THE ANATOMICAL RECORD 298:354–375 (2015)

Postsacral Vertebral Morphology in Relation to Tail Length Among Primates and Other Mammals GABRIELLE A. RUSSO* Department of Anthropology, Stony Brook University, Stony Brook, New York, 11794-8081, USA

ABSTRACT Tail reduction/loss independently evolved in a number of mammalian lineages, including hominoid primates. One prerequisite to appropriately contextualizing its occurrence and understanding its significance is the ability to track evolutionary changes in tail length throughout the fossil record. However, to date, the bony correlates of tail length variation among living taxa have not been comprehensively examined. This study quantifies postsacral vertebral morphology among living primates and other mammals known to differ in relative tail length (RTL). Linear and angular measurements with known biomechanical significance were collected on the first, mid-, and transition proximal postsacral vertebrae, and their relationship with RTL was assessed using phylogenetic generalized leastsquares regression methods. Compared to shorter-tailed primates, longertailed primates possess a greater number of postsacral vertebral features associated with increased proximal tail flexibility (e.g., craniocaudally longer vertebral bodies), increased intervertebral body joint range of motion (e.g., more circularly shaped cranial articular surfaces), and increased leverage of tail musculature (e.g., longer spinous processes). These observations are corroborated by the comparative mammalian sample, which shows that distantly related short-tailed (e.g., Phascolarctos, Lynx) and long-tailed (e.g., Dendrolagus, Acinonyx) nonprimate mammals morphologically converge with short-tailed (e.g., Macaca tonkeana) and long-tailed (e.g., Macaca fascicularis) primates, respectively. Multivariate models demonstrate that the variables examined account for 70% (all mammals) to 94% (only primates) of the variance in RTL. Results of this study may be used to infer the tail lengths of extinct primates and other mammals, thereby improving our understanding about the evolution of tail reduction/ C 2014 Wiley Periodicals, Inc. loss. Anat Rec, 298:354–375, 2015. V

Key words: tail length; vertebrae; convergent evolution; sacrum; functional morphology

INTRODUCTION The bony mammalian tail can be subdivided into three regions based on external vertebral anatomy: proximal, transitional, and distal (Fig. 1; Flower, 1876; Schmidt, 1886; Ankel, 1965, 1972; German, 1982; Lemelin, 1995; Argot, 2003; Schmitt et al., 2005; Organ, 2007, 2010; Organ et al., 2009; Russo and Young, 2011). Vertebrae in the proximal tail region typically possess a single C 2014 WILEY PERIODICALS, INC. V

Grant sponsor(s): National Science Foundation DDIG BCS1156016, Leakey Foundation General Research Grant. *Correspondence to: G. A. Russo, Department of Anthropology, Stony Brook University, Stony Brook, NY, 11794-8081. E-mail: [email protected] Revised 29 May 2014; Accepted 8 July 2014. DOI 10.1002/ar.23004 Published online 11 August 2014 in Wiley Online Library (wileyonlinelibrary.com).

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

Fig. 1. Illustration showing regional distinctions along the mammalian tail. See text for detail about labeled vertebrae. Figure modified from Russo and Young (2011), originally adapted from Kimura et al. (1986).

355

pair of transverse processes and bear neural arches, articulating with one another by way of zygapophyseal and body joint surfaces. Proceeding toward the tail’s tip, vertebral neural arches, spinous processes, and transverse processes diminish or become reduced (transverse processes may also bifurcate) until reaching the last proximal vertebra, also known as the transition vertebra (TV) (Fig. 1). The morphology of the TV is distinctive because its cranial end possesses zygapophyseal joints while its caudal end lacks them, thereby demarcating the proximal and transitional vertebral sequences. Vertebrae distal to the TV may possess one to two pairs of further reduced transverse processes, but completely lack neural arches, such that articulating vertebrae are joined only by intervertebral body joint surfaces. These vertebrae also increase in craniocaudal body length until reaching a craniocaudally longest vertebra (Fig. 1), marking the end of the transitional region. The distal tail region is comprised of all vertebrae following the longest vertebra, and they sequentially decrease in craniocaudal body length until the tail’s tip (Fig. 1; Schmidt, 1886; Ankel, 1962, 1965, 1972). Functional morphologists interested in mammalian tails have primarily focused on prehensile tails, which evolved twice in New World monkeys (once in Atelidae and once in Cebinae) (Dor, 1937; German, 1982; Emmons and Gentry, 1983; Rosenberger, 1983; Bergeson, 1992, 1995, 1996; Gebo, 1992; Lemelin, 1995; Meldrum, 1998; Garber and Rehg, 1999; Groves, 2001, 2005; Lawler and Stamps, 2002; Youlatos, 1999, 2003; Bezanson, 2005, 2006, 2009; Schmitt et al., 2005; Organ, 2006, 2007, 2010; Rylands and Mittermeier, 2009; Organ et al., 2009, 2011; Russo and Young, 2011). Functionally, prehensile tails can solely support an animal’s body weight in suspension, and assist during travel and foraging activities by allowing animals to distribute their body weight over more branch supports and access food resources located in terminal branch settings (Emmons and Gentry, 1983; Gebo, 1992; Garber and Rehg, 1999; Bezanson, 2005, 2006, 2009). Structurally, prehensile tails are distinguished from nonprehensile tails by several bony anatomical features, including a greater number of craniocaudally shorter proximal tail vertebrae having relatively higher neural arches (i.e., greater dorsoventral breadth of the neural canal) (Ankel, 1965, 1972), distal tail vertebrae with more laterally projecting transverse processes and more ventrally projecting hemal processes (Organ 2010), and, in all three tail regions, greater average bending strength (Organ, 2010; see also German, 1982). These features permit greater proximal tail flexibility (Lemelin, 1995; Schmitt et al., 2005), accommodate increased innervation to welldeveloped tail musculature (Ankel, 1965, 1972; see also Grand, 1977, Organ et al., 2011), afford increased muscular mechanical advantage of the tail abductor and ventrodorsal flexor musculature (Organ 2010), and improve structural resistance to the high magnitudes of loading likely generated during tail-suspension positional behaviors (Organ, 2010; see also German, 1982), respectively, compared to nonprehensile tails. In contrast to the extensive research devoted to prehensile tail function and structure, far fewer studies have explored other aspects of tail morphological variation. Yet, one of the most apparent sources of variation is the external length of the tail. Tail length reduction

356

RUSSO

TABLE 1. Mammal sample and relative tail lengths (RTL; see text for calculation)a

Order Carnivora

Family Felidae Hyaenidae Viverridae

Mustelidae Diprotodontia

Macropodidae

Phascolarctidae Phalangeridae Potoroidae Vombatiidae Pilosa

Primates

Atelidae Callitrichidae Cebidae Cercopithecidae

Daubentoniidae Galagidae Hominidaed

Indriidae

Species

b

Acinonyx jubatus Lynx rufus Crocuta crocuta Arctictis binturong Genetta servalina Genetta victoriae Nandinia binotata Arctonyx collaris Eira barbara Gulo luscus Dendrolagus goodfellowi Dendrolagus lumholtzi Dendrolagus matschiei Macropus giganteus Macropus robustus Phascolarctos cinereus Caluromys lanatus Spilocuscus maculatus Trichosurus vulpecula Potorous tridactylus Vombatus ursinus Bradypus tridactylus Choloepus didactylus Choloepus hoffmanni Myrmecophaga tridactyla Tamandua mexicana Ateles fusciceps Ateles geoffroyi Saguinus fuscicollis Saguinus geoffroyi Saguinus oedipus Cebus apella Saimiri boliviensis Cercopithecus mitis Erythrocebus patas Macaca arctoides Macaca assamensis Macaca cyclopis Macaca fascicularis Macaca fuscata Macaca mulatta Macaca nemestrina Macaca nigra Macaca tonkeana Macaca sinica Macaca sylvanus Mandrillus sphinx Nasalis larvatus Papio cynocephalus Presbytis rubicunda Pygathrix roxellana Trachypithecus obscurus Daubentonia madagascariensis Galago senegalensis Otolemur crassicaudatus Gorilla gorilla Homo sapiens Hylobates concolor Hylobates lar Hylobates muelleri Hylobates syndactylus Pan troglodytes Pongo pygmaeus Indri indri Propithecus diadema

N

Range (Mean) for number of proximal postsacral vertebrae

RTLc

5 5 1 3 2 1 2 1 2 1 1 4 4 3 2 10 2 3 4 1 5 5 3 2 4 7 6 1 2 3 1 8 7 6 4 7 1 1 11 2 7 8 8 8 3 1 6 6 6 2 2 2 1 7 6 7 6 2 2 2 5 8 6 2 1

5–6 (6) 3–4 (3) 3 9–11 (10) 7 6 8 2 5 3 5 5–6 (6) 4–5 (5) 5–6 (5) 5 2–3 (2) 5 8 6–7 (7) 4 4–6 (5) 1–4 (2) 0–2 (1) 1–2 (2) 15–18 (16) 16–19 (18) 8–9 (8) 8 5 5 5 5–6 (6) 6 5 4–5 (4) 3–4 (3) 4 4 4–5 (5) 4 4–5 (4) 4–5 (5) 0–2 (1) 3–4 (3) 4–5 (5) 1 2–4 (4) 4 6–7 (6) 4–5 (5) 5 4–5 (5) 5 3–4 (3) 4–5 (4) 2–5 2–4 3 2–3 2–3 1 2–5 2–3 3 5

55 20 27 92& 92 92 106 23 70 26 101 140 101 80 80 0 162& 93 66& 79 5 10 0 0 75 101& 158& 158& 142 142 142 107& 116 139 90 6 36 83 107 15 47 37 9 8 122 0 10 102 84 131 96 144 138 130 122 0 0 0 0 0 0 0 0 8 93

357

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

TABLE 1. (continued).

Order

Family Lemuridae Lorisidae Pitheciidae

Rodentia

Castoridae Caviidae Erethizontidae Heteromyidae

Scandentia

Speciesb Propithecus verreauxi Varecia variegata Nycticebus coucangd Perodicticus potto Chiropotes satanas Pithecia monachus Pithecia pithecia Castor canadensis Aplodontia rufa Hydrochaeris hydrochaeris Coendou prehensilis Erethizon dorsatum Dipodomys merriami Tupaia glis Tupaia gracilis Tupaia tana

N

Range (Mean) for number of proximal postsacral vertebrae

RTLc

6 2 10 7 5 3 4 7 4 6 6 3 6 1 1 5

5–6 (5) 5–6 (6) 0–2 2–5 (5) 5–6 (6) 4 5 4–5 (4) 4–5 (4) 2–4 (3) 12–15 (13) 6–10 (9) 4–5 (4) 5 6 5

32 120 6 20 95 106 106 42 7 0 91& 23 144 157 107 78

a

Alphabetized by Order, Family, then Species. Boldface indicates that species is part of data subset. Data compiled from: Miller (1900), Martin (1968), Fooden (1969, 2006, 2007), Wilson (1972), Napier (1981), Fa (1985, 1989), Procter-Gray and Ganslosser (1986), Fooden and Albrecht (1999), Jenkins (1990), Parker (1990), Nowak (1991), Rowe (1996), Francis (2008). A “&” following the calculated RTL indicates that this taxon’s tail is prehensile. d No mean is presented for these taxa because complete postsacral sequences were unavailable. These numbers indicate the number of postsacral vertebrae present in the museum collection, but is not necessarily representative of this taxon’s actual total number. b c

has evolved in many mammalian lineages (Table 1), including a number of primates such as lorises, Indri, and some Old World monkeys. Moreover, the complete absence of an external tail is a shared, derived trait characterizing all members (living and extinct) of the Hominoidea. As such, the ability to determine taillessness from fossil material has an important role in the attribution of hominoid status to extinct noncercopithecoid catarrhine taxa (Ward et al., 1991; Harrison, 1998; Ward et al., 1999), attesting to the phylogenetic significance of this trait for some mammalian lineages. Experimental and observational studies have demonstrated that arboreal mammals with (artificially) shortened tails cannot attain the same level of locomotor competence as their conspecifics with longer length tails (rodents, Buck et al., 1925; Horner, 1954; Siegel, 1970; primates, Igarashi and Levy, 1981), suggesting that variation in tail length is also functionally significant. For example, Igarashi and Levy (1981) demonstrated that squirrel monkeys with shortened tails performed worse in rail tests (rotating rails that require agility and high acceleration to traverse) than squirrel monkeys with longer length (i.e., not shortened) tails, and failed to achieve a stable level of performance. Additionally, mice with surgically removed tails that were trained to walk on small poles exhibited increased “hind limb prehension” (i.e., foot grasping) to maintain balance and reportedly fell significantly more often than control mice (Siegel, 1970:101). Indeed, previous work demonstrates that arboreal mammals employ their tails as a prop or counterbalance during certain postures, and as a stabilizer during quadrupedal locomotion (Horner, 1954; SprAnkel, 1965; Wilson, 1972; Rose, 1974; Grand, 1977; Rodman, 1979; Walker et al., 1998; Larson and Stern, 2006; Stevens et al., 2008). For example, arboreal quadrupeds may employ lateral tail movements (i.e., abduction) during locomotion to create oppositely directed

angular momentum that can reorient the body’s center of mass, particularly during moments of imbalance (e.g., Macaca, Larson and Stern, 2006; see also Wada et al., 1993 for tail movements in terrestrial quadrupeds). Additionally, dorsally extended tail movements during leaping or jumping may assist in reorienting the body to prepare for landing (Hatt, 1932; Emerson, 1985; Demes et al., 1996; Preuschoft et al., 1996; Hildebrand and Goslow, 2001; Essner, 2002). In accordance with these observations, anatomical data demonstrates that arboreal mammals tend to have longer tails with more well-developed musculature than their terrestrial, close-phyletic relatives that presumably do not require the tail for balance to the same extent (e.g., rodents; Horner, 1954; tree shrews, Martin, 1968; primates, Grand, 1977; see also Fleagle, 1999; Rodman, 1979). Grand (1977) found that the entire tail segment in Macaca fascicularis, which has a tail longer than the combined length of the head and body (Table 1), comprised 3.6% of total body weight, while the tails of Macaca nemestrina and Macaca mulatta, which both have tails shorter than half the length of the head and body (Table 1), comprised only 0.2% and 0.1% of total body weight, respectively. For comparison, the prehensile tail of Alouatta caraya accounted for 6.5% of its total body weight (Grand, 1977). Moreover, as a result of engaging the tail in order to stabilize the body, nonprehensile-tailed mice trained to climb exhibited more laterally expanded postsacral vertebral transverse processes compared to nonclimbing mice (Byron et al., 2011). Though the tail has a number of other functions (e.g., social and autocommunication, thermoregulation, defense, see Hickman, 1979 for a review), these aforementioned studies clearly indicate that tail length variation functionally relates to an animal’s ability to maintain balance and successfully navigate an arboreal environment, and likely has correlates in tail structure and the associated musculature.

358

RUSSO

Despite the clear functional, and in some cases phylogenetic (e.g., Hominoidea), significance of tail length variation among primates and other mammals, there is little research that examines how postsacral vertebral anatomy varies with respect to tail length. Yet, the ability to attribute tail lengths to extinct taxa from isolated postsacral vertebral elements would clearly enhance our ability to track its evolutionary changes in order to contextualize the significance of tail loss. Some previous work has focused on how to distinguish between types of postsacral vertebrae (coccygeal versus caudal) in order to determine tail absence versus tail presence. In addition to lacking a neural arch and associated prezygapophyses, Nakatsukasa et al. (2003:180) noted that coccygeal vertebrae are also characterized by an overall “T-shape”, reduction in the lateral expansion of the transverse processes, and increased dorsoventral compression of the vertebral body, compared to the first caudal vertebrae of some shorttailed mammals. Using these traits, they identified and attributed a coccygeal vertebra to the Middle Miocene (ca. 15 Ma) hominoid Nacholapithecus kerioi. The applicability of this analysis to other studies is limited, however, as these comparisons were qualitative and based on an unspecified number of individuals as representatives for the mammalian sample. More recently, Hamada et al. (2012) examined the postsacral vertebrae from a sample of hybridized long-tailed Macaca cyclopsis and relatively shorter-tailed M. fuscata. They demonstrated that craniocaudal vertebral body length increased as tail length increased (see also Fooden, 1975, 1988), though differences in the craniocaudal length of the first postsacral vertebral body was less pronounced than that of other vertebral levels (Hamada et al., 2012). While the study by Hamada et al. (2012) expands our knowledge of how postsacral vertebral structure differs in relation to tail length, it quantified only one of many biomechanically informative aspects of vertebral morphology in a taxonomically narrow primate sample.

Objectives and Predictions The objective of this study is to quantify how postsacral vertebral morphology varies in relation to tail length differences. This study focuses on postsacral vertebrae in the proximal tail region because the bulk of the musculature responsible for moving the tail is near the tail base in nonprehensile-tailed mammals (Grand, 1977; Lemelin, 1995). Though the primary interest of this study is variation in postsacral vertebrae as it relates to tail length among primates, that tail length reduction/ loss has evolved independently in nonprimate mammals offers the opportunity to validate form-function links by detecting instances of morphological convergence among distantly related taxa. Primate postsacral vertebrae were thus evaluated in the context of a broad, comparative mammalian sample (Table 1). Table 2 describes the study variables, their functional relevance, and predictions. Generally, it is expected that longer-tailed primates and other mammals will possess a greater number of features associated with increased proximal tail flexibility, increased intervertebral joint range of motion, and increased muscular mechanical advantage, compared to relatively shorter-tailed primates and other mammals, as these features would enhance the ability to actively dorsally extend, ventrally

flex, and laterally abduct the tail during locomotion (Table 2). Moreover, if longer tails represent a greater percentage of total body weight than shorter tails (Grand, 1977), then it is reasonable to assume that the proximal postsacral vertebrae in longer-tailed taxa might be structured to withstand greater loads and accommodate increased innervation to more developed tail musculature, compared to shorter-tailed taxa (Table 2). Because no study to date has comprehensively examined how tail vertebral anatomy varies in relation to tail length, predictions were delineated based on functional morphology studies of other regions of the spine (Table 2; Ankel, 1972; Shapiro, 1993, 2007; Sanders and Bodenbender, 1994; Ward, 1993; Shapiro and Simons, 2002; Argot, 2003; Shapiro et al., 2005; Organ, 2007, 2010; Russo, 2010; Russo and Shapiro, 2011).

MATERIALS AND METHODS Measurements (Fig. 2) on postsacral vertebrae were collected for 81 species from six mammalian Orders: Primates, Carnivora, Diprotodonia, Pilosa, Rodentia, and Scandentia (Table 1). The sample derives from osteological collections at the American Museum of Natural History (New York, NY, USA), the National Museum of Natural History (Washington D.C., USA), and the Field Museum of Natural History (Chicago, IL, USA). The comparative sample includes 211 primates (46 species), 23 carnivores (10 species), 39 diprotodonts (11 species), 21 pilosans (five species), 32 rodents (six species), and seven tree shrews (three species) (Total N 5 333 individuals) (Table 1). Taxonomic preference was given to closelyrelated taxa that differed considerably in tail lengths. Among the primates sampled, Macaca is a particularly useful taxonomic group for comparative purposes as tail length reduction occurs within several species groups (e.g., sinica species group, Fooden, 1988; Deinard and Smith, 2001). Carnivores, diprotodonts, rodents, pilosans, and tree shrews were chosen for comparison because these clades that contain species characterized by reduced tail lengths (Table 1). Carnivores include short-tailed felids (e.g., Lynx) and mustelids (e.g., Gulo), and long nonprehensile-tailed (e.g., Genetta) or prehensile-tailed (e.g., Arctictis) viverrids. Diprotodontia includes shorttailed vombatiformes (Phascolarctos and Vombatus), and long-tailed macropodiformes (e.g., Dendrolagus) and phalangeriformes (e.g., Trichosurus). Within Pilosa, modern two-toed (Choloepus) and three-toed (Bradypus) sloths have likely independently evolved tail loss as these living taxa are distantly related and all known fossil genera have long tails (Tito, 2008). Other modern representatives of Pilosa include anteaters, some of which have long nonprehensile (e.g., Myrmecophaga) or long prehensile (e.g., Tamandua) tails. Rodentia includes tailless caviids (e.g., Hydrochaeris), short-tailed common porcupines (Erethizon), and long-tailed heteromyids (e.g., Dipodomys). For each extant taxon, a relative tail length index (RTL 5 tail length/[head 1 trunk length] 3 100; Fooden, 1997) was calculated using morphometric data collected from the literature (Table 1). Previous studies have calculated RTL using the literature cited here and employed this index for quantifying tail length (Fooden, 1997; Youlatos, 2003; Russo and Shapiro, 2011; Hamada et al., 2012). All hominoid primates were assigned a RTL of 0 because they lack an external tail. Nonprimate

Influences bone’s ability to distribute load to minimize concentrated forces and potential for damage to cartilaginous/bony structures Reflects accommodation of innervation to tail region Influences leverage and surface area of attachment for abductors (first and midproximal) and ventral flexors (TV) Leverage for abductors Influences compartment size for tail extensor musculature Leverage for extensors Influences leverage and surface area of attachment for basal tail extensor musculature; compartment size for extensor musculature. Influences leverage for basal tail extensor musculature

Governs intervertebral movements; permits and restricts ROM to specific planes

First, midproximal, TV

First and midproximal

First, midproximal, TV

First and midproximal

First and midproximal

First and midproximal

First and midproximal

First and midproximal

First and midproximal

Cranial articular surface areab (CASA)

Area of cranial neural apertureb (ANA)

Transverse process breadthb (TPB)

TP craniocaudal orientation (TPCC) TP dorsoventral orientation (TPDV) TP positionb (TPP)

Spinous process lengthb (SPL)

SP craniocaudal orientation (SPCC)

Prezygapophyseal orientation (PZO)

Msmt 12

Msmt 11

Msmt 10; divided by BM1/3

Msmt 9; divided by BM1/3

Msmt 8

Msmt 7

(p * [0.5*Msmt 4] * [0.5*Msmt 5]), divided by BM2/3 Msmt 6; divided by BM1/3

(p * [0.5*Msmt 2] * [0.5*Msmt 3]), divided by BM2/3

Msmt 2/Msmt 3

Msmt 1; divided by BM1/3

Calculationa

Relatively more dorsal 5 smaller compartment for tail extensor musculature Relatively more ventral 5 decreased leverage for extensors Relatively less expanded 5 reduced leverage for tail extension; smaller compartment for tail extensor musculature Relatively more acute angles 5 SP more perpendicular to body; decreased leverage for tail extension in sagittal plane Relatively more acute angles/ sagittal orientation 5 decreased ROM for lateral tail bending and rotation

Relatively less expanded 5 decreased leverage for basal tail abduction and distal tail flexion Relatively more perpendicular to body 5 lateral rigidity

Relatively lower values 5 reduced innervation to a nearly diminished tail region

Relatively longer 5 Assuming equal vertebral numbers: increased flexibility and decreased resistance to bending Values closer to 1 5 more circular (more equal dimensions); ROM in more directions

Relatively shorter 5 Assuming equal vertebral numbers: reduced flexibility and increased resistance to bending Values > 1 5 more elliptical (mediolaterally wide, dorsoventrally narrow); ROM in fewer directions Relatively lower values 5 decreased surface area; decreased resistance to loads

Relatively more obtuse angles 5 SP more caudally oriented; increased leverage for tail extension in sagittal plane Relatively more obtuse angles/ oblique orientation 5 increased ROM for lateral tail bending and rotation

Relatively more ventral 5 larger compartment for tail extensor musculature Relatively more dorsal 5increased leverage for extensors Relatively longer 5 increased leverage for tail extension; larger compartment for tail extensor musculature

Relatively more expanded 5increased leverage for basal tail abduction and distal tail flexion Relatively more caudal 5 lateral flexibility

Relatively higher values 5 increased innervation to a robust tail region

Relatively higher values 5 increased surface area; increased resistance to loads

Longer tail

Shorter tail

b

Measurements obtained for variable calculations are illustrated in Figure 2. Size-corrected variable, see Calculation for size adjustment. Abbreviations: ROM 5 range of motion, Msmt 5 measurement References for functional relevance and predictions: Hartmann and Strauss (1933), Ankel (1965, 1972), Currey (1984), Jungers (1988), Saluja (1988), Ward (1993), Shapiro (1993, 2007), Lemelin (1995), Shapiro and Simons (2002), Bogduk and Twomey (2005), Kapandji (2008), Russo (2010), Russo and Shapiro (2011), Woon and Stringer (2012), Hamada et al. (2012).

a

Influences intervertebral joint potential ROM

First, midproximal, TV

Cranial articular surface shape (CASS)

Influences potential flexibility (considered with vertebral numbers and total region length)

Functional relevance

First, midproximal, TV

Vertebral levels

Vertebral body craniocaudal lengthb (VCCL)

Variables

Predictions

TABLE 2. Vertebral variables, and their functional relevance, calculations, and predictions

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

359

360

RUSSO

Fig. 2. Measurements collected on proximal postsacral vertebrae. Views are dorsal view (A), cranial (B), and lateral (C). (1) Maximum craniocaudal vertebral body length (mm) in the midsagittal plane; (2) Maximum mediolateral diameter (mm) of the cranial articular surface; (3) Maximum dorsoventral diameter (mm) of the cranial articular surface; (4) Maximum breadth (mm) of the neural aperture in the transverse plane; (5) Maximum height (mm) of the neural aperture from the dorsal edge of the vertebral body to the ventral edge of the neural canal in the midsagittal plane; (6) Maximum lateral expansion (mm) of transverse processes measured from the apex of the left transverse process to the apex of the right transverse process; (7) Transverse process craniocaudal orientation ( ), measured in the coronal plane as the angle between the caudal edge of the transverse process (defined as a line connecting the caudalmost point of the transverse process to the point of attachment on the vertebral body) and the midsagittal plane of the vertebra; (8) Transverse process dorsoventral orientation ( ), measured in the transverse plane as the angle between the ventral edge of the transverse process (defined as a line connecting the ven-

tralmost point on the tip of the transverse process and the point of attachment on the vertebral body) and the midsagittal plane of the vertebra; (9) Transverse process dorsoventral position (mm), measured as distance between a line connecting the ventralmost attachment points of the transverse processes to the vertebral body and a line drawn tangent to the ventral surface of the vertebral body; (10) Spinous process length (mm) measured as maximum distance from the dorsal edge of the vertebral canal to the apex of the spinous process; (11) Spinous process craniocaudal orientation ( ), measured as the angle between the cranial edge of the spinous process (defined as a line connecting the cranialmost point on the spinous process and the point of attachment on the dorsal edge of the neural canal) and a line connecting the dorsocranial and dorsocaudal most points on the vertebral body; (12) Prezygapophyseal orientation ( ), measured as the angle formed between the chord of the arc formed by the prezygapophyseal articular surface and a sagittal plane through the vertebra. See Table 2 for calculation of associated variables. Figure adapted from Shapiro (2007).

mammalian taxa described in the literature as vestigialtailed or as possessing diminutive tails, and thus without associated tail length measurements, were also assigned a RTL of 0 (Table 1). Although this study focused on nonprehensile-tailed taxa, prehensile-tailed taxa were included for comparative purposes because the functional morphology of their vertebrae is well understood and, additionally, prehensile-tailed taxa possess the longest tails in some clades (e.g., Primates: Ateles, RTL 5 158). This study examined six to 10 specimens per species where sufficient material was available. Small sample numbers per taxon reflect the unfavorable state of preservation of tail vertebrae, which are often missing, damaged, or fully articulated, in museum collections. Nonetheless, the sample sizes in this study are within the range of those examined in other tail studies (e.g., German, 1982; Organ, 2010). Due to small sample sizes and the large number of unknown sex cases, it was not possible to separate data collection equally across the sexes in order to statistically test for sex differences. Accordingly, pooled-sex species averages were used for

analyses. Though tail and body length measurements may vary between sexes of a single species, Fooden (1997; see also Fooden and Albrecht, 1999) has demonstrated that among macaques, for example, corresponding tail length indices do not. All examined individuals were wild-captured and classified as adult based on long bone epiphyseal fusion.

Vertebrae and Measurements This study examined the first, mid-, and transition proximal postsacral vertebrae (Fig. 1) for each individual in the mammal sample. The measurements obtained, and the variables calculated, for each vertebra are illustrated and described in Figure 2 and Table 2. The first postsacral vertebra is the vertebra immediately distal to the sacrum, regardless of form (i.e., coccygeal or caudal) (Fig. 1). The midproximal vertebra is defined numerically as halfway through the proximal tail sequence (Fig. 1), following Organ’s (2010) definition of the middistal vertebra. The TV (defined above) is the last vertebra in the proximal tail

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

sequence (Fig. 1). Measures of cranial articular surface dimensions, vertebral body craniocaudal length (VCCL), and transverse process breadth (TPB) were taken using digital calipers linked to a laptop and recorded to the nearest 0.01 mm. In some cases, measurements had to be modified: where one transverse process was broken, the lateral projection of the complete transverse process from the midline was measured and multiplied by two to obtain a reasonable estimate of total TPB; where the transverse processes had bifurcated at the level of the TV, maximum TPB was obtained (typically, the distalmost pair of transverse processes was most laterally expanded); and where some vertebrae were articulated, VCCL was measured at the body’s ventral surface. A Panasonic Lumix DMC-G2 12.1 MP (SLR) digital camera was used to photograph vertebral specimens in dorsal, cranial, and lateral views to obtain additional linear and angular measurements (Fig. 2) and for general documentation purposes. To ensure consistent orientation, vertebrae were oriented using the following guidelines: (1) In dorsal and lateral views, the cranial articular surface of each specimen was placed flush against a flat rod attached perpendicular to the end of the board such that the cranial articular surface is also perpendicular to the camera lens; (2) In cranial views, the cranial articular surface of each specimen was positioned upright and thus parallel with the camera lens. All measurements from photographs were obtained using the “line tool” in ImageJ v1.42j software (http:// rsbweb.nih.gov/ij/).

Size Adjustments Body size varies greatly across the taxa in this study. Therefore, variables need to be size-adjusted prior to comparative analyses. Previous studies of vertebral morphology have standardized measurements by the geometric mean of (nonangular) variables of the vertebrae (e.g., Shapiro, 2007). However, because tail vertebral sequences in museum collections are often partially/fully articulated, or vertebrae are missing or damaged, the ability to obtain all measurements on all vertebrae for each individual was not always possible. Excluding such cases from the analysis would have drastically reduced the study’s sample size. Body mass, an appropriate proxy of body size (Fleagle, 1984; Schmidt-Nielsen, 1984), was used instead to make size-adjustments so that partial data sets could be retained for analyses. Body mass data were available for some specimens from museum documents and applied where possible. However, this information was absent for the majority of specimens, and so body masses were obtained from the literature (primates: Smith and Jungers, 1997; nonprimate mammals: Silva and Downing, 1995; Nowak, 1991) and applied on a by-species (or bysubspecies when possible) and by-sex basis. Where sex was unknown, the average of male and female body mass values was applied. Linear measurements (mm) were standardized (i.e., divided by) by body mass1/3; area measurements (mm2) were standardized by body mass 2/3; and angular measurements ( ) were not subject to standardization (following Shapiro, 2007).

Data Sets and Analysis The sample was partitioned into two data sets. Data set #1 is comprised of all mammals in the sample (Table

361

1). For these taxa, measurements were obtained to calculate VCCL, cranial articular surface area (CASA), cranial articular surface shape (CASS), and TPB (Table 2). These variables derive from measurements taken from the vertebral body, which is typically better preserved in the fossil record than bony projections (e.g., transverse and spinous processes). Additionally, these measurements can be obtained using digital calipers and do not require a standardized orientation. In these ways, variables in Data set #1 may be more readily applicable to studies of fossil postsacral vertebrae. Data set #2 contains a subset of the total mammal sample (boldfaced taxa in Table 1). For mammals in Data set #2, measurements were obtained to calculate an additional seven variables: transverse process craniocaudal (TPCC) and dorsoventral (TPDV) orientations, transverse process position (TPP), area of the cranial neural aperture (ANA), spinous process length (SPL), and craniocaudal orientation (SPCC), and prezygapophyseal orientation (PZO) (Table 2). These variables derive from measurements obtained using calipers as well as digital photographs of specimens in multiple standardized views (Fig. 2). Data were analyzed using species means for each variable at each vertebral level. The inclusion of hominoid taxa and other mammals possessing vestigial tails or having no tail were restricted to analyses at the level of the first postsacral vertebra (and Data set #1) since these taxa do not possess tail sequences long enough to include midproximal or transition postsacral vertebrae. Each variable was examined among all mammals and among only primates. The relationship between each trait (predictor variable) and RTL (criterion variable) was evaluated using a phylogenetic generalized leastsquares (PGLS) regression technique. In this study, PGLS was used to account for the fact that closely related taxa likely have more similar vertebral morphologies than distantly-related taxa. PGLS models incorporate an error term into the standard ordinary leastsquares regression equation that considers covariation among taxa on the basis of phylogenetic similarity (i.e., phylogenetic nonindependence) (Martins and Hansen, 1997). The error term comes from a variance-covariance matrix that contains information about the phylogenetic relationships (given the inferred phylogeny) of the sampled taxa. Using Pagel’s k (Pagel, 1999; see also Freckleton et al., 2002), the covariance matrix estimates the degree to which trait evolution deviates from the Brownian motion model of evolution. The Brownian motion model of evolution predicts that k 5 1, assuming closely related species are anatomically similar and thus traits are proportional to (i.e., a direct function of) phylogenetic distance. Models that assume phylogenetic independence predict that k 5 0, indicating that there is no relationship between the trait and phylogenetic relatedness (i.e., variation in a trait among taxa is independent of evolutionary relatedness). For this study, mammalian phylogenetic information and branch lengths were taken from Bininda-Emonds et al. (2007). Branch length data were not available for Papio cynocephalus (n 5 6) or Gulo luscus (n 5 1). Therefore, for analysis, Papio cynocephalus and Gulo luscus were reclassified as Papio hamadryas and Gulo gulo, taxa for which branch length data were available. The impact of this adjustment is likely negligible as the taxa in question are more closely related to their alternative taxonomic classifications

362

RUSSO

than to any other taxon in the sample. PGLS analyses were conducted in R (R Development Core Team, 2011; Ihaka and Gentleman, 1996) using the ape and nlme packages (Paradis et al., 2004; Pinheiro et al., 2014), with code written by W. Andrew Barr. In addition to the regression models that examined the bivariate linear relationship between a given vertebral trait and RTL, the relationship between sets of vertebral variables and RTL were also considered using multivariate principal components analysis (PCA). The PCA contains only variables that were significant predictors of RTL (Table 3). PCAs were conducted for each vertebral level using both Data sets #1 and #2. PC1 scores were extracted from the PCA models and used to create predictive models using the regression equation model:

Y5mX1b where Y represents the predicted RTL, m is the slope, X is the extracted PC1 score, and b is the constant or Y-intercept. The standard error of the estimate (SEE), calculated as:

SEE5sdY冑1-r2 where sdY is the standard deviation of the Y-variable and r is the correlation between the X and Y variables, can be used to provide 95% confidence limits (i.e., 62 SEE) for tail length predictions. All PCAs were conducted in SPSS (IBM, Chicago, IL). For PGLS and PCA analyses, statistical significance was recognized at P < 0.05.

RESULTS Phylogenetic Generalized Least-Squares Vertebral body craniocaudal length (VCCL). Consistent with predictions, RTL increased as sizecorrected VCCL increased at all three vertebral levels (Tables 2 and 3; Figs. 3, 4, and 5). Vertebral body craniocaudal length explained a significant (i.e., R2 value was significantly different than 0) proportion of the variance in RTL at all three vertebral levels among all mammal and only primates, accounting for up to 63% (TV) and 74% (TV) of the variance in RTL, respectively (Table 3). Among all mammals, likelihood ratio tests show that Pagel’s k was significantly different from both 0 and 1 at the levels of the first postsacral and midproximal vertebrae, and significantly different from 1 at the level of the TV (Table 3). Among only primates, Pagel’s k was significantly different from 1 at all three vertebral levels (Table 3), indicating that the relationship between VCCL and RTL cannot be explained entirely by phylogeny.

Cranial articular surface shape (CASS). Consistent with predictions, RTL increased as ratio values of mediolateral to dorsoventral breadth decreased (i.e., CASS became more circular) (Tables 2 and 3; Figs. 3, 4, and 5). While CASS explained a significant proportion of the variance in RTL at all three vertebral levels, among all mammals and among only primates (Table 3), it accounted for only 19% and 16% of the variance in RTL at the level of the first postsacral vertebra. It is clear from the bivariate graphs of this variable that two clusters of data primarily drive the observed relationship. Box plots of these data confirm that variation in CASS may be chiefly explained by differences between tailbearing and tailless mammals and primates (Fig. 6). At

the level of the midproximal vertebra and TV, CASS explained a greater amount of the variance in RTL in all mammals and only primates than at the level of the first postsacral (Table 3). Among all mammals, likelihood ratio tests show that Pagel’s k was significantly different from 1 at the levels of the first postsacral and midproximal vertebrae, but significantly different from 0 at the level of the TV (Table 3). Among only primates, Pagel’s k was significantly different from 0 at all three vertebral levels (Table 3). That the relationship between CASS and RTL can be partly explained by phylogeny among primates makes sense in light of the primary distinction between tail-bearing and tailless taxa (Fig. 6).

Cranial articular surface area (CASA). As predicted, RTL increased with increasing size-corrected CASA at all three vertebral levels (Tables 2 and 3; Figs. 3, 4, and 5). Cranial articular surface area explained a significant proportion of the variance in RTL at all three vertebral levels among all mammals and among only primates, accounting for up to 53% (first postsacral) and 84% (TV) of the variance in RTL, respectively (Table 3). Among all mammals, likelihood ratio tests show that Pagel’s k was significantly different from 1 and 0 at all three vertebral levels (Table 3). Among only primates, Pagel’s k was significantly different from 0 and 1 at the levels of the first postsacral and transition vertebrae, and different from 1 at the level of the midproximal vertebra (Table 3). Transverse process breadth (TPB). As predicted, RTL increased with increasing measures of sizecorrected TPB at all three vertebral levels (Tables 2 and 3; Figs. 3, 4, and 5). Transverse process breadth explained a significant proportion of the variance in RTL, accounting for up to 35% (first postsacral) and 74% (TV) of the variance in RTL among all mammals and primates only, respectively (Table 3). Among all mammals, likelihood ratio tests show that Pagel’s k was significantly different from 1 and 0 at all three vertebral levels (Table 3). Among primates only, Pagel’s k was significantly different from 1 at all three vertebral levels (Table 3), indicating that the relationship between TPB and RTL cannot be explained entirely by phylogeny. Transverse process dorsoventral orientation (TPDV). As predicted, RTL increased as TPDV decreased (i.e., angle became more acute) at the level of the first postsacral and the midproximal vertebrae (the only vertebrae for which this measurement could be reliably obtained; Tables 2 and 3; Figs. 3 and 4), indicating that RTL increases with increasing ventral angulation of the transverse processes. Transverse process dorsoventral orientation explained a significant proportion of the variance in RTL at the levels of the first postsacral and the midproximal vertebrae among all mammals and among only primates, accounting for up to 56% (first postsacral) and 60% (midproximal) of the variance in RTL among all mammals and among only primates, respectively (Table 3). Among all mammals, likelihood ratio tests show that Pagel’s k was significantly different from 1 at the level of the first postsacral vertebra and statistically similar to both 0 and 1 at the level of the midproximal vertebra (Table 3). Among only primates,

Slope (SE)

First postsacral vertebra Vertebral body 26.564 (2.934) craniocaudal length Cranial articular 241.622 (9.731) surface shape Cranial articular 4.954 (0.529) surface area Transverse pro7.767 (0.874) cess breadth Transverse pro22.028 (0.349) cess craniocaudal orientation Transverse pro23.020 (0.494) cess dorsoventral orientation Transverse pro58.265 (10.386) cess position Area of cranial 21.009 (5.067) neural aperture Spinous process 26.143 (4.134) length Spinous process 20.518 (0.431) craniocaudal orientation Prezygapophyseal 21.683 (1.489) orientation Midproximal vertebra Vertebral body 18.6227 (2.3073) craniocaudal length Cranial articular 2117.747 (21.337) surface shape Cranial articular 4.685 (0.706) surface area Transverse pro6.337 (1.358) cess breadth Transverse pro22.096 (0.335) cess craniocaudal orientation Transverse pro22.929 (0.725) cess dorsoventral orientation Transverse pro11.854 (6.703) cess position Area of cranial 33.230 (8.857) neural aperture

Vertebral level variable 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.256 0.271 0.000 0.000 0.000 0.000 0.000 0.000 0.093 0.001

0.506 0.185 0.533 0.354 0.514 0.556 0.521 0.415 0.629 0.035 0.013 0.513 0.371 0.461 0.248 0.623 0.446 0.096 0.395

Adjusted R-squared R2 p value

All mammals

0.000 (1.000/0.000)

0.728 (0.177/0.001)

0.738 (0.115/0.098)

0.869 (0.024/0.157)

0.795 (0.003/0.000)

0.858 (0.002/0.003)

0.886 (0.000/0.022)

0.769 (0.000/0.000)

0.608 (0.333/0.043)

0.792 (0.332/0.134)

0.944 (0.012/0.317)

0.601 (0.572/0.021)

0.809 (0.121/0.006)

0.553 (0.052/0.001)

0.920 (0.002/0.209)

0.950 (0.000/0.042)

0.876 (0.000/0.000)

0.000 (1.000/0.000)

0.720 (0.000/0.000)

k (lower/upper bound p value)a

Y

N

Y

Y

35.530 (10.450)

21.359 (8.953)

23.565 (0.758)

21.964 (0.358)

12.092(1.906)

6.575 (0.744)

Y Y

2178.186 (27.220)

21.473 (2.702)

22.109 (1.701)

20.539 (0.487)

29.811 (3.197)

31.729 (4.306)

Y

Y

N

N

Y

Y

64.396 (11.169)

22.900 (0.606)

Y Y

22.160 (0.396)

11.043 (1.227)

6.296 (0.0641)

235.125 (11.212)

31.5861 (4.581)

Slope (SE)

Y

Y

Y

Y

Y

In PCA?

0.413

0.238

0.601

0.645

0.536

0.734

0.599

0.645

0.033

0.027

0.835

0.758

0.599

0.499

0.555

0.640

0.679

0.164

0.508

Adjusted R2

0.004

0.032

0.000

0.000

0.000

0.000

0.000

0.000

0.234

0.305

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.003

0.000

R-squared p value

Primates

TABLE 3. PGLS results for vertebral variables among all mammals and among only primates

0.000 (1.000/0.001)

0.419 (0.099/0.017)

1.000 (0.056/1.000)

0.815 (0.015/0.392)

0.000 (1.000/0.000)

0.813 (0.157/0.001)

0.946 (0.000/0.175)

0.796 (0.213/0.008)

0.589 (0.183/0.408)

0.815 (0.052/0.525)

0.000 (1.000/0.013)

0.000 (1.000/0.001)

0.519 (0.129/0.017)

0.590 (0.143/0.077)

0.937 (0.000/0.642)

0.750 (0.316/0.001)

0.761 (0.027/0.000)

0.825 (0.000/0.089)

0.739 (0.598/0.035)

k (lower/upper bound p value)

Y

Y

Y

Y

Y

Y

Y

Y

N

N

Y

Y

Y

Y

Y

Y

Y

Y

Y

In PCA?

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

363

Y

Y

Y

Y

0.427 (0.193/0.002)

1.000 (0.000/1.000)

0.733 (0.036/0.000)

0.000 (1.000/0.000)

Transverse process craniocaudal orientation (TPCC). As predicted, RTL increased as TPCC decreased (i.e., angle became more acute) at the levels of the first postsacral and midproximal vertebrae (the only vertebrae for which this measurement could be reliably obtained; Tables 2 and 3; Figs. 3 and 4), indicating that tail length increases with increasing caudal angulation of the transverse processes. Transverse process craniocaudal orientation explained a significant proportion of the variance in RTL at the levels of the first postsacral and the midproximal vertebrae among all mammals and among only primates, accounting for up to 62% (midproximal) and 65% (midproximal) of the variance in RTL, respectively (Table 3). As with CASS, it is clear from the bivariate graphs of this variable that two clusters of data primarily drive the observed relationship. Box plots of these data confirm that variation in TPCC may be chiefly explained by significant differences between tail-bearing and tailless mammals and primates (Fig. 6). Among all mammals and among only primates, likelihood ratio tests show that Pagel’s k was significantly different from 0 at all three vertebral levels (Table 3), indicating that phylogeny may partially account for the relationship between TPCC and RTL.

0.736 15.176 (1.598) Lower bound of lambda 5 0, upper bound 5 1. a

0.000 0.188

0.888 (0.002/0.006)

Y

0.837 6.677(0.554) 0.000 0.387

0.706 (0.029/0.000)

Y

0.389 2175.886 (40.475) 0.000 0.303

1.000 (0.001/1.000)

Y

0.736 9.041 (0.953) Y 0.000 0.633

0.352 (0.521/0.000)

Pagel’s k was statistically similar to both 0 and 1 at both vertebral levels (Table 3).

0.000

Y 0.662 (0.057/0.279) 0.025 0.260 0.032 0.179

0.845 (0.064/0.155)

Y

23.797(1.512)

0.000

N 0.784 (0.047/0.327) 0.217 0.115 0.222 0.059

0.850 (0.140/0.129)

N

1.050 (0.761)

0.000

Y 0.000 (1.000/0.020) 0.000 0.669 34.250 (6.115) Y 0.814 (0.156/0.115) 0.001 0.396

Spinous process 25.020 (6.66) length Spinous process 0.771 (0.592) craniocaudal orientation Prezygapophyseal 23.196 (1.381) orientation Transition vertebra Vertebral body 8.579 (0.853) craniocaudal length Cranial articular 2150.948 (32.278) surface shape Cranial articular 3.597 (0.636) surface area Transverse pro6.639 (1.734) cess breadth

k (lower/upper bound p value)a Adjusted R-squared p value R2 Slope (SE) Vertebral level variable

All mammals

0.000

In PCA? Adjusted R2 Slope (SE)

R-squared p value

k (lower/upper bound p value)

RUSSO

In PCA?

TABLE 3. (continued).

Primates

364

Transverse process position (TPP). Contrary to predictions, RTL increased with increasing distance of the position of the transverse processes relative to the ventral edge of the vertebral body at the levels of the first postsacral and the midproximal vertebra (the only vertebrae on which this measurement could be reliably obtained; Tables 2 and 3; Figs. 3 and 4), indicating that increasing RTL is associated with increasing dorsal position of transverse processes. Among all mammals, TPP explained a significant proportion of the variance in RTL at the level of the first postsacral vertebra (52%), but not at the level of the midproximal vertebra (10%) (Table 3). Among only primates, TPP explained a significant proportion of the variance in RTL at the level of the first postsacral (60%) and midproximal (24%) vertebrae (Table 3). Among all mammals and only primates, likelihood ratio tests show that Pagel’s k was significantly different from 1 at the level of the first postsacral and midproximal vertebrae (Table 3). Area of cranial neural aperture (ANA). Hominoid primates and other specimens possessing postsacral vertebrae that lack neural arches (and associated structures) are inherently excluded from this analysis and subsequent analyses (SPL, SPCC, and PZO). As predicted, RTL increased with increasing size-corrected ANA at the levels of the first postsacral and the midproximal vertebrae (Tables 2 and 3; Figs. 3 and 4). Area of the cranial neural aperture explained a significant proportion of the variance in RTL at the level of the first postsacral and midproximal vertebrae among all mammals and among only primates, accounting for up to 42% (first postsacral) and 76% (first postsacral) of the variance in RTL, respectively (Table 3). Among all mammals and only primates, likelihood ratio tests show that Pagel’s k was significantly different from 1 at the level

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

365

Fig. 3. RTL on vertebral variables at the level of the first postsacral vertebra. See bottom right for legend. For R2 values, M 5 all mammals, P 5 only primates. Larger figure can be viewed in the online issue available at wileyonlinelibrary.com.

of the first postsacral and midproximal vertebrae (Table 3), indicating that the relationship between ANA and RTL cannot be explained entirely by phylogeny.

Spinous process length (SPL). As predicted, RTL increased with increasing SPL (Tables 2 and 3; Figs. 3 and 4). Spinous process length explained a significant

366

RUSSO

Fig. 4. RTL on vertebral variables at the level of the midproximal vertebra. See bottom right for legend. For R2 values, M 5 all mammals, P 5 only primates. Larger figure can be viewed in the online issue available at wileyonlinelibrary.com.

proportion of the variance in RTL at the levels of the first postsacral and midproximal vertebrae among all mammals and among only primates, accounting for up to 63% (first postsacral) and 84% (first postsacral) of the variance in RTL, respectively (Table 3). Among all mammals, likeli-

hood ratio tests show that Pagel’s k was significantly different from 0 at the level of the first postsacral vertebra and significantly different from 1 at the level of the midproximal vertebra (Table 3). Among only primates, likelihood ratio tests show that Pagel’s k was significantly

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

367

Fig. 5. RTL on vertebral variables at the level of the transition vertebra. See bottom right for legend. For R2 values, M 5 all mammals, P 5 only primates. Larger figure can be viewed in the online issue available at wileyonlinelibrary.com.

different from 1 at the levels of the first postsacral and midproximal vertebrae (Table 3).

Spinous process craniocaudal orientation (SPCC). Results for SPCC were variable. Inconsistent with predictions, RTL increased with decreasing SPCC (i.e., angle became more acute) at the level of the first postsacral vertebra (Tables 2 and 3; Fig. 3), suggesting that RTL increased as spinous processes became more cranially oriented at this vertebral level. Consistent with predictions, RTL increased with increasing SPCC (i.e., became more obtuse) at the level of the midproximal caudal vertebra (Tables 2 and 3; Fig. 4), suggesting spinous processes become more caudally oriented as tail length increases. Nonetheless, among all mammals and among only primates, SPCC did not explain a significant proportion of the variance in RTL at either vertebral level.

Prezygapophyseal orientation (PZO). Results were inconsistent with predictions. At both vertebral levels, RTL increased with decreasing PZO (i.e., angle became more acute), suggesting that increasing RTL is associated with increasing sagittal orientation of the prezygapophyses. Prezygapophyseal orientation did not explain a significant proportion of the variance in RTL at the level of the first postsacral vertebra, but did explain a significant proportion of the variance in RTL at the level of the midproximal vertebra, accounting for

18% and 26% of the variance in RTL among all mammals and among only primates, respectively (Table 3). Among all mammals, likelihood ratio tests show that Pagel’s k was statistically similar to 1 and 0 at the level of the midproximal vertebra (Table 3). Among only primates, likelihood ratio tests show that Pagel’s k was significantly different from 0 at the level of the midproximal vertebra (Table 3).

Comparisons Among Mammalian Clades Figure 7 shows box-plot comparisons between closely related taxa that differ in RTL from three mammalian clades in the study: Primates, Diprotodontia, and Carnivora. The four chosen variables accounted for a significant proportion of the variance in RTL among all mammals and among only primates (Table 3). These variables exemplify changes in total vertebral morphology because they include a linear measurement of the vertebral body (VCCL), an area measurement of the vertebral body (CASA), a linear measurement from the neural arch (SPL), and an angular measurement of bony projections (TPDV). Results are similar among all three clade comparisons: the longer-tailed taxon (e.g., Macaca fascicularis, Dendrolagus sp., and Acinonyx jubatus) exhibits craniocaudally longer postsacral vertebral bodies having greater cranial articular surface area, possesses longer spinous processes, and has more ventrally angled transverse processes than their shorter-

368

RUSSO

Fig. 6. Box plots of (A) cranial articular surface shape (CASS) in all mammals, (B) CASS in only primates, (C) transverse process craniocaudal angle (TPCC) in all mammals, and (D) TPCC in only primates. Figure 4 shows that the relationship of these two variables with RTL appears to be driven primarily by the distinction between tail bearing and tailless taxa. See text for more detail.

tailed counterparts (e.g., M. tonkeana, Phascolarctos cinereus, and Lynx rufus; see also Fig. 8).

Principal Components Analyses (PCA) The variables included in the PCA are denoted in Table 3. The PCA results are presented in Table 4. The PCAs were conducted among all mammals and among only primates for both Data sets #1 and #2 at the levels of the first postsacral and midproximal caudal vertebra. For the TV, PCA include only Data set #1 because, as described above, the measurements obtained in Data set #2 are variably present (e.g., neural arches) at this vertebral level.

Data set #1. At all three vertebral levels for all mammals and only primates, Data set #1 included four variables that accounted for a significant proportion of the variance in RTL (Table 3): VCCL, CASA, CASS, and TPB. At the level of the first postsacral vertebra, PC1 accounted for approximately 74% and 77% of variation among all mammals and among only primates, respectively (Table 4).

At the level of the midproximal vertebra, PC1 accounted for approximately 64% and 65% of variation among all mammals and among only primates, respectively (Table 4). At the level of the TV, PC1 accounted for approximately 53% and 66% of variation among all mammals and among only primates, respectively (Table 4).

Data set #2. At the level of the first postsacral vertebra for all mammals and primates only, Data set #2 included 9 variables that accounted for a significant proportion of the variance in RTL (Table 3): VCCL, CASS, CASA, TPB, TPDV, TPCC, ANA, and SPL. PC1 accounted for approximately 57% and 62% of variation among all mammals and among only primates, respectively (Table 4). PCA at the level of the midproximal vertebra included the same variables as that at the level of the first postsacral except for the addition of PZO among all mammals and among primates only, and, the omission of TPP among mammals (see Table 3). PC1 accounted for approximately 42% and 50% of variation

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

Fig. 7. Comparisons of values for vertebral body craniocaudal length (A–C), cranial articular surface area (D–F), spinous process length (G–I), and transverse process dorsoventral orientation (J–L), among closely related taxa that differ in relative tail length from three mammalian clades. See text for description. See also Figure 8.

369

370

RUSSO

Fig. 8. First postsacral vertebrae of taxa that differ in relative tail length (RTL) from three mammalian clades: Primates, Diprotodontia, and Carnivora. Illustration exemplifies the morphological differences in relation to RTL between closely related taxa, and morphological paral-

lels in relation to RTL among distantly related taxa. Taxa are organized by rows from top to bottom (labeled on the right). Vertebral views from left to right are cranial, dorsal and lateral views. Scale bar is approximately 1cm. See text for description. See also Figure 7.

371

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

TABLE 4. Summary of results from principal components analyses Vertebral level First postsacral vertebra

Midproximal vertebra

Transition vertebra

Data set

PC1 %

Eigenvalue

All Mammals Data set 1 All Mammals Data set 2 Primates Only Data set 1 Primates Only Data set 2 All Mammals Data set 1 All Mammals Data set 2 Primates Only Data set 1 Primates Only Data set 2 All Mammals Data set 1 Primates Only Data set 1

74.12 57.38 77.09 62.32 64.18 41.68 65.24 50.13 52.75 66.20

2.97 5.16 3.08 5.61 2.57 3.33 2.61 4.01 2.11 2.65

TABLE 5. Least-squares regression coefficients and model summaries for RTL on PC1 species mean scores Vertebral level

Data set

R2

Adjusted R2

Pearson’s r

SEE

Slope

Constant

First postsacral vertebra

All Mammals Data set 1 All Mammals Data set 2 Primates Only Data set 1 Primates Only Data set 2 All Mammals Data set 1 All Mammals Data set 2 Primates Only Data set 1 Primates Only Data set 2 All Mammals Data set 1 Primates Only Data set 1

0.62 0.68 0.80 0.94 0.49 0.70 0.78 0.93 0.38 0.86

0.62 0.66 0.79 0.93 0.48 0.68 0.78 0.92 0.37 0.86

0.79 0.82 0.89 0.97 0.70 0.84 0.89 0.96 0.62 0.93

33.37 29.48 26.20 14.19 34.58 27.22 22.42 14.39 36.10 18.46

43.15 43.97 53.41 53.11 33.43 39.95 43.55 46.96 26.63 42.75

64.36 74.91 79.07 80.81 80.52 84.05 92.70 91.48 88.54 100.89

Midproximal vertebra Transition vertebra

among all mammals and among only primates, respectively (Table 4).

RTL on PC1 Scores Regression equation coefficients and model summaries for all mammals and only primates are provided in Table 5. At the level of the first postsacral vertebra, species mean PC1 scores can account for 68% (Data set #2) and 94% (Data set #2) of the variance in RTL among all mammals and among only primates, respectively (Table 5). The standard error of the estimate for Data set #2 (95% confidence intervals 5 628) is nearly half that of Data set #1 (95% confidence intervals5 652). At the level of the midproximal vertebra, species mean PC1 scores can account for 70% (Data set #2) and 93% (Data set #2) of the variance in RTL among all mammals and among primates only, respectively (Table 5). At the level of the TV, species mean PC1 scores can account for 38% (Data set #1) and 86% (Data set #2) of the variance in RTL, among all mammals and among primates only, respectively (Table 5).

DISCUSSION The primary objective of this study was to quantify postsacral vertebral morphology as it relates to differences in relative tail length (RTL) among living primates and other mammals. The results support the study’s predictions that longer-tailed taxa would generally exhibit a greater number of features associated with increased proximal tail mobility, increased intervertebral joint range of motion, and increased leverage of, and attachment area for, tail musculature, compared to shorter-tailed taxa, which should exhibit features associated with relatively reduced

proximal tail mobility, reduced intervertebral joint range of motion, and reduced leverage of tail musculature (see also Table 2). That these anatomical associations with RTL are characteristic of distantly-related mammalian taxa (Figs. 7 and 8) is indicative of a strong form-function link that underscores the utility of a comparative mammalian sample for examining the correlates of tail reduction/loss. Below, the functional significance of some convergent postsacral vertebral features is discussed. This study found that RTL increased with increasing vertebral body craniocaudal length (VCCL). Like Hamada et al. (2012), VCCL explained a greater proportion of the variance in RTL at more distal vertebral levels, specifically at the TV, followed by the midproximal vertebra. Nonetheless, RTL accounted for a significant proportion of variance at all three vertebral levels. Functionally, VCCL must be considered in the context of its interaction with vertebral numbers in a given spinal region, and that region’s total length, to understand potential range of motion (Ward, 1993; Shapiro and Simons, 2002). For example, in a comparison of two spinal regions with the same total length but different numbers of vertebrae, the sequence with the greater number of relatively shorter (i.e., reduced craniocaudal length) vertebral bodies can achieve a greater arc of flexion; in a comparison of two spinal regions with different total lengths and different vertebral body craniocaudal lengths, the longer vertebral sequence achieves a greater arc of flexion irrespective of vertebral number (Ward, 1993; Shapiro and Simons, 2002). Though shorter-tailed mammals may possess the same number of proximal postsacral vertebrae as mammals with longer tails (e.g., Macaca, Table 1), the summed length of craniocaudally shorter vertebrae produces an absolutely shorter proximal tail region. Consequently, the proximal tail region of

372

RUSSO

a longer-tailed mammal can theoretically achieve a greater arc of flexion than that of a shorter-tailed mammal (Ward, 1993; Shapiro and Simons, 2002; contra Hamada et al., 2012), suggesting that tail length reduction is associated with decreased proximal tail flexibility. Tail length reduction is also associated with increasingly elliptically-shaped (i.e., wider mediolaterally than dorsoventrally) cranial articular surfaces (CASS) with less surface area (CASA). This particular surface shape trend has been reported for the caudal articular surface of the sacrum in catarrhine primates (Russo and Shapiro, 2011). However, as noted by Russo and Shapiro (2011) and shown here (Figs. 3, 4, 5, and 6), much of the relationship between CASS and RTL is driven by the differences between tail-bearing and tailless taxa. Though they did not provide quantitative measures of articular surface dimensions in their mammalian sample, Nakatsukasa et al. (2003) asserted that dorsoventral compression of the postsacral vertebral body is rare in nonhominoid anthropoids, and that even short-tailed cercopithecoids exhibit a cranial articular surface that is wider dorsoventrally than mediolaterally. Yet, the results presented here show that all CASS ratio values are above 1 for the first postsacral and midproximal vertebrae (but not necessarily the TV) in the sample. Thus, the cranial articular surfaces of the first postsacral and midproximal vertebrae among all mammals sampled are actually slightly wider mediolaterally than dorsoventrally. Moreover, since hominoids do not possess midproximal (or transition vertebrae), that the CASS ratio values generally have an inverse relationship to tail length at all three vertebral levels suggests dorsoventral vertebral body compression is not unique to hominoid primates (Figs. 3, 4, 5, and 8). Even so, it is notable that hominoid primates exhibited the highest values, and thus the most dorsoventrally compressed vertebral bodies, at the level of the first postsacral vertebra, and there is also a considerable amount of scatter in the data among mammals. Clinical work shows that a dorsoventrally compressed sacrococcygeal joint is functionally associated with a reduced range of motion limited to few directions (Saluja, 1988; Kapandji, 2008; Woon and Stringer, 2012). For example, the range of dorsal extension movement at the (unfused) sacrococcygeal joint is between 5 and 15 (with average total mobility [extension and flexion] of 9 ) (Maigne and Bertrand, 1996; Woon and Stringer, 2012). By comparison, an increasingly circular CASS should permit an increased ranged of motion that allows for multidirectional movements (Russo and Shapiro, 2011). Wilson (1972) has shown that macaque primates can dorsally extend their tails up to at least 155 (though he maintains that shorttailed macaques should be able to produce greater tail extension than long-tailed macaques). Russo and Shapiro (2011) demonstrated that longer-tailed catarrhine primates have more dorsally tilted sacrocaudal articular surfaces than shorter-tailed catarrhines, which would facilitate greater dorsal extension in the former group. With respect to CASA, relatively smaller joint surfaces indicate a reduced ability to distribute high loads over a wider area (Currey, 1984; Jungers, 1988; Kapandji, 2008), suggesting that lower force magnitudes may transmitted across the postsacral intervertebral body joint surfaces in shorter-tailed taxa compared to longer-tailed taxa (see also Nakatsukasa et al., 2003; Russo et al., 2012). Prelimi-

nary data from trabecular and cortical bone analyses of the last sacral vertebra (whose distal surface articulates with the first postsacral vertebra) show some support for this idea (Russo et al., 2012). For example, the last sacral vertebra in macaques with longer tails exhibits features associated with increased bone strength, such as thicker cortical bone and more numerous trabeculae, compared to macaques with shorter tails (Russo et al., 2012). Taken together, more elliptically-shaped cranial articular surfaces with reduced surface area likely reduces the range of proximal tail motion and resistance to forces, respectively, in shorter-tailed mammals compared to longer-tailed mammals. Nonetheless, comparative biomechanical analyses of the sacrocaudal joint and proximal postsacral intervertebral joints are needed to test the functional hypotheses proposed here. Shorter-tailed primates are further characterized by proximal postsacral vertebrae having smaller cranial neural apertures (ANA), shorter spinous processes (SPL), and less laterally projected transverse processes (TPB) that are angled more dorsally (TPDV) and less caudally (TPCC), and positioned more dorsally on the vertebral body (TPP), compared to longer-tailed primates (Figs. 7 and 8). Smaller neural apertures likely reflect a decrease in the size of the pathway for innervation to a reduced tail region (Ankel, 1965, 1972), and a reduction in the projection of the spinous and transverse processes decreases the leverage of, and surface area of attachment for, tail musculature. For example, at the levels of the first and often midproximal postsacral vertebrae, the transverse processes serve as attachment points for abductor musculature (e.g., ischiocaudalis mm.) that laterally flexes the tail, while at the level of the TV and more distal tail vertebrae, the transverse processes serve as attachment points for ventral flexors (e.g., flexor caudae longus mm.). Further, their ventral orientation creates a deep compartment for the extensor musculature (e.g., extensor caudae lateralis mm., the caudal extension of longissimus), and their caudal angulation (along with narrower apices; see Fig. 8) allows for increased lateral flexibility (Shapiro, 1993). The length of the spinous process influences the compartment size for the tail’s extensor musculature as well their leverage (e.g., extensor caudae medialis mm., the caudal extension of the multifidus) (Shapiro, 1993). Therefore, the results presented here are suggestive of relatively reduced leverage of, and attachment area for, musculature that produces dorsal extension, ventral flexion, and lateral abduction tail movements in shorter-tailed taxa compared to longertailed taxa (Table 2). However, it was contrary to predictions that tail length would increase as the transverse processes became more dorsally positioned. Dorsally positioned transverse processes should increase leverage for dorsal extension (Shapiro, 1993). However, their positioning as it is observed here is likely a consequence of changes in dorsoventral height of the vertebral body’s cranial articular surface (CASS). That is, as tail length decreases, dorsoventral compression of the vertebral body likely reduces the available space for the origin of the transverse processes. Given changes in particular surface dimensions, it is possible that future measures of transverse process position should consider TPP relative to dorsoventral breadth of the cranial articular surface. The morphology of the distal sacrum and proximal postsacral vertebrae varied considerably in mammals with

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES

373

Significance

Fig. 9. First postsacral vertebrae in dorsal views of (A) Tonkean macaque (Macaca tonkeana, RTL 5 8); (B) Barbary macaque (Macaca sylvanus, RTL 5 0); (C) capybara (Hydrochaeris hydrochaeris, RTL 5 0); and (D) slow loris (Nycticebus coucang; RTL 5 6). Illustration highlights the variation in first postsacral vertebral anatomy among the mammals sampled. In the macaques (A and B), arrows point to grooves formed by partially developed neural arches, prezygapophyseal articulations are retained; in the capybara (C), a neural arch is fully developed, but zygapophyseal articulations are vestigial; in the slow loris (D), the neural arch is not fully formed, and the prezygapophyses are absent. Scale bar is approximately 1 cm.

This study represents the most comprehensive quantitative examination to date of how postsacral vertebral morphology varies among primates and other mammals that differ in tail length. The results presented here have the potential to greatly improve our understanding of tail loss in the primate fossil record. Though the mechanisms by which tails were reduced or lost in different primate clades are unclear (and it is possible that each lineage evolved tail length reduction for a different reason), appropriately contextualizing the occurrence of tail loss, either ecologically or behaviorally, is not possible until there is a better understanding of which fossil taxa are actually characterized by reduced tail lengths. In other words, a prerequisite to generating testable hypotheses for evaluating the evolutionary timing and functional significance of tail length reduction is the ability to first confidently identify it in the fossil record. However, complete tail sequences are rarely preserved. As such, the ability to confidently attribute tail lengths to extinct primates depends on a more complete knowledge of tail length morphological correlates. When taken together, the postsacral vertebral variables examined here can explain 94% and 93% of the variance in RTL among primates at the level of the first postsacral and midproximal vertebrae, respectively (Tables 3 and 5). Results from the study’s multivariate analyses can be used to create predictive models for estimating the tail lengths of extinct primates represented by only one, or few, caudal vertebral elements present (Table 5). Future research should test the utility of the extant primate data for making inferences about the tail lengths of extinct primates.

ACKNOWLEDGEMENTS reduced tail lengths. For example, some slow lorises (Nycticebus) possessed postsacral vertebrae with neural arches that were partially formed, with the laminae appearing as two bony ridges along the dorsal aspect of the body (Fig. 9). Zygapophyses were absent, and in many cases the postsacral vertebral series were fully ankylosed. Where the laminae were fused, the neural aperture was completely compressed (i.e., closed off). Other mammals had postsacral vertebrae lacking zygapophyseal articular facets, but bearing fully formed neural arches (e.g., Hydrochaeris, Fig. 9). The neural arches of the first postsacral vertebrae of several short-tailed macaques, including the Barbary macaque (Macaca sylvanus) and the Tonkean macaque (Macaca tonkeana), were also sometimes only partially formed (e.g., a neural arch ring was not formed), resulting in a deep groove between two laminar ridges (Fig. 9; but see Nakatsukasa et al., 2003). Among very short-tailed nonhominoid mammalian taxa, the presence of postzygapophyseal facets on the distal sacrum was variable, and there were a number of instances of absent sacral postzygapophyses/present first postsacral prezygapophyses or present sacral postzygapophyses/absent first postsacral prezygapophyses. In this study, these cases sometimes complicated vertebral counts or prohibited the collection of some measurements. Elsewhere, the pervasiveness of this inter- and intra-specific morphological variation must be taken into consideration when attributing tail lengths to fossil taxa on the basis of qualitative features of individual vertebrae.

The author wishes to thank Liza Shapiro, Chris Kirk, Laurie Godfrey, Roberto Fajardo, Carol Ward, and John Kappelman for providing constructive comments on earlier versions of this manuscript. The author is also grateful to W. Andrew Barr, who provided statistical advice and guidance, and the NEOMED journal club who offered advice on format and presentation. Thanks are also owed to William Stanley (FMNH), Eileen Westwig (AMNH), and Darrin Lunde (NMNH) who provided access to the osteological collections in their care, as well as the AR associate editor and two anonymous reviewers for assistance in the revision of this manuscript. This research was funded by National Science Foundation DDIG BCS-1156016 and a Leakey Foundation General Research Grant. This work was also made possible by a University of Texas at Austin Graduate Research Fellowship and a William C. Powers Graduate Fellowship. This project was awarded the Mildred Trotter Award at the 82nd annual meeting of the American Association of Physical Anthropologists.

LITERATURE CITED Ankel F. 1962. Vergleichende Untersuchungen iiber die Skelettmorphologie des Greifschwanzes sudamerikanischer Affen. Z. Morph. Oekol. Tiere 52:131–170. Ankel F. 1965. Der canalis sacralis als indikator f€ ur die l€ ange der caudal region der primaten. Folia Primatol. 3:263–276.

374

RUSSO

Ankel F. 1972. Vertebral morphology of fossil and extant primates. In: Tuttle, R, editor. The functional and evolutionary biology of primates. Chicago: Aldine, p 223–240. Argot C. 2003. Functional-adaptive anatomy of the axial skeleton of some extant marsupials and the paleobiology of the paleocene marsupials Mayulestes ferox and Pucadelphys andinus. J Morph 255:279–300. Bergeson DJ. 1992. The use of the prehensile tail in Alouatta and Cebus. Am J Phys Anthropol 14:48. Bergeson DJ. 1995. The ecological role of the platyrrhine prehensile tail. Am J Phys Anthropol 20:64–65. Bergeson DJ. 1996. The positional behavior and prehensile tail use of Alouatta palliata, Ateles geoffroyi, and Cebus capucinus. PhD Dissertation, Washington University, St. Louis. Bezanson MF. 2005. Leap bridge or ride? Ontogenetic influences on positional behavior in Cebus and Alouatta. In: Estrada A, Garber PA, Pavelka MSM, Luecke L, editors. New perspectives in the study of Mesoamerican primates: distribution, ecology, behavior and conservation. New York: Springer. p 333–348. Bezanson MF. 2006. Ontogenetic patterns of positional behavior in Cebus capucinus and Alouatta palliata. PhD dissertation, University of Arizona, Tucson, AZ. Bezanson MF. 2009. Life history and locomotion in Cebus capucinus and Alouatta palliata. Am J Phys Anthropol 140:508–517. Bininda-Emonds ORP, Cardillo M, Jones KE, MacPhee RDE, Beck RMD, Grenyer R, Price SA, Vos RA, Gittleman JL, Purvis A. 2007. The delayed rise of present-day mammals. Nature 446: 507–512. Bogduk N, Twomey L. 2005. Clinical anatomy of the lumbar spine and sacrum. Livingstone: Churchill. Buck C, Tolman N, Tolman W. 1925. The tail as a balancing organ in mice. J Mammal 6:267–271. Byron C, Kunz H, Matuszek H, Lewis S, Van Valkinburgh D. 2011. Rudimentary pedal grasping in mice and implications for terminal branch arboreal quadrupedalism. J Morph 272:230–240. Currey JD. 1984. The mechanical adaptations of bones. Princeton, N.J.: Princeton University Press. Deinard A, Smith, DG. 2001. Phylogenetic relationships among the macaques: evidence from the nuclear locus NRAMP1. J Hum Evol 41:45–59. Demes B, Jungers WL, Fleagle JG, Wunderlich RE, Richmond BG, Lemelin P. 1996. Body size and leaping kinematics in Malagasy vertical clingers and leapers. J Hum Evol 31:367–388. Dor M. 1937. La morphologie de la queue des mammiferes. Paris: Pierre Andre. Emerson SB. 1985. Jumping and leaping. In: Hildebrand M, Bramble DM, Liem KF, Wake DB, editors. Functional vertebrate morphology. Cambridge: Belknap Press. p 58–72. Essner RL. 2002. Three-dimensional launch kinematics in leaping, parachuting and gliding squirrels. J Exp Biol 205:24–69. Fa JE. 1985. Baby care in barbary macaque. In: Macdonald D, editor. Primates. New York: Torstar Books. p 92–93. Fa JE. 1989. The genus Macaca: a review of taxonomy and evolution. Mammal Rev. 19:45–81. Fleagle JG. 1984. Size and adaptation in primates. In: Jungers WL, editor. Size and scaling in primate biology. New York: Plenum Press. p 1–19. Fleagle JG. 1999. Primate adaptation and evolution. San Diego: Academic Press. Flower WH. 1876. An introduction to the osteology of the mammalia: being the substance of the course of lectures delivered at the Royal College of Surgeons of England in 1870. Macmillan. Fooden J. 1969. Taxonomy and evolution of the monkeys of Celebes (Primates: Cercopithecidae). Karger: Basel. Fooden J. 1975. Taxonomy and evolution of liontail and pigtail macaques (Primates: Cercopithecidae). Fieldiana Zool 67:1–169. Fooden J. 1988. Taxonomy and evolution of the sinica group of macaques: Interspecific comparisons and synthesis. Fieldiana Zool 45: 1–44. Fooden, J., 1997. Tail length variation in Macaca fascicularis and M. mulatta. Primates 38:221–232.

Fooden J. 2006. Comparative review of fascicularis-group species of macaques (Primates: Macaca). Fieldiana Zool 107:1–44. Fooden J. 2007. Systematic review of the barbary macaque, Macaca sylvanus (Linnaeus, 1758). Fieldiana Zool 113:1–58. Fooden J, Albrecht GH. 1999. Tail length evolution in fascicularis group macaques (Cercopithecidae: Macaca). Int J Primatol 20: 431–440. Francis CM. 2008. A field guide to the mammals of southeast asia. London: New Holland. Freckleton RP, Harvey PH, and Pagel M. 2002. Phylogenetic analysis and comparative data: a test and review of evidence. Am Nat 160:712–726. Garber PA, Rehg JA. 1999. The ecological role of the prehensile tail in white-faced capuchins (Cebus capucinus). Am J Phys Anthropol 110:325–339. Gebo DL. 1992. Locomotor and postural behavior in Alouatta palliata and Cebus capucinus. Am J Primatol 26:277–290. German RZ. 1982. The functional morphology of caudal vertebrae in New World monkeys. Am J Phys Anthropol 58:453–459. Grand T. 1977. Body weight: its relation to tissue composition, segment distribution, and motor function. I. Interspecific comparisons. Am J Phys Anthropol 47:211–239. Groves CP. 2001. Primate taxonomy. Washington, DC: Smithsonian Institution Press. Groves CP. 2005. Order primates. In: Wilson DE, Reeder DM, editors. Mammal species of the world: a taxonomic and geographic reference, 3rd ed. Baltimore: Johns Hopkins University Press. Vol. 1: p 111–184. Hamada Y, Yamamoto A, Kunimatsu Y, Tojima S, Mouri T, Kawamoto Y. 2012. Variability of tail length in hybrids of the Japanese macaque (Macaca fuscata) and the Taiwanese macaque (Macaca cyclopis). Primates 53:397–411. Harrison T. 1998. Evidence for a tail in Proconsul heseloni. Am J Phys Anthropol 26:93–94. Hartman CG, Straus WL. 1933. The anatomy of the rhesus monkey (Macaca mulatta). Baltimore: The Williams & Wilkins Company. Hatt RT. 1932. The vertebral columns of ricochetal rodents. Bull AMNH 63, article 6. Hickman G. 1979. The mammalian tail: a review of functions. Mammal Rev 9:143–157. Hildebrand M, Goslow. 2001. Analysis of vertebrate structure. New York, NY: Wiley. Horner B. 1954. Arboreal adaptations of Peromyscus, with special reference to use of the tail. Contrib Lab Vert Biol 61:1–84. Igarashi M, Levy J. 1981. Locomotor balance performance of shorttailed squirrel monkeys. J Med Primatol 10:136–140. Ihaka R, Gentleman R. 1996. R: a language for data analysis and graphics. J Comp Graph Stats, 5(3):299–314. Jenkins PD. 1990. Catalogue of Primates in the British Museum (Natural History) and Elsewhere in the British Isles. Part V: The Apes, Superfamily Hominoidea. The British Museum (Natural History), London. Jungers WL. 1988. Relative joint size and hominoid locomotor adaptations with implications for the evolution of hominid bipedalism. J Hum Evol 17:247–265. Kapandji IA. 2008. The physiology of the joints (V3): the spinal column, pelvic girdle and head. Paris: Lavoisier. Kimura K, Konishi M, Takahashi Y, Iwamoto. 1986. The skeletal system of Macaca fascicularis—description and measurement, Part 3: the vertebrae. J Natl Def Med Coll 11:19–34. Larson S, Stern J. 2006. Maintenance of above-branch balance during primate arboreal quadrupedalism: coordinated use of forearm rotators and tail motion. Am J Phys Anthropol 129:71–81. Lawler RR, Stamps C. 2002. The relationship between tail use and positional behavior in Alouatta palliata. Primatol 43:147–152. Lemelin P. 1995. Comparative and functional myology of the prehensile tail in New World monkeys. J Morphol 224:351–368. Maigne JY, Tamalet B. 1996. Standardized radiologic protocol for the study of common coccygodynia and characteristics of the lesions observed in the sitting position: Clinical elements differentiating luxation, hypermobility, and normal mobility. Spine 21: 2588–2593.

POSTSACRAL VERTEBRAL MORPHOLOGY IN PRIMATES Martin R. 1968. Reproduction and ontogeny in tree-shrews (Tupaia belangeri), with reference to their general behaviour and taxonomic relationships. Zeitschrift f€ ur Tierpsychologie 25:409–495. Martins EP, Hansen TF. 1997. Phylogenies and the comparative method: a general approach to incorporating phylogenetic information into the analysis of interspecific data. Am Nat 149:646–667. Meldrum DJ. 1998. Tail-assisted hind limb suspension as a transitional behavior in the evolution of the platyrrhine prehensile tail. In: Strasser E, et al., editors. Primate locomotion: recent advances. New York: Plenum. p 145–156. Miller GS. 1900. Key to the land mammals of northeastern North America. Albany: University of the State of New York. Nakatsukasa M, Tsujikawa H, Shimizu D, Takano T, Kunimatsu Y, Nakano Y, Ishida H. 2003. Definitive evidence for tail loss in Nacholapithecus, an East African Miocene hominoid. J Hum Evol 45:179–186. Napier PH. 1981. Catalogue of Primates in the British Museum (Natural History) and Elsewhere in the British Isles, Part 2: Family Cercopithecidae, Subfamily: Cercopithecinae. British Museum (Natural History), London. Nowak RM. 1991. Walker’s Mammals of the World, 5th ed., Vol. 1. Baltimore and London: J. Hopkins University Press. Organ JM. 2006. To grasp or not to grasp? Structure and function of platyrrhine caudal vertebrae. Am J Phys Anthropol (S42):142. Organ JM. 2007. The functional anatomy of prehensile and nonprehensile tails of the Platyrrhini (Primates) and Procyonidae (Carnivora). PhD Dissertation. The Johns Hopkins University, Baltimore, MD. Organ JM. 2010. Structure and function of platyrrhine caudal vertebrae. Anat Rec 293:730–745. Organ JM, Teaford MF, Taylor AB. 2009. Functional correlates of fiber architecture of the lateral caudal musculature in prehensile and nonprehensile tails of the Platyrrhini (Primates) and Procyonidae (Carnivora). Anat Rec 292:827–841. Organ JM, Muchilinski MN, Deane AS. 2011. Mechanoreceptivity of prehensile tail skin varies between atelines and Cebus. Anat Rec 294(12):2064–2072. Pagel M. 1999. Inferring the historical patterns of biological evolution. Nature 401:877–884. Paradis E, Claude J, and Strimmer K. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289– 290. Parker SP. 1990. Grzimek’s encyclopedia of mammals, Vol. 2. New York: McGraw Hill. Pinheiro J, Bates D, DebRoy S, Sarkar D, and R Core Team. 2014. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-117. Preuschoft H, Witte H, Christian A, Fischer M. 1996. Size influences on primate locomotion and body shape, with special emphasis on the locomotion of ‘small mammals’. Folia Primatol. 66: 93–112. Procter-Gray E, Ganslosser U. 1986. The individual behaviors of Lumholtz’s tree-kangaroo: repertoire and taxonomic implications. J Mammal 67:343–352. R Development Core Team. 2011. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Rodman PS. 1979. Skeletal differentiation of Macaca fascicularis and Macaca nemestrina in relation to arboreal and terrestrial quadrupedalism. Am J Phys Anthropol 51:51–62. Rose MD. 1974. Postural adaptations in New and Old World monkeys. In: Jenkins FA, editor. Primate locomotion. New York: Academic Press. p 201–222. Rosenberger AL. 1983. Tale of tails: parallelism and prehensility. Am J Phys Anthropol 60:103–107. Rowe N. 1996. The pictorial guide to the living primates. New York: Pogonias Press. Russo GA. 2010. Prezygapophyseal articular facet shape in the catarrhine thoracolumbar vertebral column. Am J Phys Anthropol 142:600–612.

375

Russo GA, Shapiro, LJ. 2011. Morphological correlates of tail length in the catarrhine sacrum. J Hum Evol 30:223–232. Russo GA, Young JW. 2011. Tail growth tracks the ontogeny of prehensile tail use in capuchin monkeys (Cebus albifrons and C. apella). Am J Phys Anthropol 146:465–473. Russo GA, Fajardo RJ, Schmitz JE. 2012. Internal bone structure of the last sacral vertebra and its relationship to tail length. Am J Phys Anthropol 147:255–256. Saluja PG. 1988. The incidence of ossification of the sacrococcygeal joint. J Anat 156:11–18. Sanders WJ, Bodenbender BE. 1994. Morphometric analysis of lumbar vertebra UMP 67–28: implications for spinal function and phylogeny of the Miocene Moroto hominoid. J Hum Evol 26:203– 237. Schmidt-Nielsen K. 1984. Scaling: why is animal size so important? Cambridge: Cambridge University Press. Schmitt D, Rose MD, Turnquist JE, Lemelin P. 2005. Role of the prehensile tail during ateline locomotion: experimental and osteological evidence. Am J Phys Anthropol 126:435–446. Schmidt, E. 1886. Uber die Wirbelsaule der Primaten. Korr Bl Ges Anthrop Urgesch 17:5–6. Shapiro LJ. 1993. Functional morphology of the vertebral column in primates. In: Gebo D, editor. Postcranial adaptation in nonhuman primates. DeKalb: Northern Illinois University Press. p 121–149. Shapiro LJ. 2007. Morphological and functional differentiation in the lumbar spine of lorisids and galagids. Am J Primatol 69:86– 102. Shapiro LJ, Simons, C. 2002. Functional aspects of strepsirrhine lumbar vertebral bodies and spinous processes. J Hum Evol 42: 753–783. Shapiro LJ, Seiffert C, Godfrey L, Jungers W, Simons E, Randria G. 2005. Morphometric analysis of lumbar vertebrae in extinct Malagasy strepsirrhines. Am J Phys Anthropol 128:823–839. Siegel M. 1970. The tail, locomotion and balance in mice. Am J Phys Anthropol 33:101–102. Silva M, Downing JA. 1995. CRC handbook of mammalian body masses. Boca Raton, Florida: CRC press. Smith R, Jungers W. 1997. Body mass in comparative primatology. J Hum Evol 32:523–559. Sprankel H. 1965. Untersuchungen an Tarsius. Folia primatologica, 3:153–188. Stevens N, Wright K, Covert H, Nadler T. 2008. Tail postures of four quadrupedal leaf monkeys (Pygathrix nemaeus, P. cinerea, Trachypithecus delacouri and T. hatinhensis) at the Endangered Primate Rescue Center, Cuc Phuong National Park, Vietnam. Viet. J Primatol 2:13–24. Tito G. 2008. New remains of Eremotherium laurillardi (Lund, 1842)(Megatheriidae, Xenarthra) from the coastal region of Ecuador. J South Am Earth Sci 26:424–434. Wada N, Hori H, Tokuriki M. 1993. Electromyographic and kinematic studies of tail movements in dogs during treadmill locomotion. J Morph 217:105–113. Walker C, Vierck C, Ritz L. 1998. Balance in the cat: role of the tail and effects of sacrocaudal transection. Behav Brain Res 91: 41–47. Ward CV. 1993. Torso morphology and locomotion in Proconsul nyanzae. Am J Phys Anthropol 92:291–328. Ward CV, Walker A, Teaford MF. 1991. Proconsul did not have a tail. J Hum Evol 21:215–220. Ward CV, Walker A, Teaford MF. 1999. Still no evidence for a tail in Proconsul heseloni. Am J Phys Anthropol 28:273. Wilson, DR, 1972. Tail reduction in Macaca. In: Tuttle R, editor. The functional and evolutionary biology of primates. New York: Aldine Press. p 241–261. Woon JT, Stringer MD. 2012. Clinical anatomy of the coccyx: a systematic review. Clin Anat 25:158–167. Youlatos D. 1999. Tail-use in capuchins. Neotrop Prim 7:16–20. Youlatos D. 2003. Osteological correlates of tail prehensility in carnivorans. J Zool 259:423–430.