On the type material and evolution of North American mammoths

Quaternary International 443 (2017) 14e31

Contents lists available at ScienceDirect

Quaternary International journal homepage: www.elsevier.com/locate/quaint

On the type material and evolution of North American mammoths Adrian M. Lister Department of Earth Sciences, Natural History Museum, London SW7 5BD, UK

a r t i c l e i n f o

a b s t r a c t

Article history: Received 26 August 2016 Received in revised form 7 February 2017 Accepted 21 February 2017 Available online 4 April 2017

The type material (holotypes, paratypes, syntypes and neotypes) and nomenclatural history of North American mammoth species are described in detail, focusing on names that have been in recent use: Mammuthus columbi, M. imperator, M. jeffersonii, M. meridionalis, M. hayi, M. haroldcooki and M. primigenius. Biometric study of the type specimens of M. meridionalis nebrascensis, M. hayi and M. haroldcooki shows them to be within the range of variation of M. columbi. These and other specimens referred to these species have a misleadingly ‘primitive’ appearance that is due to advanced individual age or, in the case of M. hayi, to inaccurate reconstruction of fragmentary fossils. The type material of M. imperator is also indistinguishable from M. columbi, but this taxon has been used to categorise mammoth fossils thought to be of intermediate grade between M. meridionalis and M. columbi. Biometric data indicate no clear morphocline in North American mammoths through the Pleistocene, except for ‘advancement’ in some Late Pleistocene samples that have been categorised as Mammuthus jeffersonii. Genetic and morphometric data suggest that these represent part of a complex metapopulation that arose with the immigration of M. primigenius into the continent, followed by varying degrees of hybridization with endemic M. columbi. Where adequate single-site Late Pleistocene samples are available they span the whole range of morphologies from ‘typical’ M. columbi to ‘M. jeffersonii’. It is difficult to impose taxonomic boundaries on a complex evolutionary process, but a suggested compromise is to treat the whole range of Late Pleistocene variation as M. columbi but informally, if desired, using ‘Jeffersonian’ as a descriptive term for the more advanced individuals or samples. © 2017 Elsevier Ltd and INQUA. All rights reserved.

Keywords: Taxonomy Mammuthus Mammuthus Mammuthus Mammuthus Mammuthus

columbi imperator jeffersonii hayi meridionalis

1. Introduction The taxonomy of North American mammoths has had a confused history. In part this is due to issues of nomenclatural priority, inadequate type material and disputes on synonymy, but in retrospect it can be seen also to result from the complexities of the evolutionary process itself. In this contribution I review in some detail the type specimens of the most important taxa, which have their own historical interest, and then briefly assess the evolutionary models which these taxa have been thought to embody. I do not attempt to review all the numerous names that have been given

to North American mammoths over the past 200 years, or their type material, but restrict myself to those taxa that have been employed in more recent literature. These are the species columbi, imperator, jeffersonii, hayi, haroldcooki, and meridionalis, with shorter notes on primigenius. Lists of the numerous other historical names for North American mammoths, and their likely synonyms, can be found in Osborn (1942), Maglio (1973) and Madden (1981). I exclude from consideration Mammuthus exilis of the California Channel Islands; see Roth (1982, 1996) and Agenbroad (2003, 2012) for taxonomic discussion of this form. 2. Materials and methods

Abbreviations: ABDSP, Anza-Borrego Desert State Park; AMNH, American Museum of Natural History, New York; ANSP, Academy of Natural Sciences, Philadelphia; DMNS, Denver Museum of Nature and Science; MTA, General Directorate of Mineral Research and Exploration, Ankara; NHMUK, Natural History Museum, London; SMG, Sedgwick Museum of Geology, Cambridge, UK; UDSH, Utah Division of State History, Salt Lake City; UNSM, University of Nebraska State Museum, Lincoln; M3 and M3, upper and lower third molars, respectively. Further abbreviations are given in the caption to Table 1. E-mail address: [email protected]. http://dx.doi.org/10.1016/j.quaint.2017.02.027 1040-6182/© 2017 Elsevier Ltd and INQUA. All rights reserved.

All of the specimens discussed here have been examined by the author with the exception of the type material of Elephas jacksoni, Elephas americanus and Archidiskodon haroldcooki, which are lost, and the Brunswick Canal molars at ANSP. Dental measurements are based on Lister and Sher (2015), modified after Maglio (1973) and Lister and van Essen (2003), and are in mm. Enamel thickness is the mode of several measurements across the occlusal surface.

A.M. Lister / Quaternary International 443 (2017) 14e31

Lamellar frequency is averaged top and bottom of crown in uppers (LF), bottom of crown only (LFB) in lowers. LF is inverted to give average length of one lamella-cement interval using the formula LL ¼ 100/LF or LLB ¼ 100/LFB, where LL and LLB are lamella length and basal lamella length, respectively. Crown height, lamella length and enamel thickness are standardised to a crown width of 100 mm using the formulae HI ¼ 100  H/W, LLI ¼ 100  LL/W, LLBI ¼ 100  LLB/W, and ETI ¼ 100  ET/W, respectively, where H ¼ maximum unworn crown height, W ¼ maximum crown width including cement, ET ¼ modal enamel thickness, HI ¼ hypsodonty index, LLI ¼ lamella length index, LLBI ¼ basal lamella length index and ETI ¼ enamel thickness index. The standardizing procedure allows these variables to be compared irrespective of molar size, while not affecting the comparison between molars of the same size. However, data and plots for the unstandardized variables can be found in Lister and Sher (2015, supplementary materials). In the case of matched left/right pairs, either the most complete of the pair was measured, or in some cases, available measurements were combined from both teeth. Particular care was taken to assess the completeness or incompleteness of preserved molars. Much of the confusion or misidentification in past assessment of elephantid molars has arisen due to the misinterpretation of incomplete specimens. The peculiar mode of dental eruption and wear in elephantid molars means that, starting around the middle of the lifetime of the tooth, the anterior end of the crown has worn to the root and that, with continuing forward movement of the molar, it gradually loses length and lamellae from the front end. Molars thus affected retain progressively shorter length and fewer lamellae than they started with (Lister and Sher, 2015, Fig. 3). Their measured width will also often become reduced from its original value, as wear extends behind the point of original maximum width (usually in the anterior part of a third molar), and eventually also below the maximum width of each lamella (usually at some height above its base). At the time of writing the latter article I was unaware that precisely the same point had been forcibly made by Graham (1986: 168e169) thirty years previously, when he wrote: “variation in dental characteristics as a result of tooth wear must be considered in mammoth taxonomy” and went on to state, “As mammoth teeth wear the number of plates is reduced, the spacing between plates is increased and the enamel thickness is greater”. Osborn at one point (Osborn, 1942, p. 1087) showed recognition of this issue, although he elsewhere usually ignored it. Methods of recognizing, and where possible compensating for, wear effects are discussed in detail in Sher and Garutt (1987) and Lister and Sher (2015, supplementary materials). Both Sher and I were also unaware that in a remarkable but obscure paper, Hay (1922) had presaged some of these observations, writing: “a complete lower molar of an elephant possesses a strong anterior root which is distinctly separated from the more or less coalesced hinder roots. This root supports three, four, or possibly five plates. When the tooth is worn down so that this root is gone, one can no longer be certain just how many front plates are missing”. This corresponds closely to the observations on woolly mammoths by Sher and Garutt (1987), who introduced a method of estimating the number of missing plates, provided the anteriormost set of paired roots and their ‘marker plate’ (behind the isolated anterior and second roots) are preserved. In less derived elephants, the number of plates above the first root is fewer (see Lister and Sher, 2015). All of the species here discussed were originally placed in the genus Elephas by their founders. Osborn (1924) subsequently created the genus Parelephas for columbi and jeffersonii, while other species (meridionalis, imperator, hayi and haroldcooki) were transferred to the genus Archidiskodon (Osborn, 1942). Since Maglio (1973), most authors have included all these species in the genus

15

Mammuthus. To use Mammuthus throughout the historical accounts, however, would obscure the pattern of name changes which is part of the story; I have therefore kept to the original terminology, switching to Mammuthus when describing recent research. I trust this will not cause any significant confusion as the species names remain constant throughout. 3. Results 3.1. Mammuthus columbi The Columbian mammoth owes its name to the endeavours of two Scotsmen. In January 1846 the celebrated geologist Charles Lyell (1797e1875), on a tour of the United States, visited the excavations for the ill-fated Brunswick Canal, intended to link the Altamaha and Turtle rivers near Darien, Georgia. Local planter and scholar Hamilton Couper had there collected fossil remains from a superficial clay deposit, including a partial skeleton of Megatherium and elements of Mammut, Mylodon, Equus and Bison (Couper, 1843; Lyell, 1849), strongly suggesting Late Pleistocene age. Most of the remains were presented by Couper to the Academy of Natural Sciences in Philadelphia, where they were identified by Richard Harlan, but some (including an elephant tooth) were given to Lyell who took them back to Britain and handed them in turn to Scottish paleontologist Hugh Falconer (1808e1865), a specialist of fossil elephants. In 1846 Falconer examined but did not describe the elephant molar. The name Elephas (Euelephas) columbi first appears in a synoptic table for proboscidean species in Part I of Falconer's monograph on fossil mastodons and elephants (Falconer, 1857a). The species is said to occur in Mexico, Georgia and Alabama, but no specimens are named or illustrated. Falconer notes a questionable synonymy with E. jacksoni of Mather (1838) (see later). Later in 1857 Falconer presented Part II of his account to the Geological Society of London, but only a short abstract was published (Falconer, 1857b), including, however, the line ‘In the southern United States and Mexico a distinct fossil species, E. (Euelephas) columbi, hitherto undescribed, occurs’. Not until several years later did Falconer publish a full account of E. columbi including detailed description of the Brunswick Canal molar (Falconer, 1863). Thus, although E. columbi has become universally accepted as the valid name for the American mammoth, with Falconer (1857a) as its source, the situation is not entirely clear due to the lack of any description or type specimen in that publication. However, we know from Falconer (1863) that Lyell had shown him the Brunswick Canal fossils in 1846, and that the reference to ‘Georgia’ in Falconer (1857a) did refer to these specimens, which is sufficient to establish the priority of the name. The Code of Zoological Nomenclature (Art. 72.4.1.1) states: ‘For a nominal species or subspecies established before 2000, any evidence, published or unpublished, may be taken into account to determine what specimens constitute the type series’ (ICZN, 1999). The principal challenger is E. texianus, a name first coined by Richard Owen (1859, p. lxxxvi) for the elephant of the ‘warm and temperate latitudes of North America’, but without
ge
Charles any indication of its material basis. In 1862 Owen's prote Carter Blake described E. texianus with reference to a molar from San Felipe de Austin on the Brazos River, Texas (Blake, 1862; the specimen still exists as NHMUK PV OR 33218). The trajectory of the two names is similar but all subsequent authors (Osborn, 1942; Maglio, 1973; Madden, 1981) have accepted that Falconer (1857a) takes priority over Owen (1859) or Blake (1862). They have also all agreed (as did both Falconer and Blake) that texianus and columbi are synonyms. Lister and Sher (2015) measured and plotted the San Felipe ‘E. texianus’ molar, together with others from the same locality in the NHMUK collection, and found them fully

16

A.M. Lister / Quaternary International 443 (2017) 14e31

conformable with both the holotype of E. columbi and other samples referred to that species. Blake (1862) attempted to shore up his claim for E. texianus with the ‘grave charge’ that the name columbi was ambiguous e ‘was it named in honour of Columbus, or because it is found in Columbia?’ Falconer (1863) was scathing in his response: not only was Colombia (by which Blake had meant present-day Venezuela and adjacent countries) ‘nowhere in question as a habitat of the species’, but ‘no educated naturalist’ could be so ignorant of Latin grammar as to mistake columbi (named for the great explorer) for columbiae (of Colombia). The holotype of E. columbi is shown in Fig. 1. The label indicates that after Falconer's death it was presented to the British Museum by his brother Charles, and it is now at the Natural History Museum, no. NHMUK PV OR 40769. Falconer had the specimen sliced and polished longitudinally, as was his habit with many elephantid molars, the better to view its internal structure. Fortunately the cast at the American Museum of Natural History (no. 1747) appears to have been made before this event, so allows the tooth's full dimensions and external morphology to be seen (Osborn, 1942, fig. 887). A maximum crown width of 89 mm measured by both Falconer (1863) and Osborn (1942, p. 1072) compares with a combined width of only 83 mm if the two halves are placed together today; hence around 6 mm was lost during sectioning. The main label reads ‘Euelephas columbi’ in a round hand that is certainly not Falconer's spidery scrawl, but it was very likely directed by him, as Euelephas was the subgenus he erected to accommodate E. columbi, E. antiquus (the European straight-tusked elephant), E. primigenius (woolly mammoth) and E. maximus (living Asian elephant) (Falconer, 1857a). The specimen is a lightly worn right lower molar, almost certainly M3, lacking both anterior and posterior ends and showing 10 lamellae from the middle region of the tooth. In 1922 Henry Fairfield Osborn published a short account of North American mammoths which was the source of much of the subsequent taxonomic confusion in this group. He introduced, without any explanation, neotype specimens for both E. columbi and E. imperator, declaring ‘We … find by the characters of the type and neotype specimens that the real Elephas columbi is not the

animal we have been describing under this name; it is a dwarf form … of the animal which we have been describing under the name Elephas imperator’. For clarity, he concluded ‘The American mammoth heretofore widely known as “Elephas columbi”, the Columbian mammoth, will hereafter be known as Elephas jeffersonii, the Jeffersonian mammoth’ (Osborn, 1922). Within two months, Hay (1922) had issued a rebuttal: “Professor Osborn takes up first Elephas columbi and announces that the real E. columbi is not the animal that we have been describing under this name. Inasmuch as the elephant which has hitherto borne this honorable title is one well known and widely distributed, it is imperative that the name shall not be disturbed except on evidence that cannot reasonably be disputed”. Hay's comment embodies the confusion engendered by Osborn's wording, although with careful deciphering Osborn's intentions were not unreasonable. He evidently considered Falconer's type undiagnostic, which according to ICZN (1999, Article 75) is a potentially valid reason for establishing a neotype, although the holotype then loses any special status. He selected as neotypes of E. columbi upper and lower third molars (AMNH 13707) from the ‘Phosphate Beds’ of Charleston, South Carolina. Later, Osborn (1942) added data on a larger series of molars from localities in the same region, whose fauna is considered largely Rancholabrean in age
n and Anderson, 1980). The neo(Hay, 1923, pp. 155e6, 363; Kurte types were chosen to match as closely as possible the holotype in its preserved part, and thereby taking the whole neotype molar as representative of Falconer's species, Osborn perceived that many of the North American molars that had been until that point assigned to E. columbi were in fact more ‘advanced’. This was his meaning in suggesting that the ‘mammoth heretofore widely known as “Elephas columbi” will hereafter be known as Elephas jeffersonii’, not that E. columbi itself would cease to exist. He should have written: ‘Many of the specimens heretofore referred to Elephas columbi will hereafter be referred to a new species, Elephas jeffersonii’. While the columbi molars from Charleston, preserved at AMNH, are a valuable sample of the species, there are grounds for preserving the Brunswick, Georgia remains as the type material. First, the

Fig. 1. The holotype right M3 of Elephas columbi Falconer, 1857a,b, NHMUK PV OR 40769. A, lateral view of molar; B, inner view of same, showing vertical section; C, close-up of labels on medial side of molar. The lower label reads ‘Euelephas columbi’ and below it, in another hand that may be Falconer's, ‘Brunswick Canal, Georgia’. Scale bar for A & B, 5 cm. Photo: Harry Taylor, ©NHMUK.

2.57e2.70 2.70e2.84 2.60e2.80 19.80e20.79 22.52e23.71 19.12e21.24 2.7 2.7 2.3 20.79 21.37 19.12

292

94.0 85.0 84.0

140

>190 117.5 97.0 83.0

14p ∞17p ∞19p ∞28½p ∞24p x22p x22p ∞8p ∞11p ∞14 RU LU RL L&RU L&RL LL RL L&RL RL LL Guadalajara, Mexico Jonesboro, IN Jonesboro, IN Zanesville, OH Zanesville, OH Pendennis, KS Twin Creek, KY Crete, NE Frederick, OK Angus, NE

19 15

19 22 31 27 22 22

355

248 122.0

6.00 8.41 5.76 6.03 6.50 4.81 4.68 5.23

11.89 17.36 16.58

1.6 1.5 2.3

>162

140.0-147.4

10.12 17.90 19.98

1.36 1.55 2.77

Osborn's imperator neotype Osborn's jeffersonii holotype Osborn's jeffersonii holotype Osborn's jeffersonii paratype Osborn's jeffersonii paratype Osborn's jeffersonii 'ideotype' Osborn's jeffersonii 'ideotype' Barbour's hayi holotype Hay & Cook's haroldcooki holotype Osborn's nebrascensis holotype 1.72 13.31 203.3 2.1 6.16

197

16.23

193.1

status

Falconer's columbi holotype Osborn's columbi neotype Osborn's columbi neotype Owen's texianus holotype Leidy's imperator holotype 2.17 2.90 1.96 2.73 2.52

ETI LLI or LLBI

21.40 21.42 14.70 20.95 14.97 173.7

HI ET

1.9 3.0 2.0 3.0 3.3 18.73 22.17 14.99 23.04 19.31

LL or LLB

5.34 4.51 6.67 4.34 5.18 152

89.0 103.5 102.0 110.0 129.0

LF or LFB H W P

L 10∞14p x1815p ∞8-

PF R/L, U/L

RL LL RU LL RU Brunswick Canal, GA Charleston SC Charleston, SC San Felipe de Austin, TX Loup R, NE

AMNH AMNH AMNH AMNH AMNH AMNH AMNH UNSM DMNS DMNS

PV OR 40796 13707 13707 PV OR 33218 2491 (cast) 11871 9950 9950 10457 10457 21892 13225 1315 1057 CM1359 NHMUK AMNH AMNH NHMUK UNSM

Locality Number Collection

Table 1 Measurements of type specimens of North American mammoths. All are third molars. Key: PF¼ Plate Formula; ‘∞‘ ¼ anterior loss through wear; ‘-’ ¼ anterior loss through breakage; x ¼ anterior talon; p ¼ posterior platelet; P ¼ observed or reconstructed complete lamellar number excluding x or p; L ¼ crown length; W ¼ crown width including cement; H ¼ crown height; LF ¼ lamellar frequency (uppers); LFB ¼ basal lamellar frequency (lowers); LL ¼ lamella length (uppers); LLB ¼ basal lamella length (lowers), ET ¼ enamel thickness. The last three columns standardize other variables (H, LL or LLB, ET) to a nominal crown width of 100 mm (see Methods). Estimated values in italics. Measurements taken by the author on original specimens except: columbi holotype W from Falconer (1863), imperator & meridionalis nebrascensis holotypes from casts at AMNH & UNSM, jeffersonii holotype and Kentucky ideotype from Osborn (1922, 1942), and haroldcooki holotype from Hay and Cook 1930 and their photographs.

A.M. Lister / Quaternary International 443 (2017) 14e31

17

holotype molar is not without diagnostic features. It lacks total lamellar number, but preserves crown height, lamellar frequency and enamel thickness, plus originally measured width as an index of molar size (Table 1). In all these dimensions it falls within the range of m3s from various localities that are considered typical M. columbi (Fig. 2A; see also Lister and Sher, 2015). Osborn's (1922) suggestion that E. columbi, or at least its type material, was a ‘dwarf form’, moreover, cannot be sustained e he may have meant this in comparison with the remarkably wide type molar of E. imperator (see below), but it is that specimen that is an outlier, not the E. columbi material. Second, there is a small but valuable collection of additional mammoth molars from the Brunswick, GA excavation. These were mentioned by Leidy (1869) but were then forgotten until Madden (1981) noted that they remained conserved at the Academy of Natural Sciences in Philadelphia. The specimens, four in number (Fig. 3), comprise left and right, upper and lower third molars, probably of a single individual. Measurements of these specimens are given in Table 2, and the most complete M3 and M3 have been plotted in Fig. 2, where they conform closely to the holotype. These specimens, referred to as topotypes by Madden (1981) - specimens that are not formal types but that hail from the type locality e have additional value in characterizing the taxon. If a neotype for M. columbi were ever to be considered necessary, then these specimens, from the original type locality, should be selected according to the Code (ICZN, 1999, Article 75.3.6 and Recommendation 75A). Finally, the Charleston ‘neotype’ M3 and M3 are themselves incomplete. Although Osborn cited their observed lamellar counts (17 or 18 in M3 and 16 or 17 in M3) as diagnostic, examination of the originals at AMNH shows that the M3 has lost an uncertain number of lamellae at the back, so the original count was 19 or 20, while the M3 has lost a significant anterior portion through wear so that its lamellar count can be given only as >14. This point was clearly made by Hay (1922), noting the lack of the first root and, with it, an uncertain number of lamellae, although he conceded that the remains were clearly referable to E. columbi (cf. Fig. 2). 3.2. Mammuthus imperator The ‘Imperial mammoth’ was announced at a meeting of the Academy of Natural Sciences of Philadelphia on March 2nd 1858, in which celebrated paleontologist Joseph Leidy (1823e1891) “directed the attention of the members to some fossils on the table, being part of the collection obtained by Dr F.V. Hayden in the valley of the Niobara river, Nebraska”. Dr Leidy “exhibited part of an upper molar tooth of an elephant from the Niobara, which he assumed to be a species distinct from those previously indicated, though it does not present sufficient characters to establish the opinion. It is the broadest tooth he had ever seen, being almost five inches [125 mm], and it has fewer plates of enamel than in any variety of teeth of Elephas americanus [effectively any other North American elephant] that had come under his inspection. The species he proposed to distinguish by the name Elephas imperator” (Anonymous, 1858). It was the exceptional size of the holotype, therefore, that suggested its ‘imperial’ designation, although from the outset its inadequacy was recognized. Leidy (1858) later suggested that the fragment might be a new species only because ‘it was found in association with a fauna very distinct from any previously noted’. The specimen (Fig. 4) is conserved at the Smithsonian Institution (no. 185), with casts in several museums (e.g. AMNH no. 2568; UNSM no. 2491). It is a mid-crown fragment of a right upper molar, very likely from its size to be an M3 although the diagnostic posterior end is lacking. In the author's experience the specimen

18

A.M. Lister / Quaternary International 443 (2017) 14e31

Fig. 2. Lamella length (inverse of lamellar frequency) and enamel thickness of North American mammoth third molars (indexes standardized to crown width 100 mm). A, M3s; B, M3s. Key to scatterplots: turquoise squares, Early Pleistocene; yellow diamonds, Middle Pleistocene; pink symbols, Late Pleistocene (diamonds: Aucilla R, FL; triangles: Lamb Spring, CO; squares: Jones and Trolinger Springs, MO; stars: Hot Springs, SD; circles, other localities); blue open circles, Beringian M. primigenius; blue closed circles, continental N American M. primigenius; red ‘C’: M. columbi holotype from Brunswick Canal, GA, NHMUK PV OR 40796; black ‘c’, additional M. columbi from Brunswick Canal, ANSP 13046 and 13048; black ‘C’, M. columbi neotype from Charleston, SC, AMNH 13707; red ‘I’, M. imperator holotype from Niobara R, NE, AMNH 2568; black ‘I’, M. imperator neotype from Guadalajara, Mexico, AMNH 11871; black ‘J’, M. jeffersonii ‘ideotype’ from Pendennis, KS, AMNH 21892; black ‘j’, M. jeffersonii paratype from Zanesville, OH, AMNH 10457. Cross hairs represent estimated bounds for worn holotype specimens: blue, M. hayi from Crete, NE, UNSM 1315; black, M. haroldcooki from Frederick, OK, DMNS 1057; red, M. meridionalis nebrascensis from Angus, NE, DMNS 1359. Specimens above dashed line are Early and Middle Pleistocene plus some Late Pleistocene; this represents the approximate range of ‘typical’ Mammuthus columbi. Late Pleistocene specimens below and to the left of this line represent more advanced, ‘Jeffersonian’ M. columbi. Modified from Lister and Sher (2015, Figs. S34 and S43) where plots of other variables and full data can be found. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Niobrara River, before I had seen the account of the former, which induced me to refer the Niobrara fossil to a species with the name of E. imperator’, and went on to declare that, on the evidence available, they were the same species. He reiterated (p. 255): ‘The notice of E. imperator was published … before I had seen Dr. Falconer's paper … published the previous year. The specimen assigned to E. imperator … exhibits the characters attributed by Dr. Falconer to E. Columbi, compared with the supposed American variety of E. primigenius’. (Leidy, in fact, considered both columbi and imperator to be junior synonyms of E. americanus DeKay, 1842, but the latter is now synonymized with Mammuthus primigenius (see below).) Osborn (1922) attempted to rectify the situation by defining as neotype for E. imperator an M3 from Guadalajara, Jalisco, Mexico (AMNH 11871; Fig. 5), presumably selected because of its large size (crown width 122 mm), although it is incomplete anteriorly so both its original length and lamella number (>14) are unknown. Even after furnishing each with a neotype, Osborn (1922) admitted the possibility that columbi and imperator could be synonyms, recognizing that this could only be determined when better material was available. He concluded, however (oddly to our eyes), that it was ‘eminently desirable’ that future evidence might show them to be separate. By 1942 he had decided that they were not only separate but should be assigned to different genera, Parelephas and Archidiskodon, respectively. As far as molar morphology was concerned, he cited only the larger size and greater investment of cement in distinguishing the imperator from the columbi holotype (Osborn, 1942, p. 1000) e both features subject to considerable intraspecific variation. Madden (1981, p. 69) considered the imperator holotype distinct from columbi because its enamel thickness (which he gives as 3.4 mm) was significantly different from a sample of columbi molars. However, similar values can be found among Late Pleistocene molars referred to that species, including one from Osborn's (1922, 1942:1075) M. columbi ‘neotype’ sample from Charleston, SC, at 3.4 mm (Lister and Sher, 2015, Supplementary Table 1). In its other measurable dimension, the lamellar frequency of 5.0, the holotype is unremarkable once scaled for the large size of the molar. The Jalisco neotype of imperator also plots within columbi range, if anything toward its ‘advanced’ end (Fig. 2B). If E. columbi Falconer, 1857a,b is considered a synonym of E. imperator Leidy, 1858, Falconer once again claims priority ‘by the skin of his teeth’. Several key works have accepted the synonymy
n and and removed imperator from consideration (e.g. Kurte Anderson, 1980; Agenbroad, 2003, 2005). Others have retained it (Aguirre, 1969; Maglio, 1973; Madden, 1981; Graham, 1986), although they have differed on the matter of Osborn's neotype: Maglio (1973) accepted that it compared well with Leidy's type as far as comparisons were possible, while Aguirre (1969) and Madden (1981) rejected it as indistinguishable from M. columbi. The question of whether specimens referred to M. imperator might represent an earlier stage of North American mammoth evolution than M. columbi is addressed in the Discussion. 3.3. Mammuthus jeffersonii

remains one of the largest (or at least widest) known elephantid molars, considering that it is incomplete and may not show the maximum width. Most authors, from Falconer (1863, p. 67) onwards, have recognized the inadequacy of the imperator type, and the consequent uncertainty over its distinction from M. columbi. Even Leidy (1869, p. 252), in a passage full of praise for Falconer, stated that ‘Nearly the same characters assigned to the teeth referred to E. Columbi I had recognized in the fragment of a molar from the

Osborn (1922) created the new species Elephas jeffersonii to accommodate North American mammoth fossils (M. primigenius excluded) that were more ‘advanced’ than M. columbi as defined by Falconer's type and his own neotypes. As type specimen he designated the skeleton excavated at a farm south-east of Jonesboro, Grant County, Indiana in 1903, and still standing in the AMNH, no. 9950. To this he added as paratypes associated left and right M3 and M3 from Zanesville, Ohio, originally described by Warren (1855) and acquired by the AMNH in 1906. Further specimens listed in

A.M. Lister / Quaternary International 443 (2017) 14e31

19

Fig. 3. Additional molars of M. columbi (probably associated) from the type locality of Brunswick Canal, Georgia. A, left M3, ANSP 13046 in occlusal and medial views; B, right M3, ANSP 13047 in occlusal and medial views; C, left M3, ANSP 13048 in occlusal and lateral views; D, right M3, ANSP 13049 in occlusal and medial views. Scale bar 10 cm. Photos courtesy of Ned Gilmour, © Academy of Natural Sciences, Philadelphia.

Table 2 Measurements of molars from Brunswick Canal, Georgia (type locality of M. columbi). For ANSP 13049, only five posterior lamellae remain, so values of LFB, LLB and LLBI are considered unreliable and are placed in brackets. Key as in Table 1. Specimen

R/L, U/L

PF

W

H

LF or LFB

LL or LLB

ET

HI

LLI or LLBI

ETI

NHMUK PV OR 40796 (holotype) M3 ANSP 13048 M3 ANSP 13049 M3 ANSP 13046 M3 ANSP 13047 M3

RL LL LL LU RU

e10e 13p 5p ∞10p 9p

89.0 96.0 >90 106.0 >85

152 [125 [125 >160 >150

5.34 4.59 (4.01) 5.63 6.33

18.73 21.79 (24.94) 17.75 15.80

1.9 2.0 2.5 2.1 2.0

173.7 [130 e >147 e

21.40 22.70 (<27.71) 16.75 <18.59

2.2 2.1 <2.8 2.0 <2.4

20

A.M. Lister / Quaternary International 443 (2017) 14e31

Fig. 5. Osborn's (1922) neotype right M3 of Elephas imperator, from Guadalajara, Jalisco, Mexico (AMNH 11871) in lateral view. Scale bar 10 cm.

Fig. 4. The holotype right M3 of Elephas imperator Leidy, 1858. A-B, Medial and occlusal views of AMNH cast 2568; C, Leidy (1869, Plate 25 fig 3) original drawing. Scale bar 5 cm.

Osborn (1922) - skulls from Washington, Texas and Ohio - technically also have the status of paratypes (ICZN, 1999, Art. 72.4.5). The species was named in honour of President Thomas Jefferson, ‘in commemoration of his long-continued devotion to mammalian palaeontology’. The Zanesville paratypes, preserved at AMNH (no. 10457), are very remarkable mammoth molars (Fig. 6). The M3 has a lamella formula of ∞28½p, with the marker lamella, second root, and back edge of first root all visible, but with only a single lamella preserved in front of the marker. This indicates 2e3 lamellae lost through wear (Hay, 1923, p. 134; Lister and Sher, 2015, supplementary materials), so the original formula was x30½p or x31½p, one of the highest counts ever recorded for an elephantid molar. Osborn (1922) gave a value of 30, although his method of counting differed from the one used here (Lister and Sher, 2015). In the M3, ∞24p is preserved, and it is unclear if the anteriormost preserved root is the true first root; if not, the original count is at least x27p (24½-26 for Osborn). Osborn (1942, p. 1084) was at pains to point out that his naming of E. jeffersonii had been published on July 8, 1922 while Hay's contending presidential epithet, E. roosevelti, that Osborn (1942, p.

1095) considered synonymous, had been published on September 30th of the same year. Notwithstanding, Osborn immediately started to play havoc with his own type material. Having designated the Zanesville, Ohio molars as paratypes of E. jeffersonii in Osborn (1922), he re-allocated them to a new species, E. washingtonii, in Osborn (1923), returned them to jeffersonii as the types of a new subspecies Parelephas jeffersonii progressus in Osborn (1924), and finally elevated that to species level as P. progressus in Osborn (1942), with the Zanesville molars as the only referred specimens. Osborn (1923) explained their removal from E. jeffersonii as a consequence of comparing them with the molars of the Jonesboro holotype, which by then had been cut out of the skull and mandible (Fig. 7). The illustrations of the holotype molars show that they are themselves far from complete, with the front root and an uncertain number of lamellae worn away anteriorly; this was recognized by Osborn in giving their lamellar formulae as ‘þ17/þ20’, although this included the posterior platelets so should now be written ∞16p for the M3 and ∞19p for the M3. He reconstructed their original counts as 25 and 24 respectively, which is possible but uncertain; the total loss of the anterior root implies at least three lamellae lost at the front, so values of 19 for the upper and 22 for the lower are the best estimate. Having removed the Zanesville molars as paratypes of E. jeffersonii, Osborn (1942, p. 1087) replaced them with mandibles from Pendennis, Lane Co., Kansas (AMNH 21892) and Twin Creek,

A.M. Lister / Quaternary International 443 (2017) 14e31

21

technically a type; the term is now mainly used in crop breeding (Martre et al., 2014)). Both jaws contain complete M3s, with plate counts given by Osborn as 24, but taking account of the anterior talon and posterior platelet, would now be written x22p. The Pendennis ‘ideotype’ M3 is plotted in Fig. 2A, and other specimens from the same locality were plotted by Lister and Sher (2015); they fall in the middle of the range of variation of M. columbi as understood here. Aguirre (1969), Maglio (1973, p.62e3) and Graham (1986) rejected M. jeffersonii as a species distinct from M. columbi, although they recognized a phyletic lineage within the latter species, of which specimens that had been referred to M. jeffersonii represented a ‘progressive’ form with increased lamellar number
n and Anderson (1980), and frequency, and thinner enamel. Kurte conversely, followed Osborn in separating the two species. In this they were followed by Madden (1981), although he considered E. jacksoni Mather, 1838 to take name priority over Osborn's jeffersonii. This species was founded on a mandible from Salt Creek, Jackson County, Ohio, and is known only from a crude sketch of the mandible (Mather, 1838) and a drawing of the occlusal surface of one of the molars (Foster, 1839). They are inadequate for species diagnosis and the original is long lost, so the name is best forgotten, as also concluded by Osborn (1942, p. 1084). 3.4. Mammuthus hayi

Fig. 6. Osborn's (1922) original paratype M3 and M3 of Elephas jeffersonii from Zanesville, Ohio (AMNH 10457), subsequently reallocated to Elephas washingtonii, then to Parelephas jeffersonii progressus, later to Parelephas progressus, and here identified as Mammuthus primigenius. A, lateral view of left M3; B, lateral view of left M3; scale bar 10 cm.

Kentucky (AMNH 13225) which, recognizing that true paratypes must be part of the original type series (ICZN, 1999, Art. 72.4.5), he designated as ‘ideotypes’ (meaning typical of the species while not

Fig. 7. Upper and lower third molars extracted from the type skeleton of Elephas jeffersonii (AMNH 9050) (from Osborn, 1942, fig. 959).

This species was introduced as Elephas hayi by Barbour (1915) in honour of O.P. Hay. The holotype is a mandible (UNSM 1315) found in the Hurlbert sand-pit on the Blue River at Crete, Saline County, Nebraska (Fig. 8). The site was indicated by Hay (1923, p. 101) to be at the western edge of the ‘Kansan drift’ and of likely ‘Aftonian’ age (possibly Middle Pleistocene), but there is no clear dating evidence associated with the specimen. The name hayi has been commonly used for North American mammoth specimens thought to be of primitive grade. Osborn (1942) placed it in the genus Archidiskodon, which he reserved for the most primitive Pleistocene elephantids within his ‘Mammontinae’ (mammoths). Maglio (1973, p. 62), who rejected the genus Archidiskodon, still treated the hayi holotype as primitive by referring it to Mammuthus meridionalis known from the European Early Pleistocene. Madden (1981, p. 49), however, preferred to keep it distinct as Mammuthus (Archaeomammuthus) hayi. The name continued to be used until quite recently for North American mammoths thought to be of primitive grade (e.g. Webb and Dudley, 1995; Morgan and Hulbert, 1995; Arroyo-Cabrales et al., 2003). However, concepts of the place of ‘M. hayi’ in mammoth evolution have varied significantly, e.g. Madden (1981, p. 57) considered it more primitive even than European M. meridionalis, while Webb and Dudley (1995) saw it as considerably more advanced than the latter species. All accounts of the type mandible have described it as having a long, low corpus, and M3s complete with only 10e11 lamellae (Barbour, 1915; Osborn, 1942, p. 1023; Madden, 1981, p. 49). This would place it on a level with the most primitive Eurasian Mammuthus, of Late Pliocene to early Pleistocene age (Lister and van Essen, 2003). Indeed Madden (1981, p. 52), on this basis, eschewed an origin from European M. meridionalis (typically with 12e14 plates in M3) proposed by Maglio (1973) and others, and opted for an earlier stage of European mammoth evolution, identified at the time as M. gromovi (Alexeeva and Garutt, 1965). However, detailed examination of the Mammuthus hayi holotype (Fig. 8) shows it to be largely a fabrication. The ‘mandible’ is constructed mainly of plaster, into which have been inserted the posterior ends of matching left and right worn M3s, and in front of each of them, a fragment of another tooth which appears to be the

22

A.M. Lister / Quaternary International 443 (2017) 14e31

Fig. 8. Holotype mandible and dentition of Elephas hayi Barbour, 1915, UNSM 1315. A, mandible (largely fabricated) in right lateral view; B, posterior portion of left M3 with fragment of upper molar inserted in front; C, posterior portion of right M3 with fragment of upper molar inserted in front; D, right M3 removed from mandible to show fragment of upper molar glued in front. Scale 10 cm.

posterior end of an upper molar placed back-to-front. The left dentition lifts out of the ‘mandible’, whence it can be seen that the two pieces of tooth have been glued together. In part this fits Barbour's (1915) description of the specimen having been retrieved in fragments, including pieces of tusk. It is therefore plausible that the upper molar fragments belong to the same individual as the lower molars, though this is not certain. However, it is impossible that these fragments, which show the subdivided apices near the top of a barely-worn portion of crown, could be the anterior parts of the M3s that are more deeply worn than them. The published lamellar count of 10e11, based on the 7e8 lamellae remaining of the M3s plus the 3 lamellae of the artificially added fragments, is totally spurious. Since the ‘mandible’ is largely fabricated, and the anterior parts of the ‘molars’ do not belong with the M3s, Lister and Sher (2015) recommended that the holotype of E. hayi be restricted to the left and right M3 remnants only. This is in accordance with Article 73.1.5 of the Code, which states: ‘If a subsequent author finds that a holotype which consists of a set of components (e.g. disarticulated body parts) is not derived from an individual animal, the extraneous components may, by appropriate citation, be excluded from the holotype’ (ICZN, 1999). Restricting description to the M3s, these show paired roots at the front, indicating that the single anterior root, which is invariably present in complete molars, has been lost though wear, so this is not the natural front end. The specimens in fact comprise the posterior remnants of very worn molars, with preserved lamellar formula ∞8p. The number of lamellae lost at the front, and hence the original full lamellar count, cannot be reconstructed, but it is likely to have been much greater than 8. The measured basal lamellar length and enamel thickness, for a reasonable estimate of

original molar width (preserved 94 mm, estimated original, 100e105 mm), are consistent with identification as Mammuthus columbi (Fig. 2A, blue cross-hairs). M. hayi should therefore be considered as a junior synonym of M. columbi, or else as indeterminate. Lister and Sher (2015) recommended that the name hayi, while ‘available’ in the sense of the Code, is not a valid taxon and should be dropped. 3.5. Mammuthus haroldcooki This species, originally Elephas haroldcooki, was based on a mandible found with other faunal remains at Holloman's Quarry, Frederick, Oklahoma (Hay and Cook, 1928). It was named in honour of Harold J. Cook (1887e1962), a noted fossil collector and briefly Curator of Paleontology at the Colorado Museum of Natural History (SNAC, 2016). The Holloman local fauna is thought to be early Irvingtonian (Dalquest, 1977), and has been suggested as 1.3 Ma in age (Cassiliano, 1999). The specimen was originally registered at the Denver Museum of Nature and Science (no. 1057), but its whereabouts are currently unknown, so the following is based on published photographs and measurements. The mandible (Fig. 9) was transferred to the genus Archidiskodon as A. haroldcooki by Osborn (1942). The name was retained by Dalquest (1977), but the specimen was referred to M. meridionalis by Maglio (1973, p. 62), and to M. imperator by Aguirre (1969) and Madden (1981, p. 71e2). Hibbard and Dalquest (1966) described it as M. imperator haroldcooki. Further description and illustration of the holotype were given by Hay and Cook (1930) and Osborn (1942, p. 1029). Hay and Cook (1930) indicated the molars to be M3s, presumably due to a tapering back end and bony plug behind. The molars

A.M. Lister / Quaternary International 443 (2017) 14e31

23

Fig. 9. Type mandible of Archidiskodon haroldcooki (Hay and Cook, 1928) from Holloman's Quarry, Frederick, Oklahoma; DMNS 1057. A, right mandibular ramus in occlusal view; B, enlargement of right molar; C, medial view of posterior part of molar; D, right ramus in lateral view. Photographs from Hay and Cook (1930), © Denver Museum of Nature and Science.

preserve 11 lamellae plus a posterior platelet, but have worn to the root at the front (Fig. 9A and B). Hay and Cook (1930), followed by Osborn (1942) and Madden (1981), interpreted the empty alveolus in front of the tooth as the alveolus for the true first root, and reconstructed the original lamellar number as 14, significantly lower than typical M. columbi. The interpretation of the molar as M3 seems likely to be correct, and if the empty anterior alveolus did indeed house the natural anterior (first) root of the molar, then the estimate of 14 for the original lamellar count would be reasonable and the interpretation of the specimen as ‘primitive’ validated. However, two features suggest a different interpretation. The first is the length/width proportion of the crown. The abraded surface of the preserved molar was said to be 182 mm long and the crown width including

cement 85 mm (Hay and Cook, 1930). From these data the length from the back of the molar to the front of the empty alveolus (i.e. the proposed original length of the tooth) can be estimated from the photograph as ca. 216 mm, giving a L/W ratio of 182/85 ¼ 2.54. This would make the molar too short for an M3, even of M. meridionalis. Second, the position of the alveolus, medial to the midline of the crown (Fig. 9A), is not correct for a true anterior root, especially as the crown is bending strongly laterally in its preserved anterior part. An anterior root would have been more laterally placed, but a more posterior root could have had this position. In this case more of the crown has been lost, but it is not possible to estimate how much, and the lamellar formula of the molar cannot be reconstructed beyond ∞11p, or taking account of the missing lamellae up to the front of the alveolus, >14.

24

A.M. Lister / Quaternary International 443 (2017) 14e31

Other features of the molar, and the mandible in which it is housed, do allow us to assess its taxonomic position, however. First, the molar is high-crowned and the mandible short and deep (Fig. 9C and D). Hay and Cook (1930) indicated that 'the hind end of the crown is almost 6 inches [15 cm] high'. A crown of this height is possible for the posterior end of an M3, but only of an ‘advanced’ mammoth species. Moreover, in an M3 the maximal height, further forward in the crown, would have been even greater. The photograph (Fig. 9C) also shows a pronounced ‘kink’ in the lamellae in medial view; this is not seen in mammoths of the grade of M. meridionalis but is characteristic of ‘advanced’ species like M. primigenius and M. columbi, especially in M3. The tooth has evidently worn very obliquely, since the back is very high and the front worn to the root. Enamel thickness was given as ‘about 3 mm’ by Hay and Cook (1930); it can be measured it at several points on their photograph (Fig. 9B), producing an average value of 2.87 mm. However, because of the oblique angle of the lamellae to the occlusal surface (ca. 20 ; Fig. 9C), this should be multiplied by the cosine of 20 (0.9397), producing an approximate value of 2.7 mm normal to the lamella. This is within range for an M3 of M. columbi of original molar width ca. 100 mm (Fig. 2A, black cross-hairs). The oblique wear, together with the divergence of lamellae toward the base, also accounts for the low LF of 4.4 measured by Hay and Cook (1930). Dividing by the cosine of 20 gives an approximate LF, parallel to the lamellae, of 4.68 and an LL of 21.37. When standardised for approximate molar size, lamellar length falls within M. columbi range (Fig. 2A). Even if this tooth were an M2, the high crown (and HI of ca. 150/ 85 ¼ 1.76), maximal at the posterior end of the crown in an M2, implies an advanced mammoth of M. columbi grade. Similarly, the minimal reconstructed lamellar count of 14 would fit M. columbi (typically 12e15: Lister and Sher, 2015, table S2), not a mammoth of the grade of M. meridionalis (9e11: Lister and van Essen, unpublished). Significantly, Hay and Cook (1930) described two additional dental remains of Mammuthus from the Holloman gravel pit. One (DMNS 1059) is part of an M3, the other (DMNS 1060) part of an M3. Both were considered by Hay and Cook to be M. columbi on the basis of their dental features. In the M3, an LF of 6 and enamel thickness of 2.5 mm were recorded; on the M3, an LF of 7, all typical M. columbi values. Madden (1981, p. 72) redetermined the upper molar as an M2 and referred the specimens to M. imperator. Dalquest (1977) supposed that these specimens were erroneously recorded as being from Holloman's quarry since ‘the sympatry of Mammuthus haroldcooki and M. columbi would be most unlikely’. A more parsimonious interpretation is that all three specimens did come from the quarry and all (the mandible included) are referable to M. columbi. Madden (1981, p. 71) added information about a further important find from the same locality. An M3 conserved at the University of Texas, Austin has ‘about 17 plates and a plate ratio [LF] of roughly 6.6’ (E. Lundelius, quoted by Madden, 1981). These features are far from the ‘primitive’ condition previously reconstructed for the haroldcooki type and are consistent with M. columbi. 3.6. Mammuthus meridionalis nebrascensis A complete skeleton (minus cranium) was excavated on a tributary of the Little Blue River near Angus, Nuckolls County, Nebraska in 1931 (Holen et al., 2011), and is mounted at DMNS (no. 1359). The remains occurred in sands and clays sandwiched between deposits originally interpreted by C.B. Schultz as Kansas Gravels below and Loveland Formation above (Holen et al., 2011). It was therefore considered to be Middle Pleistocene in age. Osborn (1932) described the remains, proclaiming “The Elephas

meridionalis stage arrives in America”, and named the skeleton E. m. nebrascensis (later transferred to Archidiskodon meridionalis nebrascensis: Osborn, 1942, 1033). This appeared to corroborate an early age for the specimen, in turn fuelling the controversy over whether Angus was an archaeological site (Holen et al., 2011). The specimen was accepted as M. meridionalis by Maglio (1973, p. 62) and McDaniel and Jefferson (2003), although Martin (1969), Aguirre (1969) and Madden (1981) referred it to M. imperator. Its significance lay not in the subspecific epithet, but in supposedly illustrating the dispersal into North America of the primitive Eurasian species M. meridionalis, as putative ancestor to later forms such as imperator, columbi and jeffersonii (Maglio, 1973). The mandible (Fig. 10) contains deeply-worn M3s. Osborn (1932) counted 13 lamellae; Madden (1981, p. 73) reconstructed the original number as 14 on the left, 16 on the right. As preserved, the left M3 shows ∞13p, the right ∞10p with a long anterior dentine platform the length of three worn-out lamellae, but the original plate-count cannot be estimated with confidence. The Angus mammoth is another example of a deeply-worn lower molar taking on an apparently more ‘primitive’ appearance due to wear, as suggested by Holen et al. (2011). The measured basal lamellar frequency (5.23), lamellar length (19.1 mm) and enamel thickness (2.3 mm), however, normalized for a reasonable estimate of original molar width (ca. 90e100 mm; preserved 84 mm), are consistent with identification as Mammuthus columbi (Fig. 2A, red crosshairs). Mandible shape also played a significant part in the allocation of the Angus skeleton to M. meridionalis: Osborn noted its ‘very prominent rostrum and relatively elongate and shallow ramus’. Most North American mammoth mandibles, by contrast, are morphologically similar to Eurasian M. trogontherii, with relatively short, high horizontal ramus and symphysis, and wide, upright coronoid. While primitive mammoths do tend to be characterized by a more elongate corpus and rostrum (Rabinovich and Lister, 2016), mandible shape is subject to considerable individual, including ontogenetic, shape variation (see further, online Supplementary Data). The Angus mandible shows M3 in late wear, indicating advanced age. The same is true of several other North American mammoth mandibles showing an apparently more ‘primitive’ morphology, such as the specimen from the Ocotillo Formation (1.1 Ma) assigned to M. meridionalis by McDaniel and Jefferson (2006; see Lister and Sher, 2015). Moreover, similar mandibular morphology can be seen in old individuals of demonstrably Late Pleistocene age, suggesting that an ontogenetic effect to some extent replicates the phylogenetic one. An example is the mandible of M. columbi from Huntington, Utah (Gillette and Madsen, 1993), which is compared to the Angus mandible in Fig. 11. McDaniel (pers. comm. 2006) considered the Huntington mandible a possible M. meridionalis, but as it has a direct radiocarbon age of 11,220 ± 110 BP, such late survival seems highly improbable and I agree with Gillette and Madsen (1993) in assigning the specimen to M. columbi. Holen et al. (2011) relocated the Angus site and conducted new excavations, finding a fragment of mammoth rib at a similar depth to the original skeleton. The deposit overlying the mammoth was reinterpreted as terrace alluvium rather than Loveland Loess, and luminescence dates directly above the bone bed gave ages in the region 70e60 ka, the mammoth considered to be not more than a few thousand years older than this. The last-glacial (Wisconsinan) age of the specimen provides further evidence that this kind of 'primitive' mandible morphology can be found in Late Pleistocene material that is almost certainly ontogenetically old Mammuthus columbi. The mammoth skeleton is the only paleontological find from the original Angus locality. However, at a second location about one

A.M. Lister / Quaternary International 443 (2017) 14e31

25

Fig. 10. Mandible and molars of the Angus I mammoth, type of Archidiskodon meridionalis nebrascensis Osborn, 1932 (DMNS 1359). A-B, occlusal and right lateral view of original mandible, from Osborn (images © Denver Museum of Nature and Science); CeF from cast at UNSM: C, mandible in left lateral view; D, molars in left lateral view; E, left M3 in lateroocclusal view; F, right M3 in occlusal view. Scale bars 10 cm.

mile distant, named ‘Angus II’ by Madden (1981, p. 73 and appendix B), further faunal remains were discovered (Schultz and Tanner, 1957; Martin, 1969), including a mammoth mandible (UNSM 1506; Fig. 12). The fauna at ‘Angus II’ occurred in green silts underlying buried soils genuinely of the Loveland Loess Formation, and was considered by Schultz and Tanner (1957) and Martin (1969) to be no younger than early Illinoian. These authors demonstrated the close similarity of the Angus fauna to that of the Sheridan Beds represented at the sites of Hay Springs, Rushville, and Gordon Quarry, Nebraska, whose mammoths form a major part of the Middle Pleistocene sample in Fig. 2 (yellow diamonds). These assemblages also lie below the Loveland Loess and lack Bison (Schultz et al., 1978). Subsequent TL dating placed the loess between 200 and 120 ka (Bell et al., 2004, p. 285). The M3s (Fig. 12) were not amenable to detailed measurement, but show an advanced morphology with 19 lamellae and a high crown, and are closely comparable to M. columbi; Schultz and

Tanner (1957) and Madden (1981, p. 73) referred it to M. imperator. Madden (1981) considered the specimen part of the same ‘sample’ as the Angus I skeleton, but as it is now recognized as being older than the latter, its clearly advanced morphology argues even more strongly against the ‘primitive’ status of Angus I. 3.7. Mammuthus primigenius The woolly mammoth has had a complex nomenclatural history of which only the salient points will be given here; detailed accounts can be found in Osborn (1942, pp. 1117e1124) and Garutt et al. (1990). The name Elephas primigenius was coined in 1799 by celebrated physician and naturalist Johann Friedrich Blumenbach (1752e1840). A few months later Cuvier (1799) proposed the name E. mammonteus, but E. primigenius has priority and was the name adopted by subsequent authors, including Cuvier himself (Cuvier, 1806). Blumenbach (1799) had mentioned finds from various

26

A.M. Lister / Quaternary International 443 (2017) 14e31

Osborn (1942, p. 1122) e a partial M3 from Siberia and an upper molar from Osterode, Germany, identified as dP4 by Osborn but later redetermined as an M1 (Reich et al., 2007). Since only a single specimen can be a lectotype, Gromova (1965) nominated the Siberian molar, the Osterode molar thereby becoming the paralectotype. Both specimens, however, were believed to have been destroyed during WWII, along with other material in the Blumenbach collection (Reich et al., 2007). As a result, Garutt et al. (1990) designated a complete skeleton of woolly mammoth, excavated in the Taimyr Peninsula of northern Siberia in 1948, as the neotype of Mammuthus primigenius, and this was approved by the ICZN Commission (ICZN, 1991). The skeleton was described by Garutt and Dubinin (1951) and is mounted in the Zoological Museum in St Petersburg. Subsequently, during cataloguing work €ttingen Museum in 2005, the paralectotype molar from at Go Osterode was rediscovered (Reich et al., 2007). Because a paralectotype has no status as a name-bearing type, this does not disturb the status of the Taimyr skeleton as neotype (ICZN, 1999, Art. 74.1.3), even if the survival of a Blumenbach specimen is noteworthy. Were the lectotype to be rediscovered, it would immediately supplant the neotype (Art. 75.8). In North America, mammoth remains from Alaska and the Yukon have generally been referred to M. primigenius, although Osborn (1930, 1942, p. 1159) proposed a subspecific designation M. primigenius alaskensis. The type specimen of Elephas americanus DeKay, 1842, an upper molar from Irondiquoit River, New York (now lost), was shown to be metrically indistinguishable from M. primigenius by Osborn (1942, p. 1156e7). 4. Discussion Fig. 11. Comparison of mandibles from: A, old individual of end-Pleistocene M. columbi from Huntington, Utah (UDSH 88.18, from Gillette and Madsen (1993)); B, ‘Archidiskodon meridionalis nebrascensis’ from Angus I, Nebraska (DMNS 1359, from Osborn (1932), image © Denver Museum of Nature and Science). The images have been adjusted to the same length, and vertical lines drawn at the posterior limit of the ascending ramus, the peak of the coronoid process, the anterior edge of the molar alveolus, and the anterior tip of the rostrum, to emphasize the similarity in shape.

localities in Germany, including a famous skeleton discovered at Burgtonna in 1695. These have the status of syntypes, and include woolly mammoth teeth from Blumenbach's personal collection €ttingen Unithat were preserved at the Zoological Institute of Go versity (Garutt et al., 1990). The Burgtonna skeleton has since been recognized as that of a straight-tusked elephant, Palaeoloxodon antiquus, so Blumenbach's species was ‘composite’ in nomenclatural terms. In an attempt to resolve this issue, two of the €ttingen were declared ‘lectotypes’ by mammoth molars at Go

The evolutionary and migrational patterns shown by North American mammoths are ultimately more significant and more interesting than mere names. Nonetheless, ideas to date on these matters have been strongly influenced by taxonomic considerations. Maglio's (1973) groundbreaking study was largely centered on the Old World, but his brief coverage of North American mammoths established an evolutionary paradigm that has been followed by most subsequent authors, with some modifications of terminology (Fig. 13). Briefly, this paradigm envisages an evolving in situ lineage M. meridionalis e imperator e columbi e jeffersonii, between them spanning much of the Pleistocene. 4.1. M. meridionalis to M. imperator Osborn (1942, p. 998) at first assumed ‘Archidiskodon’ imperator to be derived from Eurasian ‘A.’ meridionalis, but after his ‘discovery’

Fig. 12. Mandible of M. columbi from Angus II, Nebraska (UNSM 1506). A, mandible in left latero-occlusal view; B, right M3 in postero-medial view. Scale bar in A, 10 cm.

A.M. Lister / Quaternary International 443 (2017) 14e31

27

Fig. 13. Simplified representation of taxonomic schemes promoted by various authors. The species durations for Osborn (1942) are a consensus of the various statements in his monograph; he was working with different stratigraphic concepts from today's, and many specimens were in any case undated. Osborn (1942, p. 1540e1) additionally recognized six further mammoth species in continental North America. Green shading represents advanced, ‘Jeffersonian’ morphology recognized by authors but not given formal taxonomic separation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

of North American meridionalis at Angus, the origin of imperator within that continent became plausible (Osborn, 1932, 1942 p. 1037). This link from supposed North American meridionalis to later species was accepted by Maglio (1973) and many subsequent authors. However, as discussed above, the Angus mammoths cannot be referred to Mammuthus meridionalis, nor does the type material of M. hayi or M. haroldcooki indicate the presence in North America of mammoths of meridionalis grade. Lister and Sher (2015) reviewed other North American fossils that have been allocated to these species, and in all cases molars are either indeterminate (lowers, in particular, often appearing ‘primitive’ because of advanced ontogenetic age), or where measurable show characters of plate number and hypsodonty well in advance of European M. meridionalis and within M. columbi range. On present data, the occurrence of meridionalis-grade mammoths in North America cannot be demonstrated. Although this contradicts Osborn's identification of M. meridionalis in North America, in another respect his ideas are corroborated, for he derived North American M. columbi not directly from M. meridionalis but from the European M. trogontherii lineage, signified by placing both in his genus Parelephas. The close relationship between columbi and trogontherii has been supported by Lister and Sher (2015) on morphometric and biogeographic grounds, the retention of separate species names largely a pragmatic decision based on long historical usage in North America and Eurasia, respectively. It is also precautionary, since apart from the molar teeth, detailed morphological comparisons have yet to be

made. Currently, therefore, the Beringian source population is here identified as M. trogontherii, while M. columbi is recognized in the continental United States. Lucas et al. (2017) have challenged the findings of Lister and Sher (2015), affirming the presence of M. meridionalis in North America and its ancestry to later North American mammoths. A detailed response to their concerns is given in the online Supplementary Data. 4.2. M. imperator to M. columbi As discussed above, the holotype molars of M. imperator and M. columbi are not distinguishable on their limited preserved morphology. Osborn (1922, 1942) admitted, and Maglio (1973) complained, that even the neotype molars of the two species were barely distinguishable, and the summary dental metrics given by Osborn (1942, pp 1088, 1585e6) show strong overlap between the two. Nonetheless, the idea that M. imperator represented a dentally more primitive mammoth than M. columbi began to take hold. It formed part of the evolutionary succession proposed by Maglio (1973), and may have been influenced by Osborn's (1942) allocation of the former to the ‘primitive’ genus Archidiskodon, the latter to the more ‘progressive’ Parelephas, even though Osborn himself saw them as separately immigrating taxa, not as successive parts of a single lineage. Maglio (1973), Madden (1981), Graham (1986) and at one time the present author (Lister, 1993) assumed a direct phyletic transition from imperator to columbi within North America

28

A.M. Lister / Quaternary International 443 (2017) 14e31

(Fig. 13). For these authors, therefore, the concepts of imperator and columbi had become divorced from their type material and had come to represent successive stages of ‘advancement’ in North American mammoth evolution, generally assumed to be of broadly Middle and Late Pleistocene age, respectively. The use of a species name to represent a morphology not embodied in its type material is a common occurrence and not excluded by the Code, although in such a case the designation of a neotype or the erection of a new species is to be preferred (ICZN, 1999, Art. 75). The more significant issue, however, is whether there really is an evolutionary trend in North American mammoths, however its successive stages might be labelled. The supposed differences were clearly stated by Madden (1981), who referred both species to Parelephas but stated ‘P. imperator is clearly distinct from P. columbi. Samples identifiable as the former species have more primitive skulls, significantly fewer plates on the molars, longer mandibles, postcrania larger’. However, the problem with the morphological separation of ‘imperator’ and ‘columbi’ by Maglio (1973), Madden (1981) and others, is that it was not based on stratified statistical samples, but largely on the allocation of individual specimens to one species or the other based on ‘primitive’ or ‘advanced’ appearance. It therefore ran the risk of artificially subdividing a single range of variation into typological ‘taxa’ (Fig. 14A). This was compounded by poor to non-existent dating for much of the material, and an apparently confirmatory association between ‘primitive-looking’ teeth and mandibles, both however due to advanced ontogenetic age (e.g. Figs. 10 and 11). In Fig. 2A and B, lamella length and enamel thickness are plotted, the dashed line drawn such that the area above it includes all Early and Middle Pleistocene specimens (yellow and turquoise symbols). This morphological boundary would represent an

approximation of the imperator-columbi divide for authors such as Maglio (1973) and Madden (1981). However, in all Late Pleistocene sites where there is an adequate sample size of well-preserved molars (Lamb Spring, CO; Jones and Trolinger Springs, MO; and Aucilla River, FL), the morphological range spans both sides of the line (Fig. 2, pink symbols), underlining the difficulty of defining an imperator-columbi division on molar evidence. For other paleontologists, notably museum curators faced with unidentified mammoth molars, species allocation could be largely arbitrary e hence mammoths dated to the latest Pleistocene were at some localities named M. imperator, at others M. columbi (Agenbroad, 1984, Table 3.4). The problem in identifying isolated specimens is exacerbated by the fact that even supporters of an imperator-columbi distinction published overlapping ranges for key variables (e.g. Osborn [1942, p. 1071] gave plate counts of 17e18 and 18e19 for M3s of the two species, respectively). These are the principal reasons that led Agenbroad (2005) and others to synonymize the two species as M. columbi. 4.3. M. columbi to M. jeffersonii Many Late Pleistocene samples, while clearly not pertaining to M. primigenius, do show some ‘advancement’ over ’typical’ M. columbi. These are the specimens that have generally been
n and classified as M. jeffersonii by Osborn (1922, 1942), Kurte Anderson (1980), Madden (1981) and others, or as advanced, ‘Jeffersonian’ M. columbi by Maglio (1973). The type material of M. jeffersonii is slightly more helpful than in the case of imperator and columbi; the holotype (from Jonesboro, Indiana) has a minimum of 22 plates, and three specimens from the replacement ‘paratype’ locality (Pendennis, Kansas), 20e23 (Table 1; Lister and

Fig. 14. Schematic morphological distributions with taxonomic divisions superimposed. The x-axis represents morphological ‘advancement’ from right to left. A, the normal distribution of a single species divided artificially. B, Normal distributions illustrating the Late Pleistocene reality suggested in this study (cf. Fig. 2): populations of unaltered columbi morphology (represented by pale blue curve) share the continent with others of unaltered primigenius morphology (dark blue curve) and intermediate, often widely variable, populations (e.g. red curve). The dashed line corresponds to the one in Fig. 2 on a columbi-jeffersonii taxonomic division, but due to overlap between the actual distributions, many individuals would be misclassified. The dotted line represents the division from primigenius, not marked on Fig. 2 but also subject to occasional problems of identification (overlap between red and dark blue curves). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A.M. Lister / Quaternary International 443 (2017) 14e31

Sher, 2015, tables S1eS2). Many other individual specimens or small samples (e.g. a molar from Leikem, Arizona with 24 plates: Saunders, 1970) are also ‘advanced’ compared to typical M. columbi. These fossils have been taken to reflect phyletic development of the M. columbi lineage, approaching M. primigenius molar morphology in some cases and sometimes rendering distinction from that species difficult (Madden, 1981; Saunders et al., 2010). Geographical separation has also been suggested, between jeffersonii in the north of the range and columbi in the south (Osborn, 1942, pp. 1067, 1586), or with jeffersonii distributed primarily in the Midwest of the USA (Saunders et al., 2010). Based in part on the array of observed Late Pleistocene morphologies, in part on the close relationship between North American M. primigenius, columbi and ‘jeffersonii’ on mitochondrial DNA (Enk et al., 2011, 2016), Lister and Sher (2015) suggested that hybridization among endemic M. columbi and incoming M. primigenius had led to a patchwork of populations showing different degrees of ‘advancement’ (Fig. 2, pink symbols; Fig. 14B). The broad range seen in many of these samples is consistent with hybrid origin, since hybrid populations among living species are often found to be more variable in metric characters than their parent populations (Neff and Smith, 1979). The entire Late Pleistocene assemblage could be termed
n and Anderson ‘M. jeffersonii’; this approach was adopted by Kurte (1980), who treated all Late Pleistocene mammoths (M. primigenius excepted) as M. jeffersonii, reserving M. columbi (syn. imperator) for their Middle Pleistocene precursors. However, such a policy would have to be based on stratigraphic age and not on observable characters, for there are Late Pleistocene samples that retain pure columbi morphology, while other samples span (in terms of this taxonomy) from the ‘primitive’ end of typical M. columbi to the ‘advanced’ end of ‘jeffersonii’ morphology (Fig. 2). An alternative strategy could be to recognise individual specimens as M. jeffersonii on the basis of morphology. According to this taxonomy, the diagonal lines in Fig. 2 would divide columbi (above) from jeffersonii (below). This, however, would again imply the division of natural populations by ‘cherry picking’ in a way that violates biological reality (cf. Fig. 14). Two alternative strategies have the advantage of combining a recognition of population integrity with a workable taxonomy. The first would be to note that the centroid of Early and Middle Pleistocene mammoth variation (turquoise and yellow symbols in Fig. 2) is displaced from that of most Late Pleistocene samples (pink symbols). The range or mean value of a sample could therefore be taken to distinguish a columbi from a jeffersonii population. However, a substantial sample size of molars would be required to identify the species, and given the total overlap of the two groups in the range of the older one (Fig. 2, above the dashed line), many individual specimens or small samples would remain indeterminate. The second strategy, here suggested as probably the best compromise, is to follow Maglio (1973) in treating the whole range of variation as M. columbi but informally, if desired, using ‘jeffersonii’ or ‘Jeffersonian’ as a descriptive term for the more advanced individuals. The latter term is preferable as it could not be confused with a formal taxonomic category. 4.4. Jeffersonian versus woolly mammoths A final problem arises with the separation of advanced, ‘Jeffersonian’ individuals of Mammuthus columbi (as defined above) from woolly mammoths, M. primigenius. We should first dispense with the suggestion that these two species should be synonymized on the basis of their likely hybridization in central latitudes of North America (see above). Late Pleistocene mammoths across the whole of northern Eurasia and eastern Beringia (Alaska and Yukon) are

29

identified as M. primigenius; those from the Early and Middle Pleistocene of continental North America, according to the scheme presented here, as M. columbi; in both cases without ambiguity. To lump all of these together as M. primigenius (the name that takes priority) would impede faunal and taxonomic research, cause unnecessary confusion and would probably be unenforceable. In any case, strict application of the Biological Species Concept in the face of hybridization is not generally considered necessary or advisable provided the two ‘parent’ species maintain their separate identities over most of their range (Harrison and Larson, 2014). Hybridization is thought to occur in at least ten percent of living animal species and to be particularly common in relatively young species, as reproductive isolation may be incomplete for several million years (Mallet, 2005). Considering the ranges of M. columbi and M. primigenius across both space and time, they were distinct and separate over large areas and for most of their existence, and taxonomy should reflect that. Identification to species will nonetheless continue to cause occasional problems for individual specimens in the region of morphological overlap (Fig. 2, pink vs. blue symbols). In many cases, the identification of woolly mammoth molars at sites in the continental US has proved unproblematic, based on a combination of high lamella count and lamellar frequency, thin enamel, and sometimes relatively small size. Examples from the recent literature include the upper levels of the Hot Springs Mammoth Site, South Dakota (Agenbroad et al., 1994); Scarborough, Maine (Hoyle et al., 2004); and Lincoln College and Painter Creek, Illinois (Saunders et al., 2010). In other cases, especially for isolated or incomplete remains, definite identification may not be possible beyond ‘Mammuthus sp.’ (e.g. Joyce, 2006; Saunders et al., 2010). To return to the exceptional molars from Zanesville, Ohio (Fig. 6), first named as paratypes of Elephas jeffersonii by Osborn (1922) and finally as Parelephas progressus, they have to be allocated to Mammuthus primigenius on the basis of their morphology (as originally suggested by Falconer, 1863, p. 55): the lamella number of 27e30 is within the upper limit of that species in Eurasia even if thus far unknown elsewhere in continental North America (Lister and Sher, 2015, figs. 4 and S36B), and the same is shown in lamella length and enamel thickness (Fig. 2, ‘j’). They are large for M. primigenius, but molars of similar size can be found in the samples from Alaska and elsewhere (Lister and Sher, 2015, tables S1 and S2). 5. Conclusion In tandem with advances in molecular data (especially nuclear DNA), progress on these issues will come with further work in two areas: (1) morphometric comparison of statistical samples of fossils reliably dated to different stratigraphic levels, and (2) the incorporation of data other than that of molars, notably skull characters. A start on the former has been made by Lister and Sher (2015) and in the present contribution, comparing molars of North American mammoths from Early, Middle and Late Pleistocene contexts. When complete or largely complete molars have been compared, there are found to be no significant changes through the sequence. Mammoths from the Early Pleistocene (e.g. Leisey Shell Pits, Florida: Morgan and Hulbert, 1995) have the same plate number, crown height and mandible form as those from Middle Pleistocene (e.g. sites in Sheridan County, Nebraska sites: Schultz and Tanner, 1957) and many specimens of Late Pleistocene age. Thus far, this provides no evidence for recognizing separate ‘imperator’ and ‘columbi’ stages of evolution, nor for the existence of a ‘meridionalis’ stage in North America, so on the basis of nomenclatural priority these samples are all referred to M. columbi. The additional existence of more ‘advanced’ mammoths in the

30

A.M. Lister / Quaternary International 443 (2017) 14e31

Late Pleistocene is likewise corroborated by the biometric data. There is considerable variation within and among these samples, corresponding to the ‘metapopulation’ concept outlined above. The different populations of ‘advanced’ mammoths named by Osborn (1942), P. jeffersonii, washingtonii and progressus, are a reflection of this diversity. The second line of evidence will come from the morphology of crania and mandibles. Osborn (1922, 1942) based the distinction between M. imperator and M. columbi in particular (since the molars were so similar), but also between these and M. jeffersonii, on skull morphology, and this was broadly accepted by Madden (1981). However, it is unclear to what extent ontogenetic, sexual and individual variation is confounding the observed differences. An objective morphometric study, taking into account all sources of variation and also using stratified, statistical samples where possible, is necessary to test these ideas. For example, a long rostrum length is often taken to be a ‘primitive’ feature in elephantid mandibles, yet Osborn (1942, p. 1089) himself noted that ‘in certain specimens referred to M. primigenius the rostrum is quite prominent, so we cannot place too great reliance on this character’ (see also, online Supplementary Data, Figs. S3-S4). The tendency of aged individuals to remold the main body of the mandible into a more ‘primitive’ form has already been discussed (Fig. 11). Concerning crania, Hay (1922) showed that the gross proportions of four US mammoth skulls all differed significantly one from the other. Although this could signify that they were all different species as Hay (1922) supposed, it is at least as likely to reflect individual variation, and needs to be investigated. Such a study may in the future throw up some surprises that force us to reevaluate the patterns based on the molar data. Acknowledgements This paper was written in memory of Larry Agenbroad, in appreciation of his generosity and friendship, especially during my month as visiting scientist at the Hot Springs Mammoth Site in 1991. For access to other collections I thank Pip Brewer (NHMUK), Jeff Saunders (Illinois State Museum), Teresa Hsu (Smithsonian Institution), Judy Galkin (AMNH), Richard Hulbert (University of Florida) and George Corner (UNSM). I am particularly grateful to Steve Holen of the Centre for American Paleolithic Research, Hot Springs, SD, for discussion of the Angus site, Liz Trevethick of the Falconer Museum, Forres, for historical insights, and Ned Gilmore for providing photographs and measurements of the Brunswick canal molars at ANSP. William Sanders and an anonymous referee provided valuable suggestions for improving the manuscript. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.quaint.2017.02.027. References Agenbroad, L.D., 1984. New World mammoth distribution. In: Martin, P.S., Klein, R.G. (Eds.), Quaternary Extinctions. University of Arizona Press, Tucson, pp. 90e108. Agenbroad, L.D., 2003. New absolute dates and comparisons for California's Mammuthus exilis. Deinsea 9, 1e16. Agenbroad, L.D., 2005. North American proboscideans: mammoths: the state of knowledge, 2003. Quat. Int. 126e128, 73e92. Agenbroad, L.D., 2012. Giants and pygmies: mammoths of Santa Rosa Island, California (USA). Quat. Int. 255, 2e8. Agenbroad, L.D., Lister, A.M., Mol, D., Roth, V.L., 1994. Mammuthus primigenius remains from the mammoth site of Hot Springs, South Dakota. In: Agenbroad, L.D., Mead, J.I. (Eds.), The Hot Springs Mammoth Site: a Decade of Field and Laboratory Research in Paleontology, Geology, and Paleoecology. Mammoth Site, Hot Springs, SD, pp. 269e281.


n sistema
tica de los Elephantidae por su morfología y Aguirre, E.E., 1969. Revisio morfometría dentaria. Parts 2-3. Estud. Geol. 25 (123e177), 317e367. Alexeeva, L.I., Garutt, V.E., 1965. New data on the evolution of the elephant genus Archidiskodon. Byull. Kom. Izuc. Chetvertignogo Perioda 30, 161e166. Anonymous, 1858. In: Meeting of January 19th. Proceedings of the Academy of Natural Sciences of Philadelphia 1858, p. 2. Arroyo-Cabrales, J., Polaco, O.J., Aguilar-Arellano, F., 2003. Remains of the genus Mammuthus housed in the collections of the Instituto Nacional de Antropología
xico. Deinsea 9, 17e25. e Historia, Me Barbour, E.H., 1915. A new Nebraska mammoth, Elephas hayi. Am. J. Sci. 40, 129e134. Bell, C.J., Lundelius Jr., E.L., Barnosky, A.D., Graham, R.W., Lindsay, E.H., Ruez Jr., D.R., Semken Jr., H.A., Webb, S.D., Zakrzewski, R.J., 2004. The Blancan, Irvingtonian, and Rancholabrean mammal ages. In: Woodburne, M.O. (Ed.), Late Cretaceous and Cenozoic Mammals of North America: Biostratigraphy and Geochronology. Columbia University Press, New York, pp. 232e314. Blake, C.C., 1862. On a fossil elephant from Texas (E. texianus). Geologist 5, 57e58. Blumenbach, J.F., 1799. Handbuch der Naturgeschichte, sixth ed. (Dieterich, €ttingen). Go Cassiliano, M.L., 1999. Biostratigraphy of Blancan and Irvingtonian mammals in the Fish Creek - Vallecito Creek section, southern California, and a review of the Blancan-Irvingtonian boundary. J. Vertebr. Paleontol. 19, 169e186. Couper, J.H., 1843. On fossil bones found in digging the New Brunswick Canal in Georgia. Proc. Geol. Soc. Lond. 4, 33e34.
moire sur les espe ces d’ e
le
phans vivantes et fossiles. Me
moires Cuvier, G., 1799. Me
m. Cl. Sci.Math. Phys. 2, 1e22. de l’Institut national des sciences et des arts. Me
le
phans vivans et fossiles. Ann. Mus. Hist. Nat. 8, 1-58, 93Cuvier, G., 1806. Sur les e 155, 249e269. Dalquest, W.W., 1977. Mammals of the holloman local fauna, pleistocene of Texas. Southwest. Nat. 22, 255e268. DeKay, J.E., 1842. Natural History of New York. Zoology, Part 1, Mammalia. Appleton and Wiley & Putnam, New York. Enk, J., Devault, A., Debruyne, R., King, C.E., Treangen, T., O'Rourke, D., Salzberg, S., Fisher, D., MacPhee, R., Poinar, H., 2011. Complete Columbian mammoth mitogenome suggests interbreeding with woolly mammoths. Genome Biol. 12, R51. Enk, J., Devault, A., Widga, C., Saunders, J., Szpak, P., Southon, J., Rouillard, J.-M., Shapiro, B., Golding, G.B., Zazula, G., Froese, D., Fisher, D.C., MacPhee, R.D.E., Poinar, H., 2016. Mammuthus population dynamics in Late Pleistocene North America: divergence, phylogeography, and introgression. Front. Ecol. Evol. 4, 42. http://dx.doi.org/10.3389/fevo.2016.00042. Falconer, H., 1857a. On the species of mastodon and elephant occurring in the fossil state in Great Britain. Part I. Mastodon. Q. J. Geol. Soc. Lond. 13, 307e360. Falconer, H., 1857b. On the species of mastodon and elephant occurring in the fossil state in England. Part II. Elephas [Abstract] Q. J. Geol. Soc. Lond. 14, 81e84. Falconer, H., 1863. On the American fossil elephant of the regions bordering the Gulf of Mexico (Elephas columbi Falc.), with general observations on the living and extinct species. Nat. Hist. Rev. 3, 43e114. Foster, J.W., 1839. Head of mastodon giganteum. Am. J. Sci. 36, 189e191. Garutt, V.E., Dubinin, V.B., 1951. On the skeleton of the Taimyr mammoth. Zool. J. 30, 17e23. Garutt, V.E., Gentry, A., Lister, A.M., 1990. Mammuthus Brookes, 1828 (Mammalia, Proboscidea): proposed conservation, and Elephas primigenius Blumenbach, 1799 (currently Mammuthus primigenius): proposed designation as the type species of Mammuthus, and designation of a neotype. Bull. Zool. Nomencl. 47, 38e44. Gillette, D.D., Madsen, D.B., 1993. The Columbian mammoth, Mammuthus columbi, from the Wasatch Mountains of central Utah. J. Paleont 67, 669e680. Graham, R., 1986. Taxonomy of North American mammoths. In: Frison, G.C., Todd, L.C. (Eds.), The Colby Mammoth Site: Taphonomy and Archaeology of a Clovis Kill in Northern Wyoming. University of New Mexico, Albuquerque, pp. 165e169. Gromova, V.I., 1965. Kratkii obzor chetvertichnykh mlekopitayushchiich Evropy (Summary of Quaternary Mammals of Europe). Nauka, Moscow, 144 pp. Harrison, R.G., Larson, E.L., 2014. Hybridization, introgression, and the nature of species boundaries. Heredity 105, 795e809. Hay, O.P., 1922. Further observations on some extinct elephants. Proc. Biol. Soc. Wash. 35, 97e102. Hay, O.P., 1923. The Pleistocene of North America and its Vertebrate Animals from the States East of the Mississippi River and from the Canadian Provinces East of Longitude 95 . Carnegie Institution, Washington, D.C., 499 pp. Hay, O.P., Cook, H.J., 1928. Preliminary descriptions of fossil mammals recently discovered in Oklahoma, Texas, and New Mexico. Proc. Colo. Mus. Nat. Hist. 8, 33. Hay, O.P., Cook, H.J., 1930. Fossil vertebrates collected near, or in association with, human artifacts at localities near Colorado, Texas; Frederick, Oklahoma; and Folsom, New Mexico. Proc. Colo. Mus. Nat. Hist. 9, 4e40. Hibbard, C.W., Dalquest, W.W., 1966. Fossils from the Seymour Formation of Knox and Baylor Counties, Texas, and their bearing on the late Kansan climate of that region. Paleontol. Contrib. Mich. Univ. Mus. 21, 1e66. Holen, S.R., May, D.W., Mahan, S.A., 2011. The Angus mammoth: a decades-old scientific controversy resolved. Am. Antiq. 76, 487e499. Hoyle, B.G., Fisher, D.C., Borns Jr., H.W., Churchill-Dickson, L.L., Dorion, C.C., Weddle, T.K., 2004. Late Pleistocene mammoth remains from Coastal Maine, USA. Quat. Res. 61, 277e288. International Commission on Zoological Nomenclature, 1999. International Code of Zoological Nomenclature, fourth ed. International Trust for Zoological

A.M. Lister / Quaternary International 443 (2017) 14e31 Nomenclature, London. Joyce, D.J., 2006. Chronology and new research on the Schaefer mammoth (? Mammuthus primigenius) site, Kenosha County, Wisconsin, USA. Quat. Int. 142e143, 44e57.
n, B., Anderson, E., 1980. Pleistocene Mammals of North America. Columbia Kurte University Press, New York, 442 pp. Leidy, J., 1858. Notice of remains of extinct vertebrata, from the valley of the Niobara River, collected during the exploring expedition of 1857, in Nebraska, under the command of Lieut. G.K. Warren, U.S. Top. Eng., by Dr. F.V. Hayden, geologist to the expedition. Proc. Acad. Nat. Sci. Phila. 1858, 20e29. Leidy, J., 1869. The extinct mammalian fauna of Dakota and Nebraska. J. Acad. Nat. Sci. Phila. (Second Series) 7, 1e472. Lister, A.M., 1993. Evolution of mammoths and moose: the Holarctic perspective. In: Martin, R.A., Barnosky, A.D. (Eds.), Morphological Change in Quaternary Mammals of North America. Cambridge University Press, Cambridge, pp. 178e204. Lister, A.M., Sher, A.V., 2015. Evolution and dispersal of mammoths across the northern hemisphere. Science 350, 805e809. nescu, the earliest Lister, A.M., van Essen, H., 2003. Mammuthus rumanus S¸tefa mammoth in Europe. In: Petculescu, A., S¸tiucǎ, E. (Eds.), Advances in Palaeontology ‘Hen to Panta’. Romanian Academy, Bucharest, pp. 47e52. Lucas, S.G., Morgan, G.S., Love, D.W., Connell, S.D., 2017. The first North American mammoths: Taxonomy and chronology of early Irvingtonian (early Pleistocene) Mammuthus from New Mexico. Quat. Int. http://dx.doi.org/10.1016/ j.quaint.2016.12.017 (in press). Lyell, C., 1849. A Second Visit to the United States of North America, vol. 1. Harper, New York, 273 pp. Madden, C.T., 1981. Mammoths of North America (PhD thesis). University of Colorado, 271 pp. Maglio, V.J., 1973. Origin and evolution of the Elephantidae. Trans. Am. Philos. Soc. New Ser. 63, 1e149. Mallet, J., 2005. Hybridisation as an invasion of the genome. Trends Ecol. Evol. 20, 229e237. Martin, L.D., 1969. A Medial Pleistocene Fauna from Near Angus, Nuckolls County, Nebraska (MSc thesis). University of Nebraska, Lincoln. Martre, P., Quilot-Turion, B., Luquet, D., Ould-Sidi Memmah, M.-M., Chenu, K., Debaeke, P., 2014. Model-assisted phenotyping and ideotype design. In: Sadras, V., Calderini, D. (Eds.), Crop Physiology: Applications for Genetic Improvement and Agronomy. Academic Press, London, pp. 349e373. Mather, W.W., 1838. Remarks in addition to and explanation of the review of the Report of the Geological Survey of Ohio in a letter to the Editor. Am. J. Sci. 34, 362e364. McDaniel, G.E., Jefferson, G.T., 2003. Mammuthus meridionalis (Proboscidea: Elephantidae) from the Borrego Badlands of Anza-Borrego Desert State Park, California: phylogenetic and biochronologic implications. Deinsea 9, 239e252. McDaniel, G.E., Jefferson, G.T., 2006. Mammoths in our midst: the proboscideans of Anza-Borrego Desert State Park, southern California, USA. Quat. Int. 142e143, 124e129. Morgan, G.S., Hulbert, R.G., 1995. Overview of the geology and vertebrate biochronology of the Leisey Shell Pit local fauna, Hillsborough County, Florida. Bull.

31

Fla. Mus. Nat. Hist. 37, 1e92. Neff, N.A., Smith, G.R., 1979. Multivariate analysis of hybrid fishes. Syst. Biol. 28, 176e196. Osborn, H.F., 1922. Species of American Pleistocene mammoths. Elephas jeffersonii, new species. Am. Mus. Novit. 41, 1e16. Osborn, H.F., 1923. New subfamily, generic, and specific stages in the evolution of the Proboscidea. Am. Mus. Novit. 99, 1e4. Osborn, H.F., 1924. Parelephas in relation to phyla and genera of the family Elephantidae. Am. Mus. Novit. 152, 1e7. Osborn, H.F., 1930. Parelephas floridanus from the Upper Pleistocene of Florida compared with P. jeffersonii. Am. Mus. Novit. 443, 1e17. Osborn, H.F., 1932. The “Elephas meridionalis” stage arrives in America. Proc. Colo. Mus. Nat. Hist. 11, 1e3. Osborn, H.F., 1942. Proboscidea, vol. 2. American Museum of Natural History, New York, pp. 805e1675. Owen, R., 1859. In: Address. Report of the Twenty-eighth Meeting of the British Association for the Advancement of Science, Held at Leeds in September 1858, xlix-cx. Murray, London. Rabinovich, R., Lister, A.M., 2016. The earliest elephants out of Africa: taxonomy and taphonomy of proboscidean remains from Bethlehem. Quat. Int. http:// dx.doi.org/10.1016/j.quaint.2016.07.010 (in press). Reich, M., Gehler, A., Mol, D., Lister, A., 2007. The rediscovery of type material of Mammuthus primigenius (Mammalia: Proboscidea). In: Abstracts, IV International Mammoth Conference, Yakutsk. 18-22 June, 2007. Institute of Applied Ecology of the North, Yakutsk, pp. 190e191. Roth, V.L., 1982. Dwarf mammoths from the Santa Barbara Channel Islands: Size, Shape, Development, and Evolution (Ph. D dissertation). Yale University, New Haven, CT. Roth, V.L., 1996. Pleistocene dwarf elephants from the California Islands. In: Shoshani, J.H., Tassy, P. (Eds.), The Proboscidea. University of Oxford Press, Oxford, UK, pp. 249e253. Saunders, J.J., 1970. The Distribution and Taxonomy of Mammoths in Arizona (MSc thesis). University of Arizona. Saunders, J.J., Grimm, E.C., Widga, C.C., Campbell, G.D., Curry, B., Grimley, D.A., Hanson, P.R., McCullum, J.P., Oliver, J.S., Treworgy, J.D., 2010. Paradigms and proboscideans in the southern Great Lakes region, USA. Quat. Int. 217, 175e187. Schultz, C.B., Tanner, L.G., 1957. Medial Pleistocene fossil vertebrate localities of Nebraska. Bull. Univ. Neb. State Mus. 4, 59e81. Schultz, C.B., Martin, L.D., Tanner, L.G., Corner, R.G., 1978. Provincial land mammal ages for the North American Quaternary. Trans. Neb. Acad. Sci. 5, 59e64. Sher, A.V., Garutt, V.E., 1987. New data on the morphology of elephant molars. Trans. Acad. Sci. USSR Earth Sci. Sect. 285, 195e199. SNAC (Social Networks and Archival Context), 2016. Cook, Harold James, 1887e1962. Biographical Notes. http://socialarchive.iath.virginia.edu/ark:/ 99166/w6wm1nzw (Accessed 4 August 2016). Warren, J.C., 1855. The Mastodon Giganteus of North America, second ed. J. Wilson, Boston. 219 pp. Webb, S.D., Dudley, J.P., 1995. Proboscidea from the Leisey Shell Pits, Hillsborough County, Florida. Bull. Fla. Mus. Nat. Hist. 37, 645e660.