North American Caballines and Amerhippines of the Past 1 Million Years (Part 1)

Abstract

Horses were widely distributed in North America during the Pleistocene epoch and their fossil remains are common in sedimentary deposits of this age. Despite their rich fossil record, the systematics and taxonomy of North American Pleistocene horses remain unresolved. We evaluated a large sample of cranial and postcranial horse fossils of Irvingtonian and Rancholabrean North American Land Mammal Age. In this study, we present Part 1 of our evaluation, which centers on caballine horses, Equus (Equus). We present data (measurements and photographs) and analyses (Simpson’s ratio diagrams, scatter diagrams, and anatomical comparisons) that reveal morphological variation in North American caballine horses. These analyses serve as the basis for recognizing different morphospecies: E. (E.) scotti, E. (E.) alaskae, E. (E.) lambei (the latter two possibly representing “ecological variants” of a single species), E. (E.) niobrarensis, E. (E.) pacificus, and E. (E.) complicatus. How these morphospecies (or chronospecies or ecological variants) were phylogenetically related remains to be evaluated. Equus (E.) hatcheri may be considered as a morphological variant or chronological variant of E. (E.) niobrarensis. Equus holmesi is considered a junior synonym of E. (E.) scotti, while E. bautistensis may be regarded as a junior synonym of E. (E.) pacificus. Equus laurentius is a junior synonym of E. (E.) caballus, a synonymy proposed previously in other studies. We are uncertain about the nature of E. midlandensis. In addition, we identify morphometric and anatomical features that distinguish between Equus (Equus), North American Equus (Amerhippus), and Equus (Hesperohippus) mexicanus. This study aims to advance our understanding of the taxonomy of North American Pleistocene horses, providing a thoroughly documented catalogue as a basis for further studies.

1. Introduction

A number of studies have attempted to clarify the taxonomy of North American Pleistocene equids (e.g., [1,2,3,4,5,6,7,8]); despite these efforts, the taxonomy of this group remains unresolved. This is particularly problematic for Irvingtonian and Rancholabrean North American Land Mammal Age equids, as exemplified (Figure 1) in a recent literature survey of North American fossil horses [8]. The present work represents another attempt at advancing our understanding of the systematics and taxonomy of North American Pleistocene equids, particularly caballine equids. At least, it provides a thoroughly documented catalogue as a basis for further studies.

We analyzed cranial morphology (especially cranial proportions) to characterize North American caballine forms, but we also studied other skeletal elements including metapodials and dentition. We discuss anatomical features that allow discrimination between caballine, Equus (Equus), and amerhippine, Equus (Amerhippus), equids. We also discuss the peculiar case of Equus (Hesperohippus) mexicanus, which occupies an intermediate position between caballines and amerhippines. Our study is based on extensive comparisons of North American fossils with Old-World fossils, South American fossils, and extant species.

Previous morphological work recognized that extant species of Equus and several fossil forms possess a relatively large brain case [9]. This characteristic has been used to distinguish the genus Equus from two other phylogenetically close genera: Plesippus and Allohippus [9,10,11]. Some studies support the distinction between Equus, Plesippus, and Allohippus (e.g., [12]), but others do not (e.g., [13]). There is, however, agreement about the inclusion of caballines within Equus (e.g., [9,10,11,12,13,14,15,16]). Some recent studies also suggest that amerhippine equids belong within the genus Equus (e.g., [12,14]). We provide further comments on the evolutionary relationship of caballine and amerhippine equids in Section 3.

In this article, we consider caballine and amerhippine equids as distinct subgenera: Equus (Equus) and Equus (Amerhippus). Morphological features that differ between caballines and amerhippines are discussed in Section 3. We are currently working on the second part of this study concerning amerhippines that will provide detailed descriptions, measurements, and an evaluation of the taxonomic status of Haringtonhippus and other slender-legged equids. Our study is divided in two parts to make it more manageable for ourselves and for the reader: one for North American caballines and Equus (H.) mexicanus and a second for North American amerhippines and Equus (Sussemionus) [17].

2. Preliminaries

2.1. Material

For our analyses and comparisons, we used data (including photographs and drawings) from the literature and data/photographs kindly provided by colleagues and museum curators (as indicated in the species accounts below and acknowledgements), in addition to data and photographs that we have personally collected. The data and figures in the main text and information and figures in the Supplementary Materials (Supplementary Files and Tables) present all relevant data used in our study. Other information (listings and complete data sets) may be found at https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES (accessed on 1 July 2024) and/or https://drive.google.com/drive/folders/1LXQTUaU54E9OaBzBlVkoOIeUBgDrQqs6?usp=drive_link (accessed on 1 July 2024).

2.2. Methods of Description and Classification

The “corner stone” of our study is the cranial morphology (mandibular morphology is of a much lesser diagnostic value), but we also compare lower cheek teeth, upper cheek teeth, and limb bone proportions to distinguish groups of species. Recent studies have shown that the morphology of the cheek teeth (e.g., [16,18,19,20]) and metapodials (e.g., [19]) can be useful in the discrimination of extant and fossil equid species.

2.2.1. Cranial Morphology

We primarily used the relationships between various cranial measurements that may characterize specific groups. Supplementary File S1 is a schematic of an Equus cranium with all the measurements we may use; Figure 2 presents the fundamental basilar ones.

Comparisons were performed using scatter diagrams and Simpson’s diagrams.

2.2.2. Lower Cheek Teeth Morphological Types (Figure 3)

We will refer to the morphology of lower cheek teeth as caballine (1); hemionine (2); zebrine (3); and asinine (4) (Figure 3). It must be stressed, however, that not all teeth belonging to true horses are caballine in morphology, nor are all teeth belonging to zebras zebrine. Actually, types 1, 2, 3, and 4 may be found in any species and even in the same tooth row. We use these terms as a short description of a morphological pattern.

The term “caballine” (1) stands for round metaconid + pointed metastylid + angular linguaflexid; the “1-2” type has a less angular linguaflexid than 1.

“Hemionine” (2) stands for round and elongated metaconid + pointed metastylid + shallow and flared (not angular) linguaflexid. A variant of the hemionine type 2 (2′ in Figure 3) shows a slight “bridge” between metaconid and metastylid.

“Zebrine” (3) stands for round metaconid + pointed metastylid + deep and pointed linguaflexid. A variant of the zebrine type 3 (3′ in Figure 3) has an elongated metaconid.

“Asinine” (4) stands for round metaconid + round metastylid + deep and pointed linguaflexid. A kind of “bridge” (4′ in Figure 3) appears between metaconid and metastylid on some Wild African Asses (E. (Asinus) africanus somaliensis).

Figure 3. Lower and upper cheek teeth of extant and extinct equids (not to scale). The numbers and letters denote the morphological types discussed in the text (numbers for lower cheek teeth and letters for upper cheek teeth). Refer to Section 2.2.2 and Section 2.2.3 for a description of morphological types. Lower cheek teeth are p3 except AMNH-M 204210 top and NMB 3267 which are p4, and PIN 657, AMNH-M 204210 bottom, and MNHN-ZM-MO 1977-65 which are m1. Upper cheek teeth are P4 except AMNH 8601 which is a P3.

Figure 3. Lower and upper cheek teeth of extant and extinct equids (not to scale). The numbers and letters denote the morphological types discussed in the text (numbers for lower cheek teeth and letters for upper cheek teeth). Refer to Section 2.2.2 and Section 2.2.3 for a description of morphological types. Lower cheek teeth are p3 except AMNH-M 204210 top and NMB 3267 which are p4, and PIN 657, AMNH-M 204210 bottom, and MNHN-ZM-MO 1977-65 which are m1. Upper cheek teeth are P4 except AMNH 8601 which is a P3.

2.2.3. Upper Cheek Teeth Morphological Types (Figure 3)

The morphology of Equus (we consider Allohippus and Plesippus as separate genera; see Introduction above) upper cheek teeth is extremely variable, even inside a tooth series. Overall, it seems that upper cheek teeth are not very reliable for species discrimination, unless teeth are compared at similar stages of wear (e.g., [16]).

For example, post-protoconal grooves are much deeper on little worn teeth. The fossettes of worn teeth lose enamel plications, the protocones lose any previous indentations or points become somewhat shapeless (USNM 8426).

Some features, however, may be of some value for describing and sometimes even characterizing New World species in teeth with different stages of wear (but still excluding teeth in early and late wear): the post-protoconal groove is very deep in FC 21-26, IGM 56.76-80, and USNM 667, although they are clearly not little worn.

Groups A and B (Figure 3) have in common well-developed plis caballin and grooved protocones, but they differ from each other in the relative length and shape of the protocones. In Group C, the protocones have a marked gibbosity (arrow on Figure 3C) and are less grooved. In Groups D and E, protocones are flat or very little grooved. In Group D, the post-protoconal valley is very deep.

2.2.4. Limb Bones

Comparisons use Simpson’s diagrams, mostly of metapodial size and proportions but also of limb bones’ relative lengths.

2.2.5. Morphological Characters and Species Determination

In extant species of Equus, characteristic cranial, dental, and skeletal features are usually linked and it is their very coexistence that characterizes a species (https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES/MORPHOLOGICAL%20CHARACTERS%20and%20SPECIES (accessed on 1 July 2024)). We started our study by assuming that we would find the same for fossil species, but that assumption was often shaken. The question is how to choose the correct grouping when groupings based on cranial, teeth, and limb bone morphology do not coincide. In our paper’s chapters and their subdivisions, we give preeminence to cranial characters followed by limb bone proportions, in particular when crania are lacking.

2.3. Techniques

2.3.1. Scatter Diagrams

We created scatter diagrams of different cranial proportions. Of particular relevance in this study are the Palatal, Franck’s, Cheek, and Vertex indices (Figure 2; Supplementary File S1):

Palatal Index—Ratio of the palatal length sensu stricto (measurements 2-5) and the hormion–staphylion distance (measurement 3).

Franck’s Index—Ratio of the distance between staphylion and basion (measurement 4) and the hormion–staphylion (measurement 3).

Cheek Index—Ratio of the cheek length (measurement 32) and the length of the naso-incisival notch (measurement 31).

Vertex Index—Ratio of the vertex length (measurement 18) and the basilar length (measurement 1).

2.3.2. Simpson’s Diagrams

Simpson’s ratio diagrams [21] provide rapid and easy comparisons, both of size and shape, for a single bone or a group of bones. A reference is used as the baseline for comparisons. The reference is provided by a single bone (or a group of bones), or means of a bone sample, the dimensions of which are converted into decimal logarithms. By convention, logarithms of these dimensions are placed on the “0” line of the graph. In our study, we used an extant equid, Equus (Hemionus) hemionus onager, as reference. Dimensions of the material under study are also converted into decimal logarithms. Arithmetic differences between the reference logarithms and the logarithms of the studied dimensions are placed above the “0” line if they are positive (larger dimensions), or below if they are negative (smaller dimensions).

In such a logarithmic diagram, proportions remain unchanged whatever the absolute dimensions: diagrams of two bones differing by their size but identical by their proportions will appear one above the other but on parallel lines. The diagram interpretation is easier when pairs of dimensions describing main proportions are placed side by side. For instance, “gracile–robust” proportions appear immediately when “widths” are placed beside “lengths”. More information on Simpson’s diagrams can be found at https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES/OUTILS-TOOLS/DIAGRAMMES%20de%20SIMPSON (accessed on 1 July 2024).

2.4. Abbreviations

dP/dp: Upper/lower deciduous premolars.

I/i: Upper/lower incisors.

M/m: Upper/lower molars.

MC: Third metacarpal.

MT: Third metatarsal.

P/p: Upper/lower premolars.

Ph1, Ph2, Ph3: First, second, third, phalanges.

Prot.: Protocone.

Lo: Anteroposterior length of tooth.

lo: Transversal length of tooth.

Institutional acronyms are found in the Supplementary Materials (Document S1).

2.5. Chronostratigraphic Framework

A chronostratigraphic chart of the last 1.5 million years for North America is found in the Supplementary Materials (Document S2).

3. Distinction Between Caballine and Amerhippine Equus

3.1. Conceptual Framework and Assumptions

3.1.1. Cranial Proportions and the Discrimination of Caballine Equids

There is general agreement in the taxonomic separation of caballine (true) horses from all other equids. It is based on anatomical characters (e.g., [10]) as well as on biomolecular results (e.g., [22]). However, there is no consensus on the taxonomy of caballine equids and different taxonomic schemes have been proposed (e.g., [2,3,4,5,6,7,16]).

A number of factors affect taxonomic interpretations of fossil equids, including caballine equids. Some of these factors are inherent to the fossil record and affect the nature of the samples available for study, such as organismic incompleteness (some skeletal elements are more likely to be preserved than others) and temporal and spatial incompleteness in the record. Other factors are biological in nature, such as convergence/parallelism, morphological constraints, and ecomorphological variation in time and space. And yet others are methodological in nature and relate to the manner in which we collect, analyze, and interpret morphological and molecular data, and in particular label (with formal specific or subspecific names) the fossils. The underlying assumptions are often not explicitly stated. The result is that some factors can potentially lead to overestimating (taxonomic splitting) or underestimating (taxonomic lumping) equid diversity. Since the true diversity is not known, all taxonomic schemes represent hypotheses that may be testable only with additional study.

For the present study, the primary assumption is that various axial lengths of the cranium (e.g., muzzle, naso-incisival length, palate, distance from palate to vomer, and from vomer to basion) are taxonomically informative and can be used to characterize caballine equids; they seem to maintain reasonably stable reciprocal relationships and are not as susceptible to taphonomical distortions as other measurements. Our assumption is based on morphometric studies of a large sample of extant and fossil equid skulls ([9,10]). Indeed, crania of true horses (Palatal Index: Figure 4 and Supplementary File S2 [here Alaska and Yukon fossil caballines are included]), whether large (Draft) or small (Shetland ponies), have long palates (measurement 2-5) relative to the hormion–staphylion distance (measurement 3), while all other extant species have relatively shorter palates [10]. As is usual in any kind of biometrical comparison, there is some overlap between the measurements of true horses and other equids. Fortunately, it is not too pronounced.

Accordingly, we assumed, as a starting point, that all fossil crania plotting together well inside the range of variation of modern true horses in a scatter diagram of the Palatal Index are caballine. They are as follows: several crania from northwestern North America (Yukon and Alaska) including the types of E. (E.) lambei and E. (E.) alaskae (Supplementary File S2); specimens UNSM 5980, UNSM 5981, and UNSM 1346 from Hay Springs, Nebraska, and SMNH 1515-8306 from Fort Qu’Appelle, Saskatchewan (Supplementary File S3); four crania from Rock Creek, Texas, referred to E. (E.) scotti and UMMNH 46899 from the Seymour Formation (Supplementary File S4); the cranium AMNH 95588 from Pool Branch, Florida (Supplementary File S5); the type of E. (H.) mexicanus (IGM 4009), but not UMPE-521 (Supplementary File S5) referred to E. mexicanus by Jiménez-Hidalgo et al. [23], both from Mexico. The specimen UCMP V-3605/32879 from Irvington, California, reported by Azzaroli [7] as a caballine, plots outside the range of variation of caballines (Supplementary File S5). Several other specimens may or may not be caballine and will be discussed later.

The Franck’s Index is another way to discriminate caballine and non-caballine crania, at least in the Old World. In caballines, the distance between the staphylion and the basion (measurement 4) is relatively longer than the staphylion–hormion distance (measurement 3). Once again (Figure 4 and Supplementary File S6), there is an overlap between caballine and non-caballine equids. We will see later that the Palatal and Franck’s indices sometimes give different results regarding the determination of a particular cranium.

3.1.2. Taxonomical Remarks

Equus (E.) ferus Boddaert, 1785 [24], is based on the recently extinct Russian Tarpan. Osteologically, it is represented by the skulls of two castrated males and the associated skeleton of one of them (ZIN 521). They were studied by Gromova [25] under the name of E. caballus gmelini. The Moscow MGU specimen S 94535 has extremely worn upper cheek teeth and lacks the mandible; when one of us (V.E.) saw it, it was falling to pieces—making it impossible to photograph. In consequence, the material is far from satisfactory. Both skulls mainly differ from E. (E.) przewalskii by more caballine Franck’s and Palatal indices. On ZIN 521, the protocones are very short, like in E. (E.) missi. The skeleton has exceptionally short metapodials and short first posterior phalanges. The proportions and size are close to an Island pony. To our knowledge, no equivalent of this conformation was ever found in the Old-World Late Pleistocene (no associated bones at Missy). All of the common Late Pleistocene horses (such as E. (E.) latipes, E. (E.) gallicus, etc.) have completely different conformations and should not be referred to E. (E.) ferus.

The International Commission on Zoological Nomenclature has accepted the proposition of [26] to adopt the name of E. ferus for wild horses, while the name E. caballus is restricted to domestic forms; as both Tarpans have been castrated, their correct representation of a “wild” species is questionable. Azzaroli ([7], p. 2) probably had misgivings about this new nomenclature. For one thing, he wondered whether E. przewalskii was to be considered in that frame a subspecies of E. ferus. And while he assigned to E. ferus many remains of the Late Pleistocene or Holocene of Alaska, he did not refer to E. lambei or E. niobrarensis as subspecies of E. ferus but as distinct caballine species.

Nevertheless, the name of E. ferus has become commonly used, even reluctantly by two of us [11,12,16]. Here, however, we will not refer any North American caballine to E. ferus or to a subspecies of E. ferus.

Though we use specific names when mentioning caballine forms, we do not imply that these forms were true species under a biological species concept. After all, who can tell if the 40,000-year-old African E. (E.) algericus could produce fertile offspring with the 700,000-year-old North American E. (E.) scotti? Specialists already have trouble deciding whether E. (E.) przewalskii is specifically distinct from E. (E.) caballus (testicular anomalies were observed in 6 out of 12 hybrids studied by [27]). The fact remains that many fossils have in common characters we find in modern caballines and distinguish them as a subgenus of the genus Equus. Whether the morphological differences existing between caballine forms (for example, E. (E.) lambei and E. (E.) scotti) precluded them from having fertile offspring will remain undecided, but such differences may be described and it is convenient to use well-known specific names in doing so. Nonetheless, other questions remain: Are these “species” ecomorphs? Chronomorphs? Are resemblances an indication of parallel evolution? Or of close ancestral connections? This question is particularly pregnant when several caballines appear to be sympatric, as in the case of Hay Springs; although E. (E.) niobrarensis and E. (E.) hatcheri may have been simple morphs, we decided to use their names to describe and compare morphologies.

3.1.3. Evolutionary Relationship Between Caballine and Amerhippine Equus

Orlando et al. [14] found similarities between short mitochondrial DNA sequences of South American equids from Argentina referred to Equus (Amerhippus) neogeus and those of Equus caballus. Furthermore, they found that the Equus (Amerhippus) haplotype appears among typical Equus (Equus)—i.e., caballine horse—haplotypes in median-spanning network analyses [14]. In consequence, there may be some doubt about the validity of the subgenus Amerhippus. In their conclusions, Orlando et al. [14] state that, according to currently available ancient DNA data, at least two equid lineages were present in North America during the Late Pleistocene, namely, caballine horses (sensu stricto) and “stilt legged horses” sensu lato—a group of slender-limbed horses named by [28].

The study by Orlando et al. [14] provided a first attempt at evaluating the evolutionary relationships of Equus (Amerhippus) neogeus and the validity of the subgenus Amerhippus using molecular data. Here, we addressed the issue from the anatomical point of view by comparing extant caballine specimens, amerhippine fossils, and other American fossils which either by cranial, dental, or skeletal characters cannot be included into Equus (Equus) based on the characters discussed below.

3.2. Observations

3.2.1. Cranial Proportions of Caballine and Amerhippine Equus and the Particular Case of Equus (H.) mexicanus

Caballines can be well enough discriminated from Old-World non-caballine Equus by the Palatal and Franck’s indices (Figure 4). The discrimination is more difficult in the New World as shown by the scatter diagrams of the Palatal, Franck’s, Cheek, and Vertex indices (Supplementary Files S7–10, Figure 5, Figure 6, Figure 7 and Figure 8). The best separation between caballines and other crania is afforded by the Cheek Index—comparing the relative lengths of the narial notch and cheek (Figure 5, Supplementary File S7); in this plot, E. (H.) mexicanus falls within the caballines. On the scatter diagram corresponding to the Franck’s Index, only E. (H.) mexicanus, E. (A.) occidentalis (including the cranium UMPE-521 from Oaxaca), UNSM 5978 from Hay Springs, and FAM 42811 from Papago Springs plot apart from the rest (Figure 6, Supplementary File S8).

Supplementary File S11 summarizes our observations. It must be stressed that E. (A.) occidentalis almost never overlaps with other species on all of these scatter diagrams and that it is the only species to do so. We have already remarked on the particular case of E. mexicanus: caballine according to the Palatal and Cheek indices, but not caballine by the Franck’s and Vertex indices. This may well grant it the subgeneric rank of Hesperohippus bestowed by Hibbard [29].

3.2.2. Upper and Lower Cheek Teeth

South American amerhippines have variable cheek teeth patterns (Figure 9), some of which may be found in North American caballines (Group E in upper teeth, Figure 3) and amerhippines (Groups 3 and 4 in lower teeth, Figure 3).

North American upper teeth belonging to caballines fall into Groups A and B; protocones with gibbosities (Group C) are exceptional. The association of deep post-protoconal valleys with blunt, roundish, mesial and distal ends of protocones (Group D) never occurs in caballines. Protocones with blunt, roundish mesial and distal ends (Group E) occur in the very worn teeth of the type of E. (E.) lambei (Figure 10) and the type of E. (E.) hatcheri (Figure 11).

Most lower teeth belonging to caballines fall into Groups 1 and 1-2. Group 3 is sometimes observed in some teeth, but Group 4 is never present in caballines. Plis protostylid in American lower teeth south of Alaska and Yukon are a sure indication of amerhippine affinity; protostylids are very rare in caballines [31], common in South American amerhippines [31], and are known in Beringian sussemiones [11,17]. Lower m1 molars appear to be the most reliable for discrimination between caballine and amerhippines.

3.2.3. Lower Incisors

Caballine lower incisors have infundibula (“cups”) at least on i1 and i2 (we have observed only one exception!). Part of the definition of amerhippines is the lack of these infundibula [32], but exceptions do exist (Figure 12): there is an infundibulum on the i1 of AMNH 1753 and on the i1 and i2 of Ecuador # 8, both from Punin, Ecuador.

3.2.4. Metapodials

In MC and MT of extant and most fossil Old-World caballines, the distal articular breadths are larger than the supra-articular ones; in MC, the keel depth may be relatively larger than the distal articular breadth. However, the distal articular breadths may be larger than the supra-articular ones in some non-caballine fossil MT, as well as in caballines.

3.3. Comments

Our analyses identified morphological similarities between caballines and amerhippines. For example, the two groups overlap in the scatter plot of the Franck’s Index and, to a lesser degree, in the scatter plot of the Palatal Index. They may also have similar tooth morphologies (e.g., Group E in upper teeth and Group 3 in lower teeth) and limb bones in some individuals as will be discussed later. The similarities between caballines and amerhippines is one of the reasons for the existence of problematic forms, which will be addressed in Part 2 of our study.

However, there are some differences and morphological tendencies that can be used to discriminate between caballines and amerhippines. For example, the Cheek Index and, to some degree, the Vertex Index may be used to separate caballine and amerhippine crania. In the upper teeth, Group D teeth do not occur in caballines, and Group C teeth are rarely found in caballines; both of these morphologies are often seen in amerhippines. In the lower teeth, Group 4 teeth are not found in caballines, whereas this morphology is present in amerhippines; in addition, it is not uncommon for amerhippine lower cheek teeth to possess a pli protostylid [31]. Cups in the i1 and i2 are almost always absent in amerhippines but present in caballines [33].

Our observations point to a great diversity among amerhippines, strongly suggesting the presence of more than one lineage or form. Equus (A.) occidentalis is certainly one of them. We are currently progressing on Part 2 of our study which will provide more detailed descriptions and measurements of North American amerhippines.

The singular position of the Tequixquiac cranium, holotype of Equus (H.) mexicanus, leads us to support that it belongs not just to another lineage, but rather to the subgenus Equus (Hesperohippus) erected by [29]. This will be discussed further in Section 6.

4. North American Caballines

4.1. Well-Represented North American Caballines

Equus (E.) scotti Gidley, 1900, from the middle Irvingtonian of Rock Creek, Texas, is the first well-represented caballine in North America and one of the best represented Middle Pleistocene caballine horses in the world. The late Irvingtonian crania from Hay Springs, including the types of E. (E.) niobrarensis Hay, 1913a, and E. (E.) hatcheri Hay, 1915, are all badly preserved and/or very old or juvenile. The other relatively well represented North American caballine forms—E. (E.) alaskae Hay, 1913b, and E. (E.) lambei Hay, 1917—are of Rancholabrean age and primarily inhabited Alaska and Yukon.

4.1.1. Equus (E.) scotti Gidley, 1900 [34]

The Rock Creek fauna, Texas, is supposed to be of an early Irvingtonian age, but the E. (E.) scotti fossils come from another quarry ([35], p. 35). According to [36], it is situated only 4.5 m below the Lava Creek B Ash dated at 0.61 Ma. Repenning ([37], Figure 8.1) places it in the Irvingtonian II.

Equus (E.) scotti is represented by several skulls (Supplementary Table S1), more or less fragmentary crania, and limb bones [34]. The type AMNH 10606, illustrated by ([1], Pl. XX), is a mounted skeleton, the cranium of which is impossible to study in detail (Supplementary File S12); however, ([38], Table 1) gave several measurements of the cranium and ([1], Figure 25) provided a drawing of the left M1. AMNH 10612 illustrated by ([34], Figure 3) is not the type skull.

Both Gidley ([1], plate XVIII A and table in p. 136) and Hay ([39], Figures 68 and 69 and table in p. 182) provided illustrations and some measurements of another specimen (AMNH 10628) recovered from the type locality; however, we note that there are slight differences in the enamel pattern of the upper cheek teeth illustrated by these authors (Supplementary File S13). There are also some differences (Supplementary Table S1) between the measurements reported by [39] and the measurements that can be calculated using the figure published by ([1], plate XVIII A). Because of these discrepancies, we omit AMNH 10628 (and another specimen measured by [39]) from the calculations of mean values of crania of E. (E.) scotti (Supplementary File S14).

Johnston [40] provided photographs and measurements of a cranium and associated mandible (Supplementary File S15) recovered in 1936 from the type locality. Both the latter and AMNH 10612 (Figure 13) may be laterally compressed which would account for their very narrow choanae. CMN 2381 (Figure 14) seems to be the best-preserved cranium of this species known to date.

The Simpson’s diagram (Supplementary File S14) shows that the cranium of E. (E.) scotti has a moderately long muzzle and rather small choanae.

The enamel of the upper cheek teeth is plicated (Figure 15); the protocones are moderately long and symmetrically grooved. Post-protoconal valleys are deep. Pli caballins are very developed except on the worn teeth of the specimen reported by [40]. We refer them to Group A.

On the lower molars (Figure 16), the double knots are clearly caballine on m1 and m2, less so on other teeth; ectoflexids are shallow. Molars belong to our Group 1; some lower premolars belong to Groups 2 and 3.

Cups are well developed at least on i1 and i2.

The limb bones are robust. Rock Creek metapodials are very variable in proportions (Supplementary File S16) while the samples of anterior and posterior first phalanges are quite homogeneous (Supplementary File S17).

Two other crania (Supplementary File S14) may be clearly referred to, or related to, E. (E.) scotti:

-

A large skull broken in two fragments (UMMNH 46899a and b; Supplementary Table S1) from the Seymour Formation, Texas, referred to E. (E.) scotti by [41]. The Franck’s Index cannot be evaluated, but the Palatal Index is caballine. The upper dentition of another specimen (UMMNH 46898) illustrated by these authors can also be easily referred to E. (E.) scotti.

-

Eisenmann ([11,42], Figure 60 in [11]) referred to E. (E.) scotti a very well-preserved skull (Supplementary Table S1) collected at Ulakhan Bet Sular, along the Adycha River (affluent of the Iana, northern Siberia). It is preserved at the Severtsov Institute of Moscow as N° 160-455 (ex “Bet 55”). Size, proportions, and location of the posterior palatine foramen are very close to E. (E.) scotti, but the frontal region is flat instead of bulging. Both Franck’s and Palatal indices are caballine. The upper cheek teeth also resemble those of E. (E.) scotti; the lowers are not known. According to Sher (personal communication), it comes from magneto-positive deposits probably inside the lower Brunhes. Many large metapodials collected by Sher at Chukochya and Adycha may also be referred to E. (E.) scotti.

Many Pleistocene large caballines may possibly be related or referred to E. (E.) scotti, in particular the large horses from Old Crow (Yukon), and the slightly smaller MT YG 148-2 from Thistle Creek (Yukon), dated to around 700,000 years ago and which produced a full genome sequence [15]. Other large fossils from Cedral, Mexico, described as E. mexicanus by [43] and as E. ferus by [16], have morphologies similar to E. (E.) scotti. This issue will be discussed below in the chapter devoted to E. (H.) mexicanus. Equus midlandensis also has a morphology similar to E. (E.) scotti; it will be discussed below in Section 5.

Equus (E.) scotti was considered by [7] as a synonym of E. excelsus Leidy, 1858 (Niobrara River Valley, Nebraska), an opinion that has been adopted by some researchers (e.g., [44]). Azzaroli [7] argued that the two equids were about the same size and both had pointed protocones and plicated enamel. One of us (V.E.) provided evidence rejecting this synonymy [10]. The enamel pattern of the holotype of E. excelsus (USNM 667) ([10], Figure 11) is quite different from E. (E.) scotti ([10], Figure 9). In E. excelsus, the P4, M1, and M2 are not very plicated, have no pli caballin, and the protocones are not pointed. On the P4, the post-protoconal valley is very deep. By all these characters, USNM 667 much more closely resembles E. (A.) occidentalis of Rancho La Brea than any E. (E.) scotti.

4.1.2. Equus (E.) alaskae and Equus (E.) lambei (Supplementary File S2, Supplementary Table S2)

The caballines E. (E.) alaskae and E. (E.) lambei are also well represented. They are Rancholabrean in age and are known primarily from Alaska and Yukon.

-

Equus (E.) alaskae Hay, 1913b, type: cranium of a young male, USNM 7700 (Figure 17), from Sullivan’s Creek, Tofty, Central Alaska, described as E. niobrarensis alaskae [45]. The muzzle is rather long relative to the palate (Supplementary File S18).

The Simpson’s diagram of cranial dimensions (Supplementary File S18) shows that in addition to the type USNM 7700, other specimens that may be referred to E. (E.) alaskae are as follows: FAM 60001, 60010, 60011, and 60071 (Figure 18) from Alaska; CMN (formerly NMC) 17254 and 17905 from Yukon; and IA 5059, PIN 301-5 (Figure 19), and PIN 301-533 from North Eastern Siberia. Basilar lengths are 475–495 mm long.

The upper cheek teeth of the holotype and referred specimens are polymorph: the North American USNM 7700 (holotype) and FAM 60071 fall into Groups B and E, whereas the Siberian PIN 301-5 and IA 5059 fall into Groups A and E, respectively (Figure 20).

The lower cheek teeth fall into Group 1. Cups are present on lower incisors.

-

Equus (E.) lambei Hay, 1917 [46], type: skull of an old mare, USNM 8426, from Gold Run Creek (=Dawson 32), Yukon, ca. 15,000 years old (Figure 21). The skull L 1-1222 probably comes from the same locality [47]. Although referred to Onager by Quinn [3], E. (E.) lambei is a caballine [10,48].

The type cranium of E. (E.) lambei is smaller than other crania assigned to E. (E.) lambei and E. (E.) alaskae. Relative to the palate, the muzzle is shorter than in E. (E.) alaskae (Supplementary File S19). The Simpson’s diagram (Supplementary File S19) shows crania with similar proportions. They are as follows: L 1-1222; CMN (formerly NMC) 9924, 33992, 17262, and 34803a; FAM 60012, 60013, 60019, 60020, 60023, 60038, 60045, 60048, and 60066. They were found in different localities in Alaska and Yukon (Supplementary File S19; see photographs at “Localities → Alaska–Yukon”: https://drive.google.com/drive/folders/1lXhM3clTcZXn3XroVh6PNph58BYMw1CS?usp=share_link (accessed on 1 July 2024)).

Relatively similar skulls were also found in NE Siberia (E. (E.) lenensis IA 33) and in France (Jaurens; E. (E.) gallicus LY 302-486/7) (Supplementary File S19).

Crania assigned to E. (E.) lambei are small and the type skull is the smallest. A larger specimen, heavily restored (BEG 31058-2) and found at Blanco Creek, Texas, was referred to Onager lambei by [3]. The cranial preservation is so bad that it is hard to compare it to other North American fossil crania. The cheek teeth of BEG 31058-2 show some similarities to caballine teeth such as the morphology of the protocones (Figure 10); this specimen will be discussed in a section on problematic forms in Part 2 of our study.

The very worn upper cheek of the type fall into Group E. The others fall mostly into Groups A and B (Figure 10).

The lower cheek teeth fall into Groups 1 and 1-2; the ectoflexid (vestibular valley) is shallow even on the molars as in E. (E.) alaskae (Figure 22). Cups are present on lower incisors.

There are a few metapodials that by their small size may have belonged with the type cranium (Supplementary File S20). But since all other crania are much larger, these small specimens cannot be considered as typical.

According to the correlations between basilar lengths and metapodial distal breadths in extant Equus (https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES/OUTILS-TOOLS/CORRELATIONS%20between%20Basilar%20Length%20and%20distal%20widths%20of%20Metapodials (accessed on 1 July 2024)), the latter should measure 42 to 47 mm in both E. (E.) lambei and E. (E.) alaskae. Among the dozens of metapodials from Yukon and Alaska of that size and their many different proportions, we were unable to differentiate E. (E.) lambei from E. (E.) alaskae metapodials, but it may be assumed that, in both, the MC lengths ranged from 198 to 228 mm with an average of 211 mm (Supplementary File S21) and the MT lengths ranged from 237 to 262 mm with an average of 250 mm (Supplementary File S22). Recently, Landry et al. [49] identified overlap in the MT dimensions of a large sample of Pleistocene horse specimens from the Yukon. The continuous variation in MT size and degree of slenderness led these authors [49] to question the reliability of metapodials as taxonomic indicators for Beringian horses when analyzed in isolation from other skeletal elements.

Most of the caballine metapodials from Dry Cave [50] could belong to some E. (E.) alaskae–E. (E.) lambei kind of horse (Supplementary Files S21 and S22).

Equus (E.) lambei from Natural Trap Cave, Wyoming

Natural Trap Cave has produced a large collection of horse fossils, but only a small portion of the horse assemblage is caballine [16,20]. John A. Howe kindly communicated to one of us (V.E.) his data on horse fossils from Natural Trap Cave (dated to ca. 20,000 years ago). Several fossils from this site may be referred to E. (E.) lambei: two fragmentary mandibles UNSM 47238 and KUVP 51079 (the latter associated with a palate) (Figure 23, Supplementary File S23), two MC, six anterior Ph1, five anterior Ph2, two femora, two tibiae, three tali, five posterior Ph1 and seven posterior Ph2, and possibly two Ph3 (Supplementary File S24).

According to the correlations existing between mandibles and crania dimensions (https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES/OUTILS-TOOLS/CORRELATIONS%20between%20CRANIA%20and%20MANDIBLES (accessed on 1 July 2024)), the muzzle width of KUVP 51079 was ca. 71 mm and the muzzle length ca. 118 mm (Supplementary File S23); the basilar length was around 500 mm, just as in E. (E.) lambei.

The lower cheek teeth have the usual caballine pattern. In UNSM 47238, infundibula are well developed on all lower incisors but only on i1 and i2 in KUVP 51079. Protocones are long and grooved (Figure 24).

4.1.3. Comments on E. (E.) scotti, E. (E.) alaskae, and E. (E.) lambei

To sum up, E. (E.) alaskae and E. (E.) lambei were much smaller than E. (E.) scotti, and their cranial proportions are completely different (Supplementary File S25). Figure 25 and Supplementary File S23 compare the main anatomical features of E. (E.) alaskae and E. (E.) lambei. Particularly, they illustrate the very large relative protocone lengths of E. (E.) lambei and E. (E.) alaskae, particularly in E. (E.) lambei. We propose that E. (E.) alaskae and E. (E.) lambei are “ecological variants” of a single species. It appears that E. (E.) alaskae and E. (E.) lambei differ mainly by muzzle proportions. Since usually muzzles tend to shorten in cold conditions (Allen’s rule), it may be supposed that E. (E.) lambei lived during very cold times and E. (E.) alaskae during milder times. This is a hypothesis that may be tested through radiometric dating, stratigraphic work, stable isotope analysis, and population level genetic analysis.

4.2. Hay Springs Sheridan Equus Beds Caballines

The late Irvingtonian crania from Hay Springs, Nebraska, including the types of E. (E.) niobrarensis and E. (E.) hatcheri, are all badly preserved and/or very old or juvenile.

Almost all data were communicated to one of us (V.E.) by John Howe. Almost all measurements of limb bones (hundreds of them) were taken by John Howe and two of his students. Many field and accession numbers are the same for several bones, and many bones seem to have been measured several times by the same person or by different persons; in that case, means of different dimensions (labelled x) were computed. Photographs of dentitions are without scale and often bear field numbers while in the table of measurement specimens are mostly referred to by accession numbers. It is not always clear whether the number is “field” or “accession”. Data and photographs communicated by John Howe are available in HOWE HAY SPRINGS: https://drive.google.com/drive/folders/1yeU9HtK2iG4Cv8HVWHbCMgTH7KyKKhl5?usp=sharing (accessed on 1 July 2024).

The fossils of Sheridan Beds come from several localities—Sh 1 to Sh 6 Equus beds—but the bulk were found in Sh 5 Gordon Quarry 5. There is no information about the provenance of quite a number of fossils (“Sh 0”). According to [51], most of the fossils from Sheridan County Fossil Quarries can be attributed to the late Kansan and Yarmouth (Pre-Illinoian).

4.2.1. Crania (Supplementary Files S3 and S26, Supplementary Table S3)

There are at least 12 more or less preserved crania from Hay Springs and 2 fragmentary mandibles. Nine show caballine characters although some were referred to the subgenus Asinus by [52]. Specimens from Equus Sheridan Beds (excluding Sh 5 Gordon Quarry 5) are as follows:

-

USNM 4999, skull, type of Equus (E.) niobrarensis.

-

USNM 7868, skull, type of E. (E.) hatcheri; referred to E. niobrarensis by [7].

Azzaroli [7] remarked that “the type skulls of E. niobrarensis and E. hatcheri are of restricted interest because of the extensive restorations they underwent” and it is true indeed. Moreover, their very different enamel patterns may be, at least in part, the consequence of wear. We consider, however, that these patterns may be used to classify other fossils and will employ them for further morphological descriptions.

-

UNSM 1346, fragmentary juvenile skull; referred to E. niobrarensis by [7].

-

AMNH 2725, fragmentary skull; referred to E. complicatus by [1] and to E. excelsus (our E. (E.) scotti) by [7].

-

AMNH 11642, fragmentary mandible; referred to E. excelsus (our E. (E.) scotti) by [7].

Specimens from the locality Sh 5 Gordon Quarry 5 include the following: UNSM 5980, UNSM 5981, and UNSM 5982 crania; UNSM 9417, a fragmentary mandible referred to E. niobrarensis by [7]; and UNSM 5989, a fragmentary cranium possibly also from the locality of Gordon.

We have used modified photos from [7] of UNSM 5980, 5982, 1346, and 9417 and AMNH 2725 and AMNH 11642; we have no photographs for UNSM 5981 nor 5989 crania, and only Howe’s measurements for the latter.

Due, at least in part, to bad preservation, reconstructions, and juvenile or very old age, the cranial dimensions and proportions are rather variable (Supplementary Files S26 and S27). The type cranium and mandible of E. (E.) niobrarensis belong to a young adult mare; the type cranium and mandible of E. (E.) hatcheri belong to an old stallion. Possibly the differences between muzzle lengths and breadths may thus be explained. The long muzzle of the old stallion UNSM 5981 may also be due to its age. Two specimens, however, stand apart: UNSM 1346 and UNSM 5989.

UNSM 1346 is the cranium of a 2-year-old colt. It is laterally compressed, but otherwise sufficiently well-preserved. Its large size is remarkable. Its basilar length (560 mm) would be about 580 mm as an adult (https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES/OUTILS-TOOLS/CRANIAL%20GROWTH%20Extrapolations%20in%20Equus (accessed on 1 July 2024)). This would correspond to metapodial distal breadths of around 54 mm (Supplementary File S28). Another remarkable character is the uncommonly large basion–staphylion length (measurement 4). The fragmentary UNSM 5989 appears to have been even larger.

4.2.2. Dentition

The upper cheek teeth of AMNH 2725 resemble those of E. (E.) scotti (Figure 15). Those of USNM 4999 have shorter protocones and lesser plicated enamel; similar patterns are found at Dry Cave [50], Rancho La Brea, and on the teeth from the Fossil Lake area referred to E. pacificus by [1] as will be illustrated later. They fall into Groups A, B, and E.

There is nothing particular about the lower cheek teeth from Hay Springs except for the slightly larger size of E. (E.) hatcheri and AMNH 11642 (Figure 26). Ectoflexids are often shallow even on molars. Most fall into Groups 1 and 1-2.

Cups are present on i1 and i2; not perfectly developed on i3 of USNM 4999.

Upper cheek teeth of UNSM 5980, UNSM 5982, USNM 7868, and Frankfurt M-1487 belonged to old animals (Figure 11). The shorter and/or flatter protocones of E. (E.) hatcheri type (USNM 7868) and UNSM 5982 and 5980 may thus be accounted for; but the differences between the ‘E. (E.) scotti-like’ AMNH 2725 and the type of E. (E.) niobrarensis (USNM 4999) cannot be ascribed to wear. If they were sympatric–contemporaneous, they must be considered as representing different species, as suggested by Azzaroli ([7], Plate 18, p. 10). If they were not contemporaneous, it may be supposed that the ‘scotti-like’ morph evolved into a ‘niobrarensis-like’ morph and that they are both chronospecies of a caballine lineage.

4.2.3. Limb Bones

Except for a few large specimens possibly belonging to the large Equus, the bulk of well-preserved caballine MC from Hay Springs (n = 46) is relatively homogenous, the main differences being the larger depth of the keel in 13 of them (Supplementary File S29). The matter is different for the MT. First, there are more of them, and secondly, two groups may be distinguished: 40 are slightly smaller and deeper in the diaphysis than the other 50; we do not know how much significance these differences may have. The distal articular breadths of the MC and MT would be consistent with basilar lengths of 500 to 540 mm for the corresponding crania (https://veraeisenmann.com/index.php?ref=/public/WEB/EQUIDES%20MONODACTYLES/OUTILS-TOOLS/CORRELATIONS%20between%20Basilar%20Length%20and%20distal%20widths%20of%20Metapodials (accessed on 1 July 2024)); these are the dimensions of all Hay Springs crania, except the juvenile UNSM 1346 whose basilar length is ca. 580 mm. A large MT (HS 6515) may have belonged to this latter form.

4.2.4. Comments of Hay Springs Sheridan Equus Beds Caballines

We refer UNSM 1346 and 5989 to Equus (E.) sp. and AMNH 2725 and Hay Springs 6006 3268-38 to E. (E.) scotti or E. aff. E. (E.) scotti (Figure 15); several other upper cheek teeth have the same enamel pattern but, unfortunately, the available photographs have no scale.

We refer to E. (E.) niobrarensis UNSM 5980, UNSM 5982, and M-1487 of the collections in Frankfurt. Equus (E.) hatcheri may be considered as a morphological variant or chronological variant of E. (E.) niobrarensis.

Crania of E. (E.) niobrarensis are smaller than those of E. (E.) scotti. They differ mainly by their wider choanae (Supplementary File S30, measurement 10).

4.3. Hay Springs-like Fossils

Slaton, Lubbock County, Texas

Among the Early Illinoian equid fossils found at Slaton, the TMM 6643 upper and TMM 4800 lower series were referred to E. niobrarensis by Dalquest (Figure 2 in [30]), although he noted the greater resemblance of the upper series with the type of E. hatcheri. Other caballine remains are several series or associated teeth and several associated limb bones. The metatarsal TMM 5249 could belong to E. (E.) scotti (Supplementary File S16) but, unlike for E. (E.) scotti skeletons, it is longer relative to the associated tibia (see comparison of limb proportions in Section 6.2). This is a very rare occurrence; as far as we know, such proportions happen only in E. (E.) przewalskii—22 out of 35 skeletons.

Dry Cave, Eddy County, New Mexico

In addition to the published data, we are indebted to A.H. Harris for sharing detailed measurements (Supplementary Files S21 and S22).

Although most of the Late Pleistocene caballine metapodials from Dry Cave [50] could belong to some E. (E.) alaskae–E. (E.) lambei kind of horse, there are some MC that fall inside the range of variation of Hay Springs caballines and some which are larger. The size of MC UTEP 31-57 and 22-692 and of MT 31-64 would be consistent with the size of UNSM 1346 from Hay Springs.

Ingleside, Texas

Among the Late Pleistocene equid material from Ingleside, the mandible TMM 30967-410, probably associated with the large, badly preserved cranium TMM 30967-401 (Figure 27), was referred to E. complicatus by ([53], Figures 54 and 56, Table 45, p. 64). Both cranium and mandible resemble the type of E. (E.) hatcheri. Lundelius [53] did not illustrate the teeth of the cranium but gave their occlusal length and width dimensions in his Table 46 together with other specimens he also referred to E. complicatus. Unfortunately, he did not give any dimension for the upper P2-M3 series TMM 30967-379 he illustrated in his Figure 55 (Figure 28), but they seem rather similar to the type of E. (E.) hatcheri.

Also from Ingleside, Lundelius ([53], Figure 58) referred to E. pacificus four upper premolars (TMM 30967-242, 1487, 1487A, and 379F) that resemble AMNH 2725—another specimen from Hay Springs—and three first anterior phalanges.

The enamel of all Ingleside upper cheek teeth is very plicated as at Hay Springs and in E. (E.) scotti to which it was probably related.

4.4. Caballine Equus from Fort Qu’Appelle (Supplementary File S3, Supplementary Table S4)

Khan [54] referred to E. scotti—which he considered a senior synonym to E. niobrarensis—a cranium, several dentitions, and limb bones from the Echo Lake gravel beds near Fort Qu’Appelle, Saskatchewan. He believed them to belong in the Sangamon Interglacial or, at the latest, to a Wisconsin interstadial. Azzaroli [7] referred these fossils to E. niobrarensis.

The cranium SMNH 1515-8306 (Figure 29) belonged to an old mare. It has a long muzzle, remarkably short choanae, and a remarkably long distance between the basion and staphylion. The cranium is smaller than both E. (E.) scotti and E. (E.) niobrarensis (Supplementary File S30), and has a much longer basion–staphylion distance, and is narrower at the middle of the diastema (measurement 17bis).

On the upper very worn cheek teeth, the protocones are slightly or not grooved unlike the type of E. (E.) niobrarensis; the lower cheek teeth have the usual Group 1 caballine pattern (Figure 30).

Unfortunately, the samples of limb bones from Fort Qu’Appelle are highly heterogeneous (Supplementary File S31): one MC (SMNH 8004-1410) could be referred to an Equus as large as E. (E.) scotti, while SMNH 7737-1042 must certainly belong to E. (E.) alaskae or E. (E.) lambei; all MT can be referred to some large E. (E.) alaskae or E. (E.) lambei form.

4.5. Equus (E.) pacificus Leidy, 1868 (Figure 31)

Leidy [55] created E. pacificus based on one upper P3 from Martinez, Contra Costa, California; it was not illustrated but the measurements are those of a large Equus: 35 mm and 32 mm of occlusal length and width. According to these dimensions, E. (E.) pacificus would be one of the largest Equus in the world [56]. Gidley [1] referred to this species (based on size and Leidy’s [55] description of the holotype) several specimens from the Fossil Lake area, Oregon, including three upper cheek teeth that are illustrated in his Figure 11 (Figure 31, top row). The teeth illustrated by Gidley [1] resemble the holotype of E. (E.) niobrarensis (Group B) but are much larger. Some authors consider E. pacificus a valid name (e.g., [3,53]) and refer to Gidley’s [1] description and illustration when discussing this species, but others consider it a nomen vanum or nomen dubium (e.g., [2,8]); resolving this nomenclatural issue is beyond the scope of our study. We recognize a very large form of Equus (Equus) known primarily from the western USA that we tentatively refer to as E. (E.) pacificus, pending further consideration of the nomenclatural status of this species name.

Also, from the Fossil Lake area, Oregon, Cope [57] illustrated and referred to E. occidentalis three clearly caballine lower cheek teeth (Figure 31, bottom row). These teeth are similar in size to the teeth of E. (E.) pacificus described by Gidley [1], and we refer them to that taxon. Data on large limb bones from Fossil Lake, referred to E. (E.) pacificus, are in Supplementary File S32. Other Equus from Fossil Lake will be discussed in a further study.

In addition to the numerous fossils of E. (A.) occidentalis, the Rancho La Brea collections include associated upper and lower dentitions UCMP 9G 14-10.5, dated at 12,000 or 35,000 years ago, that we refer to E. (E.) pacificus based on tooth morphology, although they are smaller than most specimens referred to this taxon (Figure 31). We also refer to E. (E.) pacificus the P3-M3 series UTEP 22-1608 from Dry Cave, New Mexico (Figure 31). This series was identified as E. niobrarensis by [50] and as E. ferus by [16].

Figure 31. Upper and lower cheek teeth referred to E. (E.) pacificus: AMNH 8601-1, 2, and 3, and unnumbered teeth from the Fossil Lake area; UCMP 23238 from Bautista Creek; UTEP 22-1608 from Dry Cave; and UCMP 9G 14-10.5 from Rancho La Brea. Occlusal view. AMNH 8601-3 modified from [1]. Unnumbered lower teeth from the Fossil Lake area modified from [57]. UTEP 22-1608 modified from [50].

Figure 31. Upper and lower cheek teeth referred to E. (E.) pacificus: AMNH 8601-1, 2, and 3, and unnumbered teeth from the Fossil Lake area; UCMP 23238 from Bautista Creek; UTEP 22-1608 from Dry Cave; and UCMP 9G 14-10.5 from Rancho La Brea. Occlusal view. AMNH 8601-3 modified from [1]. Unnumbered lower teeth from the Fossil Lake area modified from [57]. UTEP 22-1608 modified from [50].

The holotype of E. bautistensis Frick, 1921 [58] (three upper cheek teeth catalogued as UCMP 23238) is similar in size and morphology to teeth of E. (E.) pacificus (Figure 31). Equus bautistensis may be regarded as a junior synonym of E. (E.) pacificus, pending further consideration of the nomenclatural status of the species name of the latter.

The upper teeth of specimens we refer to E. (E.) pacificus fall into Group B, and the lower teeth fall into Groups 1 and 1-2.

From Ingleside, Texas, Lundelius [53] referred to E. pacificus four upper cheek teeth, mentioned above (TMM 30967-242, 1487, 1487A, and 379F) (Figure 28); we already noted their resemblances with the ‘E. (E.) scotti-like’ AMNH 2725 from Hay Springs.

4.6. Equus (E.) complicatus Leidy, 1858 [59], (Figure 32)

Equus (E.) complicatus (initially named E. americanus Leidy, 1847) was the first named Equus in the USA. As for many other old names, its history is exceedingly complicated [60]. Although Savage [2] considered it a nomen vanum and Azzaroli [7] a nomen dubium, we will follow Gidley’s [1] observations based on the fossils described and illustrated by Leidy ([61], Plate 8, p. 265). A historical account of the processes leading Leidy to describe E. complicatus and the reasons for distinguishing it, or not, from E. fraternus and E. leydyi will be provided in a separate paper on North American amerhippines (Part 2 of our study).

The lectotype chosen by Gidley [1] is an M2 from the vicinity of Natchez, Mississippi, now bearing the accession number ANSP 11424 in the Academy of Natural Sciences of Drexel University, Philadelphia; the three other specimens illustrated by Leidy [61] certainly belong to the same species although both M3 were referred to E. fraternus (Figure 32, top row). On the upper molars, the enamel is plicated and protocones have pointed mesial and distal ends (Group A); the lower premolar has an angular caballine double knot with a pointed metaconid and metastylid (Group 1).

Figure 32. Lectotype of E. (E.) complicatus (ANSP 11424) and other teeth referred to this morphospecies. The top row reproduces original illustrations published by [61]: ANSP 11424 (lectotype [1]), ANSP 11433, and two M3 from Natchez Island originally referred by Leidy to E. fraternus. The next four rows show photographs of teeth from the same sample studied by Leidy [61] including the lectotype specimen, and one upper M3 from Big Bone Lick (ANSP 11496); we refer these teeth to E. (E.) complicatus except ANSP 11440 which we refer to E. littoralis. The last row shows upper teeth from Florida, South Carolina, and Iowa referred to E. (E.) complicatus. All teeth in occlusal view. Teeth in top row modified from [61]. Photos of teeth in middle section kindly provided by Dr. Daeschler with the exception of ANSP 11496, which was modified from [60]. Photos of WFIS 4086b E and WFIS 4086b F kindly provided by Ms. Dorwaldt. Drawing of WFIS 4086 modified from ([62], Figure 11). AMNH 9203 and AMNH 9217 modified from [1]. IU 122b modified from [39].

Figure 32. Lectotype of E. (E.) complicatus (ANSP 11424) and other teeth referred to this morphospecies. The top row reproduces original illustrations published by [61]: ANSP 11424 (lectotype [1]), ANSP 11433, and two M3 from Natchez Island originally referred by Leidy to E. fraternus. The next four rows show photographs of teeth from the same sample studied by Leidy [61] including the lectotype specimen, and one upper M3 from Big Bone Lick (ANSP 11496); we refer these teeth to E. (E.) complicatus except ANSP 11440 which we refer to E. littoralis. The last row shows upper teeth from Florida, South Carolina, and Iowa referred to E. (E.) complicatus. All teeth in occlusal view. Teeth in top row modified from [61]. Photos of teeth in middle section kindly provided by Dr. Daeschler with the exception of ANSP 11496, which was modified from [60]. Photos of WFIS 4086b E and WFIS 4086b F kindly provided by Ms. Dorwaldt. Drawing of WFIS 4086 modified from ([62], Figure 11). AMNH 9203 and AMNH 9217 modified from [1]. IU 122b modified from [39].

Although there are some resemblances between E. (E.) complicatus and E. (E.) scotti, as noted by Gidley [1] and Azzaroli [7], we follow Gidley [1] and consider E. (E.) complicatus as a different species in particular because of differences in the double knot patterns, including the presence of an indentation on the lingual side of the metaconid.

Photos of teeth belonging to the same sample in the collections of the Academy of Natural Sciences of Drexel University (ANSP) were kindly communicated to us by Dr. Daeschler (Figure 32). The difference in size inside this sample is very large and there are morphological differences too.

We tentatively distinguish the smaller upper ANSP 11429, 11431, and WFSI 4086b E and F, and the smaller lower ANSP 11432, 11436, and 11438 as a small form of E. (E.) complicatus and refer ANSP 11440 to E. littoralis Hay, 1913a [62], which will be discussed in Part 2 of our study. We also refer to the small form of E. (E.) complicatus AMNH 9217 from near Charleston (South Carolina), AMNH 9203 from some locality in Florida, and an upper molar from Alafia Bay (Florida). The M3 ANSP 11496 and the very large MC and MT from the late Wisconsin of Big Bone Lick (Kentucky) may belong to E. (E.) complicatus [60]. The upper premolar IU 122b from the Aftonian (Pre-Illinoian) of Turin (Iowa) may belong to an early E. (E.) complicatus.

Apart from what has been published by [60], we have found no information on the limb bones of E. (E.) complicatus.

Numerous fossils from various locations in North America were referred to E. complicatus by different authors [1,3,39,53,63,64,65,66]. We consider some of them as amerhippines and we will discuss them in Part 2 of our study.

4.7. Equus holmesi Hay, 1920 [65]

This species was described by [65]. The type is a series of upper cheek teeth (P3-M2), numbered USNM V 8642 from Afton, Oklahoma.

Equus holmesi has rarely been discussed in the recent literature of North American fossil horses ([8], Supplementary Text and Supplementary Table S6). Our study suggests that the holotype of E. holmesi possesses morphological traits that identify it as a caballine (e.g., upper cheek teeth with Group A and B morphologies), but it is not remarkable enough to grant a specific name; overall, it resembles teeth of E. (E.) scotti (Figure 15). We regard E. holmesi as a junior synonym of E. (E.) scotti.

4.8. Equus laurentius Hay, 1913a [62] (Supplementary Table S4)

The holotype of E. laurentius is a cranium and mandible, KUMNH 347, recovered from what were thought to be Upper Pleistocene deposits near Lawrence, Kansas. Additional data were published in [67]. In 1957, Quinn [3] discussed the question of the fossilization of the holotype and gave arguments in favor of an age of more than 1200 years. Discussions about that topic lasted for years until Scott et al. [68] settled the problem by proving that the cranium and mandible belonged to two separate individuals of the modern domestic horse (Equus (E.) caballus). Taylor et al. [69] contend that the cranium and the mandible belong to the same individual. Regardless of whether they belong to the same individual or not, both the cranium and mandible show pathologies that are indicative of bridling and riding, supporting the interpretation that they belong to a domestic horse [69]. Equus laurentius is a junior synonym of E. (E.) caballus [68].

The cranium belongs to a middle-aged stallion (Figure 33; composite from [62]) of moderate size. The upper cheek teeth have protocones that are rather short and are not grooved; on the lower cheek teeth, double knots are caballine (Group 1) and ectoflexids are deep on the molars (Figure 34; composite from [62]).

Hay [39] also referred to E. laurentius a partial palate ([39], Plate XIII); the protocones are longer on this series than on the cranium KUMNH 347. The teeth are too worn to compare them in more detail to other teeth of North American fossil caballines.

Quinn [3] referred to E. caballus laurentius two cheek teeth (TMM 937-191 upper and TMM 937-252 lower molar) from Blackwater Draw, Portales, New Mexico (Figure 35). The protocone of the upper molar is much longer than in the corresponding tooth of KUMNH 347. The ectoflexid of the lower molar is shallow. These teeth were identified as E. niobrarensis by [70] and as E. ferus by [16].

Mooser [71] also referred to E. caballus laurentius one upper premolar FC 110 and one lower premolar FC 113 from the Fauna Cedazo, Aguascalientes, Mexico (Figure 35). Both teeth differ from the corresponding teeth of KUMNH 347. The protocone of FC 110 is longer and pointed; the double knot of the lower tooth corresponds to Group 1-2.

4.9. Summary

We summarize below our observations and suggest some interpretations.

-

The best documented and most ancient North American caballine species, Equus (E.) scotti, was about the size of an Arabian horse but had heavier bones, wider third phalanges, and less cursorial limb bone proportions (see comparison of limb proportions in Section 6.2.). Its weight may be estimated at 500–600 kg [56]. The maximal observed metapodial distal breadth is 59 mm, and the longest MC and MT are 246 and 291 mm long, respectively. The basilar skull length is 530 to 600 mm.

-

The later E. (E.) niobrarensis was slightly smaller but probably had the same limb proportions; it differs, however, by cranial and dental features. The horse from Fort Qu’Appelle may be referred to E. (E.) niobrarensis, and the western E. (E.) pacificus may be related to E. (E.) niobrarensis, but this needs to be evaluated further.

-

The juvenile cranium UNSM 1346 from Hay Springs belongs to a caballine, but it is unique compared to other North American caballine crania we studied by its large size and the long distance between basion and staphylion; we refer to it as Equus (E.) sp. We also refer to Equus (E.) sp. the large cranium UNSM 5989 also recovered from Hay Springs.

-

Both E. (E.) alaskae and E. (E.) lambei primarily inhabited Yukon, Alaska, and NE Siberia during the Late Pleistocene. They both were about the size of E. (E.) przewalskii but with heavier bones. The associated limb bones FAM 71467 from Fox 33 and (not certainly associated) from Natural Trap Cave have the same relative lengths as E. (E.) scotti. Basilar lengths of fossil skulls of E. (E.) alaskae and E. (E.) lambei are similar (averages of 489 mm and 483 mm). Metapodial distal breadths measured 42 to 47 mm in both E. (E.) lambei and E. (E.) alaskae. The cranial proportions (Figure 25) differ only by the muzzle length and breadth: E. (E.) lambei crania have shorter and broader muzzles probably adapted to cold conditions according to Allen’s rule; it is remarkable that there are no intermediate morphologies. We hypothesize that E. (E.) alaskae and E. (E.) lambei are “ecological variants” of a single species.

-

Equus (E.) complicatus was a large form, probably restricted to southeastern North America. Unfortunately, only its upper and lower cheek teeth are well documented.

Thus, several discrete entities we treat as morphospecies existed in North America during the last 700,000 years: E. (E.) alaskae–E. (E.) lambei in the North, E. (E.) complicatus in the East, and E. (E.) pacificus in the West. Equus (E.) scotti and E. (E.) niobrarensis are morphologically different enough to qualify as distinct morphospecies if they were sympatric at Hay Springs, or as chronospecies if they were not contemporaneous. Whatever the case, and whatever their polymorphism, all these forms were undoubtedly caballine.

Our identification of E. (E.) scotti, E. (E.) lambei, and E. (E.) alaskae as caballine equids is consistent with recent morphological and molecular studies [8,15,16,72], even if the specific nomenclature used in those studies is different from the one presented here. Likewise, E. (E.) niobrarensis has been recognized as caballine in recent studies, but its taxonomic validity has been questioned [16], ([8], Supplementary Text and Supplementary Table S6).

Regarding E. laurentius, recent radiocarbon dating and osteological analyses have conclusively shown that its holotype belongs to a domestic horse [68,69]. Thus, E. laurentius is a junior synonym of E. (E.) caballus. Similarly, the holotype of E. holmesi has morphological traits that identify it as a caballine; however, our evaluation suggests that the holotype resembles teeth of E. (E.) scotti. Equus holmesi is considered a junior synonym of E. (E.) scotti. The holotype of E. bautistensis is similar in size and morphology to teeth of E. (E.) pacificus. Equus bautistensis may be regarded as a junior synonym of E. (E.) pacificus.

Now, however, we have to face a challenging problem, that of the identity of E. midlandensis and E. mexicanus.

5. Equus midlandensis

Quinn [3] described E. midlandensis Quinn, 1957, from Scharbauer Ranch, Midland County, Texas, and referred to this species teeth and bones from Baggett ranch, Odessa, Texas, and from Clovis site, 7 miles north of Portales, New Mexico.

Some other teeth and bones found in New Mexico and Texas (Blackwater Draw and Lubbock) were referred by [3] to E. caballus caballus, by [70] to E. scotti, and by [16] to E. ferus. Other relevant specimens from Late Pleistocene localities in Texas and Mexico are also discussed below.

5.1. Equus midlandensis Holotype

The type material includes a fragment of a very young adult male mandible, associated upper cheek teeth, and several limb bones, all believed to belong to one individual (TMM 998, various elements) [3]. The material dates to the Late Pleistocene, post Sangamon [3].

The type mandible TMM 998-1 (Figure 36) has a height of 101.5 mm at the middle of p4. The overall size is large, close to that of the mandibles AMNH 10612 and CMN 2381 of E. (E.) scotti from the Irvingtonian of Rock Creek, Texas, but the diastema and muzzle lengths are much shorter (which may be because of the subadult age of the Texan specimen). Infundibula are present on i1 and i2, open on the unerupted i3.

A right ramus from Tequixquiac (IGM 71-401) was tentatively referred to E. mexicanus by ([29], p. 73). It has the same height beneath the middle of p4 as in the type of E. midlandensis and the pronounced concavity between the ascending and horizontal ramus usually found in true horses. The pattern of the teeth is typically caballine, especially on the molars: angular lingual valley, rounded metaconid, pointed metastylid (Group 1). The isthmus of the double knot is very high, like in E. (E.) scotti (Figure 16), and the ectoflexid is shallow on all teeth (Figure 36). The lower p3 from Arroyo Paso Hondo, Fauna Cedazo, certainly belongs to a caballine but maybe not to E. midlandensis.

The associated upper cheek teeth of E. midlandensis resemble E. (E.) scotti (Figure 15) more than E. (H.) mexicanus (Figure 37).

The metatarsal TMM 998-3 is robust. It has about the same size and proportions as E. (E.) scotti and the MT DP 4118 from Cedral referred to E. mexicanus by [43] (Figure 38 and Supplementary File S33).

5.2. Referred Specimens

-

Baggett Ranch, Odessa, Texas, and Clovis site, 7 miles north of Portales, New Mexico

There are no measurements nor drawings of the cheek teeth from Baggett Ranch (BEG 30422) and Clovis site lower cheek teeth (TMM 937) mentioned by [3]. Although supposed to belong to a single individual [3], it is doubtful that the caballine limb bones from Baggett Ranch (Figure 39) may even belong to the same species based on their proportions (Supplementary File S33).

5.3. Other Relevant Specimens

-

Blackwater Draw, Roosevelt County, New Mexico and Lubbock, Texas

The fossils from the Wisconsin of Blackwater Draw referred by [3] to E. caballus caballus were referred by [70] to E. scotti and by [16] to E. ferus.

The upper cheek teeth (TMM 937-170, Figure 37) are very similar to the holotype of E. (H.) mexicanus: the protocones are long, symmetrically grooved, and pointed like in sectioned premolars of E. (E.) scotti. Post-protoconal valleys are deep. Pli caballins are very developed. Like in E. (E.) scotti, the lower cheek teeth have caballine double knots with high isthmus and shallow ectoflexids even on the molars. Lundelius [70] referred TMM 937-170 and other specimens from Blackwater Draw to E. scotti; furthermore, he considered E. midlandensis as a junior synonym of E. scotti.

In addition to a damaged radius (TMM 937-240), there is an astragalus from Blackwater Draw and an MT (TMM 892-11) from Lubbock. The latter (Figure 38) more closely resembles the one from Baggett Ranch by its large distal supra-articular width.

-

Boatwright and Trinity River, Texas

We are uncertain about the taxonomic assignment of the upper and lower teeth BEG 30907-24D and 30907-8G from Boatwright, Texas, and the upper BEG 30907-46B from Trinity River referred by [3], respectively, to E. complicatus and E. fraternus (Figure 36 and Figure 37).

Slaughter [73] referred to E. midlandensis two upper cheek teeth from Hill-Lower Schuler, Trinity River, Texas, and noted resemblances with E. scotti. As in E. (E.) scotti and E. midlandensis, the enamel is plicated but the wide and grooved styles are very different. Moreover, the referred MC, SMUMP 60079, is smaller and more gracile (Supplementary File S33).

-

Ingleside, Texas

We tentatively refer to E. midlandensis three first anterior phalanges from Ingleside, Texas (originally referred by [53] to E. pacificus), based on their smaller size and proportions compared to specimens we refer to E. (E.) pacificus (Supplementary Files S32 and S33).

5.4. Comments

To sum up, we see caballine characters in most fossils from Texas and New Mexico referred to E. midlandensis. The type mandible, however, has a striking resemblance to a mandible found in Tequixquiac—type locality of E. (H.) mexicanus. Is E. midlandensis a junior synonym of E. (H.) mexicanus? Or were they two large Equus species, one of them caballine (E. midlandensis), sympatric in Texas, New Mexico, and Mexico?

6. Equus (Hesperohippus) Mexicanus

As mentioned in Section 3, the cranial proportions of the holotype of Equus mexicanus are unique. This leads us to support that it belongs to the subgenus Equus (Hesperohippus) erected by [29].

6.1. Equus (Hesperohippus) mexicanus Holotype

Hibbard [29] described as Equus (Hesperohippus) mexicanus a cranium (Supplementary Table S5; Supplementary File S5) from Tajo de Tequixquiac, No. 48 (HV-3), now catalogued as IGM 4009 [74] (Figure 40). There are no associated fossils. Hibbard did not mention any lower incisors. No limb bones are known. Hibbard considered its age as Late Pleistocene.

This cranium was first referred to E. excelsus by [75] and to E. occidentalis by [76]. Because of its size, E. (H.) mexicanus has been considered a junior synonym of any of the large North American species, in particular E. (E.) scotti and E. (A.) occidentalis whose crania are well documented (see detailed discussion in [43,44]). Winans [6] included E. mexicanus in her E. laurentius group. Equus laurentius is a typical caballine and so is E. (E.) scotti, whereas E. (H.) mexicanus is not. Other authors (e.g., [43,53]) have considered E. mexicanus and E. pacificus as synonyms.

The cranial morphologies of E. (E.) scotti, E. (A.) occidentalis, and E. (H.) mexicanus are distinct from one another (Figure 41 and Supplementary File S34). A recent phylogenetic analysis of derived Equini also identified E. mexicanus as distinct from E. occidentalis and caballine equids [12]. Other recent studies also recognize E. scotti, E. occidentalis, and E. mexicanus as distinct taxa [8].

Thorough observations and comparisons led Hibbard ([29], p. 73) to state that “the skull does not show any close relationship to the true horses or to zebrine forms, a fact noted by Cope”. Hibbard [29] rightly stressed this point, which seems to have escaped proper notice. Equus (H.) mexicanus plots with caballines according to the Palatal Index and Cheek Index (Supplementary Files S5 and S7) but not according to the Franck’s Index, where it is rather aligned with E. (A.) occidentalis (Supplementary File S8). Hibbard ([29], p. 73) stated also that “the majority of the skull characters resemble those of the asses and are, in some respects, nearer to the Asiatic (Hemionus) than to the African (Asinus).” Indeed, the Franck’s Index is close to that of hemiones (Supplementary File S35), and the Vertex Index (greatest length/basilar length proportions) is close to that of E. (Amerhippus) occidentalis (Supplementary File S10). There is another remarkable point: of all the monodactyl equid crania that we have ever seen, E. (H.) mexicanus is the only one where the palatine foramina are situated at the level of the posterior end of M3.

The upper cheek teeth are rather worn (Figure 42); the enamel is plicated, the protocones are long and pointed, and the styles are grooved, like in our Group A.

6.2. Referred Specimens

-

Tequixquiac, Estado de México

We have already mentioned the right ramus, IGM 71-401, tentatively referred to E. (H.) mexicanus by [29]. The enamel pattern of the teeth is typically caballine, especially on the molars: angular lingual valley, rounded metaconid, pointed metastylid (Group 1). The isthmus of the double knot is very high, and the ectoflexid is shallow on all teeth (Figure 36).

Hibbard [29] also questionably referred to E. (H.) mexicanus four associated P3-M2, N° 49-51 (Figure 42), and one m1 (N° 49-57). The protocones of the specimen N° 49-51 are very different from the holotype, being shorter and not pointed ([29], Figure 4c), and the measurements given by Hibbard ([29], Table VIII) do not agree with the drawing.

-

Barranca del Muerto, Estado de México

The mandible (Figure 43; Supplementary Table S5) LACM 308-123901 [7] fits well with the holotype cranium in size and proportions. The concavity between the ascending and horizontal ramus is less marked than in IGM 71-401 (Figure 36, black arrow). Infundibula are present on i1 and i2 and absent on i3.

-

Cedazo, Aguascalientes

Cranial and mandibular fragments FC 678 and FC 679 (Supplementary Table S5) may also belong to E. (H.) mexicanus as suggested by [77].

-

Cedral, San Luis Potosí

From the Late Pleistocene of Central Mexico, Alberdi et al. [43] described three species of equids. The largest, referred to as E. mexicanus, is represented by upper and lower cheek teeth and limb bones. The enamel patterns of both uppers and lowers are extremely polymorphic. A few upper cheek teeth like DP 2576 and DP 2574 resemble the holotype of E. (H.) mexicanus from Tequixquiac, although the protocones are less pointed, but some others have very small protocones like DP 2595 (Figure 44) that do not seem compatible with the teeth of the holotype of E. (H.) mexicanus. Differences between the lower cheek teeth are even more marked (Figure 45). DP 2322, 4605, and 4610 resemble the E. mexicanus from Barranca del Muerto, the others less so, or not at all. A morphometric analysis of upper and lower cheek teeth of Cedral specimens referred to E. mexicanus indicated a caballine affinity for this morph [16]. Unfortunately, there is no cranial material from Cedral referred to E. mexicanus that can be compared to the holotype of E. (H.) mexicanus.

Comparisons of the large metapodials and first phalanges from Cedral (Supplementary File S36) with those of E. (E.) scotti from Rock Creek and E. (A.) occidentalis from Rancho La Brea show that the Cedral specimens are very similar to E. (E.) scotti in overall size and proportions, and less so to E. (A.) occidentalis (Supplementary Files S37–S40). Equus (A.) occidentalis stands completely apart by the proportions of its limb segments (Supplementary File S41).

-

Río Sabinal, Oaxaca

• A partial cranium from Oaxaca State, Mexico, referred to E. mexicanus (UMPE 521) by ([23], Figure 4-A) has basicranial proportions that are different from the holotype of E. (H.) mexicanus (Supplementary Files S8 and S9). In fact, UMPE 521 (Figure 46) is similar to specimens of E. (A.) occidentalis (Supplementary File S42) and its cheek teeth also show the morphology observed in this species (Figure 46); UMPE 521 may be referred to E. (A.) occidentalis.
• The enamel pattern of the partial mandible UMPE 554 from Oaxaca ([23], Figure 4-B and C) is very perplexing. In addition to caballine patterns on other teeth, a non-caballine bridge affects the p4 (Figure 45). We have already commented on this kind of “bridge” (Figure 3); let us just stress that we have never seen one so pronounced in association with caballine double knots.

-

Other referrals

Priego-Vargas [44] referred to E. mexicanus an upper cheek tooth series (LACM-9063 A 82797.64) from San Marcos, Jalisco, Mexico. Miller et al. [78] referred to E. mexicanus the juvenile P3-M3 USPLA 012 and the very worn p2-m3 USPLA 039 from Rancho Carabanchel, near Cedral, Mexico.

6.3. Taxonomic Remarks

We are certain that the holotype cranium of E. (H.) mexicanus shows a unique blend of amerhippine and caballine characters. The upper cheek teeth of the holotype are caballine in overall morphology.

Several mandibles and lower cheek teeth referred to E. mexicanus may, or may not, belong to this species. Molars of IGM 71-401 from Tequixquiac, LACM 308-123901 from Barranca del Muerto, and UMPE 554 from Oaxaca have caballine double knots; others from Cedral are variable. The overall polymorphism is astonishing (Figure 47).

Large metapodials and first phalanges from Cedral are caballine, but more robust than E. (E.) scotti, and so is the MT UMPE 489 from Oaxaca (Supplementary Files S36 and S38).

We are thus confronted with an entirely original, large cranium, with caballine teeth, not associated but found with more or less caballine lower teeth and limb bones. If not for the cranial proportions, most of these fossils could be considered caballine and indeed comparative studies of tooth morphology found them to be so. Cranial proportions, however, should not be ignored and we agree with [29] that the cranium IGM 4009 deserves the name of E. (Hesperohippus) mexicanus. Analysis of DNA of the holotype of E. (H.) mexicanus will be helpful to clarify its relationship to other North American equids.

In the meantime, we consider with caution all fossils lacking crania assigned to this taxon (e.g., some material from Tequixquiac, Cedral, Oaxaca), and suggest referring to them as E. “mexicanus”.

7. General Conclusions

We studied cranial, dental, and postcranial remains of North American fossil equids of Irvingtonian and Rancholabrean age to characterize caballine forms and Equus (H.) mexicanus.

Our analyses suggest that in the Old World, caballines, Equus (Equus), can be discriminated from other equids using basicranial proportions, specifically the Palatal Index and Franck’s Index. In the Americas, these two indices do not completely discriminate caballines from other equids as some amerhippines, Equus (Amerhippus), overlap with caballines. However, other cranial proportions (i.e., Cheek Index and Vertex Index) allow better discrimination between caballines and amerhippines. We also documented some differences between these groups in metapodial dimensions and dental morphology.

We found morphological differences within North American caballines. These differences allow us to identify different forms. Though we use specific names when mentioning caballine forms, we do not imply that these forms were true species under a biological species concept. Our intent is to bring attention to the fact that there are morphological differences within North American caballines. Such differences may be described, and it is convenient to use well-known specific names in doing so. Nonetheless, other questions remain: Are these “species” ecomorphs? Chronomorphs? Are resemblances indication of parallel evolution? Or of close ancestral connections? In our study, we also remarked on problems associated with the use of the name Equus (E.) ferus for fossil caballine horses. Thus, we did not refer any North American caballine to E. (E.) ferus or to a subspecies of E. (E.) ferus.

The best represented caballines in North America are E. (E.) scotti of Irvingtonian age and E. (E.) alaskae and E. (E.) lambei of Rancholabrean age. These forms are represented by well-preserved cranial and postcranial remains. We hypothesize that E. (E.) alaskae and E. (E.) lambei are “ecological variants” of a single species. Equus (E.) niobrarensis cranium and upper cheek teeth are intermediate by size and morphology between the Irvingtonian E. (E.) scotti and the Rancholabrean E. (E.) alaskae and E. (E.) lambei; cranium proportions are similar to E. (E.) scotti and E. (E.) alaskae, while cheek teeth characters are similar to E. (E.) lambei and E. (E.) alaskae (Supplementary File S43). Other North American caballine forms that we recognize are E. (E.) pacificus and E. (E.) complicatus. A caballine cranium from Fort Qu’Appelle (SMNH 1515-8306) may be referred to E. (E.) niobrarensis. Two large caballine crania from Hay Springs (UNSM 1346 and UNSM 5989) show unique features that prompted us to identify them as Equus (E.) sp. Equus laurentius is a junior synonym of E. (E.) caballus. Equus holmesi is considered a junior synonym of E. (E.) scotti, while E. bautistensis may be regarded as a junior synonym of E. (E.) pacificus. We are uncertain about the nature of E. midlandensis.

On the whole, although there are more localities in North America, the species are less documented than in the Old World. Another distinction is the apparent lack of difference in the limb bone length proportions between E. (E.) scotti and the only documented specimen of E. (E.) alaskae–E. (E.) lambei (FAM 71467) with associated limb bones (Supplementary File S41).

The cranial proportions of the holotype of E. (H.) mexicanus show a unique blend of amerhippine and caballine characters. This led us to support that it belongs not just to another lineage, but rather to the subgenus Equus (Hesperohippus) erected by [29]. Metapodials that can be confidently referred to E. (H.) mexicanus are not known. In the absence of cranial remains, we considered with caution fossils assigned to this taxon and suggest referring to them as E. “mexicanus”. The partial cranium UMPE 521 from Oaxaca, Mexico, described as E. mexicanus by [23] may be referred to E. (A.) occidentalis.

We hope that the data, analyses, and interpretations provided in this study will stimulate further research into the taxonomy and systematics of Pleistocene horses and will motivate DNA studies of the holotype of E. (H.) mexicanus and other North American fossil horses.