Narial Novelty in Mammals: Case Studies and Rules Of

NARIAL NOVELTY IN MAMMALS: CASE STUDIES AND RULES OF

CONSTRUCTION

A thesis presented to the faculty of

the College of Arts and Sciences of Ohio University

In partial fulfillment of the requirements for the degree

Master of Science

Andrew B. Clifford

August 2003

This thesis entitled

NARIAL NOVELTY IN MAMMALS: CASE STUDIES AND RULES OF

CONSTRUCTION

ANDREW B. CLIFFORD

has been approved for the Department of Biological Sciences and

the College of Arts and Sciences by

Lawrence M. Witmer

Associate Professor, Biomedical Sciences

Leslie A. Flemming

Dean, College of Arts and Sciences

CLIFFORD, ANDREW B. M.S. August, 2003. Biological Sciences

Narial Novelty in Mammals: Case Studies and Rules of Construction (pp. 128)

Director of Dissertation: Lawrence M. Witmer

Both anatomy and function of the enigmatic proboscis of moose and the nasal cavity of saiga are described. Dissection, sectioning, and skeletonisation of study specimens and related outgroups are supplemented with CT scans and other software-generated imaging to describe the structure of apomorphic narial tissues and skeletal modifications. Anatomy is used to assess previously suggested functions of these probosces and to advance new hypotheses based on novel anatomy. Moose noses possess elaborated nostril musculature and nasal cartilages which contribute to a nostril closing mechanism. Saiga have evolved an elaborated nasal vestibule which cleans air destined for lungs. Both probosces modify the bony naris in ways that have justified tapir-like trunks in fossil species. Integrating data from many different probosces, mammals follow limited rules of construction in proboscis- building. Outgroup anatomy constrains proboscis anatomy, and exaptation produces narial novelty. Muscular hydrostats and maxillolabial probosces leave the fewest osteological correlates, limiting proboscis reconstruction.

Approved: Lawrence M. Witmer

Associate Professor, Biomedical Sciences

Dedication

To Rosie, who always knew it is not what you take with you when you leave but what you leave behind when you go. She has left behind an immeasurable wealth that shall be cherished forever by everyone that knew and loved her.

Acknowledgments

I owe my sincerest thanks to so many individuals, many of whom I have never met, who saw fit to donate or provide specimens for the purpose of this research project. In the death of such beautiful animals, they saw the inestimable value to science these animals held. H.

Mayle, D. Pratt, and R. Ridgely donated precious time and skill to CT scanning and illustration. J. Sedlmayr, C. Holliday, P. O’Connor, B. Beatty, and T. Hieronymus provided much-needed assistance in specimen preparation. A. Biknevicius and S. Reilly were excellent, patient, and gracious committee members. My advisor, L. Witmer, gave freely many hours of guidance and advice that I hope to carry with me for the rest of my career.

Finally, the grounding given to me by my friends and family, their tolerance of my eccentricities, makes this work all the more fulfilling.

Table of Contents

Page Abstract 3

Dedication 4

Acknowlegments 5

List of Tables 8

List of Figures 9

List of Abbreviations 12

Chapter One: The Enigmatic Nose of Moose (Artiodactyla: Cervidae: Alces alces) 15 Abstract 15 Introduction 17 Materials and Methods 21 Results 22 External anatomy 22 Integument 23 Connective tissue pad 24 Musculature 24 Major nerves and vessels 34 Nasal cartilages 40 Overview of nasal cavity 43 Glands 46 Osteology 47 Discussion 51 Novel aspects of narial anatomy in moose 51 Moose narial anatomy with respect to fossil alcines 56 Functions of moose noses based on anatomical specialisations 58

Chapter Two: Structure and Function of the Nasal Cavity of Saiga (Artiodactyla: 63 Bovidae: Saiga tatarica) Abstract 63 Introduction 65 Materials and Methods 68 Results 69 Overview of nasal cavity 69 Nasal vestibule 70 Main nasal cavity 74 Nasal septum 80 Osteological correlates of the proboscis in saiga 81 Discussion 85 Reorganisation of the nasal cavity 85 Proboscis function in saiga 87

Lateral recess homology and function 90

Chapter Three: Rules of Construction in Mammalian Proboscis Building 92 Abstract 92 Introduction 94 Proboscis types within Mammalia 96 Phylogenetic constraint and exaptation 100 Rules of construction in proboscis building 104 Reliable osteological correlates resulting from proboscis building 109 Further tests of construction hypotheses 111 Applications to extinct taxa 113

References 116

Appendix One: Homolgy and Nomenclature of Facial Musculature in Mammals 123

List of Tables Page

Appendix Table 1. Facial musculature in mammals. Sources for muscles are given in column titles, and muscle groups are given in the leftmost column. Muscles in a single row are homologous. Note the disparities in homology between sources and the disparate nomenclature sometimes used. a—These muscles were grouped separately in a “lateralis nasi group” by Boas & Paulli (1908a, b). b—Except in Carnivora (Evans, 1993). 127

List of Figures

Figure Page 1.1. Lateral view of male moose (Alces alces) in velvet (a) and lateral view of male moose head (b) showing unique muzzle. (a) Courtesy of M. Reichmann; used with permission. (b) Courtesy of G. and B. Corsi and the California Academy of Sciences; used with permission. 18

1.2. Phylogenetic relationships of taxa and clades referred to in this study. Topology based on Novacek, Wyss & McKenna (1988) and Groves & Grubb (1987). 19

1.3. Superficial dissection of the face of Alces alces shown in (a) left lateral view and in (b) oblique left rostrodorsolateral view, based on OUVC 9559. Scale bars = 10cm. 25

1.4. Deep dissections of the face of Alces alces. Maxillolabial muscles are intact in (a). In (b), the maxillolabial muscles have been reflected to reveal underlying structures, based on OUVC 9559. Scale bars = 10cm. 26

1.5. Drawings of selected computerized tomographic (CT) images of the face of Alces alces (OUVC 9559) showing narial structures in successive transverse sections (a-h). (i) Skull in left lateral view to show the rostrocaudal levels of sections depicted in (a-h). Scale bars = 5 cm. 38

1.6. Oblique left rostrodorsolateral view (a) of a skull of Alces alces with cartilages in place to show the cartilaginous framework of the nose. (b) Skull in left lateral view. Drawings based on OUVC 9559. Scale bars = 10cm. 41

1.7. Medial view of right side of sagittally sectioned head (a) and skull (b) of Alces alces (OUVC 9559) with nasal septum and vomer removed to show internal nasal structures. Scale bars = 10cm. 44

1.8. White-tailed deer (Odocoileus virginianus). Left lateral view of superficial (a) and deep (b) dissections showing musculature of the face and nostril. (c) Medial view of right side of sagittally sectioned head with nasal septum and vomer removed to show internal nasal structures. Skull in left rostrodorsolateral view with nasal cartilages intact (d) and in left lateral view (e). Scale bars = 5cm. 52

1.9. Cladogram of Odocoileus virginianus (a), Cervalces scotti (b), and Alces alces (c) and their skulls in left lateral view to illustrate transformation of the bony naris. Skull in (b) redrafted from Scott (1885). 57

2.1. Left lateral views of reconstructions of AMNH 202492 using Amira. (a) Lateral view of isosurface of the intact head. (b) Voxel reconstruction of intact head to simulate a lateral radiograph. (c) skull isosurface. Scale bars = 5cm. 66

2.2. Phylogenetic relationships of taxa and clades referred to in this study. Topology based on Hassanin & Douzery (2003). 69

2.3. (a) Right medial view of Amira-generated isosurface of AMNH 202492. (b) Stereopairs of specimen in (a). Scale bars = 5cm. 71

2.4. Drawings of selected computerized tomographic (CT) images of the face of AMNH 202492 showing narial structures in successive transverse sections (a-i). (j) Skull in left lateral view to show the rostrocaudal levels of sections depicted in (a-i). Scale bars = 5cm. 73

2.5. (a) Left lateral view of Amira-generated isosurface of skull of AMNH 202492. (b) Stereopairs of specimen in (a). Scale bars = 5cm. 77

2.6. (a) Right medial view of Amira-generated isosurface of skull of AMNH 202492. (b) Stereopairs of specmen in (a). Scale bars = 5cm. 82

3.1. Skulls (left) and reconstructions (right) of extinct, putative, proboscis- bearing taxa. (a) Diprotodon, a marsupial. (b) Moeritherium, a basal proboscidean. (c) Glyptodon, a xenarthran. (d) Astrapotherium, an astrapothere. (e) Homalodotherium, a notoungulate. (f) Pyrotherium, a pyrothere. (g) Theosodon, a liptoptern. Skulls in (a), (b), and (g) from Carroll (1988). Skull and reconstruction in (c) from Gillette and Ray (1981). Skulls in (d) and (e) from Riggs (1935, 1937). Skull in (f) from Colbert et al. (2001). Reconstructions in (a), (b), (d), (e), (f), and (g) from Dixon et al. (1993). 95

3.2. Cladogram of proboscis-bearing mammals. * indicates taxa for which there is description of narial anatomy. Overall topology from Novacek (1993). Topology for Phocidae from Bininda-Emonds, Gittleman, & Purvis (1999). Topology for Ruminantia from Hassanin & Douzery (2003). 97

3.3. Skulls (left) and facial anatomy (right) of extant proboscis-bearing taxa. (a) Sus scrofa (from Dyce, Sack, & Wensing, 1987). (b) Tapirus terrestris (from Witmer et al., 1999). (c) Alces alces (from Clifford & Witmer, in review). (d) Cystophora cristata (original drawing by Ryan Ridgely, Ohio University). (e) Saiga tatarica and (f) Madoqua guentheri (from Frey & Hofmann, 1997). 99

3.4. Exaptation of maxillolabial musculature (mmax) in perissodactyls. Horses (a) and tapirs (b) have separated the origins of maxillolabial musculature, whereas these muscles primitively share a common origin. Modified from Boas & Paulli (1908a). 102

3.5. Nasal cartilages in a generalized phocid (a) and a hooded seal (b) showing elaboration of mobile lateral accessory cartilages. Modified from Brønsted (1932). 103

3.6. Nasal cavity of moose (a), tapir (b), elephant (c), and saiga (d) to illustrate rotation of the maxilloturbinate (mt) out of the main airflow through the nasal cavity resulting from vestibular enlargement. (a) from Clifford & Witmer (in review). (b) and (c) from Boas & Paulli (1908a). (d) from Clifford & Witmer (in prep). 105

3.7. Figure 3.7. Testing rules of construction in Phocidae. Determination of common anatomical substrates between taxa in the clade shown in dotted line, and hypothesis testing between proboscis-bearing taxa shown in light arrows. Topology from Bininda-Emonds et al. (1999). Modified from Witmer (1995a). 112

List of Abbreviations

ac cartilage of the alar fold (Fig. 1.5) af a. facialis (Figs 1.3 &1.4) ai a. infraorbitalis (Fig. 1.4) bo bulbus oculi (Figs 1.5 & 2.4) ci canalis infraorbitalis (Figs 1.5, 2.4 & 2.6) cnd concha nasalis dorsalis (Figs 1.7, 1.8, 2.3 & 2.4) cne conchae nasalis ethmoidale (Figs 1.7, 1.8, 2.3 & 2.4) cnl canalis nasolacrimalis (Fig. 2.4) cnla cartilago nasi lateralis accessoria (Figs 1.5, 1.6 & 1.8) cnld cartilago nasi lateralis dorsalis (Figs 1.5, 1.6, 1.8 & 2.4) cnlv cartilago nasi lateralis ventralis (Figs 1.6 & 1.8) cnm concha nasalis media (Figs 1.7, 1.8, 2.3 & 2.4) cns cartilago nasi septalis (Figs 1.5 & 1.8) cnv concha nasalis ventralis (Figs 1.5, 1.6, 1.8, 2.3 & 2.4) csn cartilago septi nasi (Fig. 2.4) csnd cartilago septi nasi, processus dorsalis (Fig. 2.4) csnv cartilago septi nasi, processus ventralis (Fig. 2.4) dei dentes incisives (Fig. 1.5) dnp ductus nasopharyngeus (Figs 2.3 & 2.4) dp ductus parotideus (Fig. 1.3) fin foramen incisivum (Fig. 1.7) fnm fissura nasomaxillaris (Fig. 1.5) gbd glandulae buccalis dorsalis (Fig. 1.4) gpo glandula preorbitale (Fig. 2.4) l lingula (Figs 1.5 & 2.4) lec lateral expansion of cnld (Figs 1.5 & 1.6) ln limen nasi (Fig. 1.7) mb m. buccinator (Figs 1.3, 1.4 & 1.8) mc m. caninus (Figs 1.3 & 1.4) mdli m. depressor labii inferioris (Fig. 1.3) mdli´ m. depressor labii inferioris, cut edge (Fig. 1.4) mdls m. depressor labii superioris (Figs 1.3 & 1.4) mdna m. dilator naris apicalis (Figs 1.3 & 1.8) mdna´ m. dilator naris apicalis, cut edge (Fig. 1.4) mdnm m. dilator naris medialis (Figs 1.3 & 1.4) mii m. incisivus inferioris (Fig. 1.4) mis m. incisivus superioris (Fig. 1.7) mlls m. levator labii superioris (Figs 1.3 & 1.4) mllst m. levator labii superioris tendons (Fig. 1.3) mln m. lateralis nasi (Figs 1.3, 1.4 & 1.8) mlnc m. lateralis nasi, portio caudalis (Fig. 1.3) mlnl m. levator nasolabialis (Figs 1.3, 1.4 & 1.8) mm m. malaris (Figs 1.3, 1.4 & 1.8) mmax maxillolabial muscle group (Fig 1.8) mmax´ maxillolabial muscle group, cut (Fig. 1.4) mms m. masseter (Figs 1.3 & 1.5)

mn m. nasalis (Figs 1.4 & 1.8) mnc meatus nasi communis (Fig. 1.5) mnd meatus nasi dorsalis (Figs 1.5, 1.7, 1.8 & 2.3) mnm meatus nasi medialis (Figs 1.5, 1.7, 1.8 & 2.3) mnpa m. nasalis pars alaris (Figs 1.3 & 1.4) mnv meatus nasi ventralis (Figs 1.5, 1.7, 1.8 & 2.3) mooc m. orbicularis oculi (Figs 1.3, 1.4 & 1.8) moor m. orbicularis oris (Figs 1.3, 1.4 & 1.8) mp mystacial pad (Fig. 1.8) mrn m. rectus nasi (Fig. 1.4) mz m. zygomaticus (Figs 1.3 & 1.8) mz´ m. zygomaticus, cut edge (Figs 1.4 & 1.8) n naris (Figs 1.5 & 2.4) ni n. infraorbitalis (Figs 1.4 & 1.5) ocn ostium canalis nasolacrimalis (Fig. 1.7) oee os ethmoidale, ethmoturbinate (Figs 1.5 & 1.7 & 2.6) oee´ os ethmoidale, ethmoturbinate underlying cnm (Fig. 1.7) oelb os ethmoidale, lamina basalis (Fig. 1.7) oelc os ethmoidale, lamina cribrosa (Figs 1.7 & 2.6) oelp os ethmoidale, lamina perpendicularis (Fig. 2.6) oen os ethmoidale, nasoturbinate (Figs 1.5 & 1.7) oent os ethmoidale, nasoturbinate (Figs 2.5 & 2.6) of os frontale (Figs 1.6 & 1.8) offs foramen supraorbitalis (Figs 1.6 & 1.8) ogln ostium glandulae lateralis nasi (Fig. 1.7) ol os lacrimale (Fig. 1.6) olfl os lacrimale, fossa lacrimale (Figs 1.6 & 1.8) ollf os lacrimale, foramen lacrimale (Figs 1.6 & 1.8) ollt lacrimal tubercle (Fig. 1.6) olm os lacrimale, tuberculum malaris (Fig. 2.5) olpn os lacrimale, processus nasolacrimalis (Fig. 2.5) olt os lacrimale, tuberculum (Fig. 2.5) om os maxilla (Figs 1.6-1.9 & 2.6) omcf os maxilla, crista facialis (Figs 1.6 & 2.5) omfi os maxilla, foramen infraorbitalis (Figs 1.6, 1.8 & 2.5) omit infraorbital tubercle (Fig. 1.6) ommi os maxilla, margo interalveolaris (Fig. 1.5) omn os mandibula (Figs 1.5 & 2.4) omt os maxilla, turbinate (Figs 1.5-1.7, 1.9, 2.5 & 2.6) omti os maxilla, tuberculum infraorbitale (Fig. 2.5) on os nasale (Figs 1.5-1.9, 2.4-2.6) onp os nasale, processus (Fig. 2.5) op os premaxillare (Figs 1.6-1.9) oppn os premaxillare, processus nasalis (Figs 1.5-1.9, 2.4-2.6) oppp os premaxilla, processus palantine (Figs 2.4 & 2.6) oppr os premaxilla, processus rostralis (Figs 2.4, 2.5 & 2.6) opt os premaxilla, tubercule (Fig. 1.6) orvl ostium recessus vestibularis lateralis (Figs 2.3 & 2.4)

14 ovn organum vomeronasale (Fig. 2.4) oz os zygomaticum (Fig. 1.6) pa plica alaris (Figs 1.5, 1.7, 1.8, 2.3 & 2.4) pa´ nostril connective tissue pad (Figs 1.3-1.5 & 1.7) pb plica basalis (Figs 2.3 & 2.4) pr plica recta (Figs 1.5, 1.7, 1.8, 2.3 & 2.4) pvl processus ventrolateralis of cnld (Figs 1.5 & 1.6) ra recessus naris alaris (Figs 1.5 & 1.7) rb n. facialis, ramus buccolabialis (Figs 1.3, 1.4 & 1.8) rmm n. facialis, ramus marginalis mandibulae (Figs 1.3 & 1.4) rnc recessus narialis caudalis (Fig. 1.5) rnr recessus narialis rostralis (Figs 1.5 & 1.7) rnv recessus nasi ventralis (Figs 2.3 & 2.4) rvl recessus vestibularis lateralis (Fig. 2.4) sm sinus maxillaris (Figs 1.5, 1.7, 2.4 & 2.6) sncc septum nasi, corpus cavernosum (Fig. 2.4) snpm septum nasi, pars membranacea (Fig. 2.4) vao v. angularis oculi (Fig. 1.4) vdn v. dorsalis nasi (Fig. 1.4) vf v. facialis (Figs 1.3, 1.4 & 1.8) vi v. infraorbitalis (Fig. 1.4) vln v. lateralis nasi (Figs 1.3, 1.4 & 1.8) vls v. labialis superioris (Figs 1.3 & 1.4) vn vestibulum nasi (Figs 2.3 & 2.4) vno organum vomeronasale (Fig. 1.5)

CHAPTER 1: THE ENIGMATIC NOSE OF MOOSE (ARTIODACTYLA:

CERVIDAE: Alces alces)

Abstract

The facial region of moose (Alces alces) is highly divergent relative to other cervids and other

ruminants. In particular, the narial region forms an expanded muzzle or proboscis that

overhangs the mouth. The nose of moose provides a case study in the evolution of narial

novelty within a phylogenetically highly-constrained group (Cervidae). The function of the nasal apparatus of moose remains enigmatic, and we propose new hypotheses based on our anatomical findings. Head specimens of moose and outgroup taxa were subjected to medical imaging (CT scanning), vascular injection, gross anatomical dissection, gross sectioning, and skeletonization. Moose noses are characterised by highly enlarged nostrils

accompanied by specialised musculature, expanded nasal cartilages, and an increase in the

connective-tissue pad serving as the termination of the alar fold. The nostrils are widely

separated, and the rhinarium that encircles both nostrils in outgroups is reduced to a tiny

central patch in moose. The dorsal lateral nasal cartilage is modified to form a pulley

mechanism associated with the levator muscle of the upper lip. The lateral accessory nasal

cartilage is enlarged and serves as an attachment site for musculature controlling the aperture

of the nostril, particularly lateralis nasi, apical dilators, and rectus nasi. Despite reduction in

bony support for narial structures, moose show greatly enlarged nasal cartilages, and the

entire osseocartilaginous apparatus is relatively much larger than in outgroups. The nasal

vestibule of moose is very large and houses a system of three blind recesses: one rostral and

one caudal to the nostrils, and one associated with the enlarged fibrofatty alar fold. As a

result of the expanded nasal vestibule, osseous support for the nasal conchae (i.e., turbinates)

16 has retracted caudally along with the bony nasal aperture. The nasoturbinate and its mucosal counterparts (dorsal nasal concha and rectal fold) are reduced. The upturned maxilloturbinate, however, is associated with a greatly enlarged ventral nasal concha and alar fold. Moose are the only species of cervid with these particular characteristics, indicating that this anatomical configuration is indeed novel. Although functional hypotheses await testing, our anatomical findings and published behavioural observations suggest that the novel narial apparatus of moose probably has less to do with respiratory physiology than with functions pertaining specifically to the nostrils. The widely separated and laterally facing nostrils may enhance stereolfaction (i.e., extracting directional cues from gradients of odorant molecules in the environment), but other attributes of narial architecture (enlarged cartilages, specialised musculature, blind recesses, fibrofatty pads) suggest a mechanical function, namely, an elaborated nostril-closing system.

INTRODUCTION

Modern moose, Alces alces, the largest living member of Cervidae, are characterised by uniquely palmated antlers and a distinctively enlarged muzzle (Nowak, 1999; Fig. 1.1). As many as six subspecies of moose have been recognised, reflecting the broad range of habitats this species occupies (Franzmann, 1981). Moose (also known as elk in Europe) are now recognised as a single species, and subspecies ranks have been abandoned in favor of recognition of somewhat distinct ecotypes, differing in features such as body size, antler size

and shape, and colouration (Geist, 1999).

Alces is a member of the New World deer clade Odocoileinae, and as such, possesses

a telemetacarpal structure in the feet (Webb, 2000; Fig. 1.2). However, they are believed to

have originated from a group separate from other New World deer given their shortened

neck (Scott, 1885), hair structure (Peterson, 1978), horizontally extending palmate antlers

(Churcher & Pinsof, 1988), and vomerine septum (Groves & Grubb, 1987). Alces is a relatively young taxon, first appearing in the fossil record in middle-to-late Villafranchian deposits (approximately 1.5 million years ago) in Europe (Franzmann, 1981). Although other members of Alcini crossed from Eurasia to North America before the appearance of

Alces (e.g., Cervalces, occurring as far back as Kansan deposits; Churcher & Pinsof, 1988),

Alces alces apparently did not arrive in North America until the late Wisconsinian

(Franzmann, 1981). Some authors, however, believe that there may have been two separate crossings of modern moose into North America (Peterson, 1978).

Although their range has been much reduced in the face of human incursion into their native habitats (Heptner, Nasimovich & Bannikow, 1988), moose live throughout northern Eurasia and northern North America, occupying taiga and boreal forests. Some

Figure 1.1. Lateral view of male moose (Alces alces) in velvet (a) and lateral view of male moose head (b) showing unique muzzle. (a) Courtesy of M. Reichmann; used with permission. (b) Courtesy of G. and B. Corsi and the California Academy of Sciences; used with permission.

previously unoccupied North American areas, such as Rocky Mountain Wyoming and Idaho and Newfoundland have seen thriving human-augmented moose populations in the past 150 years (Karns, 1997). Moose are naturally limited in the south by habitats that exceed 27°C for long periods and do not provide adequate access to water and shade (Franzmann, 1981)

Figure 1.2. Phylogenetic relationships of taxa and clades referred to in this study. Topology based on Novacek, Wyss & McKenna (1988) and Groves & Grubb (1987).

and in the north by snow depths exceeding about 1m (Geist, 1999). They do not form herds

or permanent groups, preferring habitats with access to abundant water and medium to

dense cover (Peterson, 1978). Moose are selective browsers, or concentrate selectors

(Hofmann, 1989), although their diet changes considerably during the year. In winter,

moose subsist on tree bark, young branches, and lichens while they minimise the energy

spent on foraging to conserve energy stores (Geist, 1999). During the northern growing

season, moose forage for young branches of early successional deciduous trees, herbs, forbs,

and aquatic plants, feeding often during daytime prior to the mating season (Peterson, 1978).

At this time, moose never venture far from a source of aquatic vegetation, spending much of

their feeding time partially or even completely submerged. Moose attain their largest size in

forests where recent fires have given way to early-successional boreal vegetation (Geist,

1999). They are excellent exploiters of new environments, and although they do not thrive

20 in mature habitats, they are often among the first species to establish in a new taiga or boreal habitat after a disturbance (Peterson, 1978; Geist, 1999).

The distinctive skulls and antlers of moose have been examined in numerous accounts (e.g., Scott, 1885; Folkow, 1952; Peterson, 1978; Geist, 1999). Beyond their antlers, the elongate fleshy muzzle of moose is probably their most remarkable attribute (Fig. 1.1).

Very few studies, however, have been carried out on the soft-tissue cephalic anatomy in moose. Boas & Paulli (1908) included moose in their seminal work comparing the facial musculature of several different mammals to that of elephants. Meinertz (1955) carried out a detailed study of cranial muscles and their innervations in moose. Both of these studies, however, focused on musculature and lacked reference to osteological, vascular, and other soft-tissue features of the narial regions in moose. This study attempts to analyse the functional anatomy of the narial region in moose in reference to key differences from outgroups and with a mind on the enigmatic function of the nose of moose.

This research is part of a larger study attempting to describe the functional anatomy of unusual narial structures in modern amniotes (Witmer, Sampson & Solounias, 1999;

Witmer, 2001a, b; Clifford & Witmer, 2001, 2002a, b, 2003, in prep.). Peculiar narial structures are the product of natural selection acting upon a set of anatomical substrates, and this project attempts to determine how these differences result in causally-associated bony features of apomorphic taxa. A goal is to assess the presence or absence of novel soft-tissue structures in extinct taxa by appeal to the osteological correlates of these soft tissues as determined by their anatomy in extant taxa. Soft-tissue structures can be inferred in fossil taxa based upon common osteological correlates for certain structures shared between extant and extinct taxa (Witmer, 1995a), and ultimately this study seeks to determine how the

functional anatomy of modern moose may bear on the reconstruction of soft-tissue narial structures in related fossil taxa.

MATERIALS AND METHODS

Four head specimens of Alces alces supplied by the Department of Natural Resources (DNR)

in Newfoundland, Canada, were the primary source of data in this study. All four were wild

animals that were killed by accidental collision with vehicles; two heads show evidence of

final euthanasia by gun shot by DNR officials. These specimens include a bull, a cow, a

male calf, and a female calf (OUVC 9559, 9560, 9561, 9742, respectively). Two additional

skulls (AMNH 207705, OUVC 9587) were examined prior to work on the study specimens

in order to reveal modified bony narial structures. To determine anatomical structures in an

outgroup, specimens of Odocoileus virginianus (white-tailed deer) of various ages and sexes

(OUVC 9471, 9540, 9542, 9543, 9544, 9551, 9552, 9554, 9555, 9577, 9589, 9691, 9702, 9705, and 9742) were examined either as intact heads or as skulls.

Three moose (OUVC 9742, 9559, 9560) and one Odocoileus doe (OUVC 9741) were

subjected to X-ray computed tomography (CT) at O’Bleness Memorial Hospital in Athens,

Ohio, prior to dissection. Additionally, the arterial system of one moose cow (OUVC 9741)

was injected with a radio-opaque barium/latex solution (per Sedlmayr and Witmer, 2002)

and CT scanned again to visualise arterial branches occurring within and outside the nasal

vestibule. Please see the online supplemental information

(http://digimorph.org/private/Alces_alces/) for details on scanning protocols, as well as

movies and animations of the CT data. The carotid arteries were injected while the occipital

arteries were occluded to reduce extravasation of injection medium from bullet wounds in

the cranial cavity. Both sectional anatomy and skeletal/arterial reconstructions were

analysed in this specimen.

All moose study specimens were dissected, sagittally sectioned, and skeletonized using Terg-

A-Zyme (Alconox, Inc.) in lightly boiling water. Dissections were recorded with digital

photography and videotaping. The nasal cartilages of the cow and bull were removed, fixed

in 10% neutral buffered formalin, and preserved in 70% ethanol. Outgroup specimens were

obtained either as skulls or as heads and then dissected, subjected to various kinds of

sectioning (coronal or sagittal), and ultimately skeletonized. Veterinary anatomy texts (Getty,

1975; Nickel et al., 1986; Schaller, 1992) and Nomina Anatomica Veterinaria (NAV; 1994)

were used to standardise nomenclature of anatomical structures.

RESULTS

External anatomy

The enlarged nose of moose is characterised by a fleshy muzzle and mobile upper lip that,

when relaxed, sags over the mouth. Rostrally, the muzzle takes on a flat shape, contoured

only by the rostralmost portion of the nasal septum and the bulging lateral cartilages. In

other cervids, the rhinarium covers much of the front of the nose, usually encircling both

nostrils, but moose are highly divergent in this regard. The rhinarium of moose is a small

triangular hairless patch of skin on the external surface of the nasal septum occupying

approximately 5 cm2 in adults. The ventral apex of the rhinarium is continuous with the median upper lip cleft. The nostrils in moose are displaced such that they face almost entirely laterally (Fig. 1.1). Rostrally, the nostrils have a circular orifice bounded by skin with

short dark hairs. As the nostril extends caudally, it takes on a slit-like shape for two-thirds of

its length. The dorsal portion of the fleshy nostril is supported internally by the rostralmost

23 extent of the alar fold (plica alaris), and a large fatty connective tissue pad here contributes to the slit-like shape of the caudal portion of the nostril. Within this terminal pad in the alar fold, the lateral accessory cartilages can be palpated, sending a thin process rostrally. When at rest, only the circular orifice of the nostril appears to be open, and this orifice is at a level below the occlusal plane and is displaced laterally.

The nasal cavity of moose is an extremely long structure. Bubenik (1997) reported the length of the face in moose to be 70% of total head length, largest among any cervid.

This number agrees well with our measurements of the nasal cavity which, as measured from cribriform plate to the tip of snout (see Witmer et al., 1999), accounts for 65% of total head length. This elongate nose is well-supported by extensive nasal cartilages. The upper lip and nostrils are very mobile, indicative of a herbivore specialised for browsing (Hofmann, 1989).

Aside from the enlarged and modified nostrils, the upper and lower lips in moose also contribute to the increased length rostral to the orbit.

Integument

Moose are generally characterised as being covered in thick, dense fur. On the underside of the head, the fur is indeed thick, and the underlying skin approaches 3-4 mm in thickness.

The skin overlying the nose and muzzle is much thinner, and the fur is much shorter.

Although the skin remains fur covered on the entire exterior of the head, the fur on the nose is less than 5mm thick. In comparison, the fur of the bell on males may approach 10-12cm.

The skin surrounding the nostril remains completely fur covered. In fact, the first few centimeters within the nasal vestibule are invested with hair-lined skin. One conspicuous feature missing from moose is a discrete mystacial pad. In many mammals, including the

Odocoileus examined here, there is a dense subcutaneous connective tissue pad caudal to the

nostrils containing follicles of thick bristles (vibrissae). A tightly packed mystacial pad is

apparently replaced in moose by a series of longer hairs that grow directly away from the

skin at regular intervals along the dorsal and lateral aspects of the nose.

Connective tissue pad

At the termination of the alar fold of the nasal vestibule and forming the caudodorsal aspect of the fleshy nostril there is a fatty connective tissue pad (Figs 1.3-1.5 & 1.7; pa´). At the

nostril, the pad is directed mediolaterally, but, traveling caudally within the vestibule, the pad

is directed dorsoventrally. Also traveling caudally, the mucosa of the pad changes (i.e., at the

limen nasi) from a pigmented skin covered thinly in fur to the skinless moist mucosa

characteristic of the rest of the nasal cavity and vestibule. The pad is supported ventrally by

cartilage, and dorsally and medially the pad moves freely within the enlarged nasal vestibule.

Musculature

Boas & Paulli (1908) grouped mammalian facial muscles on anatomical grounds whereas

Huber (1930) used innervation (i.e., branches of the facial nerve, CN VII). These two groupings coincide almost entirely, and we have kept the groupings for organisational

purposes. Nomenclature of the muscle groups follows Boas & Paulli (1908) whereas

nomenclature of individual muscles largely follows the NAV (1994).

Orbicularis oculi group

M. orbicularis oculi (Figs 1.3 & 1.4; mooc)—The fibers encircling the orbit superficial to the

protruding orbital margin belong to orbicularis oculi. This muscle attaches on a tubercle on

the lacrimal bone at the margin of the rostromedial angle of the orbit. The fibers of

Figure 1.3. Superficial dissection of the face of Alces alces shown in (a) left lateral view and in (b) oblique left rostrodorsolateral view, based on OUVC 9559. Scale bars = 10cm.

Figure 1.4. Deep dissections of the face of Alces alces. Maxillolabial muscles are intact in (a). In (b), the maxillolabial muscles have been reflected to reveal underlying structures, based on OUVC 9559. Scale bars = 10cm.

27 orbicularis oculi then fan out to completely surround the orbit and the bones comprising the orbital margin. The orbital portion (M. orbicularis oculi pars orbitalis) forms a nearly complete circle around the orbit, except for its brief tendinous origin and insertion on the lacrimal bone. The palpebral portion (M. orbicularis oculi pars palpebralis) runs parallel to the margin of the eyelids, attaching to the palpebral ligaments medially and laterally. The action of this muscle is to approximate the eyelids and hence narrow the palpebral fissure

(i.e., close the eye).

M. levator nasolabialis (Figs 1.3 & 1.4; mlnl)—Originating from a connective tissue raphe along the midline of the head rostral to the orbit, levator nasolabialis sends fibers laterally and ventrally. This muscle is very thin and lies just deep to the skin for about two-thirds the length of the nose, and no subcutaneous structures pass superficial to this muscle. As the muscle courses along the nose laterally, its rostrocaudal dimension decreases such that its widest point occurs at its origin and its narrowest point occurs at its insertion. Rostrally, the fibers form an arc roughly parallel to the extent of the lateral accessory cartilages and maintain an even distance from the nostril. Caudally, the muscle follows a path parallel to a line from the rostral margin of the orbit to the corner of the mouth. Levator nasolabialis inserts along the upper lip, attaching to the skin and interdigitating with fibers of other muscles around the mouth. Thus, the muscle sends fibers from a large attachment along the dorsal midline of the nose to converge on a much smaller area just dorsal to the margin of the upper lip. Levator nasolabialis, as its name suggests, elevates the upper lip, especially its caudal portions.

M. malaris (=M. præorbicularis [Boas & Paulli, 1908], M. retractor anguli oculi medialis profundus [Meinertz, 1955]) (Figs 1.3 & 1.4; mm)—Malaris is a thin muscle sharing an origin with orbicularis oculi on a tubercle of the lacrimal bone at the rostromedial margin

of the orbit. The muscle then fans out ventrally, widening as it courses to its insertion as a

series of interdigitations with buccinator caudal to the corner of the mouth. It lies deep to

levator nasolabialis near the lacrimal tubercle and deep to zygomaticus, the facial vein, and the facial nerve ventrally. The fibers forming the caudal portion of the muscle are parallel to the ventralmost fibers of orbicularis oculi. Malaris is an elevator of the corner of the mouth and of the lateral walls of the oral vestibule (i.e., cheeks).

Maxillolabialis group (=M. maxillolabialis [Boas & Paulli, 1908, Meinertz, 1955])

M. levator labii superioris (Figs 1.3 & 1.4; mlls)—The rostralmost member of the maxillolabialis group, this muscle originates from a common origin for the group at a point just caudodorsal to the infraorbital foramen on the lateral aspect of the maxilla. A tubercle here serves as the origin for the group, and the levator labii superioris occupies its dorsalmost portion. From this origin, the muscle courses, deep to levator nasolabialis, rostrally and slightly dorsally, the fibers uniting into a single tendon approximately halfway between its origin and the caudalmost aspect of the nostril. Its tendon then passes rostrodorsally over the hump formed by the lateral nasal cartilages, using the hump as a pulley. Here, on the dorsal aspect of the nose deep to levator nasolabialis, the tendon splits into several smaller tendons that ultimately insert on the dorsal aspect of the nostril and the flattened muzzle between the nostrils. Small muscle slips from levator nasolabialis insert on the tendons of levator labii superioris, and the tendons themselves sometimes possess small muscle bundles rostral to the main muscle bundle. Considerable overlap occurs between contralateral muscle tendons of levator labii superioris between the nostrils at the continuation of the dorsomedian raphe from which the levator nasolabialis originates.

Bilateral contraction of levator labii superioris results in elevation and eversion of the upper

29 lip and dorsal aspect of the nostril, rostral to the point of support by the lateral nasal cartilages. Unilateral contraction obliquely elevates the ipsilateral lap and elevates the dorsal portion of the nostril.

M. caninus (Figs 1.3 & 1.4; mc)—This muscle originates from the common origin of the maxillolabial muscles, namely, a tubercle on the lateral aspect of the maxilla just caudodorsal to the infraorbital foramen. Sending out two or three muscle bundles, caninus courses more or less directly rostrally from its origin. Each bundle of the muscle ends in a tendon at different rostrocaudal points. The differentiated tendons ultimately insert on the caudal aspect of the nostril and onto the lateral accessory cartilages caudal to the nostril.

The action of caninus is to retract the nostril and accessory cartilages caudally.

M. depressor labii superioris (Figs 1.3 & 1.4; mdls)—The third member of the maxillolabialis group originates on the ventral portion of the common origin of the group.

As with levator labii superioris, depressor labii superioris sends a single large muscle belly rostrally. Taking a course more ventral than any other maxillolabial muscle, depressor labii superioris unites into a single tendon more rostrally than any of the others, just caudal and ventral to the nostril. This single tendon courses rostrally ventral to the nostril and sends several small tendons dorsally to the ventralmost portion of the nostril. It ultimately ends in an area of the upper lip ventral to the rostralmost portion of the nostril. These small tendons serve as the point of attachment of some fibers from lateralis nasi, in much the same way that the tendons of levator labii superioris received inserted fibers from levator nasolabialis. The action of this muscle is to depress and retract the ventral aspect of the nostril and parts of the upper lip. Taken together, the maxillolabial muscles overlie the contents of the infraorbital foramen and lie superficial to the deepest members of the buccinator group. They are all retractors of the nostril and upper lip and, depending on the

angle of the muscle, have either a dorsal, ventral, or caudal action on the upper lip and

nostril.

Buccinator group

M. buccinator (Figs 1.3 & 1.4; mb)—The buccinator complex in moose consists of at least

three, and possibly four, layers. Papp (2000) described buccinator of moose and other deer

in detail, describing the layers of the buccinator and their attachments. Only those parts

relevant to narial anatomy are discussed here. The superficial portion of buccinator (M. buccinator pars buccalis [NAV, 1994], M. buccinator pars superficialis [Boas & Paulli, 1908],

M. buccinatorius pars mandibulo-maxillaris superficialis [Meinertz, 1955]) is a thin sheet of muscle passing dorsoventrally between its attachments to the maxillary and mandibular alveolar processes and the maxillary tubercle. This superficial sheet extends rostrally to the corner of the mouth, where it interdigitates with orbicularis oris, and caudally to the ramus of the mandible. The buccal salivary glands are abundant deep to this sheet. Malaris interdigitates with buccinator pars buccalis near its origin at the maxillary alveolar process, especially at the caudal end of the sheet. Rostrally, near the corner of the mouth, the insertions of zygomaticus and depressor labii inferioris converge upon the rostralmost fibers of the pars buccalis, making an area taken up by interdigitations of muscles coming from three different directions. This portion of buccinator acts to compress the oral vestibule by apposing the cheek to the molar teeth.

The deep portion of buccinator (M. buccinator pars molaris [NAV, 1994], M. buccinator pars profundus [Boas & Paulli, 1908], M. buccinatorius pars mandibulo- maxilllaris profundus [Meinertz, 1955]) is thicker than the superficial portion, sending fibers in a rostrocaudal direction from the angle of the mouth to the ramus of the mandible

between the alveolar processes of the maxilla and mandible. Caudally, the pars molaris

interdigitates with intermediate layers of buccinator. Rostrally, the pars molaris interdigitates

slightly with the common insertions of zygomaticus, depressor labii inferioris, and the

caudalmost fibers of orbicularis oris. The action of buccinator pars molaris is to retract the

angle of the mouth and, along with the pars buccalis, compresses the oral vestibule.

M. orbicularis oris (=M. buccinator pars rimana [Boas & Paulli, 1908], M.

buccinatorius pars oris [Meinertz, 1955]) (Figs 1.3 & 1.4; moor)—Orbicularis oris originates

from the interdigitating intersection of buccinator, zygomaticus, and depressor labii

inferioris. From this point at the corner of the mouth, the muscle completely encircles the

mouth within the upper and lower lips. Caudally, at the corner of the mouth, orbicularis oris

lies deep to other muscles, and the dorsal labial vessels lie on the dorsal aspect of the muscle along the lips. On the lateral face of the muscle is a series of interdigitations with the

insertion of levator nasolabialis and lateralis nasi. Orbicularis oris acts as a sphincter of the

mouth.

M. incisivus superioris (=M. buccinator pars supralabialis [Boas & Paulli, 1908]) (Fig.

1.7; mis)—Incisivus superior is a thick, extensive, fan-like muscle arising from the

rostralmost processes of the premaxillae. It radiates out from this point to insert on the

upper lip, the connective tissue pad between the nostrils, the underside of the rhinarium, and

the septal cartilage. The muscle is best seen in sagittal section, occupying almost entirely the

underside of the muzzle between the dorsalmost rostral septum and the upper lip. The

principle action of the muscle is to elevate and evert the upper lip and to retract the muzzle.

M. incisivus inferior (Fig. 1.4; mii)—Incisivus inferior arises from the mandibular

symphysis and fans out to insert on the lower lip just ventral to the orbicularis oris muscle.

It is less conspicuous than incisivus superior, yet it performs a similar (albeit opposite)

function by depressing and curling the lower lip.

M. depressor labii inferioris (Figs 1.3 & 1.4; mdli)—The inferior labial depressor is a

band of muscle originating from the fascia of the neck and partly from the platysma muscle.

From this origin, it travels along the mandible, overlying the attachment of buccinator below

the tooth row. Its insertion interdigitates with buccinator, zygomaticus, and superficial

portions of the orbicularis oris in addition to sending some fibers rostrally from this point to

interdigitate with more rostral fibers of orbicularis oris to insert on the skin of the lower lip.

As its name suggests, it is a depressor of the corner of the mouth and the lower lip.

M. lateralis nasi (=M. naso-labialis dorsalis profundus [Meinertz, 1955]) (Figs 1.3 &

1.4; mln)—This muscle arises from the lateral aspects of the premaxillae ventral and slightly

caudal to the nostrils. It lies deep to the levator nasolabialis caudally and some portions of

the orbicularis oris rostrally. Interdigitating with orbicularis oris and the tendons of

depressor labii superioris, the muscle courses obliquely rostrodorsally to insert on the

underside of the nostril and the ventral aspects of the accessory cartilages. It is a thick,

relatively well-developed muscle that acts to depress the caudal and ventral portions of the

nostril, thus dilating them. Caudally, a thin sheet of lateralis nasi (M. lateralis nasi pars

caudalis; Fig. 1.3; mlnc) courses obliquely caudodorsally to insert on the caudolateral surface

of the lateral accessory nasal cartilages. This caudal portion is continuous with the rostral

portion at their skeletal attachments. The caudal portion, however, has a more superficial

attachment.

M. dilator naris apicalis (=M. lateralis nasi, M. transversus nasi [Boas & Paulli, 1908],

M. naso-labialis dorsalis profundus pars nasalis [Meinertz, 1955]) (Figs 1.3 & 1.4; mdna)—

Dilator naris apicalis is a strong muscle lying superficially between the nostrils and running

perpendicular to the superior incisive muscle at its dorsalmost extent. It originates in the

connective-tissue septum dividing the halves of the narial vestibules at the rostralmost

portion of the face. The muscle inserts onto the dorsal and slightly caudal portions of the

nostril. As the muscle makes its way toward its insertion, it thickens, resulting from lateral fibers coalescing before attaching onto the skin of the nostril. Dilator naris apicalis interdigitates caudodorsally with the elongate tendons of levator labii superioris. Dilator naris apicalis acts to pull the nostrils open, slightly rostrally, and dorsally.

M. nasalis (=M. naso-labialis dorsalis profundus [Meinertz, 1955]) (Figs 1.3 & 1.4;

mn)—Nasalis is a quadrangular, thick sheet of muscle lying deep to the maxillolabial muscles

ventrally and levator nasolabialis dorsally. The muscle arises from the interalveolar margin

along the maxilla and inserts on the underside of levator nasolabialis and on the fascia near

the midline of the dorsal surface of the head. The lateral and dorsal nasal veins irregularly

pierce this muscle before traveling caudally superficial to it. The rostralmost fibers of nasalis

(M. nasalis pars alaris, Figs 1.3 & 1.4; mnpa) interdigitate with the caudal portion of lateralis

nasi, and both muscles run in the same direction at this point. Nasalis is the deepest muscle

in the area of the nose between malaris and lateralis nasi, and vessels and nerves supplying

the narial musculature and skin are all superficial to nasalis at its caudal border.

M. dilator naris medialis (=M. nasolabialis [Boas & Paulli, 1908], M. naso-labialis

dorsalis profundus [Meinertz, 1955], M. lateralis nasi [Schaller, 1992]) (Figs 1.3 & 1.4;

mdnm)—Dilator naris medialis (so named in accordance with terminology in Nickel et al.,

1986, rather than function) is a well-developed muscle originating from the ventral and

caudal aspect of the lateral accessory nasal cartilages. From this origin, dilator naris medialis

courses first caudally and ventrally to the caudal margin of the nostril, where it turns to travel

in a rostrodorsal direction. This muscle inserts on the dorsal portion of the nostril, between

34 dilator naris medialis and the fibrofatty termination of the alar fold. The fiber direction at its insertion is parallel to the fiber direction of the rostral fibers of levator nasolabialis, but the fibers of the medial dilator lie deep to the tendons of levator labii superioris. Based on its attachments, the function of this muscle is ironically that of a constrictor of the nostril (not a dilator, as its name suggests), as its fibers indicate the ability to pull the lateral accessory cartilages and the caudal portion of the nostril up against the fatty pad lying at the end of the alar fold of the nasal cavity.

Intrinsic musculature of the nose

M. rectus nasi (Boas & Paulli, 1908) (Fig. 1.4; mrn)—Along the nasal vestibule, there are transverse muscle fibers. Generally, rectus nasi is a diffuse muscle concentrated on the caudal and ventral portions of the nostril, extending from the mucosa of the nasal vestibule to the underside of the skin. The action of rectus nasi is presumably that of a dilator of the nasal passages.

Major nerves and vessels

The vasculature and innervation of major structures involved with the musculature of the nose of moose are summarized below. This list is not intended to be comprehensive, but major anatomical structures contributing to the function of the nose are included here.

Arterial supply of the nose and face

A. facialis (Figs 1.3 & 1.4; af)—The lingual and facial arteries arise from a common trunk from the internal carotid artery. The facial artery branches caudally from the linguofacial trunk to pass rostrally and ventrally around the edge of the mandible just rostral to the

35 attachments of masseter, where the artery turns dorsally to follow the rostral margin of masseter. The two major branches of the facial artery occurring caudal to the corner of the mouth are the superior and inferior labial arteries. The inferior labial artery supplies the lower lip as far rostrally as the midline of the lower lip. The superior labial artery courses parallel to its counterpart in the lower lip, supplying structures along the upper lip from the corner of the mouth to the midline. Several unnamed branches of the facial artery branch off at various points to supply structures ventral to the inferior labial artery and dorsal and caudal to the point where the superior labial artery courses into the upper lip. The facial artery is the major blood supply to muscles of the buccinator group of muscles, especially the caudal and ventral members.

A. infraorbitalis (Fig. 1.4; ai)—The infraorbital artery is a terminal branch of the sphenopalatine artery, itself a branch of the maxillary artery. The sphenopalatine artery branches into ophthalmic, septal, superior alveolar and infraorbital branches just caudal to the point where the infraorbital artery passes into the infraorbital canal. The infraorbital artery gives off no branches within the canal itself, remaining enclosed in the bony cavity coursing through the maxillary sinus. Rostral to exiting the infraorbital canal, the artery gives off numerous branches such as the dorsal nasal artery and lateral nasal artery that travel rostrally and dorsally along the exterior of the nasal vestibule, supplying muscles of the maxillolabialis group, ventral and rostral members of the orbicularis oculi group, and dorsal members of the buccinator group. Branches of the infraorbital artery follow the curvature of the nasal cartilages, and some rostral branches of the lateral nasal artery curve around the margin of the nostril to supply structures within the alar fold near the nostril.

A. sphenopalatina—The sphenopalatine artery gives off branches that supply structures on the dorsal aspect of the oral cavity and the ventral and lateral aspects of the

nasal cavity and vestibule. The palatine branches course rostrally through the soft palate and

the palatal rugae of the hard palate. Branches supplying the nasal vestibule spread out to

supply the mucosa lining of the ventral nasal concha. These branches appear to make

several anastomoses with each other while traveling in a mostly rostral direction

Venous drainage of the nose and face

V. facialis (Figs 1.3 & 1.4; vf)—The major vessel draining the facial musculature and exterior of the nasal cavity and vestibule is the facial vein. The vein is formed by a conspicuous

confluence just rostral to the margin of malaris of several branches draining the face. From

this point, the vein travels caudally and ventrally along the corner of the mouth and the

rostral edge of masseter. The facial artery and the facial vein travel parallel to each other at

this point and wind together underneath the mandible

V. labialis superior (Figs 1.3 & 1.4; vls)—The drainage of the upper lip is carried out

by the superior labial vein, which travels parallel to the superior labial artery along the outer

edge of the orbicularis oris muscle from about the midline of the nose to about halfway

along the cheek to drain into the facial vein.

V. lateralis nasi and V. dorsalis nasi (Figs 1.3 & 1.4; vnl & vdn, respectively)—The

dorsal and lateral nasal veins drain muscles and structures from the dorsal midline to the

upper lip and from the nostril to the formation of the facial vein superficial to the masseter.

Muscles of the orbicularis oculi group, maxillolabialis group, and buccinator group are

drained by these veins. There are several anastomoses along the midline between the

contralateral dorsal nasal veins. The lateral nasal vein also drains structures just inside the

nostril, such as the alar fold rostral to the limen, and one of its small tributaries wraps

around the nasal margin of the maxilla after anastomosing with vessels in the nasal vestibule.

V. angularis oculi (Fig. 1.4; vao)—The drainage of most of the muscles of the

orbicularis oculi group is carried out by the angularis oculi vein. This vein courses from the

formation of the facial vein along the rostral margin of malaris and enters the supraorbital

foramen just medial to the margin of the orbit in the frontal bone. From this foramen, the

angularis oculi vein communicates with ophthalmic vessels caudal to the orbit and ultimately with the internal jugular system at the cavernous sinus. This vein is an anastomosis between

superficial drainage of the nose with blood draining deeper structures of the cranium.

V. infraorbitalis (Fig. 1.4; vi)—The infraorbital vein drains structures deep to and

including the maxillolabialis group. It lies ventral to the lateral nasal vein and dorsal to the

superior labial vein, but it is deep to both of these veins. The infraorbital vein makes a more

or less straight line from the caudalmost margin of the fleshy nostril to the infraorbital

foramen, where it enters the infraorbital canal to join veins draining the nasal cavity.

Sensory innervation of the nose and face

N. infraorbitalis (Figs 1.4 & 1.5; ni)—The sensory innervation of the nose is carried by

branches of the trigeminal nerve (CN V), principally by the infraorbital nerve (CN V2). This

nerve travels in the infraorbital canal with the artery and vein of the same name and exits the canal with the artery. From the opening of the infraorbital canal, the infraorbital nerve gives off several branches that travel rostrally, fanning dorsally and ventrally to supply the skin from the infraorbital foramen rostrally to the midline of the nose. The nerve, artery, and vein traveling through the infraorbital canal are bound together by fascia once they are in the canal itself, and this fascia disappears to allow the vessels and nerves to travel separately rostral to the foramen (Fig. 1.5).

Figure 1.5. Drawings of selected computerized tomographic (CT) images of the face of Alces alces (OUVC 9559) showing narial structures in successive transverse sections (a-h). (i) Skull in left lateral view to show the rostrocaudal levels of sections depicted in (a-h). Scale bars = 5 cm.

Motor innervation of the nose and face

N. facialis (CN VII)—The motor nerve supply for all of the muscles of the face and nose is carried by the facial nerve. The nerve branches into a buccolabial branch (ramus buccolabialis, Schaller, 1992; dorsal buccal branch, Getty, 1975, in ruminants) and a marginal mandibular branch (R. marginalis mandibulae, Meinertz, 1955, and Schaller, 1992; ventral buccal branch, Getty, 1975). These branches split from each other behind the caudal

Figure 1.5. Continued. margin of masseter and rejoin rostral to the rostral border of masseter. Other branches, such as auricular, temporal, and zygomatic branches are not described here, as their contributions are minor in the facial region. Meinertz (1955) gave an extensive account of the course of the facial nerve and the muscles it innervates in the head. Many of our observations agree with those of Meinertz (1955), and the parts relevant to the current study are summarized here.

R. marginalis mandibulae (Figs 1.3 & 1.4; rmm)—The marginal mandibular ramus of the facial nerve branches caudal to masseter and courses ventrally along the caudal and

ventral margin of the muscle. As it follows the outline of the muscle, it supplies platysma,

depressor labii inferioris, ventral parts of the orbicularis oris, and incisivus inferior. At the

rostral limit of masseter, the marginal branch follows the facial artery and vein dorsally and

slightly rostrally to join up with the buccolabial branch once it exits its common sheath with

the parotid duct just rostral to the margin of the masseter.

R. buccolabialis (Figs 1.3 & 1.4; rb)—The buccolabial branch of the facial nerve courses laterally along masseter in a fascial sheath. A few smaller branches appear to break away from the buccolabial branch and join up with it a short time later, creating a small plexus of nerves along masseter and just rostral to its margin. Some of these smaller nerves supply zygomaticus and malaris. After rejoining the marginal branch, the buccolabial branch of the facial nerve then branches into dorsal and ventral buccolabial branches. The ventral buccolabial branch supplies superficial muscles along the upper lip, such as orbicularis oris and buccinator, and deeper muscles of the nose, such as nasalis. The dorsal buccolabial branch sends various branches to supply the muscles of the maxillolabialis group and the orbicularis oculi group except the caudoventral portion of orbicularis oculi and the dorsal portion of malaris, which receive their innervation from the R. zygomatico-orbitalis of the facial nerve, as also noted by Meinertz (1955). Further rostrally, around the nostril, branches of the facial nerve are difficult to isolate and definitively differentiate from branches of the infraorbital nerve.

Nasal cartilages

Cartilago nasi lateralis dorsalis (Figs 1.5 & 1.6; cnld)—The dorsal lateral nasal (parietotectal) cartilages of moose form an extensive framework for the support and attachment of muscles responsible for moving the nose. As a whole, the nasal cartilages of moose are elongated

Figure 1.6. Oblique left rostrodorsolateral view (a) of a skull of Alces alces with cartilages in place to show the cartilaginous framework of the nose. (b) Skull in left lateral view. Drawings based on OUVC 9559. Scale bars = 10cm.

both rostrally to accommodate the overhanging muzzle and caudally to assume much of the

support of narial structures that would normally be taken up by the skull bones comprising

the bony naris in many other mammals. Caudally, the dorsal lateral cartilage attaches to the

nasal and maxillary bones, extending directly rostrally at their attachment. The cartilage

expands laterally, creating a convex (with respect to the outside) hump supporting tendons

of levator labii superioris. At the rostral termination of the hump, the dorsal lateral cartilage

ends laterally in a prong of cartilage that extends ventrally and rostrally to support the dorsal

aspect of the caudalmost portion of the nostril. Ventral to the convexity, the dorsal lateral

cartilage receives the attachment of the lateral accessory nasal cartilage, the two being

connected by a mobile joint. Farthest rostrally, the dorsal lateral cartilage sends a median

process ventrally. This terminal process does not attach to the premaxilla or the ventral

lateral nasal cartilage as in other ruminants (Schaller, 1992). Rather, it ends just rostral to the

termination of the premaxilla, supporting the rostralmost structures of the nose, including a

blind rostromedial recess.

Cartilago nasi lateralis accessoria (Figs 1.5 & 1.6; cnla)—The lateral accessory cartilage of moose forms a rough triangle. The caudoventral apex of the cartilage attaches to the underside of the dorsal lateral cartilage, forming a movable joint that is capable of a considerable amount of excursion. The caudodorsal apex curves upward to almost meet the convexity of the dorsal lateral cartilages. The rostral apex of the accessory cartilages extends directly rostrally roughly parallel to the terminal process of the dorsal lateral cartilages.

These last two processes support the caudal end of the nostril, and movement of the lateral accessory cartilages facilitates the compression or dilation of the nostril.

Cartilago nasi lateralis ventralis (Fig. 1.6; cnlv)—The ventral lateral cartilage has an indefinite separation from the dorsal lateral cartilage. Both lateral cartilages attach to each other for a

43 considerable distance along the nose, but there is a clear distinction ventrally between structures belonging to the ventral lateral cartilage. The ventral lateral cartilage lines the bones underlying them, namely the premaxilla and maxilla, on the surfaces of these bones within the nasal cavity. Extending dorsally from this position, the ventral lateral cartilage supports a cartilaginous process (Fig. 1.5; ac) that sweeps upward to support the plica alaris and the fatty pad that blocks the entrance to a lateral recess in the nose. The cartilage supports this fatty pad only on its most ventral portions, leaving the pad apparently free to move mediolaterally.

Cartilago nasi septalis (Fig. 1.5; cns)—The septal cartilage of moose separates the halves of the nasal cavity rostrally as far as the termination of the premaxilla and caudally as far as the perpendicular plate of the ethmoid bone. At the rostralmost extent, the separation of the halves of the nasal cavity is carried out by mucosa extending between the rostralmost septal cartilage and the rostromedian process of the dorsal lateral cartilages. Between its rostral and caudal limits, the septal cartilage extends from the vomer bone dorsally to the dorsal lateral cartilages, being thicker at its base than at its apex.

Overview of nasal cavity

The nasal cavity in moose is composed chiefly of three mucosal conchae and the surrounding air spaces. The largest of these spaces is the ventral nasal meatus (Figs 1.5 &

1.7; mnv). This space communicates freely with the nostril rostrally and the choana ventrally.

Dorsal to this space is a greatly enlarged concha nasalis ventralis (Figs 1.5 & 1.7; cnv). The ventral nasal concha is supported in its caudal half by the maxilloturbinate (Figs 1.5-1.7; omt) and farther rostrally by the lateral nasal cartilages. Rostral to the maxilloturbinate, the ventral nasal concha is composed of the plica alaris (Figs 1.3-1.5 & 1.7; pa). The alar fold attaches to

Figure 1.7. Medial view of right side of sagittally sectioned head (a) and skull (b) of Alces alces (OUVC 9559) with nasal septum and vomer removed to show internal nasal structures. Scale bars = 10cm.

the lateral wall of the nasal vestibule through a robust mucosal and fibrofatty connection.

Caudal to the nostril, the alar fold changes its orientation from more or less directly medial

to rostromedial. At this point, the fibrofatty dorsal expansion of the alar fold forms the

boundary of a blind sac, termed here the recessus alaris (Figs 1.5 & 1.7; ra). Rostral and ventral to the dorsal fibrofatty pad of the alar fold, a ventral pad forms the rostralmost extension of the alar fold and the caudodorsalmost portion of the nostril.

Dorsal and caudal to the ventral nasal concha are the concha nasalis dorsalis (Fig.

1.7; cnd), concha nasalis media (Fig. 1.7; cnm), and conchae ethmoidales (Fig. 1.7; cne), which

are situated in a more or less doroventral series at the caudal end of the nasal cavity. The

space between the ventral and middle nasal conchae (meatus nasi medius, Figs 1.5 & 1.7;

mnm) extends for the entire rostral extent of the middle concha and the caudal half of the

maxilloturbinate-supported portion of the ventral concha. The ventral surface of the middle

nasal concha is open to the choana rostrally, and the ethmoid conchae are partially enclosed

by the lamina basalis of the ethmoid (Fig. 1.7; oelb) (transverse lamina of Moore 1981). The

dorsal nasal concha is much less complex than the other conchae, as it is only gently curved

ventrally and does not contain multiple scrolls. The dorsal nasal concha is supported

caudally by the nasoturbinate (Figs 1.5 & 1.7; oen) (endoturbinal I of Paulli, 1900), and

rostrally, it continues as the plica recta (Figs 1.5 & 1.7; pr). The rectal fold does not extend as far rostrally as the fibrofatty extensions of the alar fold, instead terminating just rostral to the mucosal ostium of the lateral nasal glands (Fig. 1.7; ogln). The middle nasal meatus is

between the dorsal and ventral nasal conchae rostral to the middle nasal concha. The middle

nasal meatus is continuous with the alar recess bounded by the fibrofatty dorsal extension of

the alar fold. The air passageway dorsal to the dorsal nasal concha (meatus nasi dorsalis, Figs

1.5 & 1.7; mnd) is widened rostrally between the cartilaginous roof of the nasal vestibule and

the rectal fold and becomes dorsoventrally constricted caudally ventral to the nasal bones.

The mucosa of the dorsal nasal meatus is olfactory at its caudal end and respiratory for the

rest of its rostral extent.

The nasal vestibule, the area of the nose where air first enters from the nostril, is

characterised by three distinct spaces. First, the largest space in the nasal vestibule—the

recessus narialis rostralis (Figs 1.5 & 1.7; rnr)—is a rostral expansion of the space dorsal and

46 rostral to the nostril opening. When the narial musculature is at rest, the rostral narial recess is minimised and results in the squared and broadened muzzle. Second, the blind sac dorsal to the alar fold—the recessus alaris—beginning at the caudodorsal margin of the nostril fills the nasal vestibule dorsally and caudally. It occupies the space created by the lateral expansion of the dorsal lateral nasal cartilages (i.e., the “hump” mentioned earlier that serves as a pulley for levator labii superioris tendons). This space is bordered rostrally, medially, and ventrally by the attachment of the fatty pad on the alar fold, laterally by the dorsal lateral nasal cartilages, and caudally by the ventral concha and the constriction at the entrance to the nasal cavity proper produced by the maxillae and premaxillae. Third, the recessus narialis caudalis (Figs 1.5 & 1.7; rnc) begins at the ventral portion of the nostril, ventral to the alar fold, and occupies the caudoventral portion of the nasal vestibule. This caudal narial recess is open to the circular portion of the nostril lined with hair and rostral to the fatty pad of the alar fold.

Anatomically, the main air passageway from the nostril to the choana is the ventral nasal meatus. All nasal meatuses can communicate with each other through the common nasal meatus (meatus nasi communis, Fig. 1.5; mnc), which is the open space between the cartilaginous septum and the conchae. The dorsal and middle conchae do not lie in the direct path of air passing through the nasal vestibule and conchae. Instead, they occupy a more caudodorsal position.

Glands

Moose and other deer are characterised by relatively large salivary glands to accommodate a browsing diet (Hofmann, 1989). As this study focused on the narial anatomy of moose, the only salivary glands approximating this area are the buccal glands associated with buccinator

(Papp, 2000). The dorsal group of buccal glands (Fig. 1.4; gbd) lie within and between successive layers of the muscle. Their presence is indicated by loose collections of rounded or oblong shaped collections of tissue that are smooth in texture and lighter in colour than the surrounding tissue. The ventral buccal glands are separated from their dorsal counterparts by about 2cm. The collections of glandular tissue are more discretely organised in the ventral glands, and their total mass is larger than in the dorsal glands. The saliva produced in these glands enters the oral vestibule through several minute ostia. Caudal to masseter, moose possess an enlarged parotid gland that also has been described elsewhere

(Hofmann, 1989). Although the gland does not approximate any of the other relevant tissues described here, the duct of the parotid gland (Fig. 1.3; dp) is associated with several facial structures. The duct first appears at the caudoventral portion of the gland and extends ventral to the angle of the mandible with the facial artery, facial vein, and marginalis mandibulae ramus. Rostral to masseter and caudal to the angle of the mouth, the parotid duct then courses rostrally to pierce the superficial part of buccinator and ultimately opens in the oral vestibule opposite the second upper molar. The lateral nasal glands are a diffuse series of glands that open via a common ostium within the middle nasal meatus near the rostral termination of the rectal fold.

Osteology

The skulls of moose are easily distinguishable from those of even closely related cervids by virtue not only of their size but also by the highly modified shape of the bony nasal aperture.

The aperture in moose is highly angled back toward the orbits, shortening the nasal bones and excavating the premaxillae and maxillae. Overall, the skull is easily classified as a cervid by virtue of the toothless premaxillae and large nasomaxillary fissure (Fig. 1.6; fnm). As this

study focuses on anatomical details of the nose in moose, only those facial skeletal features relevant to the nose are included here.

Os premaxilla (= os incisivum, NAV, 1994; intermaxillary, Boas & Paulli, 1908) (Figs

1.6 & 1.7; op)— The premaxilla in moose is fairly typical for a ruminant, i.e., it is thin and edentulous. Remaining shallow for its entire length, it possesses two caudally directed processes. The medial process contacts both the medial process of the contralateral premaxilla and the rostralmost extent of the vomer. Laterally, the nasal process (Figs 1.5-1.7; oppn) of the premaxilla, which contacts the nasal bones in most other mammals, forms the

margin of the bony naris up to about the rostral extent of the maxillary spine of the maxilla.

The nasal process flares laterally as it extends caudally along the bony narial aperture.

Between these two processes, an elongate incisive foramen (Fig. 1.7; fin) typical of many

ruminants conducts the incisive duct and the vomeronasal organ. The body of each

premaxilla contains two roughened tubercles (Fig. 1.6; opt) serving as the attachment of the

superior incisive muscle (rostrally) and lateralis nasi (laterally). The premaxillae are

discontinuous from each other rostrally, each sending processes directly rostrally that have

oval margins medially. Ventrally, the hard palate is continued rostrally by the premaxilla.

Os maxilla (Figs 1.6 & 1.7; om)—The largest and most robust skeletal component of

the narial anatomy of moose is the maxilla. Laterally, this bone is roughly triangular, with a

dorsal apex that contacts the nasal bone for its entire lateral suture, a caudal margin that

contacts the nasomaxillary fissure and the lacrimal and zygomatic bones, a ventral margin

containing six teeth in adults, and a rostral margin covered partially by the nasal process of

the premaxilla and partially forming the bony nasal aperture. Ventrally, the interalveolar

margin (Fig. 1.5; ommi) is the shallowest part of the bone. The palatal mucosa and the buccal

cavity meet at the interalveolar margin before they part to accommodate the tooth row.

Immediately dorsal to the root of the first molar, the infraorbital canal (Fig. 1.5; ci) exits the

skull through a large, oval foramen (Fig. 1.6; omfi). Dorsal and slightly caudal to the

infraorbital foramen, a roughened tubercle serves as the point of origin for the three

maxillolabial muscles. Caudal to this tubercle, the facial crest (Fig. 1.6; omcf) extends along

the rest of the maxilla onto the zygomatic bone. Dorsal to the facial crest and maxillolabial

tubercle, the maxilla slopes medially to form the rostral margin of the nasomaxillary fissure

and to meet the nasal bone. Ventrally, the maxilla forms the middle third of the hard palate

and supports the vomer bone. On its medial surfaces, the maxilla possesses an upturned

maxillary spine supporting a large, double-scrolled maxilloturbinate. The maxilloturbinate is

roughly oval shaped, with the rostral end projecting beyond the margin of the bony narial

aperture. Approximately halfway along the ventral surface of the maxillary spine, the bony

nasolacrimal duct opens into the nasal cavity. Caudodorsal to the maxillary spine is a large

opening into the maxillary sinus (Fig. 1.7; sm). The turbinates of moose are specialised in

that they extend rostrally past the margin of the bony nasal aperture and are visible in lateral

view (Fig. 1.6).

Os lacrimale (Fig. 1.6; ol)—The triangular lacrimal bone of moose contacts the maxillary and zygomatic bones ventrally, the nasomaxillary fissure rostrally, the ethmoid medially, and the orbit and frontal bone caudally. A ridge on its sloping rostral margin

contacts the thin cartilaginous plate overlying the nasomaxillary fissure. Along the margin of

the orbit, the lacrimal bone holds two lacrimal foramina (Fig. 1.6; ollf), serving as the

caudodorsal opening of the nasolacrimal duct. Between these two foramina, a rough

tubercle (Fig. 1.6; ollt) serves as the origin of members of the orbicularis oculi group. A concave fossa (Fig. 1.6; olfl) occupies the lateral face of the lacrimal bone. In many other

ruminants, this fossa is formed by a preorbital gland. However, no gland or glandular orifice

could be found in any of the specimens we dissected.

Os ethmoidale—The ethmoid bone in moose consists of a perpendicular plate

forming the bony portion of the nasal septum and a labyrinth of ethmoturbinates &

and nasoturbinate. The nasoturbinate (endoturbinal I of Paulli, 1900; Fig. 1.7; oen) extends from the dorsalmost area of the cribriform plate (Fig. 1.7; oelc) along the dorsal margin of the ethmo- and maxilloturbinates and terminates at the rostral edge of the nasal bone. Ventral to these, the numerous, scrolled ethmoturbinates (Fig. 1.7; oee) occupy the rest of the nasal

cavity dorsal and ventral to the maxilloturbinate and connect to the cribriform plate. The

dorsalmost of these (endoturbinal II of Paulli, 1900; Fig. 1.7; oee´) is enlarged and supports

the middle nasal concha.

Os nasale (Figs 1.5-1.7; on)—The relatively simply shaped nasal bone of moose is

unique in that its rostral extent is severely limited. As stated above, the nasal bone does not

contact the premaxilla, but rather terminates a few centimeters caudal to the nasal process of

the premaxilla. The nasal bone is roughly quadrangular in shape, taking on the appearance

of a rectangle half as wide as long that is bent at 90º along its long axis so as to contact its

fellow medially and the maxilla ventrolaterally. The lateral and dorsal faces of the nasal bone

are smooth, having no obvious specialised features except for the jagged rostral margin that

contacts the dorsal lateral cartilages. The nasofrontal suture is arched such that the frontal

bones send processes rostrally on the lateral and medial sides of each nasal bone.

DISCUSSION

Novel aspects of narial anatomy in moose

The nose of moose is an obviously distinctive structure differing in many ways from the condition found in other cervids and in other ruminants. Aside from external differences, visible from quite a distance, there are many other differences apparent upon dissection.

These anatomical specialisations have long warranted special attention (see Boas & Paulli,

1908; Jacobi, 1921; Meinertz, 1955).

The general narial musculature of moose is characteristic of other ruminants, namely, a well- developed malaris, three maxillolabial muscles, and marked specialisation of orbicularis oris muscles (Boas & Paulli, 1908; Getty, 1975; Schaller, 1992). However, as a result of enlargement of the narial structures in general, many of these muscles are modified from the condition in closely related groups. For example, whereas in other ruminants the maxillolabial muscles (caninus in particular) have bellies that separate distally along the snout, in moose the tendons that extend rostrally from each separate muscle belly are elongate. These tendons form an extensive network extending farther along the dorsal and ventral aspects of the nostril compared to other deer and ruminants, although levator labii superioris extends along the dorsal aspect of the nostril in Odocoileus (Fig. 1.8) (see also Boas

& Paulli, 1908; Getty, 1975). Interaction of the tendons with the specialised lateral nasal cartilages has produced another functional differentiation in the nose of moose. The angle taken by levator labii superioris and its tendons in deer and bovids is such that contraction of the muscles would depress and caudally displace the nostrils and the snout (Boas & Paulli,

1908; Getty, 1975; Schaller, 1992). In moose, however, levator labii superioris has become

more of a true levator by wrapping around the lateral enlargement of the dorsal lateral nasal

cartilages which act as a pulley redirecting the muscle’s line of action. In this way, the

tendon of levator labii superioris angles ventrally at its rostralmost extent, enabling

contraction of the muscle to produce elevation of the upper lip more easily in moose than in

other ruminants.

Figure 1.8. White-tailed deer (Odocoileus virginianus). Left lateral view of superficial (a) and deep (b) dissections showing musculature of the face and nostril. (c) Medial view of right side of sagittally sectioned head with nasal septum and vomer removed to show internal nasal structures. Skull in left rostrodorsolateral view with nasal cartilages intact (d) and in left lateral view (e). Scale bars = 5cm.

Muscles of the orbicularis oris group are also specialised in moose. Nasalis, which in moose is broad and very well-developed, is not even described in the veterinary literature for ruminants (Getty, 1975, and Schaller, 1992) and only briefly described in Boas & Paulli

(1908). This muscle was found in Odocoileus (Fig. 1.8), maintaining similar attachments but being considerably less developed than in moose. In Odocoileus, nasalis inserts on the caudal part of the dorsal lateral nasal cartilages, the underside of levator nasolabialis, and the toughened fascial sheath surrounding the maxillolabial muscles and the infraorbital vessels.

The action of nasalis appears to be that of a depressor and compressor of the nasal cavity, decreasing the distance between the dorsal lateral and ventral lateral nasal cartilages and potentially affecting intranasal pressures. It is not clear whether the expansion of this muscle in moose reflects simply enlargement of the cartilages themselves or enhancement of the particular actions of the muscle.

The nostrils of moose are greatly enlarged relative to those in other deer and other ruminants. Thus, it is not surprising that the muscles associated with the nostrils are enlarged as well. Moreover, the enormous nostrils of moose are directed laterally and have virtually lost the rhinarium that in other ruminants binds the nostrils together. The nostrils of moose are not only larger than in other ruminants but also are more mobile. Dorsal and ventral to the caudal, slit-like portion of the nostril, a suite of muscles such as narial dilators and lateralis nasi are much more conspicuous in moose. These muscles have been variably named by different workers or completely ignored by others (Boas & Paulli, 1908; Meinertz,

1955; Getty, 1975; Schaller, 1992), reflecting both their small size in outgroups and the difficulty in isolating particular muscles. These muscles in moose interact closely with the hinged lateral accessory nasal cartilage forming the caudal aspect of the nostril and with the tissue surrounding the fleshy nostril. We regard the enlargement of the nostril, the elaboration of the nostril musculature, and the mobile joint of the lateral accessory cartilage to be indicative of increased control over the external aperture to the nasal vestibule.

Inside the nasal vestibule and main cavity, there are several differences between

moose and closely related deer and other ruminants. Overall, the nasal vestibule of moose is

greatly enlarged both dorsoventrally and rostrocaudally, displacing structures such as the

maxilloturbinate and nasoturbinate caudally within the nasal cavity. Likewise, the nasal

ostium of the lateral nasal gland, which generally marks the caudal end of the nasal vestibule

in amniotes (Witmer, 1995b), is displaced caudally relative to the outgroups. The meatuses

are also greatly enlarged. In mammals in general, the ventral nasal meatus is the largest

(Getty, 1975), and this condition is true in the deer examined here (Fig. 1.8). Moose,

however, have relatively enlarged the middle and dorsal nasal meatuses, as well as the

common nasal meatus between the conchae and the septum. The conchae are also enlarged

but occupy relatively less space in the nasal cavity (see below). In both moose and Odocoileus,

the ventral nasal concha extends as far rostrally as the middle of the nostril. The alar fold of

the ventral concha sends a fatty process dorsally within the nasal vestibule and its

termination forms the caudodorsal portion of the nostril. In moose, however, this fatty

process is much enlarged, extending farther rostrally and farther medially than in Odocoileus.

The alar fold is supported laterally in this area by the lateral cartilages. The presence of the fatty pad, coupled to better-developed musculature, may aid in nostril closure (see below).

In both moose and Odocoileus, the ventral concha appears to undergo a rotation such that the attachment of the concha changes from lateral to ventral as the concha extends rostrally (Getty, 1975; Schaller, 1992). In moose, the result of this attachment and the enlargement of the fatty pad is an enlarged blind sac—the alar recess—caudodorsal to the nostril. Odocoileus has a homologous recess but it is much smaller. In both Alces and

Odocoileus, this recess is lateral and caudal to the alar fatty pad and communicates with the

ostium of the lateral nasal glands. Ventral to this pad and just caudal to the nostrils, the ostium of the nasolacrimal duct (Fig. 1.7; ocn) opens into the ventral nasal meatus.

Within the nasal cavity, moose show unique characters that differ from other deer,

other ruminants, and other mammals in general. The conchae within the cavity are tightly

packed in Odocoileus, filling up almost the entire available space. The meatuses between the

conchae are easily distinguished and follow a gently sloping course through the cavity

between the vestibule and choana. This pattern is repeated in other ruminants, such as oxen,

sheep, and goats, and also in horses (Getty, 1975; Schaller, 1992). Mammals outside

ruminants, such as dogs, also have tightly packed conchae within the nasal cavity (Dawes,

1952). Moose, however, have relatively larger spaces between the conchae. Moreover, the

ventral nasal concha (overlying the maxilloturbinate) is inclined somewhat dorsally going

from caudal to rostral, a situation unique among any deer examined here. Other species

possessing a proboscis, such as tapirs (Witmer et al., 1999) and saiga antelope (Frey &

Hofmann, 1996; Clifford & Witmer, in prep.) share this inclination of the ventral concha,

suggesting that enlargement of the nasal vestibule may impart some morphogenetic rotation

of more caudally located structures (Witmer et al., 1999). The already enlarged ventral nasal

meatus in the nasal vestibule continues to be enlarged in the main nasal cavity, probably by

virtue of the sloping maxilloturbinate. The conchae remain tightly packed only at the

caudodorsal portion of the cavity, at the intersection of the dorsal, middle, and ventral

conchae. As they extend rostrally, the meatuses enlarge between the conchae. Thus, the

inner structures of the nose are characterised by open spaces rostrally with an altered but

generally typical condition caudally.

Moose narial anatomy with respect to fossil alcines

Present-day Alces alces is a fairly young species, first appearing less than 100,000 years ago and

migrating to North America only 10,500 years ago (Geist, 1999). The closest fossil relatives

of Alces are species of Cervalces (synonymous with Libralces; Breda, 2001). Several specimens of Cervalces have been found throughout Eurasia and North America, but specimens possessing well-preserved facial structures are rare. For this reason, alcines are distinguished from other deer mainly by features of antlers and limb bones (Churcher and Pinsof, 1988).

Cervalces show characters of the foot and of antlers that unite them with Alces as members of

Alcini, but the facial skeletons in fossil alcines do not approach the level of specialisation seen in present-day moose (Fig. 1.9) (Scott, 1885). The nasal bones in Cervalces are much longer than in moose, and these nasals contact the ascending process of the premaxillae for at least 3cm (Scott, 1885; Breda, 2001). In addition, the articulations between the nasals and frontals and between the nasals and the nasal cartilages differ between fossil alcines and moose, particularly in that moose have a process of the articulation of the frontals that divides the caudal end of the nasals (Scott, 1885). Even specimens of Alces latifrons do not show any separation of the nasals and premaxillae or the modifications of the articulations of the nasal bones with surrounding osseocartilaginous structures (Churcher and Pinsof, 1988).

Nasal-premaxilla contact is a feature frequently used to infer the presence (but not the structure) of probosces in mammals (Scott, 1885; Jacobi, 1921), and so no fossil alcine has been reconstructed with the conspicuous narial specialisations seen in moose.

We consider additional osteological features in addition to the separation of the nasal and premaxillary bones to be indicative of highly modified narial anatomy in moose. The specimens of Odocoileus that were examined in this study showed variable contact between these bones, although the two bones were much closer to each other than in moose. The

nasals in moose are wider and have lateral and dorsal faces that meet at roughly 90°. The

nasal bones in other deer and other ruminants examined in this study are not evenly

widened, nor do they form an angle between dorsal and lateral parts. The dorsal face of the

nasal bones is almost perfectly flat in moose, whereas these bones in Odocoileus, Bos, Bison,

Ovis, and other ruminants have a slight ventral curvature to them at their rostral ends. No

ruminant examined here has the modified articulations of nasal bones that are described in

fossil alcines. Cervalces and Alces latifrons both possess nasal bones that have dorsal and lateral

faces that meet at right angles, even though the nasal and premaxillary bones contact each

other. Present-day moose are the only cervids with the particular narial specialisations

described here, although the evolution of this peculiar set of anatomical structures may be traced through the fossil record of alcines by virtue of the subtle differences in facial anatomy beyond simply a lack of articulation between the nasal and premaxillary bones.

Figure 1.9. Cladogram of Odocoileus virginianus (a), Cervalces scotti (b), and Alces alces (c) and their skulls in left lateral view to illustrate transformation of the bony naris. Skull in (b) redrafted from Scott (1885).

The cartilaginous capsule and the modified nostril musculature found in moose leave causally-associated osteological correlates that separate moose from any other known cervid.

The squared-off rostral edges and quadrangular shapes of the nasal bones serve as evidence

of the more extensive attachment of the dorsal lateral nasal cartilages. The lateral hump of

the dorsal lateral nasal cartilages modifies the nasal process of the premaxillae, causing it to

flare outward near its caudalmost extent along the bony nasal aperture. The enlarged and

highly developed lateralis nasi, which acts as a ventral dilator of the nostrils, leaves a tubercle on the lateral face of the body of the premaxillae. In addition, the superior incisive muscle leaves another, more medial, tubercle on the body of the premaxilla. Both of these muscular attachment sites are undetectable in either closely related cervids or other large-bodied ruminants, indicating the reliance upon nostril musculature in moose rather than simply allometric effects of having a large face. Admittedly, the inference of missing soft-tissue anatomy in mammalian narial anatomy is difficult (Witmer, 1995a). However, the enlarged cartilaginous capsule and highly developed nostril musculature (unique in moose among cervids) do leave causally associated bony signatures.

Functions of moose noses based on anatomical specialisations

Despite studies that deal with moose ecology (see Flerow, 1952; Geist, 1999) and narial anatomy (Boas & Paulli, 1908; Jacobi, 1921; Meinertz, 1955), no study has explained why moose have such apomorphic noses relative to very closely related outgroups or what novel functions may be carried out by these novel anatomical structures. Modified narial structures serve many adaptive functions for an animal (Witmer, 2001a), such as acting as a muscular hydrostatic organ of manipulation (Witmer et al., 1999) or as an essential water- conservation device (Langman et al., 1979). The nose of moose, however, remains enigmatic in terms of its function. We attempt here to address the adaptive significance of the modified narial anatomy in moose by appeal to its anatomy. While we hypothesize that this highly derived structure performs adaptive functions, we do not mean to imply a single,

driving “raison d’être” for these modified structures. Noses in general simultaneously serve diverse functions for animals (Witmer, 2001a), and moose are no different.

Flerow (1952) hypothesized that the enlarged narial structures in moose contribute

physiologically to an escape mechanism, allowing moose to run rapidly and for prolonged

periods in cold dry habitats. This mechanism has indeed been found in other cervids, by

which cooled blood from the nasal mucosa can bypass drainage into the facial vein and drain into the angularis oculi vein which ultimately leads to the cavernous sinus (Johnsen &

Folkow, 1988). This mechanism permits selective brain cooling during periods of heat stress, a feature that appears to be ubiquitous within Mammalia, to varying degrees (Kuhnen,

1997). However, this mechanism does not appear to be enhanced in moose for two reasons.

First, selective brain cooling in artiodactyls is dependant upon drainage of the alar fold through the dorsal nasal vein, which then in turn drains either into the facial vein to enter the external jugular system or into the angularis oculi vein to join the intracranial drainage system (Ghoshal, 1985). Our dissections and injections of the enlarged alar folds in moose do not reveal an increased vascularity of the structure. In fact, the alar fold in moose appears to be less vascularized than expected, as it is filled with loose connective tissue and fat. Additionally, prolonged running escape is not a behaviour moose preferentially employ in the wild (Geist, 1999). Although moose are capable of extended periods of “trotting,” they prefer to escape through moderately dense cover strewn with obstacles that they can easily clear (by virtue of leg length) and that their pursuers (e.g., wolves and coyotes) cannot, thus minimising their energy expenditure in fleeing (Geist, 1999).

The behavioural ecology of moose can produce several other lines of evidence concerning the evolution of the moose nose. Scott (1885) noted that although modern moose are forest-edge browsers, most fossil moose appear to have been open grazers. He

60 and other authors (e.g., Jacobi, 1921) have attributed the modified narial anatomy to the evolution of a mobile and sensitive upper lip to enhance browsing selectivity. Moose do indeed have a sensitive upper lip by virtue of rich innervation of the nose and upper lip

(Meinertz, 1955; Geist, 1999), but other deer species likewise are browsers and have similarly sensitive and mobile upper lips for selecting food items (Hofmann, 1989). Moose do have certain anatomical specialisations that suggest a lip more mobile than other ruminants. The pulley mechanism of levator labii superioris creates an angle of movement for the upper lip that is considerably more dorsoventral than other ruminants. In Odocoileus, for example, the contraction of incisivus superior is the only way to achieve directly dorsal elevation of the upper lip, based on its attachments and fiber direction. Moose, on the other hand, would be able to achieve this movement through a combination of contraction of incisivus superior and levator labii superioris. This levator in tapirs runs more dorsally than in ruminants, and this leads to more efficient elevation of the proboscis in tapirs (Witmer et al., 1999). In this way, the modified lateral nasal cartilages and levator labii superioris tendons of moose may analogously contribute to proboscis-like movement of the upper lip.

Many observers of moose have noted their keen sense of smell (Folkow, 1952;

Peterson, 1978; Geist, 1999). Indeed, moose do appear to have acquired unusual characteristics that may enhance the information content of olfactory cues. Their widely spaced nostrils permit odorant molecules to be collected from different locations in the environment. This physical separation of the two nostrils opens the possibility of stereolfaction. That is, moose may derive directional information from olfactory cues, a notion previously suggested by Geist (1999). Wide separation of the nostrils is certainly apomorphic for moose in that, in other cervids (indeed most other mammals), the two nostrils open almost on the midline, being separated only by the thickness of the septum.

In general, internarial width appears to decrease in vertebrates, indicating a reliance on

klinotaxis (orientation using one chemical sensor) rather than tropotaxis (orientation using

two sensors to derive directional information; Stoddart, 1979). Moose potentially represent

a rare exception to this rule. It is currently unknown whether moose actually engage in

stereolfaction, and, in fact, stereolfaction is a generally under-studied phenomenon (Kobal,

Van Toller & Hummel, 1989). Nevertheless, it may be significant in this regard that humans can derive directional information from olfactory stimuli, albeit by a different mechanism

(e.g., the nasal cycle; Sobel et al., 1999). Moose likewise may benefit from these characteristics of odorant detection. Moose rely heavily on “scent urination” to coordinate mating and estrous, and this information needs to be communicated over long distances

(Miquelle, 1991). This behaviour has not been found in other cervids (Miquelle, 1991), and the uniqueness of this activity in moose, coincident with anatomical diversification, is intriguing. Whether or not moose use stereolfaction awaits experimental confirmation. The point here is that moose have evolved an anatomical conformation that would enhance such a function.

One aspect of their behavioural ecology that separates moose from other large cervids is their reliance upon aquatic vegetation as a mineral-rich source of food (Geist,

1999). It has been suggested that the large antlers of male moose increase mineral demand, and foraging for aquatic vegetation is one means of increasing mineral ingestion (Peterson,

1978). Coincident with the formation of unique narial structures is a unique foraging method, and this perhaps could be an important factor in the evolution of the moose nose

(Witmer, 2001a). Moose are the only species of deer that consistently feed on aquatic vegetation; moreover, they spend a great deal of time with the head submerged as they forage for vegetation underwater (Peterson, 1978; Geist, 1999). Moose also dive periodically

to search for food, remaining submerged for a minute or more and reaching depths of at

least 5m (Peterson, 1978). These behaviours would place increased demands on narial

structures that are not faced by other non-diving deer, such as nostril closure, the ability to

detect food items underwater, etc. The nostrils of moose are specialised in several ways,

including lateral displacement, increased mobility, and proximity to an enlarged and highly

mobile alar fold. The lateral enlargement of the dorsal lateral nasal cartilages, together with

the rostral portion of the alar fold, forms a blind sac (the alar recess). This mobile portion of

the alar fold lies against the caudal two-thirds of the nostril. In addition, contraction of

levator labii superioris, resulting in dorsal displacement of the upper lip and nostril, would

align the ventral face of the nostril against the alar fold. Therefore, we suggest a possible

mechanism of nostril closure in moose consisting of evacuation of air from the rostral and

caudal recesses in the nasal vestibule, compression of the alar fold against the nostril by air in

the alar recess, and apposition of the dorsal and ventral faces of the nostril by contraction of

maxillolabial and lateral nasal musculature. Although validation of this mechanism has not

been tested experimentally, narial anatomy would suggest that this hypothesis is both feasible and merits testing.

At present, the adaptive significance of the specialised nose of moose remains enigmatic. We have presented anatomical findings that suggest that the key to deciphering

this enigma resides in the highly derived nostrils. Whether the evolutionary driving force is

enlargement and separation of the nostrils for stereolfaction or elaboration of the nostrils for

narial closure in association with underwater feeding cannot be determined. Indeed, both

explanations require substantiation. In any case, anatomical investigation has clarified the

functional components and outlined predictions that can be tested by subsequent

investigation.

CHAPTER 2: STRUCTURE AND FUNCTION OF THE NASAL CAVITY

OF SAIGA (ARTIODACTYLA: BOVIDAE: Saiga tatarica)

Abstract

Much of the narial anatomy of the enigmatic antelope Saiga tatarica has been described.

However, the anatomy of the nasal cavity and the causally-associated osteological correlates

of the proboscis structure remain undescribed. These data are integral for soundly

reconstructing probosces in fossil taxa and for understanding the functional significance of

the development of a proboscis. Saiga and outgroup specimens were subjected to CT

imaging, gross dissection, and skeletonization. CT data were analysed using eFilm and

Amira. The nasal cavity of saiga is characterised by an enlarged nasal vestibule and basal

conchal fold. Many of the structures associated with the nasal cavity proper (i.e., turbinates,

lateral cartilages, mucosal folds, nasolacrimal duct) are retracted caudally to a small area in

the caudodorsal part of the nasal cavity. The enlarged vestibule is associated laterally and

ventrally with paired sacs. The nasal septum is largely membranous and contains a large

patch of cavernous tissue. Bones comprising the narial margin have modified muscular and cartilaginous attachment sites for buccinator group muscles and reduced lateral cartilages.

The premaxilla is greatly modified by the enlarged musculature associated with nasolabial

fusion. Maintenance of the topological relationships of narial structures compared to bovid

outgroups has resulted in a nasal cavity with much larger area for seromucous glands of the

vestibule and narial musculature capable of controlling the aperture of the nasal cavity.

Maxillolabial muscles and lateralis nasi act together to both compress the nasal cavity and control the dilation of the nostrils such that air flow through the cavity is highly modified

64 relative to bovid outgroups. The lateral vestibular recess is an outpocket of the nasal vestibule that may function in the production of excess seromucous secretion and appears to have no homolog with outgroups.

INTRODUCTION

Saiga (Saiga tatarica) are a relatively little studied but morphologically disparate group of

antelopes (Fig. 2.1). Bovidae (e.g., antelopes, cattle) have undergone a dramatic radiation in

the past 18 million years, now encompassing 135 species (Vrba & Schaller, 2000). Previously

of varying affiliation (e.g., Caprinae in Nowak, 1999), saiga are now regarded as members of

Antilopinae within Bovidae (Vrba & Schaller, 2000). Saiga have been difficult to place

phylogenetically because they are bizarrely apomorphic, particularly in the head and skull as a

result of their evolution of unique narial structures. Like many other antelopes, only males

possess horns, but their most distinguishing characteristic, a conspicuous proboscis, is not

possessed by both male and female saiga. The inflated proboscis of saiga provides a case

study in novel narial anatomy occurring in otherwise morphologically and phylogenetically

well constrained Bovidae.

Saiga are a relatively young species, first occurring in middle Pleistocene deposits

approximately 1.0 million years ago, though some authors (Heptner, Nasimovich &

Bannikow, 1988) speculate that this genus may occur as far back as the late Pliocene. Only

one other species of Saiga is currently recognised (S. prisca), although the Mongolian

population previously received species rank (Heptner et al., 1988). This population is merely

an artifact of human-induced discontinuity of the range of saiga (Milner-Gulland et al., 2003).

Generally, fossil saiga are virtually identical to Recent forms, differing mostly in geographical

range (Sokolov, 1974). Occurring now only in open, arid grasslands of central Asia, their

range once extended from the British Isles to eastern Alaska (Frick, 1937).

Figure 2.1. Left lateral views of reconstructions of AMNH 202492 using Amira. (a) Lateral view of isosurface of the intact head. (b) Voxel reconstruction of intact head to simulate a lateral radiograph. (c) Skull isosurface. Scale bars = 5cm.

The inflated narial apparatus in saiga has been studied to various extents by previous

workers. Murie (1870) gave an extensive account of the anatomy of the whole animal,

discussing much of the musculature and innervation of the proboscis, as well as some

aspects of the nasal cartilages and skull, attributing the peculiar nose to an increase in tactile

sensation (Murie, 1870). Boas & Paulli (1908) figured the skull of saiga but offered few other anatomical details. Jacobi (1921) provided a diagram of the skull and nasal cartilages of saiga (seemingly based largely on Murie, 1870) in order to describe the apparent convergence between many phylogenetically disparate proboscis-bearing mammals.

Lodyshenskaya (1952) described the complex narial musculature and nasal cartilages in saiga, adding developmental and histological components. More recently, Frey & Hofmann (1995,

1997) conducted a series of morphological studies of the proboscis in saiga, focusing on the skull, glands, musculature, and anatomical differentiation from another proboscis-bearing bovid, Guenther’s dikdik (Madoqua guentheri). All of these studies differ from each other in focus, completeness, and terminology.

This study has two major aims. The first is to highlight functional aspects of the transformation of the nasal cavity of saiga, based on cross-sectional anatomy and dissection.

The second is to detail the causally-associated bony modifications of the skull resulting from the evolution of a proboscideal nose in saiga. The anatomical configuration of internal narial structures remains largely undescribed. Previous work has either omitted apomorphies occurring inside the nose or has only briefly described certain features. In addition, the osteological correlates of soft tissues comprising the nose in saiga remain undescribed as well. The present study is part of a larger effort attempting to describe the functional anatomy of apomorphic narial structures in extant amniotes (Witmer, Sampson & Solounias,

1999; Witmer, 2001a, b; Clifford & Witmer, 2001, 2002a, b, 2003, in review). As novel narial

structures are the product of selection acting upon discrete anatomical substrates, the project

as a whole seeks to explain how anatomical specialisation results in causally-associated bony features in apomorphic taxa. A goal is to assess the presence or absence of novel soft-tissue structures in extinct taxa by appeal to osteological correlates of these soft tissues found in

extant taxa (Witmer, 1995). Previous work has concentrated largely on muscular,

cartilaginous, nervous, and vascular specialisations in saiga without integrating many internal

changes into potential causally-associated features of the skull, and this study seeks to fill that

gap.

MATERIALS AND METHODS

A skull (AMNH 119649) and an intact head (AMNH 202492) were the primary source of

data for this study. The intact head was a zoo specimen later preserved as a fluid specimen.

Prior to dissection, the intact specimen was subjected twice to X-ray computed tomography

(CT) at O’Bleness Memorial Hospital in Athens, Ohio using a GE HiSpeed Fx-i Helical CT

Scanner. The first scan was set at 140.0 kV, 170.0 mA, 5.0 mm slice thickness using both

standard and bone algorithms. The second scan was set at 140.0 kV, 160.0 mA, 2.0 mm slice

thickness using a bone algorithm. CT data were exported in DICOM format using eFilm (v.

1.5.3, Merge eFilm, Toronto). Analysis and postprocessing employed the software packages

eFilm (v. 1.8.3) and Amira (v. 3.0, TGS, Inc., San Diego). Both sectional anatomy and

skeletal reconstructions were analysed in this specimen. Figures 1, 3, 5 & 6 were produced

from CT data using Amira. Study of the head also included gross dissection and sagittal

sectioning. Following sectioning, the right side was CT scanned a third time (120.0 kV,

130.0 mA, 1.0 mm slice thickness, bone algorithm). Dissections were recorded with digital

photography.

To determine bovid outgroup anatomy (Fig. 2.2), skulls of Madoqua saltiana (dikdik;

OUVC 9575), Ovis aries (domestic sheep; OUVC 9704), Bos taurus (domestic ox; OUVC

9473, 9474, 9475, 9476, 9477, 9478, 9479, 9480, 9481, 9482, 9547, 9548, 9558) and Bison bison (American bison; OUVC 9484, 9489, 9557) in addition to intact heads of Capra hircus

(domestic goat; OUVC 9744, 9746) were examined prior to dissection. One Capra (OUVC

9744) was injected in both carotid arteries with radio-opaque barium/latex (per Sedlmayr &

Witmer, 2002), CT scanned (120.0 kV, 100.0 mA, 1.0 mm slice thickness, bone algorithm), and analysed as above. Additionally, previous research (see above), veterinary texts (Getty,

1975; Nickel et al., 1973; Nickel et al., 1986; Schaller, 1992), and Nomina Anatomica

Veterinaria (NAV, 1994) were consulted to standardise terminology and homologise narial structures in saiga.

Figure 2.2. Phylogenetic relationships of taxa and clades referred to in this study. Topology based on Hassanin & Douzery (2003).

RESULTS

Overview of nasal cavity

The nasal vestibule in saiga is an extremely large structure, relative to outgroups. The nasal

passage appears to be divided into two distinct regions, more or less separated by vestibular

and respiratory/olfactory mucosa (Fig. 2.3). First, an expanded nasal vestibule at the rostral

end of the passage is continuous with a ventral space in the nasal cavity and the

nasopharyngeal duct (ductus nasopharyngeus; Figs 2.3 & 2.4; dnp). Second, the caudodorsal one-third of the nasal cavity is tightly packed with conchae. At the caudal end of the nasal vestibule, there is a blind recess or sac which opens rostrally and is lined with hair and vestibular mucosa (“nasal sac” of Murie, 1870). There is another recess whose dorsal wall forms the floor of the nasal vestibule (“shelf” of Murie, 1870). This recess opens caudally into the ventral space of the nasal cavity. The mucosa in this ventral recess is much like the mucosa of the nasal vestibule, but it is not lined with hair. The nostrils are the narrowest portion of the nasal passage, which is generally a wide-open space internally. The following description emphasises the various parts of the nasal cavity and mucosal structure in saiga.

Muscles and cartilages have been described elsewhere (Murie, 1870; Lodyshenskaya, 1952;

Frey & Hofmann, 1995, 1997). Standard veterinary nomenclature will be applied to anatomical structures in saiga.

Nasal vestibule

Vestibulum nasi (Figs 2.3 & 2.4; vn)—The nasal vestibule extends immediately caudal to the nostrils. In saiga, the nostrils are separated only by a thin continuation of the nasal septum, rather than by a rhinarium (planum nasi) which is well-developed in other bovids but not in saiga (Murie, 1870). The nostrils are oval and set close to the mouth. In the specimen described here, the nostrils did not overhang the mouth, although other workers have described a more pendulous proboscis (Murie, 1870; Sokolov, 1974; Heptner, 1988). The nostril appears tightly bound in position by musculature extending from the dorsolateral face of the premaxilla and winding around the nostril. The mucosa of the nasal vestibule is thick, fur-lined, and contains many small seromucous glands. The hairs in the vestibule are much

Figure 2.3. (a) Right medial view of Amira-generated isosurface of AMNH 202492. (b) Stereopairs of specimen in (a). Scale bars = 5cm.

smaller and much less dense than the hairs covering the head. Seromucous glands occur

ubiquitously throughout the vestibule on its dorsal, lateral, and ventral walls. The vestibule

extends caudally to the crescentic rostral edge of the basal fold (plica basalis) at the limen

nasi. At the limen, the mucosa changes to respiratory mucosa typical of other mammals.

Lateral to the limen is the ostium of the lateral recess, and medial to it is the ostium of the

nasolacrimal duct.

Recessus vestibularis lateralis (Figs 2.3 & 2.4; rvl)—The lateral recess of the nasal vestibule is roughly oval coronally and semicircular in sagittal section. Its opening into the main nasal vestibular chamber occurs in the middle of the rostrolateral edge of the basal fold about halfway along the rostrocaudal extent of the entire nasal passages. The sac then extends caudolaterally, somewhat mediolaterally compressed, over the edge of the nasomaxillary incisure. Its caudalmost extent is lateral to the maxilla, underlying musculature

(e.g. M. levator labii superioris & M. caninus) responsible for compressing the proboscis

(Murie, 1870) and situated just rostral to the preorbital gland (Fig. 2.4; gpo; Frey and

Hofmann, 1997). The mucosa lining the interior of the lateral recess is nearly identical to that in other parts of the nasal vestibule, even to the extent that it contains small hairs and minute glandular ostia. Its mucosa is thickened by fibrofatty elaboration, as elsewhere in the nasal vestibule. No muscles, cartilages, or bones could be found associated with the lateral recess; it occurs simply as an outpocketing of the nasal vestibule.

Canalis nasolacrimalis (Fig. 2.4; cnl)—The nasolacrimal canal is the primary connection between the nasal cavity and the orbit. Generally in bovids, this canal opens near the limen nasi, at the juncture of the vestibular mucosa and the respiratory mucosa of the nasal cavity proper (Nickel et al., 1973). In saiga, this delineation can be clearly seen on the basal fold. As the nasolacrimal canal exits the lacrimal bone to enter the nasal vestibule,

Figure 2.4. Drawings of selected computerized tomographic (CT) images of the face of AMNH 202492 showing narial structures in successive transverse sections (a-i). (j) Skull in left lateral view to show the rostrocaudal levels of sections depicted in (a-i). Scale bars = 5cm.

74 it next enters a space between the ventral portion of the dorsal lateral cartilages (Fig. 2.4; cnld) and the ventral lateral cartilages (“sesamoid cartilage” of Murie, 1870) within the basal fold. The mucosal ostium of the nasolacrimal canal occurs relatively far dorsal within the nasal vestibule, just medial to the rostral lip of the crescent of the basal fold.

Main nasal cavity

Cavum nasi proprium—The nasal cavity proper extends from the limen caudally to the nasopharyngeal duct. The mucosa in the nasal cavity proper is formed by a respiratory region (regio respiratoria) on the ventral concha and rostral parts of the ethmoid conchae and a pigmented olfactory region (regio olfactoria) covering the caudal portion of the ethmoid conchae. The turbinate-supported nasal conchae and the meatuses between them are displaced caudodorsally in the nasal cavity. In outgroups, the nasal cavity takes up most of the nasal passage (Nickel et al., 1973). In saiga, however, these structures do not extend far rostrally toward the nostril. The conchae are somewhat closed off by the basal fold, and the mucosal folds extending rostrally from the conchae are generally reduced.

Plica basalis (Figs 2.3 & 2.4; pb)—The basal fold is the ventralmost mucosal fold of the main nasal cavity. This fold in saiga extends from the rostralmost extension of the turbinate-supported portion of the ventral concha ventrally to the floor of the nasal vestibule. The thickened, fatty mucosa of the basal fold is covered with minute folds. Its rostral margin is crescentic. The space rostral to the basal fold is the nasal vestibule, the major space of the nasal passage in saiga. The space caudal to the basal fold is much smaller, forming the ventral meatus (meatus nasi ventralis; Figs 2.3 & 2.4; mnv). Just rostral to the crescent of the basal fold is the ostium of the lateral recess of the nasal vestibule. Within the fibrofatty basal fold, the major lateral cartilages of saiga can be found. The dorsal lateral

nasal cartilages (cartilago nasalis lateralis dorsalis; “lower lateral cartilage” and “upper lateral

cartilage” of Murie, 1870) are suspended within the rostral crescent of the basal fold. A

smaller flange of cartilage attaching to the ventralmost edge of the dorsal lateral cartilages

(cartilago nasalis lateralis ventralis; “sesamoid cartilage” of Murie, 1870) extends caudoventrally to attach on the lacrimal bone. These two cartilaginous structures frame the rostral ostium of the nasolacrimal canal, which opens onto the basal fold near the edge of

the rostral crescent.

The dorsal lateral nasal cartilages in saiga are figured both in Murie (1870) and in

Lodyshenskaya (1952), and the two major lateral components of the lateral cartilages are

given different anatomical names. Some mammals share this feature (having two

cartilaginous processes extending ventrolaterally along the lateral wall of the nasal cavity),

and the two cartilaginous laminae are referred as dorsal lateral cartilages (Nickel et al., 1973).

Thus, the two laminae of lateral cartilages in saiga are most likely homologous to, and

properly named, dorsal lateral cartilages. Attaching to the rostral portion of the dorsal lateral

nasal cartilages and ventral portions of the nasomaxillary incisure, many mammals also

possess ventral lateral cartilages. Murie (1870) identified a cartilaginous process extending

away from the caudoventral portion of the rostral lamina of the dorsal lateral cartilage,

naming it the sesamoid cartilage. In all other mammals, cartilages attaching to the ventral

portion of the nasomaxillary fissure along the lateral wall of the nasal cavity are termed

ventral lateral cartilages (Nickel et al., 1973).

Concha nasalis ventralis (Figs 2.3 & 2.4; cnv)—The ventral nasal concha is the largest

of the conchae within the nasal cavity. It is double-scrolled (characteristic of artiodactyls)

and supported by the maxilloturbinate (Figs 2.5 & 2.6; omt). The dorsal scroll makes at least

three complete turns, whereas the ventral scroll completes one and a half. Also

characteristic of other artiodactyls, the ventral concha is widest halfway along the maxilloturbinate. The ventral concha is covered exclusively by respiratory mucosa. The ventral concha is strongly angled dorsally, becoming almost vertical, which is in marked

contrast to other artiodactyls, in which it is basically horizontal.

Plica alaris (Figs 2.3 & 2.4; pa)—The alar fold in saiga is relatively very short

compared to most other ruminants. This mucosal fold is the direct rostral continuation of

the ventral nasal concha, and it ends rostrally just beyond the rostral edge of the basal fold.

Caudally, the alar fold retains a partial scroll of the ventral concha which curls first dorsally

then ventrolaterally, making a nearly complete turn. As the fold courses laterally along the

lateral wall of the nasal cavity, it becomes less scrolled. Near the rostral termination of the

straight fold (plica recta), the attachment of the alar fold migrates dorsally along the wall of

the nasal cavity. This migration continues dorsally and medially, and the alar fold ultimately

terminates on the nasal septum just inside the caudal end of the nasal vestibule and dorsal to

the ostia of the lateral recess and nasolacrimal canal. The mucosa of the alar fold is thick

and fatty, as it is in much of the vestibule, but as the fold travels on the nasal septum, there is

a collection of cavernous tissue deep to the mucosa.

Concha nasalis dorsalis (Figs 2.3 & 2.4; cnd)—Dorsally, the dorsal nasal concha,

supported by the nasoturbinate (Figs 2.5 & 2.6; oen; endoturbinal I of Paulli, 1908), courses directly rostrally from the cribriform plate just ventral to the roof of the nasal cavity. The mucosa of the dorsal concha is respiratory rostrally and olfactory caudally. This concha does not angle relative to outgroups, but its rostral extent is shortened in saiga. The dorsal concha makes a complete scroll before unwinding to continue as the straight fold.

Figure 2.5. (a) Left lateral view of Amira-generated isosurface of skull of AMNH 202492. (b) Stereopairs of specimen in (a). Scale bars = 5cm.

Plica recta (Figs 2.3 & 2.4; pr)—The straight fold is the dorsalmost mucosal fold of the nasal cavity. This is the direct rostral continuation of the dorsal nasal concha. The straight fold travels from the end of the dorsal concha to just caudal to the point where the

alar fold begins its rotation on the lateral wall of the nasal cavity. The fold is very small, and

there appear to be no glandular orifices associated with lateral nasal glands as in other ruminants (Nickel et al., 1973). Immediately lateral to the straight fold are the flattened

laminae of the dorsal lateral nasal cartilage. Between these two laminae lateral to the straight

fold, there is a sheet of dense connective (noncartilaginous) tissue commencing at the apex

of and partially overlying the dorsal lateral cartilage. This plate was described by Murie

(1870), and its function remains unknown. The straight fold does not undergo any rotation

akin to that experienced by the alar fold. The mucosa of the straight fold is covered in much

thinner mucosa more characteristic of the respiratory mucosa of the nasal cavity proper.

Concha nasalis media (Figs 2.3 & 2.4; cnm)—The middle nasal concha extends rostrally from the cribriform plate between the dorsal and ventral conchae. The middle nasal concha is roughly triangular in shape, its base commencing at the cribriform plate (Fig. 2.6;

oelc) and its apex occurring at the level of the end of the nasoturbinate. There are no

mucosal folds associated with the middle concha. As in the dorsal concha, the mucosa of

the middle concha is olfactory near its attachment to the cribriform plate. Surrounding the

middle concha dorsally, there are several smaller ethmoid conchae (Figs 2.3 & 2.4; cne) that

also are lined with respiratory and olfactory mucosa. The conchae associated with the

ethmoid do not undergo significant rotation like that in the ventral concha.

Nasal meatuses—The air spaces between the conchae are also modified in saiga relative to outgroups. The ventral nasal meatus is the largest space in the nasal cavity proper of saiga. Below the ventral concha, the ventral meatus (meatus nasi ventralis) occupies

almost the entire dorsoventral extent of the nasal cavity proper, owing to the inclination of the ventral concha. Its narrowest point occurs at the termination of the ventral concha, where the nasopharyngeal duct meets the nasal cavity. The air space dorsal to the ventral concha and ventral to the middle concha is the middle meatus (meatus nasi medius; Fig. 2.3; mnm). This space is restricted rostrally and open caudally, resulting from the rotation of the ventral concha. The middle meatus communicates with the nasal vestibule dorsally through the space between the alar fold and the outer wall of the cavity. Farthest rostrally, the dorsal nasal meatus (meatus nasi dorsalis; Fig. 2.3; mnd) occupies the space between the dorsal wall of the nasal cavity and the dorsal nasal concha. This is the smallest space in the nasal cavity.

Unlike in other bovids, this space does not broadly communicate rostrally with other air spaces in the cavity due to the rotation of the alar fold. Traversing between the nasal septum and the conchae within the nasal cavity, the common meatus (meatus nasi communis) connects the dorsal, middle, and ventral meatuses. The common meatus remains laterally

compressed, as structures extending out from the lateral wall of the nasal cavity approximate

the nasal septum.

Recessus nasi ventralis (Figs 2.3 & 2.4; rnv)— The ventral recess in the nasal cavity is the ventralmost space in the cavity. The ostium of the ventral recess opens caudally into the ventral nasal meatus just caudal to the ventral termination of the basal fold. Directly ventrally, the ventral recess is bounded by the palatine processes of the premaxilla and maxilla. Laterally, the recess is bounded by the maxilla. Medially, the ventral recess is bounded by the nasal septum. Along its ventromedial margin, the recess lies directly next to the vomeronasal organ. The dorsal relations of the ventral recess are mostly muscular. A series of strong muscle fibers from M. incisivus superior extend from the rostral face of the premaxilla to insert on the nasal vestibule near the ventral termination of the basal fold.

Organum vomeronasale (Fig. 2.4; ovn)—The vomeronasal organ of saiga courses along the ventral and lateral surface of the nasal septum. Caudally, the organ is associated with the opening of the ventral recess of the nasal cavity and the ventral attachment of the basal fold. The vomer and the palatine process of the premaxilla separate the vomeronasal organ from the ventral process of the septal cartilage. The organ itself is contained in an envelope of cartilage (cartilago vomeronasale) that extends from about the rostral extent of the palatine bone to the incisive duct. The organ communicates with the oral cavity through the incisive duct, ultimately opening on either side of the incisive papilla. The lumen of the vomeronasal organ remains patent for much of its length, clearly separated from the ventral recess of the cavity in cross-section.

Nasal septum

The nasal septum (septum nasi) in saiga consists of three major parts. Traveling caudal to rostral, the three main components are 1) lamina perpendicularis of the ethmoid (Fig. 2.4; oelp) 2) cartilago septi nasi (Fig. 2.4; csn) and 3) pars membranacea (Fig. 2.4; snpm). The perpendicular plate of the ethmoid lies between contralateral ethmoturbinates. It is much reduced in saiga, accompanying the caudodorsal retraction of the ethmoturbinates. Rostral to the bony portion of the nasal septum is the cartilaginous portion. The middle portion of the septal cartilage in saiga is emarginated such that its rostral edge is roughly crescentic.

The ventral prong of the septal cartilage (Fig. 2.4; csnv) divides the contralateral ventral recesses. Near the termination of the ventral process of the septal cartilage, the palatine processes of the premaxillae widen along their dorsal faces to accept a widened bulb of cartilage lying just caudal to the nostrils. The dorsal prong of the septal cartilage (Fig. 2.4; csnd) travels rostrally to the rostral termination of the basal fold. Farther rostrally, this prong

is continued by a fibrous cord (Murie, 1870), although this could not be directly verified in

the specimen described here.

Between the dorsal and ventral processes of the septal cartilage, the septum is

membranous. The membranous portion of the septum extends from the rostral extent of

the cartilaginous septum to the opening of the fleshy nostrils. The mucosa of the septum is

considerably thinner than on the lateral walls of the vestibule on the whole, as it is not as

invested with fibrofatty tissue or seromucous glands. The membranous portion of the

septum is fairly uniformly thin, with the exception of a collection of cavernous tissue

occurring in direct opposition to the ostium of the lateral recess (Fig. 2.4 sncc). This cavernous mass is placed near the rostralmost extent of the cartilaginous septum almost exactly halfway dorsoventrally and rostrocaudally in the nasal cavity. The cavernous mass is more or less spherical and continuous between the two sides of the nasal vestibule.

Osteological correlates of the proboscis in saiga

Os nasale (Figs 2.4, 2.5 & 2.6; on)—As in many other proboscis-bearing mammals, the nasal bones are caudally retracted in saiga. The frontonasal suture is closed in adults (frontonasal bone, Frey & Hofmann, 1995, 1997). The nasal cartilages modify the nasal bones by leaving a rugose surface where they attach. The rostral, triangular process of the nasal bone, together with its contralateral process, supports the dorsalmost portions of the septal cartilages and the dorsal lateral nasal cartilage. Laterally, where the nasal bone contacts the lacrimal bone, there is a shallow invagination of the bony narial margin that is smoother than the medial processes. Ventral to this margin, the nasal bone sends a short process along the

nasolacrimal suture (Fig. 2.5; onp), and this process is again rugose. The dorsal lateral nasal

Figure 2.6. (a) Right medial view of Amira-generated isosurface of skull of AMNH 202492. (b) Stereopairs of specmen in (a). Scale bars = 5cm.

cartilages additionally attach on this process. Retraction of the nasal cartilages has led to the

presence of attachment sites on the nasal bones.

Os lacrimale—The lacrimal bone of saiga is unique among ungulates in separating

the nasal bone from the maxilla (Murie, 1870). Near the nasolacrimal suture, the lacrimal is

roughened for attachment of dorsal lateral cartilages. Dorsal to the lacrimal fossa, there is a

shallow but wide tubercle (Fig. 2.5; olm; Murie, 1870). This tubercle serves as the attachment of malaris, a fan-like muscle extending along the preorbital region. Along the bony narial margin, the lacrimal bone sends out two processes, one dorsal and one ventral to the bony

nasal ostium of the nasolacrimal canal. The dorsal process (Fig. 2.5; olpn) does not attach to

any muscular or cartilaginous structures. It appears more likely to serve as a dorsal support

for the nasolacrimal canal, as its rostral edge abuts the curved process of the ventral lateral

cartilages. This process results from the caudal relocation of the nasolacrimal canal. Ventral

to the nasolacrimal canal, a second, triangular tubercle (Fig. 2.5; olt) serves as the ventral

attachment of the ventral lateral cartilage forming the ventral support for the nasolacrimal

canal. Again, as a result of caudally relocating the nasal cartilages, their attachment has also

traveled caudally to meet the lacrimal.

Os maxilla (Figs 2.4 & 2.6; om)—The maxilla forms the majority of the bony narial

aperture in saiga. The shallow angle taken by the maxilla gradually increases caudally. On

the lateral face of the maxilla between the infraorbital foramen (Fig. 2.5; omfi) and the facial

crest (Fig. 2.5; omcf), an angled tubercle with a sharp rostral margin (Fig. 2.5; omti) serves as

the skeletal attachment of the maxillolabial muscles (e.g., levator labii superioris, caninus, and

depressor labii superioris). These muscles fan out along the lateral side of the proboscis,

acting to compress the nasal vestibule (Frey & Hofmann, 1995, 1997). Medially, the wall of

the maxilla is extremely thin, as the conchae that this wall supports are reduced in saiga

84 compared to outgroups. The basal fold takes up much of the space in the nasal cavity and has no skeletal support associated with it. As a result, the medial wall of the maxilla has less conchal mass to support, permitting replacement of what is bone in outgroups with a thin sheet of mucosa.

Os premaxilla—As described previously (Murie, 1870), the premaxilla is very short and shallow with a truncated nasal process (Figs 2.4-2.6; oppn). Interestingly, perhaps the most distinguishing feature of the premaxilla has remained undescribed. The rostral end of the premaxilla is curved ventrolaterally and possesses a semicircular rugose margin (Figs 2.4-

2.6; oppr). This area of the premaxilla serves as the skeletal attachment of incisivus superior, which itself courses rostrocaudally in saiga. The enlarged attachment site of incisivus superior reflects its greater development in saiga. However, because the skeletal attachment is ventrally displaced and the distal attachment is caudally displaced in saiga, incisivus superior makes an unusual bend around the body of the premaxilla. Incisivus superior also produces a small bony lip on the medial surface of the premaxilla near the palatine process

(Figs 2.4 & 2.6; oppp). Thus, the larger attachment site is reflective of a larger muscle, but incisivus superior now must travel over the dorsal surface of the body of the premaxilla before attaching to the vestibule near the ventral extent of the basal fold.

Near the shortened nasal process of the premaxilla, lateralis nasi (transversalis nasi of

Lodyshenskaya, 1952) leaves a conspicuous attachment site. This muscle is generally poorly understood in many taxa, and its enlargement probably reflects greater control of the aperture of the fleshy nostril. Saiga control the fleshy nostril to a much larger extent than in outgroups (Frey & Hofmann, 1997), thus its attachment to the premaxilla is much more conspicuous.

DISCUSSION

Reorganisation of the nasal cavity

The nasal cavity of saiga is highly divergent from the condition in outgroups. The nasal vestibule dominates the rostral half of the cavity rather than being restricted to the rostralmost portion immediately near the nostrils. The effects of this transformation have implications for many of the structures comprising the nose, particularly those interacting with the nostrils. The distal attachment sites for musculature, the ostium of the nasolacrimal duct, and the mucosal folds of the nasal cavity are all modified as a result of the caudal expansion of the vestibular portion of the nasal cavity.

The proximal attachment sites (origins) for musculature of the proboscis retain the same pattern as in outgroups, yet the distal attachment serves externally as evidence for the caudal displacement of the nasal vestibule. Levator nasolabialis retains the same origin as in many other mammals, yet its insertion has been relocated (Frey & Hofmann, 1997).

Typically, this muscle attaches on the upper lip caudal to the nostrils (Nickel et al., 1986;

Schaller, 1992), but its caudal retraction accompanies the caudal retraction of the vestibule.

Levator nasolabialis retains an attachment approximating the extent of the nasal vestibule.

However, when the vestibule is expanded as in saiga, the distal attachment of levator nasolabialis tracks this change. The result is that levator nasolabialis retracts the proboscis and turns the nostrils upward, creating transverse furrows in the proboscis (Frey &

Hofmann, 1997). This muscle still acts upon the nasal vestibule, but its line of action has been altered.

Similarly, the maxillolabial musculature tracks the expansion of the nostril area corresponding to the nasal vestibule. Primitively, these muscles (i.e., levator labii superioris, caninus, depressor labii superioris) attach around the dorsal, caudal, and ventral portions of

the nostril on the lateral wall of the nasal vestibule. If these vestibular walls are to be

retracted, as they are in saiga, the maxillolabial musculature will attach more dorsally than in

outgroups. The maxillolabial muscles thus act to compress the nasal vestibule in this way

(Frey & Hofmann, 1997). Their line of action is modified resulting from dorsal relocation of

the distal attachment on the lateral walls of the vestibule. In other mammals that have well-

developed musculature for compressing the nasal vestibule (e.g., hooded seals; Clifford &

Witmer, 2001; and moose; Clifford & Witmer, in review), this action is carried out by M. nasalis. However, the dorsal lateral cartilages, the distal attachment for nasalis, have retracted far caudally in saiga. The need for compression of the nasal vestibule has been taken over by the maxillolabial musculature, the dorsal angulation of which makes this action possible.

The unusual morphology of incisivus superior again reflects caudal expansion of the nasal vestibule. Primitively, this muscle attaches to ventral structures in the nasal vestibule, such the cartilaginous septum and basal fold, and fans outward to the upper lip. The caudal relocation of the ventral attachment of the basal fold, however, has taken the musculature associated with it caudally. Incisivus superior now lies in the floor of the nasal vestibule.

Thus, the ventral recess opening into the ventral nasal meatus is an epiphenomenon resulting from the caudal direction of incisivus superior.

The nasolacrimal canal is similarly affected by the caudal relocation of the basal fold.

The ostium of this canal primitively opens into the nasal cavity in the basal fold near the limen nasi (Nickel et al., 1973), and saiga retain this condition. However, when the basal fold is retracted caudally and its dorsal termination on the alar fold is retracted dorsally and caudally, the ostium of the nasolacrimal duct accompanies this transformation. The interaction of the nasolacrimal canal with the dorsal lateral and ventral lateral nasal cartilages

is most likely the result of these structures being reorganised to adjacent caudodorsal

locations.

Saiga, as bovids, are constrained to having a scrolled maxilloturbinate and a series of ethmoturbinates. Rather than lose these structures, saiga modify them to accommodate the vestibular expansion. Conchae that primitively extend far rostrally into the nasal cavity are

either reduced rostrocaudally (as is the straight fold) or rotated (as is the ventral concha).

The retraction of the basal fold forces the turbinate-supported portion of the ventral concha

into an almost vertical position. The mucosal continuation, the alar fold, is rotated around

onto the nasal septum. These changes have resulted in removal of many of the conchal

structures from the main airflow, such that air flowing through the nasal cavity will more likely pass over structures in the nasal vestibule and through the ventral meatus.

Proboscis function in saiga

The apomorphic nose of saiga has been implicated in a number of different functions. The mechanism of any of these functions has not been explained with reference to specific anatomical novelties, leading to some misinterpretation of the functions of anatomical novelty in saiga.

Murie (1870) attributed the proboscis of saiga to an improved tactile organ, by virtue

of increased innervation and vasculature to structures comprising the proboscis.

Anatomically, however, an increase in innervation or vascularity could not be verified. The

infraorbital foramen does not appear enlarged in saiga, nor does the infraorbital nerve appear

larger than in outgroups. Furthermore, the pelage covering the nose is characterised more as

dense fur than as tactile vibrissae (Heptner et al., 1988). The mobility of the proboscis would

presumably be enhanced if tactile information from the nose was essential. The musculature

of the proboscis does not contribute to enhanced movement, but rather it relates to

compression of the vestibule and control of the aperture of the nostrils, limiting mobility of

the vestibule as a whole. The extent of the proboscis does not extend far rostrally beyond

the mouth, further limiting lateral or dorsal excursion.

It is clear in the proboscis of saiga that air inhaled into the nasal cavity is destined to

interact with the enlarged nasal vestibule at the expense of the conchal structures of the nasal

cavity. Several workers (Sokolov, 1974; Heptner et al., 1988; Frey & Hofmann, 1997) attributed narial specialisations in saiga to an increased performance of the air-conditioning mechanisms of the nose. When air is inhaled through the nasal cavity, it must be warmed and humidified so as not to damage the sensitive exchange surfaces of the lungs. Upon exhalation, air is passed over the same surfaces in order to recover the heat and humidity passed to the air during inhalation. This process is dependent upon the surface area of the conchae projecting medially into the nasal cavity (Schmidt-Nielsen, Hainsworth & Murrish,

1970). Saiga, however, do not appear either to increase the surface area available for countercurrent exchange or to bring those surfaces in-line with the main airflow. Thus, a counter-current exchange mechanism is clearly not being enhanced in saiga, and, if anything, is compromised by vestibular expansion.

Frey & Hofmann (1997) advanced a second hypothesis which integrates behavioural

observations of saiga with their apomorphic narial apparatuses. Saiga live in dusty, arid

habitats and employ a highly efficient mode of locomotion in which the head is held low so that cervical musculature may be recruited into forelimb movement (Heptner et al., 1988).

As a result, saiga are constantly inhaling air filled with dust particles (Frey & Hofmann,

1997). The mucosa of the nasal vestibule is lined extensively with seromucous glands

(Murie, 1870), providing moisture and surface area for adhering suspended particulate. The

musculature of the proboscis is aligned to produce forceful compression of the vestibule,

thus ridding the vestibule of dust that has accumulated on the moist mucosa. Frey &

Hofmann (1997) described the nasal “cough” of the proboscis to expel these collections of

dust particles from inhaled air, which recruits maxillolabial muscles, levator nasolabialis, and

lateralis nasi. Our data support this hypothesis.

The mechanism by which particulate matter is adhered to the mucosa of the nasal

vestibule can be explained anatomically by virtue of the size and orientation of vestibular

structures. Because the nostrils are the narrowest portion of the nasal cavity, as particle-

laden air enters the vestibule, its speed would decrease as it passes into a space with a larger

cross-sectional area (Poiseuille’s Law). The largest space within the vestibule is rostral to the basal fold and caudal to the nostrils, hence air would be traveling slowest within the vestibule. As the air slows, the suspended particles would be more likely to precipitate out of the air and adhere to the moist mucosa of the vestibule. The dynamics of air flow within the rostral chamber of the nasal vestibule are further altered by the cavernous plexus on the membranous septum, the basal fold, and the lateral recess. The cavernous plexus, when engorged, would direct inhaled air, and the particles suspended within it, laterally to the lateral wall of the nasal vestibule around the ostium of the lateral recess. The basal fold would assist this interruption of direct passage of air traveling from the vestibule to the ventral meatus. Despite the arid environment in which they occur, saiga probosces anatomically appar to be more suited for filtering air, perhaps at the expense of water conservation and reclamation that would normally be set up by the conchae.

Lateral recess homology and function

A functional hypothesis suggested here for the evolution of an enlarged lateral recess in the vestibule is for the production of extra seromucous secretions available for the collection of inhaled particles. This recess may perform a similar function to that of lateral nasal glands in other mammals. The lateral nasal glands serve as a significant source of fluid available for evaporative cooling in the nasal passageways (Blatt, Taylor & Habal, 1972), and the ostium for these glands in other mammals is associated with the straight fold (Schaller, 1992). The straight fold in saiga, however, is nearly obliterated and any secretions from glands here would lie well outside the main air passageway. The lateral recess does not appear to have musculature capable of dilating or widening it, and so this space would presumably remain outside the airflow in the nasal vestibule. Nevertheless, the mucosa of the recess is morphologically indistinguishable to the mucosa lining the remainder of the nasal vestibule.

The secretions produced throughout the vestibule are also produced in the lateral recess, yet the secretion produced in the recess represents an excess since it is not in the main airflow.

Thus, saiga have evolved a mechanism for alternative excess production of fluids available for collecting suspended particles in inhaled air.

The homology of this recess is enigmatic, as no other bovid develops a large recess in the nasal vestibule. One difficulty in assessing potential homologues of this structure is that the majority of the nasal vestibule in saiga is homologous to a very small area just inside the nostril in other ruminants. The basal fold in bovids such as oxen and goats is developed, but the space ventral to the basal fold is very restricted dorsoventrally (Nickel et al. 1973;

Schaller, 1992). In none of these animals is a glandular ostium described, nor was one found in the specimens of Capra examined here. The ostia found in most other artiodactyls are those associated with the nasolacrimal duct and the lateral nasal glands. Although the

91 ostium of the nasolacrimal gland potentially communicates with the lateral recess in saiga, they are on opposite sides of the basal fold. The ostium of the lateral nasal gland is always associated with the straight fold, a structure nearly obliterated in saiga. The only known potentially analogous structure to the lateral recess in saiga is a nasal sac in Camelus described by Arnautovic & Abdalla (1969). This sac in camels is much longer than in saiga, yet it is capable of being compressed by musculature of the nose, specifically maxillolabial musculature. The sac is lined with vestibular mucosa, producing excess seromucous secretions to keep the mucosa of the nasal cavity moist (Arnautovic and Abdalla, 1969).

Saiga may have evolved a convergent structure not to conserve water but to utilise vestibular secretions to adhere particles inhaled in an arid environment.

CHAPTER 3: RULES OF CONSTRUCTION IN MAMMALIAN PROBOSCIS

BUILDING

Abstract

Despite the diversity of probosces in modern mammals, most narial reconstructions of fossil taxa with a modified bony naris resemble tapir-like trunks. This study attempts to outline rules of construction in proboscis building by integrating the anatomies of different proboscis-bearing taxa. Osteological correlates determined from the effects of specialised soft tissues in proboscis-bearing taxa then shed light on the narial reconstructions in extinct mammals. Based on large-scale narial changes, two types of probosces can be identified in mammals, maxillolabial and vestibular probosces. These structures result both from natural selection acting upon anatomical substrates shared by members of a clade and from fortuitous exaptation. The anatomical configuration of soft tissues in outgroups is essential for the interpretation of the evolution of proboscis-building. However, such primitive functions of narial structures, as smelling and respiratory physiology are not the driving forces in the evolution of vestibular probosces. Bony turbinates, the overlying conchae, and the mucosal folds deriving from them are all modified in vestibular probosces, suggesting a different mechanism of dealing with narial counter-current exchange. With regard to the bony support for probosces, the key osteological correlates of proboscis building are mostly related to modified musculature. Muscles belonging to the maxillolabial group are most important in maxillolabial probosces, and muscles in the buccinator group are most important in vestibular probosces. These correlates reflect changes in different structures comprising a proboscis compared to outgroups, rather than entirely novel structures. The

hypotheses advanced in this study can be further tested in several ways. First, the narial

anatomy of whole clades of mammals with a proboscis-bearing taxon remain undescribed, and their integration into hypothetical rules provides additional data. Second, a reappraisal of the bony support for putative probosces in extinct taxa would test whether narial osteological correlates are able to be found in the fossil record.

Introduction

The reconstruction of a proboscis in extinct mammals based upon a retracted nasoincisive incisure (i.e., retracted nasal bones) dates back at least to the days of Cuvier (Jacobi, 1921), but rarely are any meaningful anatomical criteria advanced beyond retracted nasals and the enlarged bony naris. Extant mammals possess an impressive diversity of narial anatomies

(Witmer, 2001a). We should expect fossil mammals to be no different, yet narial reconstruction of elaborated structures in extinct mammals inevitably results in short, tapir- like trunks (Clifford & Witmer, 2002). A proboscis is defined here as any enlargement of the narial apparatus in a species relative to outgroups, and this definition includes a diverse array

(both phylogenetically and anatomically) of narial apparatuses in which a muscular trunk is just one example.

Popular and technical paleontology texts are replete with reconstructions of mammals with generic “trunks” (Fig. 3.1; see Dixon et al., 1993; Carroll, 1988; Colbert,

Morales & Minkoff, 2001). Mammals as diverse as Diprotodon (a marsupial) and Moeritherium

(a basal proboscidean) are given remarkably similar, generic, tapir-like muscular trunks.

Other works, however, attempted to justify the reconstruction of these trunks in fossil taxa.

For example, Wall (1980) justified the reconstruction of a proboscis in derived amynodonts based on comparison with the skulls of proboscis-bearing mammals, although some of these are also known only from fossil remains, and so their probosces were more assumed than demonstrated. Gillette and Ray (1981) justified the presence of a generic muscular trunk on

Glyptodon on the basis of perceived need; i.e., of the apparent difficulty in collecting food resulting from vertebral fusion and the lack of incisors. Shoshani (1998) justified elephant- like probosces in non-elephantid proboscideans by virtue of an enlarged infraorbital foramen, in addition to a nasoincisive incisure roughly similar to that in extant elephants.

Figure 3.1. Skulls (left) and reconstructions (right) of extinct, putative, proboscis-bearing taxa. (a) Diprotodon, a marsupial. (b) Moeritherium, a basal proboscidean. (c) Glyptodon, a xenarthran. (d) Astrapotherium, an astrapothere. (e) Homalodotherium, a notoungulate. (f) Pyrotherium, a pyrothere. (g) Theosodon, a liptoptern. Skulls in (a), (b), and (g) from Carroll (1988). Skull and reconstruction in (c) from Gillette and Ray (1981). Skulls in (d) and (e) from Riggs (1935, 1937). Skull in (f) from Colbert et al. (2001). Reconstructions in (a), (b), (d), (e), (f), and (g) from Dixon et al. (1993).

Seldom in these works is mentioned the evidence of specific soft-tissue structures acting upon the bony support of the nose to produce a proboscis. But why a tapir-like trunk in the first place? Why not a proboscis as in elephant seals, with a complex interaction of muscles and cavernous tissue (Laws, 1953), or a proboscis as in moose, with elaborated nasal cartilages and nostril musculature (Clifford & Witmer, in review)?

The aim of this paper is to test the hypothesis that there exists a limited set of “rules

of construction” in proboscis-building. Within the context of the inverted pyramid of

inference (Witmer, 1995a), the potential rules of construction outlined here are intended for

more sound inference of narial novelty in the fossil record. Proboscis structure compared

between types of probosces may highlight key transformations characterising each type.

That is, the combination of anatomical novelty within a clade and analogy between proboscis

types of different clades may reveal a series of principles that govern proboscis-building in mammals and allow these principles to be applied to extinct species, for which soft anatomy is not preserved. Inherent within this hypothesis are two dependent assumptions. First, there are identifiable proboscis types within mammals, based on convergence in structure.

Second, causally-associated osteological correlates of novel soft tissues should result from specific anatomical changes required by each type. Once sound anatomical inferences about potential probosces can be made, one can test hypotheses at progressively higher levels of the pyramid more soundly (Witmer, 1995a). Appendix 1 lists the proposed homology and nomenclature of facial muscles proposed in this study and used in the following sections.

Proboscis types within Mammalia

Noses perform diverse functions apart from those characteristic of mammals, such as conditioning inspired air (i.e., filtration, humidification, etc.) or collecting odorants (Witmer,

2001a), although these may indeed be the primitive function for a narial apparatus in mammals (Clifford & Witmer, 2002). This is exemplified in proboscis-bearing mammals, which have constructed a novel narial apparatus that either supplements primitive functions or even compromises those functions to enhance some other adaptive function. It is the identification of these functional specialisations integrated with apomorphic anatomical

structure that is essential to understanding the evolution of novel narial apparatuses and

identifying more soundly these structures in the fossil record. Although proboscis-building

is species-specific to a large extent (Fig. 3.2), it is the intent of this section to reveal gross

anatomical types in probosces. Identification of anatomical types is a prerequisite for

addressing issues of homology of type and function, convergence, and reliable, causally-

associated osteological correlates of novel soft-tissue structures.

Figure 3.2. Cladogram of proboscis-bearing mammals. * indicates taxa for which there is description of narial anatomy. Overall topology from Novacek (1993). Topology for Phocidae from Bininda-Emonds, Gittleman, & Purvis (1999). Topology for Ruminantia from Hassanin & Douzery (2003).

Based on skeletal morphology, there are two basic types of probosces within

Mammalia (Fig. 3.3). The first, termed maxillolabial probosces, can be characterised as a mobile elongation of the rostral end of the nose. This type is exemplified by species as

phylogenetically diverse as elephant shrews (Macroscelidia), pigs (Suidae), peccaries

(Tayassuidae), star-nosed moles (Condylura, Talpidae), and coatis (Nasua, Procyonidae).

These extended probosces are characterised by a largely unmodified skull and an elongation

of the snout between the rostral end of the skull and the nostrils. The bony snout retains a

tubular shape, and it is usually somewhat elongated relative to outgroups (Boas & Paulli,

1908a; Nickel et al., 1986; Grand, Gould & Montali, 1998). Cartilaginous elements are

carried forward from the rostral end of the skull and into the snout in star-nosed moles

(Grand, Gould & Montali, 1998), coatis (Boas & Paulli, 1908a), and pigs (Nickel et al., 1973).

Muscular differentiation associated with this type of proboscis is largely restricted to the

maxillolabial musculature (i.e., the group composed of M. levator labii superioris, M. caninus,

and M. depressor labii superioris; see Appendix 1). For example, star-nosed moles are able

to direct motion of the nostrils as a unit by action of three paired maxillolabial muscles

(Grand et al., 1998). The nasal cavity in pigs is elongated and somewhat compressed, and

conchal structures in the cavity are similarly tightly packed as in outgroups with the nasal

vestibule restricted to the rostralmost end of the cavity.

The second, and more easily recognised, type of proboscis is that which results in

reorganisation of the bony naris due to expansion of the nasal vestibule and/or vestibular

structures. This second type, termed vetibular probosces here, will be the focus of the remainder of the paper, as it is this type that has drawn attention in reconstructions of narial structures in extinct taxa. This type is found in many different groups of mammals, reflective of the diversity of anatomies found in each proboscis-bearing taxon. In all of these taxa, the nasal bones have been retracted or reduced, and the maxilla usually forms a large portion of the margin of the bony naris (i.e., there is a loss of premaxilla-nasal contact; except in elephants; Shoshani, 1996). The narial anatomies of several of the taxa belonging

Figure 3.3. Skulls (left) and facial anatomy (right) of extant proboscis-bearing taxa. (a) Sus scrofa (from Dyce, Sack, & Wensing, 1987). (b) Tapirus terrestris (from Witmer et al., 1999). (c) Alces alces (from Clifford & Witmer, in review). (d) Cystophora cristata (original drawing by Ryan Ridgely, Ohio University). (e) Saiga tatarica and (f) Madoqua guentheri (from Frey & Hofmann, 1997).

to this type have been described (e.g., Shoshani, 1996; Frey & Hofmann, 1997; Witmer et al.,

1999; Clifford, 2000; Clifford & Witmer, in review). Characters shared by all taxa in this type occur within the nasal cavity. The nasal vestibule in all these taxa is enlarged, and it is this portion of the nasal cavity that is specialised, causing secondary modification of other parts of the nasal cavity. For example, the conchal structure within the cavity is altered, resulting in a reduction of the mucosal folds and retraction and modification of the bony turbinates in the nasal cavity proper. Cartilaginous, muscular, nervous, and vascular structures differ greatly in this vestibular type, in association with the wide array of functional and anatomical apomorphies.

100

Phylogenetic constraint and exaptation

Probosces are the result of at least two major phenomena. First, there is a set, or

organisation, of facial muscles common to mammals (Boas & Paulli, 1908b), and clades have

variably modified this ancestral organisation (Boas & Paulli, 1908a). Anatomical substrates

derived in a clade are the starting points for the development of a proboscis. That is,

probosces are the product of natural selection acting upon an anatomical configuration

shared by clade members. Second, the form and function of a proboscis also arises from

exaptation (sensu Gould & Vrba, 1982) of these anatomical substrates. Proboscis-bearing

mammals have taken advantage of fortuitous arrangements of narial soft tissues and have

recruited structures originally adapted for some other use.

Muscles of the buccinator group are responsible for controlling the aperture of the

nasal cavity, the fleshy nostrils, and the oral vestibule. Proboscis-bearing mammals have

taken advantage of this property and emphasise these muscles in one way or another to achieve movement of the proboscis. Boas & Paulli (1908b) pointed out that M. lateralis nasi

(assigned to its own muscle group by them) is only developed in ungulates. Despite the fact

that ungulates may not form a natural group (Novacek, Wyss & McKenna, 1988), their

retention of this muscle permits its elaboration in proboscis-bearing mammals. Loss of

lateralis nasi in a clade restricts the potential elaboration of a novel narial apparatus, as specialised movement of the nostrils and incorporation of the muscle into a hydrostatic organ are restricted. Similarly, a well-developed M. incisivus superior in pecoran artiodactyls

(perhaps related to loss of upper incisors) has permitted its elaboration in proboscis structure. If the portion of the muscle attaching the premaxilla to the cartilaginous septum were not developed (as it is in ruminants), then the “muscular cushion” of the proboscis of dikdiks (Fig. 4 in Frey & Hofmann, 1997) and vestibular floor in saiga (Clifford & Witmer, in

101 prep) could never have arisen. Incisivus superior is not well developed in carnivores (Evans,

1993; Clifford, 2000), thus potentially limiting proboscis structures in this clade.

Maxillolabial muscles are constrained in many mammals to share a common origin between the infraorbital foramen and the maxillary contribution to the zygomatic arch.

Proboscis-bearing ruminants share this constraint and work around it in building a proboscis. In proboscis-bearing ruminants, the rostral attachment of levator labii superioris occurs more dorsally on the nose, and multiple, split tendons of contralateral muscles cross each other rostral and dorsal to the nostrils (Nickel et al., 1986; Schaller, 1992). Dikdiks

(Frey & Hofmann, 1997), moose (Clifford & Witmer, in review), and saiga (Clifford &

Witmer, in prep) have taken advantage of this condition, although for different functions. In dikdiks, this muscular relocation is essential for the hydrostatic myostructure of the hydrostatic proboscis. Saiga have relocated the insertions of maxillolabial muscles to make them the primary compressors of the vestibule, rather than nasalis. In moose, it is the interaction of the relocation of the muscle’s attachment and laterally displaced cartilages that creates a pully mechanism aiding in nostril closure and eversion of the upper lip (Clifford &

Witmer, in review). Moose recruit other tissues of the nose, such as elaborated nasal cartilages, in order to make levator labii superioris a more effective levator of the upper lip.

Maxillolabial muscles, however, are exapted in construction of a proboscis. For example, separation of the common origin of the muscle group and rostral relocation of the skeletal attachment of M. levator labii superioris appears to be a derived condition in perissodactyls (Fig. 3.4). Horses (Nickel et al., 1986; Schaller, 1992) and rhinoceroses (Saban,

1970) share the condition noted in tapirs (Witmer et al., 1999) of uniting contralateral tendons of levator labii superioris along the rostrodorsal portion of the nose. Tapirs have exapted the mechanical advantage afforded by rostral relocation and tendinous consolidation

102

Figure 3.4. Exaptation of maxillolabial musculature (mmax) in perissodactyls. Horses (a) and tapirs (b) have separated the origins of maxillolabial musculature, whereas these muscles primitively share a common origin. Modified from Boas & Paulli (1908a).

of contralateral muscles in construction of a proboscis. This trasformation, shared by all perissodactyls, has become an integral part of the proboscis in tapirs, perhaps reflecting in the many species of fossil ceratomorphs (tapirs and rhinos) assumed to have developed a trunk (Wall, 1980). This muscle group obviously is also integral for maxillolabial probosces, as the maxillolabial musculature creates the motion carried out in these probosces.

Another example of exaptation in proboscis-building concerns nasal cartilages.

Mammals have evolved a set of cartilaginous structures supporting other soft tissues within the nasal cavity and serving as attachment sites for facial muscles. Certain features, such as

103

separation of dorsal lateral nasal cartilages, separation of septal cartilaginous elements, and

mobile lateral accessory cartilages are found in clades of mammals and used in some taxa for

proboscis-building. Lateral accessory cartilages support caudal and ventral portions of the nasal vestibule and are often connected to other cartilaginous structures of the nose by a mobile joint (Nickel et al., 1973). Hooded seals and moose both take advantage of this configuration to carry out movements of the caudoventral portions of their nasal vestibules

(Fig. 3.5; Clifford, 2000; Clifford & Witmer, 2001; Clifford & Witmer, in review). Moose have

Figure 3.5. Nasal cartilages in a generalized phocid (a) and a hooded seal (b) showing elaboration of mobile lateral accessory cartilages. Modified from Brønsted (1932).

104

an elaborated nostril-closing mechanism reliant upon increased control of the nostril

aperture, and so have enlarged the structures associated with moving and supporting the

nostril. Hooded seals have disassociated lateral accessory cartilages from the nostril in order

to support the inflatable elastic bladder comprising their proboscis rather than to support

structures immediately associated with the nostrils, thus exapting them. In order to carry out

movements of the maxillolabial proboscis, separation of dorsal lateral and septal

cartilaginous elements creates a narial joint upon which the maxillolabial musculature may

act, as in pigs (Nickel et al., 1973).

Rules of construction in proboscis building

One interesting finding in comparing proboscis-bearing mammals to outgroups is that

probosces seem not to be developed to enhance the primitive functions of the nasal cavity.

Instead, evolution of a proboscis seems to compromise these functions. Enhancement of

the air-conditioning mechanism of the nasal cavity, whereby inhaled air is warmed and

humidified, and that warmth and humidity are reclaimed upon exhalation, requires an increase in the surface area of the nasal mucous membrane, which is most efficiently accomplished by expansion of the conchal structures in the nasal cavity (Schmidt-Nielsen,

Hainsworth & Murrish, 1970). However, lengthening and narrowing the nasal cavity are also important factors in enhancing the air-conditioning properties of the nose (Langman et al.,

1979). Many proboscis-bearing species indeed possess elongated and narrowed nasal passages, yet the increase in surface area of the mucosa does not seem to be markedly increased. Further modifications of the conchae result in a configuration that suggests actually a diminished capability to condition air moving through the nasal cavity (Fig. 3.6).

For example, moose and tapirs have longer and taller nasal cavities than their outgroups, yet

105

Figure 3.6. Nasal cavity of moose (a), tapir (b), elephant (c), and saiga (d) to illustrate rotation of the maxilloturbinate (mt) out of the main airflow through the nasal cavity resulting from vestibular enlargement. (a) from Clifford & Witmer (in review). (b) and (c) from Boas & Paulli (1908a). (d) from Clifford & Witmer (in prep).

the scrolled turbinates supporting mucosal conchae are not similarly expanded to fill this enlarged space (Witmer et al., 1999; Clifford & Witmer, in review). Elephants and saiga, in the extreme of restricting nasal conchae to the caudodorsal portion of the nasal cavity proper, have dramatically rotated them, such that airflow over these surfaces appears to be restricted

(Shoshani, 1996; Clifford & Witmer, in prep.). Vestibular encroachment into areas normally occupied by surfaces available for counter-current exchange has resulted in modifications of these cochal structures, the extreme of which can be seen in saiga. There may, then, be a

106

space constraint on the structures that all mammals possess within the nasal cavity, such as

turbinates and conchae, and enlarged vestibular structures that result in a reorganisation of

those structures in the nasal cavity proper. Thus, it appears that the ability to condition air

may actually be compromised, particularly in vestibular probosces. Certainly, the functions

are not enhanced.

Although exact measurements of the surface area of nasal mucosa have yet to be

calculated for proboscis-bearing taxa, it seems doubtful, from a gross anatomical viewpoint,

that the increase in vestibular mucous membrane compensates for the displaced or even

reduced turbinates. Fluid-producing glands are more numerous in

vestibular mucosa, compared to the mucosa covering turbinates (Negus, 1958), and so

respiratory evaporative water loss may function slightly differently in vestibular probosces.

This appears to be the case in dikdik, an African antelope with considerable water-

conservation and temperature regulation demands placed on it by the environment. Rather

than relying upon the nasal cavity to conserve water, as in giraffes (Langman et al., 1979),

dikdiks tolerate water loss in the nasal cavity in order to keep the brain cool when the animal

is in heat stress (Kamau, Maina & Maloiy, 1984). The observation that vestibular mucosa

often seems less vascular than the mucosa covering turbinates (Clifford & Witmer, in review;

Clifford & Witmer, in prep.) could indicate a modification of the conditioning mechanisms of the nose by changing the mechanism from an autonomic control of the narial vasculature

(Jessen, 1998) to a mechanism relying upon mucosal glands in a less-folded but enlarged nasal vestibule. Vestibular enlargement in mammals appears then to reduce structures evolved for increasing mucosal surface area in the nasal cavity, whereas other amniotes have done the opposite (e.g., birds expand turbinates and conchal structures into the vestibule;

Witmer, 1995b).

107

Moreover, pinnepeds (as a clade) have developed larger, more complexly branched turbinates than other carnivorans, presumably to enhance their physiological role in heat and water balance (Negus, 1958; described in detail only for Mirounga angustirostris; Huntley, Costa

& Rubin, 1984). Only three species of pinnipeds have developed a vestibular proboscis

(hooded seals, elephant seals, and grey seals) even though the clade as a whole has derived nasal cavities. Turbinate expansion in these species is restricted to the nasal cavity proper and does not encroach in the vestibule. This finding further illustrates that it is vestibular enlargement, rather than enlargement of conchal structures, that affect counter-current exchange mechanisms in vestibular probosces.

The sense of smell is another primitive function of mammalian noses that does not seem enhanced by proboscis-building. Olfactory epithelium is restricted to the caudalmost portion of the ethmoturbinates within the nasal cavity proper (Nickel et al., 1973). In vestibular probosces, in particular, no proboscis-bearing taxon has been described with a larger surface area of olfactory mucosa. Dawes (1952) described airflow through the nasal cavity of dogs, noting that the olfactory mucosa experienced its greatest exposure to airflow upon exhalation rather than inhalation. Thus, for a proboscis-bearing species to realise an increased sense of smell, the portion of the nasal cavity closest to the cribriform plate would have to be either elongated, enlarged, or modified in position in some way, and this does not appear to be the case. Rather, the olfactory mucosa remains close to the cribriform plate in the caudodorsalmost portion of the nasal cavity, as in outgroups. Proboscis-bearing taxa do not seem, then, to have an increased sense of smell. Rather, it is movement of the proboscis, as in dikdiks (Frey & Hofmann, 1997) and tapirs (Witmer et al., 1999) or separation of the nostrils in moose (Clifford & Witmer, in review) that permits an increase in the directional information collected from odorants. Vestibular probosces have the anatomical

108 configuration that could enable a sort of three-dimensional smelling either by moving the nostrils together to collect directional cues from odorants, as in a hydrostatic proboscis, or by separating nostrils to make two separate odorant collectors, as in the proboscis of moose.

A general rule of construction, particularly in vestibular probosces, is that many of the topological relationships of structures comprising the nasal vestibule remain consistent between proboscis-bearing taxa and outgroups. Retraction of the nasal vestibule accompanies retraction of the osseocartilaginous (cartilaginous and bony) elements, yet muscular attachment remains fairly constant. For example, in saiga, the nasal vestibule is homologous to an area immediately surrounding the nostril in other bovids. Yet, when the nasal vestibule is expanded far caudally in the nasal cavity and the nostrils move to a position in the rostroventralmost portion of the vestibule, the nasal bones retract, the nasal cartilages remain in the caudalmost portion of the vestibule (as they are in outgroups), and the muscles attaching to the vestibule maintain attachments to the same relative portions of the vestibule.

In moose, the enlarged and widely separated nostrils result in enlargement of the vestibular spaces around the nostrils, enlarged cartilages relating to the same portions of the nostril as in outgroups, and hypertrophied muscles attaching to the same structures as in outgroups.

In these vestibular probosces, narial novelty results from changing the size and line of action of narial muscles, rather than their topology. These modifications have significant functional consequences by permitting novel movement of modified vestibular structures. In this way, vestibular probosces are the result of relative modification of the anatomical substrates shared by a clade rather than structures created de novo.

109

Reliable osteological correlates resulting from proboscis building

Muscular osteological correlates resulting from proboscis evolution provide the best

clues as to the type of proboscis. A hallmark of probosces, regardless of function, appears

to be related to movement of the nostrils or of the nasal vestibule. Maxillolabial probosces

rely upon maxillolabial musculature to fan out around the nostrils, which are usually held

together by a rhinarium (i.e., planum nasi) and move more or less as a unit. Vestibular

probosces often achieve this same sort of movement through development of a muscular

hydrostat as in elephants (Shoshani, 1996), dikdiks (Frey & Hofmann, 1997), and tapirs

(Witmer et al., 1999). Other vestibular probosces rely upon compression of the nasal cavity,

such as hooded seals (Clifford, 2000; Clifford & Witmer, 2001) and saiga (Frey & Hofmann,

1997; Clifford & Witmer, in prep.). Moose have evolved a system of elaborated cartilages and

nostril musculature (Clifford & Witmer, in review). It is the muscular correlates that provide

the best evidence for the type of a proboscis, particularly within vestibular probosces.

Key muscular osteological correlates relating to morphology of a proboscis reside on

the rostral portion of the lateral surface of the maxilla, the ventrolateral portion of the premaxilla, and the rostral surfaces or processes of the premaxilla. These correlates are related specifically to members of the buccinator musculature, specifically M. nasalis, M. lateralis nasi, and M. incisivus superior. Incisivus superior is involved in movements of the upper lip and ventral portion of the nasal vestibule. Its hypertrophy in proboscis-bearing taxa is related to fusion of tissues of the upper lip and nose (i.e., nasolabial fusion; rather than independence of the upper lip). Several taxa fuse incisivus superior to ventral portions of the nasal vestibule, such as moose (Clifford & Witmer, in review), elephants (Shoshani,

1996), dikdiks, and saiga (Frey & Hofmann, 1997; Clifford & Witmer, in prep.). Lateralis nasi

in saiga (Frey & Hofmann, 1997; Clifford & Witmer, in prep.) and moose (Clifford & Witmer,

110

in review) leaves evidence of its development in proboscis building on the lateral surface of

the premaxilla. The muscle scars of lateralis nasi are perhaps the only causally associated

skeletal feature that permits direct inference of nostril specialisation in proboscis-bearing

taxa (e.g., as in moose). The ability to infer nostril position in extinct amniotes is generally

easy, as it is almost always rostroventrally placed either within the bony narial aperture or

rostroventral to it (Witmer, 2001b). This is true for all proboscis-bearing taxa examined

here, yet the development of lateralis nasi is the only osteological correlate that permits

inference of the differentiation of nostril morphology. Osteological correlates of nasalis are

similarly indicative of a proboscis requiring enhanced action of the muscle, namely,

depression or compression of the nasal vestibule. A conspicuous attachment site for

incisivus superior, in ruminants, serves as a reliable correlate for nasolabial fusion (e.g., as in

saiga). Thus, for vestibular probosces, the correlates resulting from modification of buccinator group musculature serve as the best proxies for the structure of a proboscis.

The apparent elongation of the nasal vestibule, resulting either as a phenomenon of its own or as an epiphenomenon of the retraction of dorsal portions of the bony narial aperture, is a characteristic shared by all vestibular probosces and not found in maxillolabial probosces . Interestingly, development of a muscular trunk (e.g., in tapirs and elephants) produces fewer muscular osteological correlates than other proboscis types. These probosces are characterised by elongation of the nasal cavity well beyond the bony narial aperture, and control of the movement of this passageway and of the nostrils is carried out largely by M. rectus nasi, a muscle with no skeletal attachment whatsoever (Boas & Paulli,

1908a; Shoshani, 1996; Witmer et al., 1999). Reduction of osseocartilaginous (skeletal and cartilaginous) tissues is also a character necessary to construct a hydrostatic proboscis

(Witmer et al., 1999), yet this condition is not sufficient to describe a hydrostatic proboscis,

111

as it is also found in saiga. The strength of the relationship between soft tissues comprising a vestibular proboscis and osseocartilaginous reduction increases by describing osseocartilaginous reduction as an indicator of nasolabial fusion. Muscular and mobile trunks, ironically, are then characterised by fewer of the muscular osteological correlates identified here. Additionally, maxillolabial probosces are characterised by very few reliable osteological correlates, and so reconstruction of a maxillolabial proboscis is justifiably a rarity and a theoretically tenuous undertaking. Thus, the reconstruction of a proboscis in fossil mammals requires a detailed understanding of the morphology of the bony naris, as the strength of the relationship between causally-associated osteological correlates and soft tissues that produce them is often subtle.

Further tests of construction hypotheses

As a result of exaptation and of derived anatomical substrates shared by a clade, an analysis of outgroup anatomy is vital for the interpretation of anatomical structure in probosces.

Additionally, proper reconstruction of putative extinct proboscis-bearing species requires an extant phylogenetic bracket (EPB; Witmer, 1995a) type of approach. How many outgroups are necessary? How closely related must they be? The answers, ideally, are “many” and

“very,” respectively. However, this is often unfeasible. Assessing proboscis structure in ruminants, such as bovids (e.g., saiga) and cervids (e.g., moose), is a relatively easy task, as neither of these groups is very old (approximately 18 and 20 million years old, respectively), both have adequate fossil records, and each is represented by an enormous adaptive radiation (133 and 45 extant species, respectively; Gentry, 2000) that is relatively well- resolved phylogenetically (Hassanin & Douzery, 2003). Contrastingly, this situation is muddled in groups like Perissodactyla, in which a small number of extant taxa remain to

112

represent a once-diverse clade (Prothero & Schoch, 1989). Having closely related outroups

strengthens the hypothesis testing necessary in the assessment of anatomical novelty in the

fossil record, yet definitive statements can be made even in relatively ancient groups that

have been ravaged by extinction (e.g., Archosauria; Witmer, 1997).

Additionally, there remain many extant taxa that do not have an adequate anatomical

description of proboscis structure. Elephant seals, grey seals, elephant shrews, proboscis

monkeys, and takins all shed light on the evolution of proboscis structure in their respective

clades. Elephant seals, grey seals, and hooded seals are all phocids yet have independently

derived probosces (Fig. 3.7). It appears that elephant seals lack many of the muscular and

Figure 3.7. Testing rules of construction in Phocidae. Determination of common anatomical substrates between taxa in the clade shown in dotted line, and hypothesis testing between proboscis-bearing taxa shown in light arrows. Topology from Bininda-Emonds et al. (1999). Modified from Witmer (1995a).

cartilaginous correlates of the inflatable proboscis in hooded seals, congruent with the observation that their proboscis is largely erectile (Laws, 1953). Similarly, takins, saiga, and dikdiks are all bovids that have independently derived novel narial apparatuses. Rules of construction can be tested by integrating a description of proboscis structure in takins to the

113 already described outgroup condition for bovids and the proboscis structure of saiga and dikdiks. Addition of such taxa has bearing upon hypothetical anatomical rules of construction in proboscis-building by describing more of the EPB for each. Proboscis monkeys occur in a clade with perhaps the best understood mammal—humans. The anatomical novelties in proboscis monkeys would also provide evidence relating to hypothetical rules, as they occur in a well-understood group of mammals. Elephant shrews, however, are enigmatic in terms of their phylogenetic affiliation (Woodall & FitzGibbon,

1995) and of their narial anatomy. Thus, tests of outgroup comparison are limited for elephant shrews.

Other mammals that have retracted nasals and a modified bony narial aperture but do not possess a narial proboscis are available to test hypotheses of proboscis construction.

Manatees and dugongs have an elaborate hydrostatic structure (Marshall, Clark & Reep,

1998), yet this structure seems more involved with the lips rather than with narial structures.

The reorganisation of the bony nostril may be an epiphenomenon of both the elaboration of lip musculature and a dorsal relocation of the fleshy nostril to accommodate an aquatic lifestyle. Modern cetaceans have also relocated the fleshy nostril to accommodate their aquatic lifestyles (Berta & Sumich, 1999), potentially shedding light on the convergent condition in manatees.

Applications to extinct taxa

Many extinct taxa also serve as tests of the construction hypotheses advanced here. Taxa such as Glyptodon (Gillette & Ray, 1981) and Diprotodon (Dixon et al., 1993) have been reconstructed with short trunks, and each possesses extant outgroups. Whether muscular, cartilaginous, vascular, or other soft-tissue structures as anatomical substrates can be traced

114

within Xenarthra and Diprotodontia, respectively, remains to be investigated. Many species

of extinct, archaic South American ungulates, such as notoungulates, pyrotheres,

astrapotheres, and liptopterns are apparently closest relatives of each other, leave no extant

relatives, and possess characters suggesting a vestibular proboscis (Colbert et al., 2001).

Because members of each of these groups have been reconstructed with a short, tapir-like trunk (Carroll, 1988; Dixon et al., 1993; Colbert et al., 2001), closer examination of these groups would test the rule that muscular hydrostats rely more on intrinsic musculature of the nose (Shoshani, 1996; Witmer et al., 1999) and leave fewer muscular correlates than other vestibular probosces. Hypothesis testing of proposed rules of construction in proboscis- building requires closer examination of both the narial features of extinct taxa and their outgroups and the description of other extant proboscis bearing taxa and their outgroups.

This is indeed feasable, as the rules of construction outlined here further support the absence of a proboscis in Moeritherium (Jacobi, 1921; Shoshani, 1998), despite its common reconstruction with one. Clifford & Witmer (in review) demonstrated that interpretation of the soft-tissue anatomy of moose and cervid outgroups utilizing rules of construction in proboscis-building indicates that Cervalces, an extinct basal alcine, did not have a moose-like proboscis based on a lack of specialised osteological correlates. The identification of novel skeletal features resulting from the modification of anatomical substrates shared by all deer in Alces, such as squared-off nasal bones, a flared-out narial aperture, and recognisable attachment sites of nostril musculature permits an analysis of potential narial structures in

Cervalces. This fossil moose shares none of these features with Alces, despite a close phylogenetic affinity. The evolution of the proboscis in moose then can be more easily interpreted, as moose are the only deer with a proboscis and other extinct deer closely related to Alces did not have such conspicuous (or specialised) narial features, and by

115 inference were not able to, for example, close off their nostrils as in modern moose. Thus, the evolution of novel narial structures can provide valuable clues in more accurately interpreting both to the structure and, by establishing a form/function relationship, the life histories of many groups of extinct mammals.

116

REFERENCES

Arnautovic, I. & Abdalla, O. (1969). Unusual blind sac on the face of the one-humped camel. Acta Anat. 73: 272-277.

Berta, A. & Sumich, J. L. (1999). Marine Mammals: Evolutionary Biology. San Diego: Academic Press.

Bininda-Emonds, O. R. P., Gittleman, J. R. & Purvis, A. (1999). Building large trees by combining phylogenetic information: a complete phylogeny of the extant Carnivora (Mammalia). Biol. Rev. 74: 143-175.

Blatt, C. M., Taylor, R. C. & Habal, M. B. (1972). Thermal panting in dogs: the lateral nasal gland, a source of water for evaporative cooling. Science 177(4051): 804-805.

Boas, J. E. V. & Paulli, S. (1908a). The elephant’s head: studies in the comparative anatomy of the organs of the head of the Indian elephant and other mammals. Part I. Copenhagen: Folio, Gustav Fisher.

Boas, J. E. V. & Paulli, S. (1908b). Ueber den allgemeinen Plan der Gesichtsmuskulatur der Säugetiere. Anat. Anz. 33: 497-512.

Breda, M. (2001). The holotype of Cervalces gallicus (Azzaroli, 1952) from Sénesè (Haute- Loire, France) with nomenclatural implication and taxonomical-phylogenetic accounts. Riv. Italiana Paleontologia e Stratigrafia 107: 439-449.

Brønsted, H. V. (1932). Bygningen af snuden og ansigmuskulaturen hos nogle Pinnipedier. D. Kgl. Danske Vidensk. Selsk. Skrifter, Naturvidensk, Og Mathem. 9: 8-85.

Bubenik, A. B. (1997). Evolution, taxonomy, and morphophysiology. In Ecology and Management of the North American Moose: 77-123. Franzmann, Albert W. & Schwartz, Charles C. (Eds.). Washington: Smithsonian Institution Press.

Carroll, R. L. (1988). Vertebrate Paleontology and Evolution. New York: W. H. Freeman & Company.

Churcher, C. S. & Pinsof, J. D. (1988). Variation in the antlers of North American Cervalces (Mammalia: Cervidae): Review of new and previously recorded specimens. J. Vertebr. Paleontol. 7: 373-397.

Clifford, A. B. & Witmer, L. M. (2001). The narial anatomy of hooded seals (Cystophora cristata) with respect to other Carnivora. Annual Meeting of the Society of Integrative and Comparative Biology, Chicago, Illinois. Am. Zool. 40(6): 976.

117

Clifford, A. B. & Witmer, L. M. (2002a). Proboscis evolution in Mammalia: preliminary studies. Annual Meeting of the Society of Integrative and Comparative Biology, Anaheim, California. Am. Zool. 41(6): 153.

Clifford, A. B. & Witmer, L. M. (2002b). Not all noses are hoses: an appraisal of proboscis evolution in mammals. Annual Meeting of the Society of Vertebrate Paleontology, Norman, Oklahoma. J. Vertebr. Paleontol. 22(Suppl. to 3): 66A.

Clifford, A. B. & Witmer, L. M. (2003). Nasal structures in moose (Cervidae: Alces). Annual Meeting of the Society of Integrative and Comparative Biology, Toronto, Ontario, Canada. Integ. Comp. Biol. 42(6): 137-138.

Clifford, A. B. (2000). The narial anatomy of hooded seals (Cystophora cristata) with respect to other Carnivora. Ohio University Honors Tutorial College Thesis. Ohio University, Athens. 84pp.

Colbert, E. H., Morales, M. & Minkoff, E. C. (2001). Colbert’s Evolution of the Vertebrates. (5th edn.). New York: John Wiley & Sons, Inc.

Dawes, J. D. K. (1952). The course of the nasal airstreams. J. Laryngol. Otol. 66: 583-593.

Dixon, D., Cox, B., Savage, R.J.G. & Gardiner B. (1993). The Macmillan Illustrated Encyclopedia of Dinosaurs and Prehistoric Animals; A Visual Who's Who of Prehistoric Life. New York: Macmillan Publishing Co.

Dyce, K. M., Sack, W. O. & Wensing, C. J. G. (1987). Textbook of Veterinary Anatomy. Philadelphia: W. B. Saunders Company.

Evans, H. E. (1993). Miller’s Anatomy of the Dog. (3rd edn.). Philadelphia: Harcourt Brace & Company.

Folkow, K. K. (1952). Genus Alces Gray (1821) - Elk. In Fauna of U.S.S.R.: Mammals. I: Musk Deer and Deer: 166-192.

Franzmann, A. W. (1981). Alces alces. Mamm. Species 154: 1-7.

Frey, R. & Hofmann, R. R. (1995). Der Kopf der Saiga-Antelope (Saiga tatarica tatarica Linnaeus 1766, Mammalia: Bovidae)---Ausgwählte funktionsmorphologische Aspekte. 1. Die Speicheldrüsen, die Mandibula und die Zunge. Zoologische Beitrage 36(2): 169-198.

Frey, R. & Hofmann, R. R. (1996). Skull, proboscis musculature and preorbital gland in the saiga antelope and Guenther’s dikdik (Mammalia, Bovidae). Zool. Anz. 235: 183-199.

118

Frey, R. & Hofmann, R. R. (1997). Skull, proboscis musculature and preorbital gland in the saiga antelope and Guenther’s dikdik (Mammalia, Artiodactyla, Bovidae). Zool. Anz. 235: 183-199.

Frick, C. (1937). Horned ruminants of North America. Bull. Am. Mus. Nat. Hist. 69.

Geist, V. (1999). Moose: behavior, ecology, conservation. Stillwater, MN: Voyageur Press.

Gentry, A. (2000). The ruminant radiation. In Antelopes, Deer, and Relatives: 38-64. Vrba, E. S. & Schaller, G. B. (Eds.). New Haven: Yale University Press.

Getty, R. (1975). Sisson and Grossman’s The Anatomy of the Domestic Animals. I: General, equine, ruminant. (5th edn). Philadelphia: W. B. Sauders Company.

Ghoshal, N. G. (1985). Thermoregulatory role of the cranial circulation in cerebral temperature control. Vlaams Diergeneesk. Tijdschr. 54: 246-261.

Gillette, D. D. & Ray, C. E. (1981). Glyptodonts of North America. Smithsonian Contr. Paleobiol. 40:1-255.

Gould, S. J. & Vrba, E. S. (1982). Exaptation—a missing term in the science of form. Paleobiology 8: 4-15.

Grand, T., Gould, E. & Montali, R. (1998). Structure of the proboscis and rays of the star- nosed mole, Condylura cristata. J. Mamm. 79(2): 492-501.

Groves, C. P. & Grubb, P. (1987). Relationships of living deer. In Biology and Management of the Cervidae: 21-59. Wemmer, Christen M. (Ed.). Washington: Smithsonian Institution Press.

Hassanin, A. & Douzery, E. J. P. (2003). Molecular and morphological phylogenies of Ruminantia and the alternative postion of the Moschdae. Syst. Biol. 52(2): 206-228.

Heptner, V. G., Nasimovich, A. A. & Bannikow, A. G. (1988). Mammals of the Soviet Union. I: Artiodactyla and Perissodactyla. Washington, D. C.: Smithsonian Institution Libraries and the National Science Foundation.

Hofmann, R. R. (1989). Evolutionary steps of ecophysiological adaptation and diversification of ruminants: a comparative view of their digestive system. Oecologia 78: 443-457.

Huber, E. (1930). Evolution of facial musculature and cutaneous field of trigeminus, part I. Quart. Rev. Biol. 5(2): 133-188.

Huntley, A. C., Costa, D. P. & Rubin, R. D. (1984). The contribution of nasal heat exchange to water balance in the northern elephant seal, Mirounga angustirostris. J. Exp. Biol. 113: 447-454.

119

Jacobi, A. (1921). Die Rüsselbildung bei Säugetieren der Gegenvart und Vorzeit. Jena. Z. Naturwiss. 57: 199-218.

Jessen, C. (1998). Brain cooling: an economy mode of temperature regulation in mammals. News Physiol. Sci. 13: 281-286.

Johnsen, H. K., & Folkow, L. P. (1988). Vascular control of brain cooling in reindeer. Am. J. Physiol. 254: R730-R739.

Kamau, J. M. Z., Maina, J. N. & Maloiy, G. M. O. (1984) The design and the role of the nasal passages in temperature regulation in the dik-dik antilope (Rhynchotragus kirkii) with observations on the carotid rete. Respir. Physiol. 56: 183-194.

Karns, P. D. (1997). Population distribution, density, and trends. In Ecology and Management of the North American Moose: 125-140. Franzmann, Albert W. & Schwartz, Charles C. (Eds.). Washington: Smithsonian Institution Press.

Kobal, G., Van Toller, S. & Hummel T. (1989). Is there directional smelling? Experientia 35: 130–132.

Kuhnen, G. (1997). Selective brain cooling reduces respiratory water loss during heat stress. Comp. Biochem. Physiol. 118: 891-895.

Langman, V. A., Maloiy, G. M. O., Schmidt-Nielsen, K. & Schroter, R. C. (1979). Nasal heat exchange in the giraffe and other large mammals. Respir. Physiol. 37: 325-333.

Laws, R. M. (1953). The elephant seal: I. Growth and age. Falkland Islands Dependencies Survey Scientific Reports 8: 1-67.

Lodyshenskaya, W. I. (1952). Morphology of the upper respiratory pathways of the saiga antelope (Saiga tatarica). Uchenye Zapiski Karelo-Finskogo Universiteta, Petrazavodsk 4: 17-41.

Marshall, C. D., Clark, L. A. & Reep, R. L. (1998). The muscular hydrostat of the Florida manatee (Trichechus manatus latirostris): A functional morphological model of perioral bristle use. Mar. Mamm. Sci. 14(2): 290-303.

Meinertz, T. (1955). Das facialisgebeit beim Elch (Cervus alces). Morph. Jahrb. 96: 523-598.

Milner-Gulland, E. J., Bukreeva, O. M., Coulson, T., Lushchekina, A. A., Kholodova, M. V., Bekenov, A. B. & Grachev, I. A. (2003). Reproductive collapse in saiga antelope harems. Nature 422: 135.

Miquelle, D. G. (1991). Are moose mice? The function of scent urination in moose. Am. Nat. 138: 460-477.

Moore, W. J. (1981). The Mammalian Skull. New York: Cambridge University Press.

120

Murie, J. (1870). On the saiga antelope, Saiga tatarica (Pall.). Proc. Zool. Soc. Lond. 451-503.

Negus, V. (1958). The Comparative Anatomy and Physiology of the Nose and Paranasal Sinuses. Edinburgh: E. & S. Livinstone, Ltd.

Nickel, R., Schummer, A., Seiferle, E. & Sack, W. O. (1973). The Viscera of the Domestic Mammals. Siller, W. G. & Stokoe, W. M. (trans.). Berlin: Verlag Paul Parey.

Nickel, R., Schummer, A., Seiferle, E., Frewein, J., Wilkens, H., & Wille, K.-H. (1986). The Anatomy of the Domestic Animals. I: The Locomotor System of the Domestic Mammals. Siller, W. G. & Stokoe, W. M. (trans.). Berlin: Verlag Paul Parey.

Nomina Anatomica Veterinaria (1994). Ithaca: World Assoc. Vet. Anat.

Novacek, M. J. (1993). Reflections on higher mammalian phylogenetics. J. Mamm. Evol. 1: 3-30.

Novacek, M. J., Wyss, A. R., & McKenna, M. C. (1988) The major groups of eutherian mammals. In The Phylogeny and Classification of the Tetrapods, Volume 2: Mammals: 31-71. Benton, M. J. (ed.) Oxford: Clarendon Press.

Nowak, R. M. (1999) Walker’s Mammals of the World. (6th edn). Baltimore: Johns Hopkins University Press.

Papp, M. J. (2000). A critical appraisal of buccal soft-tissue anatomy in ornithischian dinosaurs. Ohio University Master’s Thesis. Ohio University, Athens. 229pp.

Paulli, S. (1900). Über die Pneumaticität des Schädels bei den Säugertieren. Eine morphologische Studie. II. Über die Morphologie des Siebbeins und die der Pneumaticität bei den Ungulaten and Probosciden. Gegenbaurs Morphol. Jahrb. 28: 179–251.

Peterson, R. L. (1978) North American Moose. Toronto: University of Toronto Press.

Prothero, D. R. & Schoch, R. M. (1989). The Evolution of Perissodactyls. New York: Oxford University Press.

Riggs, E. S. (1935). A skeleton of Astrapotherium. Geol. Ser. Field Mus. Nat. Hist. 6: 167-177.

Riggs, E. S. (1937). Mounted skeleton of Homalodotherium. Geol. Ser. Field Mus. Nat. Hist. 6: 233-243.

Saban, R. (1970). La musculature peaucière de la tête et du cou chez Rhinoceros unicornis Linné 1758. Gegenbaurs Morphol. Jahrb. 115: 418-473.

121

Schaller, O. (1992). Illustrated veterinary anatomical nomenclature. Stuttgart: Ferdinand Enke Verlag.

Schmidt-Nielsen, K, Hainsworth, F. R. & Murrish, D. E. (1970). Counter-current heat exchange in the respiratory passages: effect on water and heat balance. Respir. Physiol. 9: 263-276.

Scott, W. B. (1885). Cervalces americanus, a fossil moose, or elk, from the Quaternary of New Jersey. Proc. Acad. Nat. Sci. Phil. 37: 174-202.

Sedlmayr, J. C. & Witmer, L. M. (2002). Rapid technique for imaging the blood vascular system using stereoangiography. Anat. Rec. 267: 330-336.

Shoshani, J. (1996). Skeletal and other basic anatomical features of elephants. In The Proboscidea: Evolution and Palaeoecology of Elephants and Their Relatives. Shoshani, J. & Tassy, P. (Eds.). Oxford: Oxford University Press.

Shoshani, J. (1998). Understanding proboscidean evolution: a formidable task. Trends Ecol. Evol. 13(12): 480-487.

Sobel, N., Khan, R. M., Saltman, A., Sullivan, E. V. & Gabrieli, J. D. E. (1999). The world smells different to each nostril. Nature 402: 35.

Sokolov, V. E. (1974). Saiga tatarica. Mamm. Species 38: 1-4.

Stoddart, D. M. (1979). External nares and olfactory perception. Experentia 35: 1456-1457.

Vrba, E. S. & Schaller, G. B. (2000). Phylogeny of Bovidae based on behavior, glands, skulls, and postcrania. In Antelopes, Deer, and Relatives. Vrba, E. S. & Schaller, G. B. (Eds). New Haven: Yale University Press.

Wall, W. P. (1980). Cranial evidence for a proboscis in Cadurcodon and a review of snout structure in the family Amynodontidae (Perissodactyla, Rhinocerotoidea). J. Paleontol. 54: 968-977.

Webb, S. D. (2000). Evolutionary history of New World Cervidae. In Antelopes, Deer, and Relatives: 38-64. Vrba, E. S. & Schaller, G. B. (Eds.). New Haven: Yale University Press.

Witmer, L. M. (1995a). The Extant Phylogenetic Bracket and the importance of reconstructing soft tissues in fossils. In Functional Morphology in Vertebrate Paleontology: 19–33, Thomason, J. J. (Ed.) New York: Cambridge Univ. Press.

122

Witmer, L. M. (1995b). Homology of facial structures in extant archosaurs (birds and crocodilians), with special reference to paranasal pneumaticity and nasal conchae. J. Morphol. 225: 269–327.

Witmer, L. M. (1997). The evolution of the antorbital cavity of archosaurs: A study in soft- tissue reconstruction in the fossil record with an analysis of the function of pneumaticity. J. Vertebr. Paleontol. 17(Suppl. to 1): 1-73.

Witmer, L. M. (2001a). A nose for all reasons. Nat. Hist. 110: 64-71.

Witmer, L. M. (2001b). Nostril position in dinosaurs and other vertebrates and its significance for nasal function. Science 293: 850-853.

Witmer, L. M., Sampson, S. D. & Solounias, N. (1999). The proboscis of tapirs (Mammalia: Perissodactyla): A case study in novel narial anatomy. J. Zool. 249: 249-267.

Woodall, P. F. & FitzGibbon, C. (1995). Ultrastructure of spermatozoa of the yellow- rumped elephant shrew Rhynchocyon chrysopygus (Mammalia: Macroscelidea) and the phylogeny of elephant shrews. Acta Zool. 76: 19-23.

123

Appendix One: Homology and nomenclature of facial musculature in mammals

In order to describe the anatomical transformations occurring between a proboscis-bearing species and its closely related outgroups, it is first necessary to organise facial musculature into a coherent system so that homologies and convergence may be addressed. Huber

(1930) grouped facial musculature by innervation of branches of the facial nerve (CN VII), whereas Boas & Paulli (1908b) used proximity and function. These groupings are almost entirely identical. As we were not always able to trace the proximate innervation of individual facial muscles, homology based on attachment (particularly skeletal) and action is much more amenable to the present study. For this reason, we retain the groups of

orbicularis oculi, maxillolabialis, buccinator, and intrinsic musculature of the nose (rectus

group of Boas & Paulli, 1908b). Appendix Table 1 lists the homologies and nomenclature of

facial musculature proposed in this study.

Orbicularis oculi group

Muscles comprising the orbicularis oculi group are generally restricted to the dorsum of the

nose and areas rostral to and immediately surrounding the eyes. The group is presumably

named according to its proximity to M. orbicularis oculi, yet not all members are adjacent to

this muscle. Orbicularis oculi, malaris, and levator anguli oculi medialis share a common

skeletal attachment on a tubercle on the lacrimal contribution to the orbital rim. Malaris fans

out rostrally and caudally along the lateral aspect of the nose and inserts at the angle of the

mouth or in portions of the buccinator, whereas levator anguli oculi medialis takes the

opposite path to fan out on the skin rostrodorsal to the eye. Other members of the

orbicularis oculi group do not have a skeletal attachment. Rather, they attach to midline

fascia of the nose and fan out toward the lateral aspect of the nose, attaching to the dorsal

aspect of the nostrils and upper lip. Levator nasolabialis, dilator naris medialis, and dilator

124

naris apicalis possess such a morphology. Nostril dilators are undescribed by Boas & Paulli

(1908a, b) or Huber (1930), yet their presence is evident such that dilator naris apicalis is

described in much of the veterinary literature (Nickel et al., 1986; Schaller, 1992; NAV,

1994), whereas dilator naris medialis is described in Nickel et al. (1986). Muscles of the

orbicularis oculi group (with the exception of levator anguli oculi medialis found only in

perissodactyls) are levators. Rostral members of the group (nostril dilators) elevate dorsal

portions of the nostril, thus dilating them. Caudal members of the group (levator

nasolabialis, malaris) are levators of the upper lip and rostral portions of the oral vestibule.

Maxillolabialis group

Muscles comprising the maxillolabialis group are much more clearly delineated than

members of other facial muscles. In general, maxillolabial muscles originate on the lateral

face of the maxilla near the infraorbital foramen and extend rostrally to attach dorsal, caudal,

and ventral to the nostril. The maxillolabialis group constitutes longitudinal musculature of

the lateral aspect of the nose. Primitively, there are three maxillolabial muscles (from dorsal

to ventral): levator labii superioris, caninus, and depressor labii superioris. Carnivorans

(Evans, 1993), elephants (Shoshani, 1996), and perissodactyls (Witmer et al., 1999) lose

depressor labii superioris, and artiodactyls typically split caninus into at least two muscle bellies. Perissodactyls and elephants have split the attachment of levator labii superioris to a point more dorsal than the attachment of caninus. Maxillolabial muscles generally retract the nostrils and rostral portion of the nasal vestibule caudally. Levator labii superioris elevates the rostral portion of the upper lip in addition to retracting the nostril, and depressor labii superioris slightly depresses the upper lip while retracting the nostril.

125

Buccinator group

The largest group of muscles related to narial structures is the buccinator group. Generally,

this group comprises muscles on the ventral portions of the nasal cavity, attaching to the

premaxilla and maxilla or remaining continuously cutaneous. Members relating to narial

musculature are buccinator, orbicularis oris, incisivus superior, lateralis nasi, and nasalis. The

superficial portion of buccinator extends ventrally from the maxilla to the mandible,

overlying other differentiated parts of the muscle (Papp, 2000). Orbicularis oris, like its ocular counterpart, serves as a sphincter for the oral vestibule, yet it has no skeletal

attachment (unlike orbicularis oculi). The remaining members of the group have skeletal

attachments on the dorsolateral portions of the maxilla and premaxilla from the midline to the tooth row. The rostralmost member, incisivus superior, attaches to the rostral portions of the premaxilla and fans out to insert on the upper lip, the rostralmost skin of the muzzle, and the ventral portion of the septum of the nose. Next caudally, lateralis nasi attaches to the lateral face of the body of the premaxilla and fans out dorsolaterally to attach on ventral and caudal portions of the nostrils and the lateral accessory cartilages. Nasalis attaches skeletally near the premaxilla-maxilla suture on the lateral aspect of the maxilla and dorsal or just rostral to the premolars. It extends dorsally to the midline of the nose, deep to members of orbicularis oculi. Ventral members of the buccinator group can be best characterised as constrictors of the oral vestibule, and rostrodorsal members can be best described as depressors of the nasal vestibule or nostrils (thus dilating them ventrally).

Incisivus superior is involved in both functions.

Intrinsic musculature of the nose

Mammals possess diffuse musculature extending from the mucosa of the nasal vestibule and cavity to the underside of the skin. Taken together, these fibers are part of rectus nasi.

126

Different species emphasise rectus nasi to different extents, and this muscle can be regarded as an integral component of the lateral walls of the nasal passage despite the lack of description concerning its morphology. Primitively, rectus nasi is a dilator of the nasal passage, increasing the aperture of those spaces by pulling the mucosa lining them laterally.

Related facial musculature

Though perhaps not interacting with narial structures directly, other facial muscles impact the morphology and action of the above muscles. The most important group interacting with narial musculature is the platysma-sphincter group. These muscles generally run in a rostrocaudal direction, separating as strap-like muscles attaching near the corner of the mouth or the ventral aspect of the face. Muscles of this group interacting with other narial musculature are zygomaticus, depressor labii inferioris, and platysma.

127

MUSCLE GROUP NAV (1994) BOAS & PAULLI (1908) WITMER et al. (1999) PLATYSMA-SPHINCTER sphincter colli superficialis sphincter colli superficialis platysma platysma depressor labii inferioris zygomaticus platysma pars zygomaticus zygomaticus sphincter colli profundus sphincter colli profundus levator anguli oris ORBICULARIS OCULI orbicularis oculi pars orbitalis orbicularis oculi orbicularis oculi orbicularis oculi pars palpebralis levator anguli oculi medialis praeorbicularis dorsalis levator anguli oculi medialis malaris praeorbicularis ventralis malaris supraorbicularis frontalis levator nasolabialis nasolabialis levator nasolabialis MAXILLOLABIALIS levator labii superioris maxillolabialis pars superioris levator labii superioris caninus maxillolabialis pars inferioris caninus depressor labii superiorisb BUCCINATOR buccinator pars buccalis buccinator pars superficialis buccinator buccinator pars molaris buccinator pars profundus orbicularis oris pars marginalis pars rimana buccinatorii orbicularis oris orbicularis oris pars labialis incisivus superior supralabialis pars buccinatorii incisivus inferior mentalis mentalis depressor labii inferioris depressor labii inferioris nasalis lateralis nasi dilator naris apicalis

lateralis nasi lateralis nasia transversalis nasia RECTUS NASI rectus labii rectus nasi rectus nasi Table 1. Facial musculature in mammals. Sources for muscles are given in column titles, and muscle groups are given in the leftmost column. Muscles in a single row are homologous. Note the disparities in homology between sources and the disparate nomenclature sometimes used. a—These muscles were grouped separately in a “lateralis nasi group” by Boas & Paulli (1908a, b). b—Except in Carnivora (Evans, 1993).

128

SCHALLER (1992) NICKEL et al. (1986) HUBER (1930) GRAY'S ANATOMY (1995) cutaneous fascei platysma cutaneous fascei pars labialis platysma risorius depressor labii inferioris depressor labii mandibularis triangularis depressor anguli oris zygomaticus zygomaticus zygomaticus zygomaticus major malaris malaris orbicularis oculi pars orbitalis orbicularis oculi orbicularis oculi orbicularis oculi pars orbitalis orbicularis oculi pars palpebralis orbicularis oculi pars palpebralis depressor supercilii corrugator supercilii malaris malaris levator labii superioris aleque nasi levator anguli oculi medialis frontalis orbicularis oculi pars frontalis levator nasolabialis levator nasolabialis nasolabialis procerus levator labii superioris levator labii maxillaris quadratus labii superioris levator labii superioris caninus caninus caput zygomaticum quadratus labii superioris levator anguli oris depressor labii superiorisb depressor labii superiorisb buccinator pars buccalis buccinator pars buccalis zygomaticus minor buccinator pars molaris buccinator pars molaris buccinator buccinator orbicularis oris orbicularis oris orbicularis oris orbicularis oris pars peripheralis orbicularis oris pars marginalis incisivus superioris incisivus maxillaris incisivus labii superioris incisivus inferioris incisivus mandibularis incisivus labii inferioris mentalis mentalis depressor anguli oris depressor labii mandibularis transversalis menti depressor labii inferioris nasalis dilator naris apicalis dilator naris apicalis dilator naris dilator naris medialis lateralis nasi lateralis nasi depressor septi

Table 1. Continued.