at the highest unworn, preserved lamella, measurements for worn lamellae are noted. All data are from this study, with the exception of ROM 28983 and ROM IBW.83 (from Lister and Sher, Reference Lister and Sher2015).
a Cementum measured on occlusal surface.
b Based on cementum measurement from posterior portion of tooth.
c Estimated.
Table 1b. Additional data for M6 of Mammuthus recovered from Alberta, Canada. All measurements in millimeters.
Key: LF, lamellar frequency; LFcalc, calculated lamellar frequency; LL, lamella length calculated using LF; LL′, lamella length calculated using LFcalc; ET, enamel thickness; LLI, ETI and LLI′, ETI′ are variables standardized to a crown width (W and W′, respectively) of 100 mm (see “Materials and Methods”). All data are from this study, with the exception of ROM 28983 and ROM IBW.83 (from Lister and Sher, Reference Lister and Sher2015).
Table 2a. Measurements of m6s of Mammuthus recovered from Alberta, Canada. All measurements in millimeters.
Key: PF, Plate Formula; ∞, anterior loss through wear; -, anterior loss through breakage; x, anterior talonid; p, posterior platelet; P, observed or reconstructed complete lamellar number excluding x and p; ~, approximately; L, length of the preserved tooth crown perpendicular to the average orientation of lamellae; W, width of the tooth crown at the widest preserved lamella including cementum; l and l′, lamella (or lamellae) where measurements were taken (the lamellae are numbered in an anterior–posterior direction [excluding the talonid]; l indicates lamella(e) of teeth in which the anterior end of the crown is preserved or reliably estimated; l′ indicates that the anterior portion of the crown is broken or worn and the lamellae are numbered from the anterior end of the preserved crown); C, cementum thickness (estimated based on measurements of cementum preserved on medial and/or lateral sides); H, tooth crown height at the highest unworn, preserved lamella, measurements for worn lamellae are noted. All data are from this study, with the exception of CMN 17845 (from Churcher, Reference Churcher1972).
a Embedded in jaw; posterior end not visible.
b Missing plates at the end.
c To last measurable plate [l′20].
d Cementum thickness of 5.8 mm added to reported width (Churcher, Reference Churcher1972) based on average cementum width of the other Alberta teeth studied.
e Cementum measured on occlusal surface.
f Cementum measured on occlusal surface, anterior position of specimen P97.10.143.
g Based on cementum measured in area where it was sectioned.
h Cementum measured on occlusal surface, anterior portion.
Table 2b. Additional measurements of m6s of Mammuthus recovered from Alberta, Canada. All measurements in millimeters.
Key: LFB, basal lamellar frequency; LLB, basal lamella length; ET, enamel thickness; LLBI and ETI are variables standardized to a crown width (W) of 100 mm (see “Materials and Methods”). All data are from this study, with the exception of CMN 17845 (from Churcher, Reference Churcher1972).
a LFB, LLB, and LLBI is for medial side (the buccal side is embedded in the jaw).
b LFB, LLB, and LLBI estimated by calculating the average ratio between lamellar frequency of medial and lateral sides in other m6 specimens from RAM (see “Materials and Methods”); these values are presumably closer to the actual LFB, LLB, and LLBI for this specimen.
Figure 1. Locality map for specimens of Mammuthus included in this study. Inset map includes known northern margins for distribution of M. columbi and known southern margins for distribution of M. primigenius. Zone of geographic overlap is illustrated by shaded area. Distributional boundaries based on records in Neotoma Paleoecology Database (Williams et al., Reference Williams, Grimm, Blois, Charles, Davis, Goring and Graham2018). Requests for more specific locality information should be directed to the Royal Alberta Museum.
Figure 2. Upper sixth molars (M6) of Mammuthus from Alberta. (a) RAM P80.14.1, lateral and occlusal views. (b) RAM P84.5.1, lateral and occlusal views. (c) RAM P91.10.4, medial? and occlusal views. (d) RAM P94.4.3, lateral and occlusal views. (e) RAM P88.7.1, lateral? and occlusal views. (f) RAM P97.8.1, medial and occlusal views. (g) RAM P81.3.5, medial and occlusal views. (h) RAM P02.8.70, medial? and occlusal views. Anterior side of all teeth located to the left.
Figure 3. Lower sixth molars (m6) of Mammuthus from Alberta. (a) RAM P21.391.1, occlusal and medial views. (b) RAM P99.3.164, occlusal and lateral views. (c) RAM P97.10.143, occlusal and medial views. (d) RAM P18.319.4, occlusal and medial views. (e) RAM P97.10.144, occlusal and lateral views. (f) RAM P97.11.1b, occlusal and lateral views. Anterior side of all teeth located to the left.
Radiocarbon dating
Radiocarbon data for four specimens (CMN 17845, RAM P94.4.3, P02.8.70, and P97.11.1b) were compiled from published sources (Hills and Harington, Reference Hills and Harington2003; Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017). We sampled nine additional specimens (RAM P84.5.1, P91.10.4, P97.8.1, P97.10.143, P97.10.144, P97.11.1T, P99.3.164, P18.319.4, and P21.391.1) for radiocarbon analysis. P97.11.1 is inferred to represent multiple elements of a single individual. Both the original radiocarbon age assigned to the dentary and teeth (see Metcalfe et al., Reference Metcalfe, Longstaffe, Jass, Zazula and Keddie2016) and our reassessed age come from a bone fragment assigned to that specimen (P97.11.1T). All samples were collected using a Dremel 100-N/7 rotary tool fitted with a cutting disc made of a hard abrasive. The samples were sent to the A. E. Lalonde AMS Laboratory at the University of Ottawa, Ottawa, Ontario, Canada (UOC), or the W. M. Keck Carbon Cycle Accelerator Facility at the University of California, Irvine, United States (UCIAMS). Pretreatment methods (ultrafiltration) and processing for samples sent to Lalonde were described in Crann et al. (Reference Crann, Murseli, St-Jean, Zhao, Clark and Kieser2017). Equipment used for sample preparation was summarized in St-Jean et al. (Reference St-Jean, Kieser, Crann and Murseli2017). Samples submitted to the Keck Carbon Cycle AMS Facility (UCIAMS) were decalcified in 0.5N HCl, gelatinized at 60°C and pH 2, and ultrafiltered to select a high molecular weight fraction (>30 kDA; John Southon, personal communication). Radiocarbon dates were calibrated in OxCal v.4.4.4 using the IntCal 20 calibration curve (Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey and Butzin2020; Bronk Ramsey, Reference Bronk Ramsey2021).
Morphometric analysis of mammoth teeth
For measurements, we followed the methodology of Lister and Sher (Reference Lister and Sher2015) and Widga et al. (Reference Widga, Saunders and Enk2017) and a subset of their comparative data sets (Supplementary Tables 1–3). See Lister and Sher (Reference Lister and Sher2015, figs. 2E and D) for illustration of location of measurements. All measurements (except for the estimated total number of lamellae) are in millimeters and were taken using Mitutoyo digital calipers to the nearest 0.01 mm. Collected dental data and calculated indices are as follows:
(1) Total number of lamellae in an unworn molar (P). We estimated this value for sufficiently complete specimens following the methodology of Lister and Sher (Reference Lister and Sher2015). This number excludes talons, talonids, and platelets.
(2) Length (L) of the preserved tooth crown perpendicular to the average orientation of lamellae. In other words, we measured perpendicular to the most prevalent orientation of lamellae in occlusion.
(3) Width (W) of the tooth crown at the widest preserved lamella (measured with calipers held parallel to the lamella) including cementum. If cementum was weathered or missing in one or both sides of the tooth, we estimated the amount of cementum missing (C) based on measurements of cementum present in other areas of the tooth or in other teeth from the same sample (following Lister and Sher, Reference Lister and Sher2015).
(4) Width (W′) of the tooth crown at the widest preserved lamella (measured with calipers held parallel to the lamella) excluding cementum. If cementum was present, we subtracted its estimated width (following Widga et al., Reference Widga, Saunders and Enk2017). We took this measurement only in upper M6 for comparisons with the data set of Widga et al. (Reference Widga, Saunders and Enk2017).
(5) Height (H) of the tooth crown measured from the base of the crown to the apex of the highest unworn, preserved lamella.
(6) Enamel thickness (ET). We measured enamel thickness parallel to the growth axis of the lamella. We took this measurement at up to 10 points on the occlusal surface and we calculated the average of those values.
(7) Enamel index (EI and EI′). We calculated EI with the formula EI = 100 × ET/W. For upper molars, we also obtained EI′, which was calculated in the same manner as EI, but using W′ instead of W. This variable measures enamel thickness with the tooth standardized to a width of 100 mm.
(8) Lamellar frequency (LF) and basal lamellar frequency (LFB). We measured LF and LFB following Lister and Sher (Reference Lister and Sher2015). In this methodology, lamellar frequency is obtained with the formula LF = 100 × p/l; l = length of a portion of the tooth, and p = number of enamel plate–cementum intervals occupying that length. In the upper molars, lamellar frequency (LF) was measured at the top and base of the crown, on both the medial and lateral sides, and the average of these four measurements was calculated. In the lower molars, lamellar frequency was only measured at the base of the tooth crown (LFB) on both the medial and lateral sides and the average of the two measurements was calculated. Lamellae in the lower molars tend to converge towards the top of the crown, making the top measurements less consistent due to differences in tooth wear among individuals (Lister and Sher, Reference Lister and Sher2015).
(9) Calculated lamellar frequency (LFcalc). We obtained LFcalc for the upper molars in our sample following Widga et al. (Reference Widga, Saunders and Enk2017). These authors calculate lamellar frequency using the formula LFcalc = 100 × P′/L; P′ = number of preserved lamellae in the tooth.
(10) Lamella length (LL, LL′, and LLB). We calculated the average length of one lamella-cementum interval using the formulae LL = 100/LF and LLB = 100/LFB for the upper and lower molars, respectively (Lister, Reference Lister2017); LL = lamella length and LLB = basal lamella length. For upper molars, we also obtained LL′, which was calculated in the same manner as LL, but using LFcalc instead of LF.
(11) Lamellar length index (LLI, LLI′, and LLBI). We calculated LLI and LLBI for upper and lower molars, respectively, using the formulae LLI = 100 × LL/W and LLBI = 100 × LLB/W; LLI = lamella length index and LLBI = basal lamella length index. For upper molars, we also obtained LLI′, which was calculated in the same manner as LLI, but using LL′ and W′ instead of LL and W. This variable measures lamella length and basal lamella length with the tooth standardized to a width of 100 mm.
Given the state of preservation for many teeth in our sample, our study focused especially on the analysis of two variables: lamellar length index and enamel index (measurements 7 and 11 above). Other than the total number of lamellae (enamel plates P), these two variables are documented as most relevant for discriminating North American mammoth teeth (Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017; Widga et al., Reference Widga, Saunders and Enk2017). Using bivariate scatter plots, we compared lamellar length index and enamel index for the specimens we measured and specimens from Alberta reported in the literature (Churcher, Reference Churcher1972; Lister and Sher, Reference Lister and Sher2015; Widga et al., Reference Widga, Saunders and Enk2017) with a subset of the comparative data sets of Lister and Sher (Reference Lister and Sher2015) and Widga et al. (Reference Widga, Saunders and Enk2017) (Supplementary Tables 1–3).
Two of the specimens we studied (RAM P88.7.1 and P97.11.1b) preserved a specific morphology that necessitates a more detailed description of how data were collected. RAM P88.7.1 (Fig. 4) preserves what appear to be “accessory” lamellae between plates 6 and 7 and between plates 7 and 8. The “accessory” lamellae do not reach the occlusal surface, but rather terminate at approximately the midsection of the tooth crown. We excluded the “accessory” lamellae from our measurement and calculation of lamellar frequency. A second specimen, RAM P97.11.1b, is embedded in a partial dentary with most of the medial side of the tooth exposed, but the posterior and lateral sides of the tooth are obscured by the dentary. Thus, we were unable to quantify the total number of lamellae (enamel plates, P) and we could only measure the basal lamellar frequency (LFB) of the medial side. We estimated LFB for the lateral side of the tooth by calculating the mean LFB lateral/LFB medial ratio in the other m6 specimens we measured (mean ratio = 1.0546 ± 0.05; n = 5) and using this value to estimate LFB of the lateral side (RAM P97.11.1b LFB medial = 5.44; estimated LFB lateral = 5.74). This allowed us to estimate the average LFB of this specimen (estimated average LFB = 5.59). Tables 1a and b show both the LFB of the medial side and the estimated average LFB along with the respective indices we calculated (LLB and LLBI). We plot both the LLBI calculated for the medial side and the estimated LLBI in relevant figures (see “Results”).
Figure 4. RAM P88.7.1. Arrows point to “accessory” enamel plates between plates 6 and 7 and between plates 7 and 8.
Delimiting “typical” morphologies of M. primigenius and M. columbi
We used the distribution in morphospace of specimens traditionally identified as M. primigenius and M. columbi to delimit “typical” morphologies of these taxa. Late Pleistocene Mammuthus fossils from Beringia have traditionally been identified as M. primigenius (e.g., Maglio, Reference Maglio1973; Agenbroad, Reference Agenbroad, Martin and Klein1984, Reference Agenbroad2005; Harington, Reference Harington2011; Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017). Late Pleistocene Mammuthus fossils from the southern United States and Mexico (and further south into Central America) have traditionally been identified as M. columbi (e.g., Maglio, Reference Maglio1973; Agenbroad, Reference Agenbroad, Martin and Klein1984, Reference Agenbroad2005; Arroyo-Cabrales et al., Reference Arroyo-Cabrales, Polaco, Johnson and Ferrusquía-Villafranca2010; Lucas and Alvarado, Reference Lucas and Alvarado2010; Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017). These two subgroups tend to occupy distinct areas of morphospace in bivariate scatter plots of lamellar length index versus enamel index (e.g., Fig. 5), but they overlap to some degree. We identified specimens that plot below the zone of overlap (in the lower-left area of the graph) as specimens with “typical” M. primigenius morphology. Specimens that plot above the zone of overlap (in the upper-right area of the graph) were considered specimens with “typical” M. columbi morphology. Specimens that plot in the zone of overlap (shaded area) were treated as individuals with “intermediate” or overlapping M. columbi–M. primigenius morphology (see “Results” and “Discussion” for further considerations regarding specimens with “intermediate” morphology). We note one exception to the delimitation of these morphological groups. Early Pleistocene specimens from Beringia primarily plot in the morphospace region with “typical” M. columbi morphology, but they are instead identified as M. trogontherii. As mentioned in the introduction, molars of M. trogontherii, an Early to Middle Pleistocene Eurasian and Beringian species, have dental characters that closely resemble those observed in M. columbi (Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017).
Figure 5. Plots of enamel thickness index (EI) vs. lamella length index (LLI) for M6s of Mammuthus from Alberta and other sites in Eurasia and North America (following Lister and Sher, Reference Lister and Sher2015). Symbols: black circles (●) = specimens from Alberta; open circles (○) = M. primigenius from northwestern North America/Siberia (i.e., Beringia); (x) = M. trogontherii from China and northwestern North America/Siberia (i.e., Beringia); open diamonds (◊) = M. columbi from the Early and Middle Pleistocene of the contiguous United States; open triangles = (Δ) Late Pleistocene Mammuthus from the Great Plains and the Great Lakes (Canada and USA); plus signs (+) = Late Pleistocene Mammuthus from the Rocky Mountains (USA); rectangles (▭) = Late Pleistocene Mammuthus from the southern United States and Mexico; black squares (■) = Osborn's (Reference Osborn1922) neotype of M. columbi and other specimens from the neotype locality; open squares (□) = holotype (Leidy, Reference Leidy1858) and neotype (Osborn, Reference Osborn1922) of M. imperator; black triangles (▲) = specimens from one of Osborn's (Reference Osborn1942) M. jeffersonii “ideotype” localities. Shaded area = M. columbi–M. primigenius area of overlap. Data for Alberta, this study; other data from Lister and Sher (Reference Lister and Sher2015).
For further comparison, we subdivided the data sets of Lister and Sher (Reference Lister and Sher2015) and Widga et al. (Reference Widga, Saunders and Enk2017) chronologically and geographically (Supplementary Tables 1–3). The subgroups we identified are: (1) Early Pleistocene specimens from northwestern North America/Siberia (i.e., Beringia) and China; (2) Late Pleistocene specimens from northwestern North America/Siberia (i.e., Beringia); (3) Early/Middle Pleistocene specimens from the contiguous United States; (4) Late Pleistocene specimens from the northeastern United States; (5) Late Pleistocene specimens from the Great Plains and Great Lakes (Canada and USA); (6) Late Pleistocene specimens from the Rocky Mountains (USA); (7) Late Pleistocene specimens from the southern United States and Mexico.
Table 3. Radiocarbon data for last molars (M6/m6) of Mammuthus from Alberta.
Notes: Unless otherwise noted (i.e., CMN 17845) all specimens are housed at the Royal Alberta Museum. C:N represents the atomic ratio. n/a, not available. P97.11.1T is a postcranial fragment that represents part of an individual animal (P97.11.1), and the date provided is inferred to be equivalent to the age of the dentary preserving the m6 (P97.11.1b) included in this study.
a 13,073–12,740 cal yr BP (95.4% probability; calculated with OxCal v.4.4.4 using the IntCal20 calibration curve; Bronk Ramsey, Reference Bronk Ramsey2021; Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey and Butzin2020).
b 52,761–42,914 cal yr BP (88.7% probability; calculated with OxCal v.4.4.4 using the IntCal20 calibration curve; Bronk Ramsey, Reference Bronk Ramsey2021; Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey and Butzin2020).
Results
Radiocarbon dating
Table 3 presents new and previously published radiocarbon data for some of the mammoth teeth included in our analyses. Of the 13 specimens with age data, only a single specimen has a finite age (CMN 17845). We interpret the previous finite age for P97.11.1T as erroneous, given that reevaluation of that specimen with newer equipment and techniques produced a non-finite age (Table 3).
Our radiocarbon results indicate that most of the known record of mammoth teeth from Alberta consists of specimens that predate the last glacial maximum (LGM). Whether that represents a biological reality or not (i.e., mammoths were rare in Alberta in finite time leading to the LGM and following the LGM) remains a testable scenario, awaiting additional specimens and radiocarbon data. There are finite ages on other mammoth elements found in Alberta (see Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017), but many of those records represent ages that likely should be reassessed with newer pretreatment and analytical techniques, particularly given the result for P97.11.1T. Collectively, our radiocarbon results mean that our discussion of mammoth teeth and taxonomy in Alberta is primarily focused on pre-LGM time.
Taxonomic evaluation and geographic distribution
Tables 1a, b and 2a, b summarize the measurements and calculated indices for the Alberta specimens in our study. Figures 5–9 are scatter diagrams of lamellae length and enamel thickness indices for the Alberta sample and comparative data sets from Lister and Sher (Reference Lister and Sher2015) and Widga et al. (Reference Widga, Saunders and Enk2017).
Figure 6. Plots of enamel thickness index (EI′) vs. lamella length index (LLI′) for M6s of Mammuthus from Alberta and other sites in Eurasia and North America (following Widga et al., Reference Widga, Saunders and Enk2017). Symbols: black circles (●) = specimens from Alberta; open circles (○) = M. primigenius from Eurasia and northwestern North America (Alaska); (x) = M. trogontherii from China and northwestern North America/Siberia (i.e., Beringia); open diamond (◊) = M. columbi from the Early and Middle Pleistocene of the contiguous United States; black diamond (♦) = Late Pleistocene Mammuthus from the northeastern United States; open triangles (Δ) = Late Pleistocene Mammuthus from the Great Plains and the Great Lakes (Canada and USA); plus signs (+) = Late Pleistocene Mammuthus from the Rocky Mountains (USA); rectangles (▭) = Late Pleistocene Mammuthus from the southern United States. Shaded area = estimated M. columbi–M. primigenius area of overlap. Data for Alberta, this study; other data from Widga et al. (Reference Widga, Saunders and Enk2017) and Lister and Sher (Reference Lister and Sher2015).
Figure 7. Plots of enamel thickness index (EI) vs. lamella length index (LLI) for M6s of Mammuthus from the Early (EP), Middle (MP), and Late Pleistocene (LP) of the contiguous United States and Mexico, based on data reported by Lister and Sher (Reference Lister and Sher2015). Data for Alberta, this study. Shaded area = M. columbi–M. primigenius area of overlap.
Figure 8. Plots of enamel thickness index (EI) vs. lamella length index (LLI) for M6s of Mammuthus from the Early (EP) and Late Pleistocene (LP) of northern North America and Siberia (i.e., Beringia), based on data reported by Lister and Sher (Reference Lister and Sher2015). Data for Alberta, this study. Shaded area = M. columbi–M. primigenius area of overlap.
Figure 9. Plots of enamel thickness index (EI) vs. basal lamella length index (LLBI) for m6s of Mammuthus from Alberta and other sites in Eurasia and North America (following Lister and Sher, Reference Lister and Sher2015). Symbols: black circles (●) = specimens from Alberta; open circles (○) = M. primigenius from northwestern North America/Siberia (i.e., Beringia); (x) = M. trogontherii from China and northwestern North America/Siberia (i.e., Beringia); open diamonds (◊) = M. columbi from the Early and Middle Pleistocene of the contiguous United States; open triangles (Δ) = Late Pleistocene Mammuthus from the Great Plains and the Great Lakes (USA); plus signs (+) = Late Pleistocene Mammuthus from the Rocky Mountains (USA); rectangles (▭) = Late Pleistocene Mammuthus from the southern United States and Mexico; black diamond (♦) = holotype of M. columbi (Falconer, Reference Falconer1857); black squares (■) = Osborn's (Reference Osborn1922) neotype of M. columbi and other specimens from the neotype locality; black triangles (▲) = one of Osborn's (Reference Osborn1942) M. jeffersonii “ideotypes” and specimens from the same “ideotype” locality. Shaded area = M. columbi–M. primigenius area of overlap. Data for Alberta, this study; other data from Lister and Sher (Reference Lister and Sher2015). Note that there are two data points for specimen RAM P97.11.1b; the data point to the left is based on the estimated LLBI for the tooth, the data point to the right is based on the calculated LLBI for the medial side of the tooth (refer to “Materials and Methods” for more details).
In this study, we identified specimens that plot with other mammoth molars displaying “typical” M. columbi morphology as M. columbi, and we identified specimens that plot with molars displaying “typical” M. primigenius morphology as M. primigenius. We referred to specimens that plot in the M. columbi–M. primigenius area of overlap as individuals with intermediate M. columbi–M. primigenius morphology, without making any specific taxonomic assignments or interpretations about their affinity.
Some Late Pleistocene North American mammoth specimens outside of Beringia with tooth morphologies that are intermediate to “typical” M. columbi and M. primigenius morphologies were previously identified as a distinct taxon, M. jeffersonii (see Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017). Some researchers hypothesized that M. jeffersonii evolved from M. columbi and developed more advanced dental traits approaching those observed in M. primigenius (Osborn, Reference Osborn1922; Kurtén and Anderson, Reference Kurtén and Anderson1980; Lister, Reference Lister2017). Those morphological changes are now hypothesized as likely the result of introgression between M. columbi and M. primigenius (Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016), a process that led to M. columbi populations with different degrees of “advancement” relative to the “typical” M. columbi morphology (Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017). Lister (Reference Lister2017) suggested referring to the “advanced” individuals as “Jeffersonian” M. columbi, without assigning formal taxonomic separation. Intermediate morphologies from Alberta that plot in the M. columbi–M. primigenius area of overlap could represent individuals of M. primigenius, M. columbi, hybrids of the two, or possibly “Jeffersonian” M. columbi. Without a strong analytical basis for any taxonomic interpretation, we prefer to simply note the existence of specimens with morphologies that fall within a range of overlap between M. columbi and M. primigenius. Moreover, using “Jeffersonian” M. columbi for specimens with intermediate morphologies implies that introgression only morphologically affected populations of M. columbi without similar effects in populations of M. primigenius, a process that seems unlikely in Alberta, given the known biogeographic fluidity of the region throughout the Pleistocene (e.g., Heintzman et al., Reference Heintzman, Froese, Ives, Soares, Zazula, Letts and Andrews2016; Karpinski et al., Reference Karpinski, Hackenberger, Zazula, Widga, Duggan, Golding and Kuch2020).
The 10 upper molars analyzed from Alberta cluster into two morphological groups (Figs. 5–7) thought to represent two distinct taxa. We recovered this pattern whether we employed the methodology and comparative data sets of Lister and Sher (Reference Lister and Sher2015) or Widga et al. (Reference Widga, Saunders and Enk2017), although we note that the Alberta sample size decreases by two specimens (ROM IBW.83 and ROM 28983) when using the methodology of Widga et al. (Reference Widga, Saunders and Enk2017). Six of the 10 Alberta specimens (RAM P84.5.1, P88.7.1, P91.10.4, P94.4.3, P97.8.1, and ROM IBW.83) plot with molars displaying “typical” M. columbi morphology (Figs. 5–7), including the neotype specimen proposed by Osborn (Reference Osborn1922) and specimens from the neotype locality (Fig. 5). The remaining four molars (RAM P80.14.1, P81.3.5, P02.8.70, and ROM 28983) plot with specimens displaying “typical” M. primigenius morphology (Figs. 5–7). When compared with only northern North America (Fig. 8), a similar pattern exists, but morphologies consistent with M. columbi occur in Early Pleistocene records in Beringia whereas morphologies consistent with M. primigenius occur in Late Pleistocene records in Beringia. Since some North American Mammuthus are hypothesized to represent chronospecies (e.g., M. trogontherii vs. M. columbi; Lister and Sher, Reference Lister and Sher2015), some specimens that we assigned to M. columbi could represent M. trogontherii, assuming that Early Pleistocene records represent that taxon.
In contrast to the pattern observed in the upper molars, the six lower m6 molars analyzed from Alberta do not cluster into two distinct groups (Fig. 9). These specimens form a morphological gradient that ranges from molars displaying “typical” M. primigenius morphology to molars displaying “typical” M. columbi morphology (Fig. 9). Molars with “typical” M. primigenius morphology include RAM P97.10.143 and P97.10.144. Two other molars, RAM P97.11.1b and P99.3.164, plot within the M. columbi–M. primigenius area of overlap. We note that RAM P97.11.1b plots within the M. columbi–M. primigenius area of overlap whether we use the estimated LLBI or the LLBI calculated only from the medial side of the tooth (see “Materials and Methods”). Molars with “typical” M. columbi morphology include RAM P21.391.1 and P18.319.4, and CMN 17845.
The mammoth teeth we studied are distributed over a wide geographic area in Alberta (Fig. 1). Molars with “typical” M. columbi morphology range from the Peace River region in northwestern Alberta to the Medicine Hat area in the southeastern region of the province. Molars with “typical” M. primigenius morphology range from the Edmonton area in central Alberta to the Medicine Hat area in southeastern Alberta. Molars with intermediate or overlapping M. columbi–M. primigenius morphology were recovered from the Edmonton area.
Discussion
Morphological variation and taxonomic identity of Alberta mammoth molars
Our analyses of upper and lower sixth molars (M6/m6) recovered from Pleistocene deposits in Alberta demonstrate the presence of morphologies consistent with M. primigenius and M. columbi, along with the occurrence of two lower molars with intermediate morphologies. Some of the taxonomic identifications suggested by our morphometric analyses are consistent with the identifications presented in previous studies, but others are substantially different (Table 4).
Table 4. Comparison of previous taxonomic hypotheses with current taxonomic hypothesis for M6/m6 molars of Mammuthus from Alberta. Unless otherwise noted, previous hypotheses are from Jass and Barrón-Ortiz (Reference Jass and Barrón-Ortiz2017)
For the upper molars we evaluated, our results support the previous or tentative identifications of M. primigenius (Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017; Lister, Reference Lister2017) for RAM P81.3.5, RAM P02.8.70, and ROM 28983, although we note that the latter specimen was also previously identified as M. columbi (Widga et al., Reference Widga, Saunders and Enk2017). Previous or tentative identifications of M. columbi (Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017; Widga et al., Reference Widga, Saunders and Enk2017) supported by our analyses include RAM P94.4.3, RAM P88.7.1, and ROM IBW.83. We revise previous or tentative identifications (see Jass and Barrón-Ortiz, Reference Jass and Barrón-Ortiz2017) of RAM P91.10.4, RAM P97.8.1, and RAM P84.5.1 to Mammuthus columbi.
For the lower molars we evaluated, our results support the previous identification of M. columbi for CMN 17845. CMN 17845 was originally described as Mammuthus imperator (Churcher, Reference Churcher1972), but was reassigned by Hills and Harington (Reference Hills and Harington2003) to M. columbi following the species concept for this taxon proposed by Kurtén and Anderson (Reference Kurtén and Anderson1980). In more recent studies, CMN 17845 was referred to M. columbi (Widga et al., Reference Widga, Saunders and Enk2017), and this taxonomic assignment is supported by our study, as this specimen plots with molars displaying “typical” M. columbi morphology.
The remaining lower molars in our study have revised identifications based on our analysis, and we present identifications for two previously undescribed specimens. The new records (RAM P18.319.4 and RAM P21.391.1) are assigned to M. columbi. RAM P97.10.143 and RAM P97.10.144 are reassigned from Mammuthus cf. M. columbi to M. primigenius. Intermediate morphologies of M. columbi and M. primigenius are preserved in RAM P99.3.164 and RAM P97.11.1b.
The discrepancies between our results and the tentative identifications of Jass and Barrón-Ortiz (Reference Jass and Barrón-Ortiz2017) are likely due to the quantitative methodologies employed here and our comparisons with baseline data sets of Lister and Sher (Reference Lister and Sher2015) and Widga et al. (Reference Widga, Saunders and Enk2017) versus the more qualitative observations that formed the basis for the original taxonomic hypotheses. Perhaps the most surprising result is the recovery of nine specimens that represent M. columbi, but recovery of only six specimens assigned to M. primigenius. That result contradicts our hypothesis that M. primigenius would be the most prevalent form of mammoth recovered in Alberta. At the very least, our results reemphasize the importance of quantitative evaluation of mammoth teeth, rather than reliance on other factors (e.g., geographic location). As predicted, mammoth teeth recovered from Upper Pleistocene deposits in Alberta show a large range of quantitative variation, ranging from “typical” M. primigenius to “typical” M. columbi morphologies, although with fewer intermediate morphologies than we might have speculated.
Biogeographic and spatial-temporal trends in the distribution of mammoth taxa in Alberta
Although some of our taxonomic results were unexpected, the range of morphological variation observed in our sample may support the hypothesis that M. columbi and M. primigenius occupied western Canadian landscapes in the Pleistocene and is consistent with studies that indicated that animal populations of southern and northern affinities inhabited this geographic region in the past (e.g., Shapiro et al., Reference Shapiro, Drummond, Rambaut, Wilson, Matheus, Sher and Pybus2004; Wilson et al., Reference Wilson, Hills and Shapiro2008; Burns, Reference Burns2010; Heintzman et al., Reference Heintzman, Froese, Ives, Soares, Zazula, Letts and Andrews2016). In our sample, M. columbi is the most abundant form (9/17 = 53%), followed by M. primigenius (6/17 = 35%), and specimens with intermediate morphologies (2/17 = 12%).
The fossil record of M. columbi suggests that this species was particularly abundant from the North American midcontinent south to Mexico and into Central America (e.g., Siebe et al., Reference Siebe, Schaaf and Urrutia-Fucugauchi1999; McDonald and Dávila A, Reference McDonald and Dávila A2017; Smith and Graham, Reference Smith and Graham2017), but our results may show a somewhat more consistent presence at northern latitudes than we would have predicted (Fig. 1). Those geographic data may indicate that M. columbi ranged as far north as the Peace River Country of northwestern Alberta (Fig. 1). That distribution could indicate a more widespread presence of that taxon in portions of Canada in the Pleistocene, or may suggest a pattern of northward migration during specific timeframes, consistent with the northward movement of other populations of “southern” megafauna during interglacial timeframes (e.g., Mammut americanum; Karpinski et al., Reference Karpinski, Hackenberger, Zazula, Widga, Duggan, Golding and Kuch2020). Although those may both represent plausible interpretations, we would be remiss to not offer an alternative explanation for our observations.
Only one specimen (CMN 17845) that we assigned to M. columbi has a finite radiocarbon date (10,930 ± 100 yr BP; 13,073–12,740 cal yr BP at 2σ probability; Table 3), and that specimen was recovered from southeastern Alberta (Fig. 1). All other radiocarbon data for the specimens we describe as M. columbi are non-finite ages (Table 3). While those data are consistent with biogeographic hypotheses presented above, they are also potentially consistent with an interpretation that the teeth represent older records of M. trogontherii. Descriptions of camelid remains from gravel deposits in Alberta included a record of “Giant Camel,” possibly indicative of an Irvingtonian or older fauna, and therefore significantly older than many other remains recovered from sand and gravel deposits in Alberta (Jass and Allan, Reference Jass and Allan2016). The occurrence of M. trogontherii in the Alberta record would potentially be consistent with that observation, and would suggest an even greater amount of mixing of specimens of disparate age in regional gravel deposits than is already recognized. Such a scenario would be consistent with previous interpretations that portions of Alberta represented extensions of the mammoth steppe during some portions of the Pleistocene (Schwartz-Narbonne et al., Reference Schwartz-Narbonne, Longstaffe, Kardynal, Druckenmiller, Hobson, Jass, Metcalfe and Zazula2019). However, because the majority of Quaternary fauna recovered in Alberta represents isolated finds with no detailed contextual information, we have no other definitive basis for this interpretation at present.
Collectively, our records of M. columbi may indicate (a) a broader, sympatric geographic distribution for M. columbi and M. primigenius in western Canada, (b) a northward dispersal of M. columbi during interglacial time that may or may not have been sympatric with M. primigenius, or (c) previously unrecognized, geologically older records of M. trogontherii, an Early to Middle Pleistocene form with tooth morphology indistinguishable from M. columbi. All three are plausible scenarios that cannot be resolved by molar morphology alone, although given our understanding of other records (e.g., Mammut, Karpinski et al., Reference Karpinski, Hackenberger, Zazula, Widga, Duggan, Golding and Kuch2020; camelids, Jass and Allan, Reference Jass and Allan2016) either scenario (b) or (c) seems the most likely explanation. Determining which scenario is more likely could be vastly improved by the discovery of additional remains in geologic context, but that may be a rare hope. The nature of the record in Alberta is such that Quaternary fossils deeply buried in gravels do not typically reveal themselves without the work of industry, and that usually means the remains are not recovered until the primary context is impacted. Another approach to further resolution would be to evaluate the genetic character of mammoth remains described here, as each scenario could potentially be evaluated on a genetic basis, and such an analysis would allow for further integration of our observations with emerging evolutionary observations based on aDNA (e.g., Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016; van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021).
In contrast to the predominant distribution of M. columbi in Alberta, the fossil record of M. primigenius in North America suggests that this species was primarily distributed along northern latitudes ranging from Alaska and Yukon south to the Great Lakes region (Saunders et al., Reference Saunders, Grimm, Widga, Campbell, Curry, Grimley, Hanson, McCullum, Oliver and Treworgy2010; Smith and Graham, Reference Smith and Graham2017; Wang et al., Reference Wang, Widga, Graham, McGuire, Porter, Wårlind and Williams2021). In our sample, molars with M. primigenius morphology occur as far south as the Medicine Hat area in southeastern Alberta, a pattern that is not surprising given that specimens of M. primigenius were recovered further south into the northern contiguous United States. Although scarce in the western United States, records of M. primigenius are known as far south as The Mammoth Site of Hot Springs, in southwestern South Dakota (Agenbroad et al., Reference Agenbroad, Lister, Mol, Roth, Agenbroad and Mead1994).
The two specimens with intermediate M. columbi–M. primigenius morphology were recovered from the central region of Alberta in the Edmonton area (Fig. 1). Whether these morphologies reflect sympatry and introgression of these two taxa or simply represent the margins of morphological space occupied by one species or the other is an open question that could be addressed by other types of analyses (e.g., aDNA).
At a broad timescale, individuals with morphologies in the core ranges of the morphological space of M. columbi and M. primigenius, and individuals with morphologies plotting in the intermediate zone between those ranges were present in central Alberta prior to the onset of the LGM. However, we do not have the resolution to determine more specific patterns of chronological distribution within pre-LGM time. Only one directly dated specimen from Alberta, a molar with M. columbi morphology recovered from a gravel pit in Bindloss, southeastern Alberta (CMN 17845), produced a radiocarbon date that postdates the LGM (Table 3). The age of this specimen suggests that M. columbi spread into the area that is now Alberta as the ice sheets receded during post-LGM times. Other post-LGM mammoth records lack elements that permit species assignment, but have associated or direct radiocarbon data indicative of dispersal into Alberta following deglaciation (e.g., Burnco Pit; Burns, Reference Burns2010). Given our biogeographic interpretation for CMN 17845, we hypothesize that these are also likely representative of south-to-north movement and likely represent M. columbi.
Tooth morphology and evolutionary models in mammoths
Temporal and spatial morphological trends in dental characters have traditionally been used to establish species boundaries and devise models of mammoth evolution (Osborn, Reference Osborn1942; Maglio, Reference Maglio1973; Madden, Reference Madden1981; Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2022), and we utilized those taxonomic approaches in this paper. However, recent molecular studies provide evidence to suggest that the evolutionary history of mammoths is more complex than originally hypothesized based on dental characters (Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016; Chang et al., Reference Chang, Knapp, Enk, Lippold, Kircher, Lister and MacPhee2017; Palkopoulou et al., Reference Palkopoulou, Lipson, Mallick, Nielsen, Rohland, Baleka and Karpinski2018; van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021). Ultimately, integrating morphological patterns of dental variation with variation at the molecular level and accurate dating of individual specimens will be necessary to advance our understanding of mammoth evolution, systematics, and phylogeography. In this regard, mammoth teeth recovered from Alberta, or from similar areas that may serve as biogeographic corridors, may play a significant role in advancing our understanding of morphological and molecular variation in Late Pleistocene North American mammoths, given the intermediate geographic location between Beringia to the north and the North American midcontinent to the south.
The occurrence of discrepancies between evolutionary models based on dental morphology and those based on molecular data are not overly surprising. The study of molar morphology in mammoths and other elephantids has undoubtedly revealed major events in the evolutionary history of these animals (Maglio, Reference Maglio1973; Madden, Reference Madden1981; Lister and Sher, Reference Lister and Sher2015; Saarinen and Lister, Reference Saarinen and Lister2023). However, the molar dentition is one of many character complexes that are subject to evolutionary change, each of which can evolve at different times and rates. Furthermore, the evolution of molar morphology in mammoths and other elephantids appears to be largely influenced by developmental timing, the interplay between structural and functional constraints, and selective pressures associated with feeding ecology and/or environmental parameters such as aridity (Maglio, Reference Maglio1973; Herridge, Reference Herridge2010; Saarinen and Lister, Reference Saarinen and Lister2023). Therefore, it is unrealistic to expect that molar dentition by itself can inform us about the complete evolutionary history of mammoths. Nevertheless, some instances of population divergence may be recognized by dental characters and perhaps some morphologies represent instances of hybridization if interbreeding between species resulted in intermediate tooth morphologies (e.g., Lister and Sher, Reference Lister and Sher2015; Lister, Reference Lister2017). Some instances of population isolation and divergence in the absence of developmental or selection pressures on tooth morphology may remain cryptic. A potential example of this cryptic divergence in an unnamed lineage with M. trogontherii dental morphology was recently identified based on genetic evidence (van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021). Likewise, instances of hybridization between lineages with similar tooth morphologies may remain cryptic. A potential example of this cryptic hybridization is the postulated hybrid origin of M. columbi (van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021). As a result, it may be unsurprising, or even expected, that molecular phylogenies may depart from phylogenies derived from tooth morphology.
Teeth have played a significant role in developing models of mammoth evolution (e.g., Maglio, Reference Maglio1973; Lister and Sher, Reference Lister and Sher2015; Saarinen and Lister, Reference Saarinen and Lister2023), but by focusing on a single character complex, we inevitably miss grasping the whole history. We should not discard using teeth for understanding mammoth evolution, but they should continue to be analyzed in a broader morphological and molecular context. Likewise, evolutionary models based on molecular data cannot be fully understood in the absence of paleontological data about mammoths, other organisms they interacted with, and the habitats in which they lived. Rather than a final word on mammoth taxonomy and distribution in Alberta, the data we present here represent a starting point for understanding the taxonomic character and evolutionary history of mammoths across a regional, dynamic Pleistocene landscape. Ultimately, understanding the mosaic of mammoth morphology, spatial and temporal distribution, and molecular history across distinct geographic settings will provide a more complete, nuanced understanding of evolution of Mammuthus as a whole.
Conclusions
The evolutionary history of mammoths (or any taxon) is a collective of regional patterns, and our work provides a starting point for understanding the evolutionary history of mammoths across a regional landscape in northern North America that is characterized by a history of major ecological disturbances (e.g., LGM), climatic change, and consequential biogeographic change, including the arrival of humans. At least three mammoth tooth morphologies are preserved in the Alberta record (M. columbi, M. primigenius, and morphologies falling within the intermediate zone of those taxa). The presence of teeth with a morphology indicative of M. columbi may document a broader geographic range for that taxon than previously recognized, but could also represent a much deeper time component to the history of Mammuthus in Alberta (i.e., M. trogontherii).
Morphological and ancient DNA studies suggest that hybridization and introgression played an important role in the evolution of several proboscidean taxa, including M. columbi and M. primigenius (Enk et al., Reference Enk, Devault, Widga, Saunders, Szpak, Southon and Rouillard2016; Chang et al., Reference Chang, Knapp, Enk, Lippold, Kircher, Lister and MacPhee2017; Palkopoulou et al., Reference Palkopoulou, Lipson, Mallick, Nielsen, Rohland, Baleka and Karpinski2018; van der Valk et al., Reference van der Valk, Pečnerová, Diez-del-Molino, Bergström, Oppenheimer, Hartmann and Xenikoudakis2021). Our morphometric study allowed us to identify specimens with intermediate morphologies in the sample of Alberta mammoth teeth, and that observation deserves further attention to determine if those teeth simply lie at the margins of morphological space that delineate different taxa or represent a morphological expression of interbreeding of distinct taxa. The blessing and the curse of working in the Pleistocene (and on mammoths, in particular) is a record in close enough proximity to the modern that we can document the existence of complex biological and morphological patterns, but sparse enough that our understanding of underlying mechanisms driving those patterns remains elusive.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/qua.2024.47
Acknowledgments
We thank Jim Burns and many donors and industry collaborators for collection of many of the specimens discussed in this paper. Data for Figure 1 were obtained from the Neotoma Paleoecology database (http://www.neotomadb.org), and the work of data contributors, data stewards, and the Neotoma community is gratefully acknowledged. Chris Widga and an anonymous reviewer provided constructive comments that improved the manuscript. We thank the Senior Editor (Derek Booth) and the Associate Editor (Curtis W. Marean) for their editorial comments and suggestions. The authors declare no conflict of interest.
