Contributions in Science, 524:1-29

20 July 2016

Fossil Porcupine (Mammalia, Rodentia, Erethizontidae) erom El Goleo de Santa Clara, Sonora, Mexico, with a Review oe the

Q Taxonomy oe the North American Erethizontids^

11

L52X

David R. Sussman,'’^ Fred W. Croxen III,^ H. Gregory McDonald,"^ and Christopher A. Shaw^

ABSTRACT. Among the South American animals that entered North America following the establishment of the Panamanian land bridge were members of the family Erethizontidae. The early fossil record of this group m North America is sparse and so the discovery of fossil porcupines from the Middle Pleistocene (Irvingtonian) fauna of El Golfo de Santa Clara, Sonora, Mexico, provides significant new information on the history of these animals in North America. The taxonomic assignment of the new material required a review of the systematics of North American erethizontids. Based on the most common element preserved, we have restricted our study to features of the dentary that can be used to distinguish the Neotropical porcupine Coendoit from the North American porcupine Erethizon, supplemented by an examination of differences in the caudal vertebrae that also distinguish the two genera. The results of our study lead us to recognize the presence of both Coendoti and Erethizon in the fossil record of North America and to assign the El Golfo fossils to Coendoti cf. C. kleini. We also reassign the taxa E. poyeri and E. klemi, previously reported from Florida, to the genus Coendou. Further, we reassign specimens previously identified as Erethizon dorsatiim from Irvingtonian faunas in Florida and Aguascalientes, Mexico, to the genus Coendou.

In contrast to previous studies in which a linear evolution from Coendou to Erethizon was proposed, we present an alternative model for the origin of the North American genus Erethizon. We propose the existence of two different migration pathways through Mexico. In the west, a branch dispersed north along the north-south mountain ranges into northern North America, resulting in the evolution of modern Erethizon in the more temperate part of the continent. In the east, a second branch followed the Gulf Coast into Florida, resulting in the establishment of an eastern population of Coendou in the subtropical part of North America, a population that eventually became extinct.

INTRODUCTION

An Early to Middle Pleistocene fauna has been recovered from badlands near the small fishing village of El Golfo de Santa Clara (El Golfo), Sonora, Mexico. The badlands, which cover about 100 square miles, are formed by the erosion of sediments deposited in the delta of the Colorado River at the northern end of the Gulf of California over the past four or five million years (Dorsey, 2006). These sand-dominated sediments, with interbeds of silt, clay, and gravelly sands, were subjected to a series of tectonic uplift events (Colletta and Ortlieb, 1984; Pacheco et ah, 2006) followed by water and wind erosion and the consequent exposure of large numbers of vertebrate and botanical fossils (Croxen et ah, 2007).

Over the last 20 years, an organized effort to recover these fos- sils has been undertaken by Professor bred Croxen and Christo- pher Shaw under permit from the government of Mexico and under the oversight of the Reserva de la Biosfera Alto Golfo de California y Delta del Rio Colorado. At present, about 11,500 individual fossils have been recovered and identified, representing more than 130 different vertebrate taxa. Among these fossils are

’ URL: www.nhm.org/scholarlypublications

^Department of Biology, Arizona Western College, Yuma, Arizona 85367, USA.

^Department of Geology, Arizona Western College, Yuma, Arizona 85367, USA.

United States Bureau of Land Management, Salt Lake City, Utah, 84101, USA.

^Department of Rancho La Brea, La Brea Tar Pits and Museum, Natural History Museum of Los Angeles County, Los Angeles, California 90036, USA.

^ Corresponding author: David R. Sussman, E-mail: susshotch@gmail.com

partial dentaries, teeth, and a single humerus that represent six- individual erethizontids, the subject of this paper.

As discussed by Croxen et al. (2007), establishing the exact geo- logic age of these fossils is difficult. The fossil assemblage is assumed to represent a single paleobiota, but the lack of inter- bedded ashes prevents determining its age radiometrically. Although studies of the magnetostratigraphy have not produced definitive results, preliminary data suggest that the fossil-bearing sediments are reversely magnetized (F.W. Croxen, personal com- munication, 2015). If this is the case, the Matuyama reversed epoch (2.588-0.781 Ma) (Cohen et al., 2013) is the only reversal of an age appropriate to the vertebrate assemblage at El Golfo. This assemblage correlates with other faunal assemblages asso- ciated with the Irvingtonian North American Land Mammal Age (NALMA) (Bell et ah, 2004), particularly with the fauna from the Irvingtonian portion of the stratigraphic sequence at Anza- Borrego State Park in California (Cassiliano, 1999; Jefferson and Lindsay, 2006). Fossils of Mammuthus meridionalis, Megalonyx wheatleyi, Nothrotheriops texamis, and Sigmodon curtisi are present, but Bison is not and, given the large available sample size, the absence of Bison does not seem to be a collection artifact. Some Blancan taxa {Canis lepophagiis, Felts rexroadensis, Sigmodon curtisi) are present and may be considered remnant populations but suggest an earlier rather than later age for the fauna at El Golfo. However, the presence of Mammuthus suggests an age not older than about 1.35 Ma, the earliest well-dated appearance of Mammuthus in North America (Bell et ah, 2004). With this information, an age range of 0.781 Ma (youngest possi- ble Matuyama) to 1.35 Ma seems a reasonable estimate for the fauna of El Golfo.

Fossil porcupines in North America were reviewed by White (1968, 1970). Based on his examination of a variety of dental

NOV 23 2016 j

2 ■ Contributions in Science, Number 524

Sussman et al.: North American Erethizontidae

and skeletal features and on measurements in modern and fossil specimens, White concluded that the earliest of these fossil porcu- pines should be referred to the extant Neotropical genus Coendon and that the modern North American genus Erethizon evolved in North America from an immigrant species of Coendou. The fossil North American erethizontids were subsequently reevaluated by Frazier (1981), including the specimens studied by White plus newly discovered material, and Frazier reached the conclusion that, in contrast to White, all the North American fossils belong in the genus Erethizon.

Since the work of Frazier, most researchers have accepted the assignment of fossil North American porcupines to the genus Erethizon (e.g., Morgan and White, 1995; Flulbert, 1997; Alb- right, 1999). Currently, five species of North American porcu- pines are recognized from the fossil record (Frazier, 1981; Hulbert, 1997). These taxa include Erethizon bathygnathnm and E. cascoensis (which could be considered bathygnathnm) from western North America, E. poyeri and E. kleini from. Florida, and fossils of the living species, E. dorsatnm, from several local- ities in the United States and Mexico. However, there is not com- plete agreement regarding the generic identification of these fossils and the discovery of the porcupine fossils at El Golfo prompted us to reevaluate the taxonomy of the fossil North American porcu- pines, in order to assign the El Golfo fossils appropriately.

The ancestors of modern Erethizon first entered North America from South America after the formation of the Panamanian land bridge. The timing of the closure of the isthmus, which defines the earliest possible time for the porcupines’ entry into North America, is important for understanding their subsequent evolu- tion in North America. Previously, the closure was thought to have occurred around 3.5 Ma. (Webb, 1985; Flynn et ah, 2005). However, studies utilizing different dating methodologies, one by Montes et al. (2015) and the other by Bacon et al. (2015), now suggest a more complex and protracted history of closure, possibly beginning in the middle Miocene or earlier, with asso- ciated faunal dispersal events in both directions, the last major dis- persal event occurring about 6 Ma. These earlier dates, if accurate, create a many-million-year gap (6 Ma or earlier to about 2.5 Ma.) during which porcupines may have entered North America but for which we have no fossil record. Other studies related to the ances- try and early history of North American porcupines include those of Vilela et al. (2009) and Voss et al. (2013), who have proposed, based on mitochondrial molecular clocks, that the lineages ances- tral to modern Coendon and Erethizon separated from a common ancestor at some point during the late Miocene.

Two questions need to be answered regarding Plio-Pleistocene erethizontid fossils in North America. The first, addressed pre- viously by Sussman (2011), relates to the assignment of a 2.5-Ma dentary from the Uquia Formation in Argentina to the genus Erethizon by Reguero et al. (2007). The issue is whether the original immigrants to North America had enough morphological traits to assign them to the genus Erethizon, as the assignment of the Uquian fossil to Erethizon suggests is possible, or whether the distinguish- ing hallmarks of Erethizon evolved after the immigrants entered North America and were exposed to the temperate North American environment. Second, if the morphological traits characteristic of Erethizon evolved in North America and are only present in the North American taxon, do any of the North American fossils show traits that might suggest a closer generic relationship to the Neotropical porcupine Coendou than to Erethizon} The first ques- tion is discussed in the previously mentioned Sussman paper (2011), with the conclusion that the Uquian fossil more properly belongs in the genus Coendou. The second will be examined here.

The assignment of the porcupines from El Golfo to a genus requires precise definitions of the skeletal and dental characters

that distinguish the two living genera and so permit a comparative evaluation of the erethizontid fossils that have been found on both continents.

The extant North American porcupine, Erethizon dorsatnm, is monotypic. Its distribution includes boreal forests as far north as Canada and Alaska and the forests (and occasionally deserts) of the western, north-central, and northeastern United States, and extends south into the mountainous regions of northern Mexico. Although E. dorsatnm may be found in open scrub, it prefers forested habitat. It is arboreal and utilizes its muscular tail exten- sively in its tree-climbing activities. The tail, however, is not prehen- sile, in contrast to the tail of its Neotropical relative Coendon. The diet of Erethizon consists preferentially of the leaves and shoots of trees in hardwood forests when available, but it also possesses the well-known ability to eat the cambium layer of tree bark, a trait that allows it to survive winters in deciduous and coniferous forests. Notably, it is the only erethizontid known to have adapted to sub- arctic conditions and it survives the harsh winters of Canada and the northern parts of the United States not by hibernating but by continuing to forage for food (Roze, 1989).

Subsequent to the arrival of the original immigrants, winters in North America became progressively colder, with the onset ulti- mately of severe glacial conditions. The substantial metabolic and physical changes evidently necessary to allow a Neotropical porcupine to survive in such an environment form the crux of the argument we propose below that the generic name Erethizon must apply only to porcupines that have evolved in North America.

Differences of opinion exist regarding the taxonomy of living South and Central American porcupines. Morphologic and genetic evidence support the conclusion that Neotropical porcu- pines are monophyletic, with the exception of Chaetomys (a mor- phologically distinct animal with no close living relatives) (Voss and Angermann, 1997; Voss and da Silva, 2001; Voss, 2011; Voss et ah, 2013). All Neotropical porcupines (except Chaetomys) are included in the genus Coendon by Voss. Consequently, Sphig- gnrns and Echinoprocta are, in this scheme, considered junior synonyms of Coendon. In this paper we follow the nomenclature proposed by Voss.

The extant species of Coendon live in tropical and subtropical forests and savannahs of South and Central America. All living species of Coendon (13 or more) are substantially smaller in body size than Erethizon, with some being quite diminutive. Although knowledge of their natural history is generally sparse, it appears that the ecology of all species of Coendon is similar. They eat the flowers, leaves, stems, and fruit of the forests, which are available year round. They may also eat roots, tubers, and insects (Woods, 1984; Emmons and Eeer, 1997; Eisenberg and Redford, 1999). Anecdotal reports that Coendon, like Erethizon, eats bark have not been well documented (L.H. Emmons, personal communication, 2006). They are arboreal and utilize prehensile tails for clinging to branches. As discussed by Voss et al. (2013), even the short tail of C. rnfescens displays evidence of being pre- hensile, although it is at the short end of a spectrum of prehensile tail lengths possessed by the various Coendou species.

Modern Erethizon has been found as far south as southern Sonora and Chihuahua (Jones and Genoways, 1968) and Coen- don is common in Yucatan and is found as far north as San Luis Potosi on the east side of the Mexican mainland (Emmons and Feer, 1997) and in the state of Michoacan on the west (Monterru- bio-Rico et al., 2010), but a distance of several hundred miles in central Mexico separates the modern northern and southern por- cupine genera (Fig. 1). This contrasts with the geographically con- tiguous populations of porcupines that presumably existed after the original dispersal event(s) into North America.

Contributions in Science, Number 524

Sussman et al.; North American Erethizontidae ■ 3

METHODS AND MATERIALS

In order to find consistent anatomical and proportional differ- ences between the two extant genera {Erethizon and Coendou), we studied numerous examples of modern animals, including nine species of Coendou (Appendix 1). We concentrated on den- taries and teeth because these predominate among the El Golfo fossil specimens, are among the most commonly preserved parts of porcupine fossils from other localities, and would seem to have sufficient anatomical complexity to permit the recognition of consistent differences between the genera. Dentaries are also the type specimens for three of the extinct North American species [E. bathygnatlmm (USNM 13684), E. poyeri (UF 121740), and E. kleini (UF 21473)] and to our knowledge are the only erethi- zontid fossils from Uqm'a (MACN 5376) and Tarija (MNHM TAR 695, 696; Hoffstetter, 1963) in South America.

Table 1 provides the dentary measurements we made on mod- ern porcupines, with their averages, ranges, and standard devia- tions. Where useful, we also performed ratios or multiplied length X width of tooth measurements and these are also shown in Table 1. Values are also shown for the fossils from E! Golfo (Table 2), Florida (Table 3), California, Idaho, Aguascalientes (Table 4), and Cumberland Cave, Maryland (Table 5). We exam- ined the caudal vertebrae of porcupines to identify differences between the nonprehensile and prehensile porcupine tails. Although the fossil record is minimal, the differences may be revealing.

Detailed descriptions of how we performed our measurements are found in Appendix 2.

ABBREVIATIONS

ABDSP Anza-Borrego Desert State Park, California AMNH American Museum of Natural History, New York, New York AWC Arizona Western College, Yuma, Arizona

DMNS Denver Museum of Nature and Science, Denver, Colorado FMNH The Field Museum of Natural History, Chicago, Illinois F:AM Frick Collection, American Museum of Natural History,

New York, New York

IGM Instituto de Geologia, Universidad Nacional Autonoma de

Mexico, Mexico City, Me.xico

rVCM Imperial Valley College Museum Collection, Anza-Borrego Desert State Park, California

LACM Natural History Museum of Los Angeles County, Los Angeles, California

LACM Natural History Museum of Los Angeles County, California (CIT) Institute of Technology, Los Angeles, California MACN Museo de Historia Natural de Buenos Aires, Argentina MNA Museum of Northern Arizona, Flagstaff, Arizona

MNHN Museum National d’Histoire Naturelle, Paris, France PSM University of Puget Sound Slater Museum of Natural

History, Tacoma, Washington

UF Florida Museum of Natural History, Gainesville, Florida

UMMP University of Michigan Museum of Paleontology, Ann Arbor, Michigan

UO University of Oregon Condon Museum of Geology, Eugene,

Oregon

USNM National Museum of Natural History, Washington, D.C. ZMUC Zoological Museum, University of Copenhagen, Denmark

RESULTS

MODERN DENTARIES

In the modern dentaries, we found five traits that are useful in dis- tinguishing the two genera. Refer to Appendix 2 for definitions of terms and measuring techniques. The five traits are listed below.

1. The length of the cheek-tooth row (Fig. 2A): It is immediately apparent that this measurement is definitive in living porcupines.

Only the very smallest E.rethizon equals the very largest Coendou in size and there is minimal overlap at those extremes. White (1970:10) found fossil measurements of the cheek-tooth row that fall in the range of modern E.rethtzon but nevertheless assigned these fossils to the genus Coendou. Frazier (1981:24- 25) included these values (approximately the same as White’s) in a classification by discriminant analysis and concluded the fos- sils should be assigned to the genus E.rethizon.

2. Ratio of size of p4 to ml: A p4/ml ratio smaller than about 1.10 indicates Coendou, while a p4/ml ratio larger than about 1.36 indicates Erethizon. Although most Coendou tend toward smaller ratios and Erethizon toward larger, those individuals with a ratio between 1.10 and 1.36 cannot be reliably distin- guished by this trait. White (1968:8), measuring only the widths of p4 and ml, arrived at a similar conclusion, while Frazier (1981) did not address this trait.

3. Ratio of anterior-posterior (A-P) diameter to lateral width of lower incisors: In modern adult porcupines, there is an overlap in the ratio of A-P diameter to width between the two genera at 1.2, but incisors with a greater ratio are Coendou and those with a lesser ratio are Erethizon. Porcupine incisors, in contrast to molars and premoiars, continue to grow after eruption in both length and in A-P and lateral diameters for much if not all of the animal’s life; this produces a difficulty in using this trait because incisors of Coendou appear to lengthen in the A-P direc- tion at a more rapid rate than they widen laterally as the animal grows into adulthood, starting off in newborns with a square (i. e., 1.0 ratio) cross section (PSM 14197, 14201, 14202, 14203) and then progressively elongating in the A-P direction over the next 1-2 years (ratio 1.1: AMNH 262274 with m2 and m3 unerupted, FMNH 14182 with partially erupted m3, LACM 74306 with dp4 and all three molars). The incisors Erethizon retain their juvenile cross-sectional square shape as adults (71 of 75 have ratios of 0.9-1. 1). Because of this apparent discrepancy in growth patterns, only animals with permanent p4’s (i.e., adults) were included in the analysis.

While both White (1970:11) and Frazier (1981:14) measured incisors, neither discussed this trait.

4. The orientation of the cheek-tooth row relative to the incisor (Fig. 3): The results of our examination of this trait differ con- siderably from those of both White (1968, 1970) and Frazier (1981). The reasons for this may relate to the fact that the cheek-tooth row and the incisor on the same side lie in different horizontal and vertical planes and so slightly different measur- ing techniques or differing rotational positions of the dentary may result in significantly different results. Nevertheless, our technique resulted in 72 of 73 dentaries of Coendou having cheek-tooth axes directed lateral to the incisor and one-third of 58 axes directed medially in Erethizon. Thus, utilizing our technique, a cheek-tooth axis that runs medial to the incisor in an individual porcupine strongly suggests Erethizon, while a large sample of dentaries from a given population of porcu- pines, all with axes running lateral to the incisor, may suggest Coendou. Our results show that modern dentaries of Erethizon possess a distinct trend toward anterior convergence of the cheek-tooth rows, a trend not present in the dentaries of mod- ern Coendou, in which the cheek-tooth rows remain parallel.

5. The amount of anterior extension (procumbency) of the inci- sors (Figs. 2B, 2C, 4): As first described by Sussman (2011), the lower incisors of Coendou extend anteriorly an average of 7 degrees more than those of Erethizon. This difference appears to provide an important means for distinguishing between the two genera. When viewed in conjunction with the tendency for the cheek-tooth rows in Erethizon to converge medially toward the incisor and the stronger, square-shaped

4 ■ Contributions in Science, Number 524

Sussman et al.; North American Erethizontidae

cross section of the incisors in Erethizon, this trait suggests the possibility of an evolutionary process in Erethizon, first noted by White (1970), which provides a strengthened jaw mechan- ism as compared with Coendoti.

Other measurements in Table i, excepting size, do not provide distinctive differences between the genera. In particular, the length of the diastema relative to the length of the cheek-tooth row (the length of the cheek-tooth row should not change once all the teeth have erupted) cannot be used to help diagnose a genus or species because the length of the diastema is variable, being sometimes shorter and sometimes longer than the tooth row. This variation appears to be age (and therefore growth)-related: in Erethizon, our measurements show the diastema/tooth-row ratios to be 0.50 in juveniles (n=13), 0.57 in young adults (n=37), and 0.62 in older adults (n=ll); in Coendou the values are 0.44 (n=8), 0.53 (n=36), and 0.58 (n = 52). Thus, on average, older porcu- pines have longer diastemas relative to the cheek-tooth row com- pared with those of younger animals.

The angles of scratches on the enamel of the cheek-tooth occlu- sal surface due to mastication were studied by White (1968), mea- sured from the longitudinal axis of the tooth row. Differences were found between the two genera, providing another diagnostic trait. However, these differences were not considered by Frazier (1981) sufficient to assist in generic identification. Our attempts to examine the scratches suggest that scratches may have multiple angles on the same tooth; curved scratches were also noted, prob- ably because the porcupines do not always chew in a strictly straight back-and-forth or side-to-side motion, but with a circular contribution as well. If this chewing motion is the case, then multi- ple scratch directions might be expected. In addition, the long axis of individual cheek teeth frequently — but not consistently — varies from the axis of the entire tooth row, producing another difficulty in using these scratches, particularly in isolated teeth. As a result, we conclude that we are not able to utilize the enamel scratches diagnostically.

MODERN TAILS

The number of caudal vertebrae in modern porcupine tails, while variable, appears to depend on whether the tail is prehensile or not. In 19 Coendou, we saw tails with as few as 22 and 24 verte- brae, but in most specimens the number ranged from 26 to 34. Thirteen of these tails are from specimens of C. prehensilis. We did not see any Coendou for which the number of caudals is as small as the numbers observed in the nonprehensile tails of Erethi- zon: 13-17 in 19 tails described by Sutton (1972) and 8-16 by us (n=14). The series of vertebrae in tails of Coendou decreases in size gradually in more posterior positions, whereas in Erethizon the caudals show a rapid progression from larger to smaller for the length of the tail.

Caudals of Coendou have single transverse processes in the most anterior vertebrae, but beginning variably at about the 12th to 15th vertebra, foramina (probably neurovascular) appear in the transverse processes and the processes then gradually pro- gress to complete bifurcation around the 17th to the 20th verte- bra. The bifurcations persist for a few vertebrae and then foramina may reappear in the transverse processes of more poste- rior vertebrae and may or may not persist nearly to the end of the tail (Fig. 5). The exact vertebrae involved in these changes in the transverse processes varies in individual animals, but includes about one-third to one-half of the vertebrae.

We saw no foramina in the transverse processes of caudal ver- tebrae of modern Erethizon and there were no bifurcations in most animals. However, bifurcated transverse processes were

observed in one or two posteriorly located caudal vertebrae in a small number of individuals (e.g., MNA Z9.509; Fig. 5D). It is probable that these bifurcated transverse processes in the caudal vertebrae of Erethizon are equivalent to, and vestiges of, those seen in the more centrally located vertebrae found in the longer tails of prehensile-tailed ancestors of Erethizon prior to the evolu- tionary loss of the most posterior caudal vertebrae. That is, the trait in Neotropical porcupines that produces the foramina and bifurcated processes may also exist in Erethizon but is only rarely observed because of the evolutionary loss of the involved vertebrae.

In the following discussion of the various fossil erethizontid dentaries, we compare the fossils to the five traits we identified as distinguishing modern Erethizon from modern Coendou as il- lustrated in Table 6.

SIZE AS A TRAIT TO IDENTIEY A GENUS

Lagoa Santa Eossils versus Shelter Cave Fossils A large portion of Frazier’s argument for assigning all North American porcupine fossils to the genus Erethizon is based on a statistical comparison of cranial and mandibular measurements of the fossils with samples of extant Erethizon and Coendou (Fra- zier, 1981:21-26). His data demonstrate that the fossils are the same size as Erethizon. Similarly, Frazier stated (1981:28) that “(i)ncisor enamel thickness in Erethizon is significantly greater than in Coendou.” Using Frazier’s measurements of A-P diameters of incisors and his measurements of enamel thickness, we calcu- lated the percentage of the diameter of the incisors that is enamel and found it to be 5.7% for Coendou (0.21mm/3.7mm) and 5.6% (0.27mm/4.8mm) for Erethizon. Thus, since the percentage of enamel thickness is virtually identical in both genera, the signif- icantly different absolute values of enamel thickness in the genera reflect only differences in the sizes of the animals, as in the other measurements mentioned here.

In general, we do not accept that size, in and of itself, is suffi- cient to determine to which of two genera an individual specimen should be assigned. Also needed are significant differences in mor- phology and function that may reflect significant ecological differ- ences. In contrast, a species of differing size isolated from related species geographically (e.g., Mcmimuthus exilis] or in time (Coen- dou kleim as mentioned below) may be considered a distinct spe- cies (but not genus).

To support our contention that the physical differences we found between modern North American and Neotropical porcu- pines are real and actually distinguish the genera and that they are not related to size, we examined the previously unpublished late Pleistocene/Holocene fossils of Erethizon dorsatum from Shel- ter Cave, New Mexico, and contemporaneous fossils from Lagoa Santa, Brazil (Coendou magmis) (Winge, 1888; Hansen, 2012) (Table 7). It is important to note that the fossils from Lagoa Santa are the same size or somewhat larger than modern Erethizon and are much larger than any modern species of Coendou (Table 1).

Looking at the five distinguishing characters described above, the fossils from Lagoa Santa are about 10% larger than those from Shelter Cave; the single p4/ml ratio from Shelter Cave is 1.31, high but within the range of modern Coendou. The five fossils from Lagoa Santa average a p4/ml ratio of 1.08 (Coen- dou-only), with three of them falling within the range that is exclu- sively Coendou and two being in the area of overlap of the two genera.

Four out of five adult incisors from Shelter Cave have the cross-sectional shape of Erethizon and the fifth is in the area of overlap between the two genera. Three incisors in adult dentaries from Lagoa Santa are either in the area of overlap (n=2) or are

Contributions in Science, Number 524

Sussman et al.: North American Erethizontidae ■ 5

Coendou in shape (n=l). A fourth incisor is in a dentary that retains a dp4 and, as predicted, has a square shape. We examined six isolated incisors and, although we could not definitively determine the animals’ ages, the two largest incisors (larger than the others hy 2-3 mm in the A-P direction and 0.6-1. 7 mm in the lateral direction, thus probably adults) have large ratios ( 1.4, 1..5) as in Coendou and the smaller remaining incisors have ratios of 1.1, with the exception of one that has a 1.3 [Coen- doii) ratio.

The cheek-tooth axes of three fossils from Shelter Cave extend lateral to the incisor; two axes are centered on the inci- sor. In the dentaries from Lagoa Santa, the axes of all six extend laterally.

All five dentaries from Shelter Cave with measurable incisor angles of procumbency fall into the Erethizon-on\y range. Eight measurements from five individuals from Lagoa Santa all are in the Coendou-on\y range except for one buccal and one lingual measurement, both of which fall within the areas of overlap between Coendou and Erethizon (Fig. 2D).

An assembled series of disarticulated caudals from the fossils from Lagoa Santa produced a tail with 31 vertebrae. Some of the vertebrae contain foramina in the transverse processes and others show bifurcated transverse processes (Fig. 6). Except for its larger size, the tail appears to us to be identical to that of mod- ern Coendou.

In summary, the fossils from Lagoa Santa are larger than those from Shelter Cave. In the remaining traits, measurements that do not fall into the range of overlap between the two genera are, in the case of Lagoa Santa, exclusively in the range of Coendou and, for Shelter Cave, exclusively in the range of Erethizon. In addition, the porcupines from Lagoa Santa possessed a prehensile tail, a trait found only in modern Coendou.

Thus, a comparison of the fossils from Shelter Cave with the fossils from Lagoa Santa shows that our defined characteristics can distinguish the two genera, regardless of size. A large species of Coendou lived in South America into late Pleistocene/Holocene times, demonstrating that size alone is not sufficient to diagnose fossil Erethizon.

FOSSIL PORCUPINES El Golfo de Santa Clara

Fossils of porcupine from El Golfo de Santa Clara include the following:

AWC 10858: Right dentary fragment with incisor and m2 and m3; m3 less worn than m2, suggesting a young adult; ascending ramus missing; symphysis and diastema intact (Fig. 7).

AWC 12197: Left dentary fragment with a worn p4; incisor broken off at top of bony alveolus but its diameters are measur- able; ml broken off at top of bony alveolus and occlusal surfaces of m2 and m3 are broken off; most of ascending ramus missing and symphysis and diastema, while mostly intact, have missing portions (Fig. 8).

AWC 13592: Worn right ml in small fragment of mandibu- lar bone.

AWC 14810: Right dentary fragment with partially broken incisor and ml-3 but missing p4; m3 minimally worn, suggesting young adult; missing diastema, symphysis, and ascending ramus (Fig. 9).

IGM 10199 (collected by H. Garbani ca. 1980): isolated broken right incisor.

AWC 12540: Right humerus, about one-third of proximal end missing; epiphyseal suture fully closed, length of fragment 67 mm. Because of the erosion of the compact surface bone, other meaning- ful measurements cannot be made (Fig. 10).

Measurements of the fossils are provided in Table 2.

Applying the results from our study of modern porcupines to the fossil remains from E4 Ciolfo demonstrates the following: The lengths of the cheek-tooth rows of the two measurable FI Golfo fossils (AWC 10858 and 12197) are intermediate in size between £. dorsatum and C. prehensilis, equaling the smallest Erethizon and the largest Coendou.

None of the fossils had both p4 and an intact ml, so that ratio could not be calculated.

Four fossil incisors could be measured. Two had an A-P/lateral ratio of 1. 1 and two were 1.2. Both dentaries with incisor ratios of 1.1 are missing the (d)p4’s but have lightly worn m3’s (suggesting a young adult), so this measurement provides equivocal results in the fossils.

Two dentaries (AWC 10858 and 12197) have cheek-tooth row axes that run lateral to the incisor, providing inconclusive generic evidence.

AWC 10858 and 14810 have measurable incisor angles of pro- cumbency that fall well within the range of Coendou, hut are at the highest values for Erethizon.

A single p4 and six molars from four different animals all mea- sure either smaller than the smallest modern Erethizon or smaller than the mean for Erethizon. Two are the size of the largest Coendou.

Although incomplete, the humerus is from an adult, appears to be larger than modern species of Coendou, and may be about the size of a small Erethizon.

No caudal vertebrae from El Golfo have yet been recovered.

In summary, the fossils from El Golfo fall in an intermediate size range between modern Erethizon and Coendou, have anterior extension of the lower incisors as in modern Coendou, and have no traits exclusive to Erethizon. They are the same size as C. kleini from Florida (discussed below).

Although the porcupines from El Golfo have few distinctive physical traits in the fossils we have found, they do possess the combination of a Coendou-type incisor procumbency, laterally directed cheek-tooth axes, and the absence of any traits definitely restricted to Erethizon. Additional evidence regarding their iden- tity may be derived from the paleoenvironment of the El Golfo area at the time the porcupines lived. The recovery of remains of fan palm (Washmgtonia sp.), giant tortoise (Hesperotestudo sp.), crocodile (Crocodylus sp.), beaded lizard (Heloderma sp.), boa constrictor [Constrictor constrictor), crested guan [Penelope sp.), flamingo [Phoenicopterus sp.), and giant anteater [Myrmeco- phaga tridactyla) at El Golfo implies that the annual regional tem- perature supported tropical to subtropical environments (Mead and Shaw, 2011; Shaw et ah, 2013).

Steadman (2011) notes that, regarding the avifauna of El Golfo, “[t]he more tropical biogeographic affinity of some of the birds agrees with that suggested for certain associated reptile and mammal fossils.” This subtropical ecosystem, of which the El Golfo porcu- pines were a part, supports our interpretation that they are Coendou and not Erethizon, based on the known habitat differences of the two modern genera. There do not appear to be any convincing rea- sons to think that the fossil porcupines at El Golfo were preadapted for cold-weather tolerance or that they subsequently acquired it while living in a subtropical climate, cold-weather tolerance being a defining characteristic of living Erethizon.

The temporal and physical isolation of the porcupines at El Golfo from the other fossil porcupines in North America raises the possibility that they were part of a dispersal event to North America distinct from that of earlier dispersals. As a group, their size — somewhat smaller than other early to mid- Pleistocene fossil porcupines, except for C. kleini in Florida, and smaller than modern Erethizon but larger than extant Coendou — perhaps represents an intermediate stage in a

6 ■ Contributions in Science, Number 524

Sussman et al.: North American Erethizontidae

diminution in size of Coendoii that apparently occurred at some point during the Pleistocene but for which we know of no South American fossil record.

We conclude that the El Golfo porcupines should be assigned to the Neotropical genus Coendou and further, due to their overall similarity to C. kleini, we assign them to Coendou cf. C. kleini, pending the discovery of more diagnostic material to help clarify the species designation.

Florida

We examined porcupines from late Blancan and Irvingtonian fau- nas from five sites in Florida previously identified as Erethizon poyeri (found at Haile 7C, ca. 2 Ma) (Hulbert, 1997), E. kleini (Inglis lA, ca. 1.9 Ma) (Frazier, 1981), E. dorsatum (Haile 16A, ca. 1.6 Ma.) (Morgan and White, 1995), E. dorsatum (Feisey 3 A, ca. 1.5 Ma.) (Morgan and White, 1995), and E. dorsatum (Coleman 2A, 0.6-0. 3 Ma.) (Morgan and White, 1995). See Table 3 for measurements.

The lengths of the cheek-tooth rows are long, at or above those of Erethizon, except for two fossils from Inglis lA, which are somewhat smaller, falling between the sizes of modern Coendou and Erethizon.

The E. poyeri fossil (UF 121740) has a p4/ml ratio of 0.96, in the range of modern Coendou. The four other measurable fossils all have larger ratios, none of which exceed the range of Coendou, but trending toward Erethizon.

Three incisors have A-P/lateral ratios of 1.1 and one has a ratio of 1.2. Of the three with ratios of 1.1, one has worn molars (UF 21473), suggesting an older adult, one has a broken (d)p4 and lightly worn ml and m2 but is missing m3, possibly a juvenile or young adult (UF 135669), and the third has an unworn p4 and lightly worn ml and m2, a young adult (UF 21490). With the exception of the single older adult, these measurements are not diagnostic.

The axes of six cheek-tooth rows relative to the incisor, repre- senting three of the five sites we studied, are measurable (Fig. 11). All six axes run lateral to the incisor, a feature that is charac- teristic of Coendou. Our findings contrast with those of Morgan and White (1995) that the axes of the Feisey 3 A cheek-tooth rows run medial to the incisor. We found that, using our method of analysis (see Appendix 2), UF 124632 has an axis that is direc- ted laterally (Fig. 11); in UF 135669, the incisor alveolus and the alveolus of p4 are both broken, making a measurement unreliable.

Six measurements of incisor angle of procumbency (from four fossils, representing three sites) all fall in the range of Coendou, are greater than the range of Erethizon, and all but one are greater than the mean for Coendou (Figs. 2F, 2G, 12).

The sizes of the p4’s and molars in the Florida fossils, as judged by the product of length times width, all fall within or slightly above the range for modern Erethizon, with the notable exception of E. kleini. Three dentaries of E. kleini and their individual teeth have measurements close or equal to those of the fossils from El Golfo (see above) and are intermediate between modern Coendou and Erethizon. This intermediate size was the primary distinguish- ing feature used by Frazier (1981) to designate them as a new spe- cies (Fig. 2E).

A previously undescribed specimen of E. poyeri from Haile 7G was preliminarily reported by Hastings et al. (2006). It includes eleven caudal vertebrae, some of which were noted to have bifur- cated transverse processes (“dual transverse process sets”) and from those vertebrae the investigators estimated the original length of the tail at “26-32 caudal vertebrae.” In addition, eight other skeletal traits “are preserved well enough in this new speci- men for study and all plot within the range of Erethizon." While not explicitly stated, these traits presumably relate to the size of

the animal. We have not been given access to this fossil and so can- not at this point make an independent judgment regarding its taxonomic assignment.

Except for size and the one incisor ratio, none of the studied traits in the fossils from Florida fall into a range exclusive to Erethizon. The large p4/ml ratios trend towards Erethizon but there are modern Coendou with the same ratios. On the other hand, the small p4/ml ratio of E. poyeri falls well within the range of Coendou. The six tooth-row axes, all directed laterally, suggest Coendou, because in a population of Erethizon one or two axes could be expected to be medial.

The strongest evidence in the Florida fossils is found, first, in the incisor angles of procumbency, which are uniformly in the range of Coendou, and, second, as described by Hastings et al. (2006), in the presence in the caudal vertebrae of bifurcated transverse processes accompanied by the suggestion of approximately 30 total vertebrae — in other words, a prehensile tail, a distinguishing trait of modern Coendou.

The porcupines from Florida we examined show no evi- dence of the mandibular modifications present in modern Erethizon, that is, the anterior convergence of the cheek- tooth row relative to the incisor and the more acute angle of incisor procumbency in Erethizon. These modifications we interpret as helping to allow Erethizon to survive in northern climates by eating the cambium of trees when no other food is available. Except for their larger size, the fossil porcupines from Florida cannot be distinguished from mod- ern Neotropical Coendou. We therefore conclude there is suf- ficient reason to reassign the Florida porcupines found at Haile 7C, Haile 7G, Inglis lA, Feisey 3 A, and Haile 16A to the genus Coendou. We assign the nondiagnostic Coleman 2A dentary fragment to Coendou also, as a relict population, acknowledging that its young age may represent instead a south- ern range extension of Erethizon.

The fossils from Inglis lA were described as a new species (£. kleini) by Frazier (1981) based on their size, which is somewhat smaller than the other fossil porcupines from Florida. The fossils from El Golfo described above are the same size as C. kleini. However, because several hundred thousand years and two thousand miles separate the two populations, the strength of the relationship between the two groups can be questioned. Still, the existence of two separate populations of smaller fossil porcupines at least suggests that C. kleini from Florida and Coendou cf. C. kleini from El Golfo may represent distinct South American dispersals into the North America.

Western North America

Aside from the fossils from El Golfo, we examined fossils from faunas in the western United States: Grand View from Jackass Butte, Idaho (ca. 2.0-2.3 Ma) (Wilson, 1935; Shotwell, 1970; Repenning et al., 1995; Bell et al., 2004), and Vallecito Creek and El Casco, California (Table 4). Work by F. K. Murray (perso- nal communication, 2011) restricted all of the Vallecito Creek erethizontid fossils to the C2r.ln geomagnetic chron (1.9-2. 2 Ma). The fossils from El Casco have a similar age, with the excep- tion of a single broken tooth from an adjacent site (the San Timo- teo Badlands) that may, along with a tooth from Wolf Ranch in Arizona (Harrison, 1978; Findsay et al., 1990), be among the old- est of the erethizontids in North America at 2.5 + Ma (Albright, 1999; Bell et ah, 2004).

Our measurements indicate that the lengths of four cheek-tooth rows (two from Idaho and two from California) equal or are slightly larger than the lengths of modern Erethizon. Only two p4/ml ratios could be determined, both on the fossils from Idaho. One (USNM

Contributions in Science, Number 524

Sussman et al.: North American Erethizontidae ■ 7

13684) is quite large but not beyond the range of modern Coendow, the other falls into the range of overlap between the two genera.

Three incisors from Idaho have cross-sectional ratios of 1.0 and 1.1; one of these is an adult (USNM 13684), one a juvenile (UO F-16272), and the third (UO F-16271) is a young adult. Three fossils from El Casco and Vallecito Creek have ratios of 1.0, a fourth is 1.1, but the ontogenetic ages of the animals cannot be determined. Although none of these incisors have a ratio in the range of Coendon, the inability to determine the ontogenetic age in the majority of them prevents a definitive assignment to Erethizon.

Three cheek-tooth axes are all directed laterally, which is uninformative.

UO F-16271 from Jackass Butte, Idaho, has a lingual angle of incisor procumbency of 132 degrees, a trait of Erethizon