THE ANATOMICAL RECORD 00:000–000 (2013) Trigeminal Nerve Morphology in Alligator mississippiensis and Its Significance for Crocodyliform Facial Sensation and Evolution IAN D. GEORGE AND CASEY M. HOLLIDAY* Integrative Anatomy, Department of Pathology and Anatomical Sciences, University of Missouri, Columbia, Missouri, USA ABSTRACT Modern crocodylians possess a derived sense of face touch, in which numerous trigeminal nerve-innervated dome pressure receptors speckle the face and mandible and sense mechanical stimuli. However, the morphological features of this system are not well known, and it remains unclear how the trigeminal system changes during ontogeny and how it scales with other cranial structures. Finally, when this sys- tem evolved within crocodyliforms remains a mystery. Thus, new mor- phological insights into the trigeminal system of extant crocodylians may offer new paleontological tools to investigate this evolutionary transformation. A cross-sectional study integrating histological, mor- phometric, and 3D imaging analyses was conducted to identify patterns in cranial nervous and bony structures of Alligator mississippiensis. Nine individuals from a broad size range were CT-scanned followed by histomorphometric sampling of mandibular and maxillary nerve divi- sions of the trigeminal nerve. Endocast volume, trigeminal fossa vol- ume, and maxillomandibular foramen size were compared with axon counts from proximal and distal regions of the trigeminal nerves to identify scaling properties of the structures. The trigeminal fossa has a significant positive correlation with skull length and endocast volume. We also found that axon density is greater in smaller alligators and total axon count has a significant negative correlation with skull size. Six additional extant and fossil crocodyliforms were included in a sup- plementary scaling analysis, which found that size was not an accurate predictor of trigeminal anatomy. This suggests that phylogeny or soma- tosensory adaptations may be responsible for the variation in trigemi- nal ganglion and nerve size in crocodyliforms. Anat Rec, 00:000–000, 2013. V C 2013 Wiley Periodicals, Inc. Key words: alligator; crocodyliform; integument; sensation; trigeminal; allometry; evolution; brain Grant sponsors: MU Life Sciences Fellowship; University of Received 24 November 2011; Accepted 14 June 2012. Missouri Research Council; the Department of Pathology and DOI 10.1002/ar.22666 Anatomical Sciences. Published online in Wiley Online Library *Correspondence to: Casey M. Holliday, Integrative Anatomy, (wileyonlinelibrary.com). Department of Pathology and Anatomical Sciences, University of Missouri, Columbia, MO 65212. E-mail: hollidayca@ missouri.edu C 2013 WILEY PERIODICALS, INC. V 2 GEORGE AND HOLLIDAY The American alligator (Alligator mississippiensis) is through the maxillomandibular foramen. The maxillary one of 23 extant species of crocodylians, a lineage of nerve turns rostrally from the maxillomandibular fora- crocodyliforms that first appeared during the Cretaceous men, passing dorsal to m. pterygoideus dorsalis and ven- period (Brochu and McEachran, 2000; Brochu, 2003). tral to the orbit. The mandibular nerve emerges Alligators are semi-aquatic vertebrates primarily found ventrolaterally from the trigeminal ganglion into the not only in freshwater areas of the Southeastern United adductor chamber, passing between the adductor mandi- States but also venture into brackish and occasionally bulae posterior and adductor mandibulae internus, saltwater. Alligators have many specialized features for medially, and m. adductor mandibulae externus later- their semi-aquatic lifestyle including a platyrostral skull, ally. During this passage, the nerve gives off several a nictitating membrane, a palatal valve that isolates the large sensory and motor branches including rami ptery- oral cavity from the choana and pharynx, an external goideus, anguli oris, and caudalis (Holliday and Witmer, ear flap, elevated eyes, and specialized narial muscles 2007). The mandibular nerve passes caudoventrolateral that close the nostrils. In addition to these features alli- to the cartiliago transiliens, lateral to m. intramandibu- gators, as well as other living crocodylians, have highly laris where is gives off the large intermandibular nerve sensitive faces packed with receptors that are capable of (i.e., mylohyoid nerve), and then passes into the inferior detecting mechanical stimuli, such as prey or danger, alveolar canal in the mandible just dorsal to Meckel’s while submerged in water (Soares, 2002). cartilage. As the mandibular nerve continues through As in most vertebrates, the fifth cranial nerve, the tri- the mandible, it gives off various large and small axons geminal nerve (CN V) detects sensory information from that, along with vasculature, perforate the lateral sur- the face of the alligator. This large mixed cranial nerve face of the dentary. There, the axons terminate in the divides into three major branches: the ophthalmic, max- integument as normal nerve endings, as well as DPRs. illary, and mandibular divisions. The ophthalmic and Consequently, numerous neurovascular foramina, maxillary divisions transmit solely sensory information forming a “beehive” pattern, rather than the plesiomor- from the upper face whereas the mandibular division phic, lizard-like “linear” pattern (Soares, 2002); charac- also provides motor innervation to the jaw muscles as teristically pepper the facial and mandibular elements of well as sensation from the mandible and tongue (Holli- crocodylians. Soares (2002) used this pattern of facial fo- day and Witmer, 2007). Beyond somatic touch, several ramina as a proxy, or osteological correlate, to infer the vertebrates evolved trigeminal nerve-based specialized evolution of the DPR system in crocodyliforms, recon- sensory systems such as electroreceptors in the platypus structing them as being absent in the Early Jurassic (Gregory et al., 1987; Manger and Perrigrew, 1996) and Protosuchus richardsoni, a primitive crocodyliform, but infrared receptors in some snakes (Molenaar 1974, also absent in the putative terrestrial Eocene crocodyli- 1978a,b). form Sebecus icaeorhinus. Our own observations of Sebe- Likewise alligators, and likely all other living crocody- cus found, however, that the lateral portion of the lians, are characterized by a group of trigeminal-inner- symphysis is clearly perforated by numerous neurovas- vated specialized sensory organs called dome pressure cular foramina arranged in a “beehive” pattern, much receptors (DPR) (Leitch and Catania, 2012). DPRs are like that of Alligator. Moreover, in tracking the history highly sensitive mechanoreceptors which react to of the DPR system in crocodyliforms, Soares (2002) only changes pressure associated with the movement of water studied a limited number of taxa which were largely while partially submerged (von During, 1973, 1974; von sampled from crown-group crocodylians (e.g., Leidyosu- During and Miller, 1979). Soares (2002) tested the func- chus canadensis) or derived neosuchians including Dyro- tion of DPRs as mechanoreceptors involved in head-ori- saurus phosphaticus, Goniopholis sp., Pachycheilosuchus enting behavior by alligators with and without a rubber trinqui (formerly the Glen Rose Form), and Eutretaura- coating over their faces. Previous research on mechanor- nosuchus delphi. Finally, cursory analyses have found eceptor distribution showed that sensitivity is directly little evidence of a clear pattern among DPRs, other sen- proportional to receptor density (Dehnhardt and Kamin- sory nerves, and facial neurovascular foramina (Allen, ski, 1995; Nicolelis et al., 1997). Regions that have 2005; Morhardt, 2009; Morhardt et al., 2009) in archo- mechanoreceptors require additional innervation (Wine- saurs. Thus, the utility of facial neurovascular foramina ski, 1983; Nicolelis et al., 1997; Ebara et al., 2002) and as osteological correlates of a DPR system remains sus- thus the nerve supplying this region, in the case of pect. On the other hand, larger, more proximal neuro- DPRs, the maxillary and mandibular branches of the tri- vascular foramina, similar to those formed by the larger geminal nerve, will contain more axons, and should be portions of the crocodylian trigeminal nerves, have pro- larger as a result. ven useful as osteological correlates in testing ecomor- In crocodylians, the trigeminal nerve roots emerge phological and evolutionary patterns in the peripheral from the brain at the lateral corner of the medulla nervous system. Significant, albeit contentious, correla- oblongata and exit the endocranial cavity through a tion was identified between the hypoglossal nerve and short passage (the trigeminal foramen) into the trigemi- the hypoglossal canal in primates, which were used as a nal fossa (Meckel’s cave) formed by the laterosphenoid proxy for lingual function in speech evolution (Kay rostrally and prootic caudally (Fig. 1). Here the trigemi- et al., 1998; DeGusta, 1999; Jungers et al., 2003). nal ganglion is seated, surrounded by the laterosphenoid Muchlinski (2008, 2010) showed the primate infraorbital and prootic medially, and the quadrate and pterygoid foramen, a presumed proxy for whiskers and facial sen- laterally. Although the ophthalmic nerve exits rostrally sation, correlated with maxillary mechanoreception and from the ganglion through its own trough in the lateros- potentially foraging ecology. phenoid (Holliday and Witmer, 2009), the maxillary and The presence of the laterosphenoid, which represents mandibular branches emerge from the lateral part of the the ossified pila antotica, characterizes alligators, croco- trigeminal ganglion and exit the trigeminal fossa dyliforms, as well as archosauriforms as a whole  CROCODYLIFORM FACIAL SENSATION AND EVOLUTION 3 Fig. 1. Imaging and virtual model of alligator specimen. (A) Lugol’s nal ganglion in yellow, and the branches of the trigeminal nerve in yel- Iodine enhanced microCT of a juvenile alligator (described in Tsai and low. (D) Sagittal CT slice showing cutting plane from (A) and (B) in Holliday, 2011). (B) Axial CT slice from AL005 showing segmentation orange. Abbreviations: EC, endocranial cavity; Ls, laterosphenoid; Vf, of the endocast in blue and the trigeminal fossa in green. (C) The 3D trigeminal fossa; Vg, trigeminal ganglion; V3, mandibular nerve; Mn, rendering of the skull (grey) showing the endocast in blue, the trigemi- mandible. (dinosaurs, pterosaurs, and stem taxa). Thus, whereas to which these structures scale with the size of the skull the trigeminal fossa is only partially ossified in lizards, and the brain as a whole, is yet unclear. Given that turtles, and birds, it is surrounded by bone in crocodyli- larger nerves and more axons would be necessary to con- forms and nonavian dinosaurs, making it a faithful vey the additional sensory fibers necessary for the DPR “endocast” of the trigeminal ganglion. Therefore, as an system, it is expected that the size of the trigeminal archosaur endocranial endocast is generally an accurate fossa, foramen, and related structures would reflect proxy for brain size and shape (Hopson, 1979; Witmer these changes in the innervation patterns of the face. and Ridgely, 2009), it is expected that the size of the tri- Before inferences can be made about the evolution of the geminal fossa is an accurate representation of the size trigeminal system in fossil crocodyliforms, patterns must and shape of the trigeminal ganglion. The bony con- be tested among extant taxa first. struction of the trigeminal foramen is quite similar This article explores the sizes of the trigeminal gan- among extant crocodylians (Brochu, 1999) and despite glion, the relative distributions of the mandibular and several changes associated with the palate’s and epipter- maxillary nerves, and their relationships to brain and ygoid’s fusion onto the braincase during crocodyliform skull size in Alligator mississippiensis. We use these evolution, the maxillomandibular portion of the foramen metrics to identify scaling relationships among the nerv- has remained largely consistent in its construction over ous tissues to gauge the accuracy by which the bony the course of 200 million years of morphological change structures reflect the soft tissue anatomy. We then eval- (Holliday and Witmer, 2009). However, when the DPR uate these results for their implications for sensory dis- system evolved within crocodyliforms, and, if present, tribution along the face of the alligator. Finally, how the system was utilized in numerous Mesozoic ter- crocodyliforms underwent a significant diversification restrial crocodyliform taxa remains unclear. during the Mesozoic resulting in several independent Thus, insights from the trigeminal nerve may shed lineages of marine, semi-aquatic, as well as terrestrial new light on the evolution of this important cranial sen- taxa (Brochu, 2003). Thus, it would be expected that sory system. Yet, changes in the size of the trigeminal changes in the trigeminal system, such as an increase or nerve divisions, the trigeminal fossa, and maxilloman- decrease in relative trigeminal nerve size might accom- dibular foramen have not been explored, and the degree, pany life in these varying habitats. Although a 4 GEORGE AND HOLLIDAY comprehensive evolutionary analysis and transforma- 3D solid models were generated from the CT segmenta- tional hypothesis is beyond the scope of this article, we tion for measurement and analysis. The vertical diame- include relevant metrics from several extant crocodyli- ter of the maxillomandibular foramen was measured ans including the caiman Melanosuchus niger and two using both calipers on the macerated skulls and virtual crocodylids (Crocodylus niloticus, C. johnstoni [OUVC measurement tools on the segmented CT data to test 10425]). In addition, we included similarly sized fossil repeatability and similarity of measurement methods. crocodyliform taxa: the semi-aquatic basal brevirostrine An axial plane traversing the caudal edge of the foramen Leidyosuchus canadensis (ROM 1903); the putatively magnum marked the posterior extent of the endocast. terrestrial peirosaur Hamadasuchus rebouli (ROM Obliquely parasagittal planes separated each internal 52560); and the marine dyrosaur cf. Rhabdognathus cranial foramen from the endocranial cavity. The olfac- (CNRST-SUNY-190) to illustrate the potential utility of tory tract and bulbs were segmented and included as trigeminal morphometrics in archosaur and crocodyli- part of the cranial endocast. The trigeminal fossa was form cranial somatosensory evolution. segmented to include all space bounded by the lateros- phenoid, prootic, and quadrate. The trigeminal fossa was MATERIALS AND METHODS outlined by a parasagittal plane traversing the trigemi- nal canal medially, an axial plane through the caudal Nine frozen heads of Alligator mississippiensis were portion of the ophthalmic canal rostrally, a parasagittal acquired from Rockefeller State Wildlife Refuge (Grand plane traversing the edge of the maxillomandibular fora- Chenier, LA). Specimen skull lengths, measured from men laterally, and then the prootic-quadrate suture cau- the tip of the snout to the caudal edge of the skull table, dally. The ophthalmic canal was not included in the ranged from 120.90-mm long to 307.88-mm long repre- reconstruction. CT data were segmented in each ana- senting juvenile (about 120-mm skull length), subadult tomical plane. Using the CT data, volumes of the endo- (ca.190mm), and adult (246–307mm) individuals cast and the left trigeminal fossa were segmented and (Table 1). Body masses were not known. All heads were measured. CT-scanned at Cabell Huntington Hospital, Huntington, WV, or University of Missouri School of Veterinary Medi- Histology and Histomorphometry cine at 0.625-mm slice thickness prior to dissection. One large individual (AL025) was imaged with magnetic res- Specimens were thawed and dissected to expose the onance at the Brain Imaging Center at the University of left mandibular nerve and left proximal maxillary nerve Missouri, Columbia at a 0.5mm slice thickness using T1 (Fig. 3). Nerve samples were taken from four sites along and T2 weighted optimizations. After imaging and dis- the mandibular nerve and one site along the maxillary section, each head was skeletonized (Fig. 2). Additional nerve (Fig. 4). Site V3-1 was the most proximal portion extant and fossil crocodyliforms were scanned at O’Ble- of the mandibular nerve after emerging from the trigem- ness Memorial Hospital (Athens, OH), Cabell-Hunting- inal ganglion and was sampled to obtain a total count of ton Hospital (Huntington, WV) or Royal Veterinary the axons in the nerve as it emerges from the maxillo- College at 0.625-mm slice thickness. mandibular foramen with before any branches leave the All CT and MRI data were imported into Amira v5.2 main nerve. Site V3-2 was immediately distal to the (Visage Imaging) for segmentation and morphometric branching of the mylohyoid nerve and was sampled to analysis. The entire skull was segmented from the CT obtain a count of only sensory axons that pass into the series and the mandibular nerve and proximal maxillary mandible. Site V3-3 was immediately distal to the nerve were segmented from the MRI series (Fig. 1). The branching of the internal oral nerve and site V3-4 was TABLE 1. Skull length, brain, and trigeminal fossa volumes for measured Alligator mississippiensis and other crocodyliforms Skull Endocranial CN V fossa Olfactory Endocranial Specimen length (sl) volume (ev) volume (Vfv) tract volume 1 olfactory tract AL 001 120.90 3216.65 41.65 353.44 3570.10 AL 002 189.87 7468.46 164.36 1615.88 9084.34 AL 003 127.80 3391.92 58.27 370.65 3762.56 AL 004 189.51 7763.21 161.23 1193.29 8956.49 AL 005 294.37 12720.11 329.76 2982.05 15702.16 AL 015 280.83 10932.13 362.55 2939.42 13871.55 AL 016 307.88 13376.16 337.30 3214.57 16590.73 AL 017 259.28 10550.55 251.18 2009.14 12559.70 AL 025 246.16 9354.26 183.13 1951.10 11305.36 Melanosuchus niger 276.13 20969.65 388.06 Crocodylus johnstoni 299.10 10723.28 189.14 Crocodylus niloticus 516.86 30722.77 205.37 Leidyosuchus. canadensis 380.00 16210.00 430.00 cf. Rhabdognathusa 330.0* 23912.20 151.50 Hamadasuchus. rebouli 326.90 11527.70 115.30 Skull length is measured in mm. Brain and trigeminal fossa volumes are measures in mm3. a Rhabdognathus skull length measurement is 274.0 mm but is missing rostralmost portion of snout; we estimated length to be 330.0 mm.  CROCODYLIFORM FACIAL SENSATION AND EVOLUTION 5 Fig. 2. Anatomy and phylogenetic relationships of extant and fos- representation from alligator sample with corresponding endocast. sil crocodyliforms used in this study. (A) Left lateral view of skulls A. mississippiensis specimens smallest to largest are AL001, AL and associated endocasts of extant and fossil species included in 04, and AL016. Endocasts are blue and trigeminal ganglion is yel- our study. Alligator mississippiensis AL016, *Melanosuchus. niger, low. Scale bars are ten cm for skulls and five cm for endocasts. *Crocodylus johnstoni (OUVC 10425), *Crocodylus. niloticus, Lei- Nodes: 1, Neosuchia; 2, Crocodylia; 3, Crocodylidae; 4, Alligatori- dyosuchus canadensis (ROM 1903), cf. Rhabdognathus (CNRST- dae. *, specimens from personal collection of JR Hutchinson, SUNY-190), Hamadasuchus rebouli (ROM 52620). (B) Ontogenetic Royal Veterinary College. 6 GEORGE AND HOLLIDAY Fig. 3. Soft tissue anatomy of trigeminal nerve (CNV) branches within 5 and V3-1 sample sites. (D) Lateral view of V3 showing the nerve to the alligator head (AL 017). (A) Overall dissection of orbit and temporal mylohyoid branch and V3-2 sample site. (E) Medial view of mandible fossa. (B) Dissection of orbit showing the ophthalmic, maxillary, and showing the V3-3 and V3-4 sample sites. Abbreviations: V1, ophthalmic mandibular divisions of the trigeminal nerve emerging from the lateral nerve; V2, maxillary nerve; V3, mandibular nerve; EMF, external mandib- wall of the braincase. (C) Lateral view of temporal region showing V2 ular fenestra; Ls, laterosphenoid; mmf, maxillomandibular foramen; and V3 divisions emerging from the maxillomandibular foramen and V2- nMh, mylohyoid nerve (intermandibular nerve); Sp, splenial. Fig. 4. Nerve histology and relative axon proportions of the trigemi- arranged by skull length. The proximal portion (V3-1) includes both nal nerve. (A) The 3D model of alligator skull (grey), trigeminal ganglion motor and sensory axons whereas V3-2 through V3-4 portions con- and trigeminal nerve branches (yellow), and nerve sample sites. (B) tains only sensory axons. (C-E) Nerve cross sections (V3-1, V3-2, and Proportions of nerve fibers of mandibular nerve at each sample site V3-4 respectively) at 34 and nerve fibers from each section at 340.  CROCODYLIFORM FACIAL SENSATION AND EVOLUTION 7 TABLE 2. Numbers of axons counted at specific sites of trigeminal divisions in Alligator mississippiensis as depicted in Fig. 4. V3-1 V3-2 V3-3 V3-4 V2-5 Specimen Axons Density Area Axons Density Area Axons Density Area Axons Density Area Axons Density Area AL001 26619 478.25 5.065 22232 764 2.648 10661 599.25 1.619 3223 406.67 0.721 16287 574.67 2.579 AL002 26512 282.00 8.574 19531 446.75 3.987 8532 307.00 2.529 2771 227.00 1.111 20425 431.00 4.322 AL003 26162 740.75 3.221 21256 619.75 3.128 9642 647.50 1.358 2165 480.33 0.411 11262 518.75 1.980 a a a AL004 30718 266.50 10.512 14331 253.25 5.161 8842 259.875 3.103 20847 259.875 7.316 a a a AL005 19088 351.00 4.954 14876 359.00 3.775 1208 157.25 0.700 22018 350.25 5.727 AL015 34293 250.50 12.485 29564 264.75 10.184 16292 242.75 6.121 4911 234.50 1.910 17901 253.43 6.442 a a a AL016 31095 226.75 12.507 18400 231.37 7.253 2541 201.67 1.149 13971 148.00 8.609 AL017 37289 299.75 11.333 20106 427.86 4.281 14775 484.00 2.781 5310 462.00 1.047 25895 352.25 6.697 AL025 36228 262.75 12.575 17722 174.00 9.289 10020 200.00 4.569 3868 276.00 1.278 20737 258.75 7.309 Density is equal to the number of axons per ROI as described in Materials and Methods. Area is measured in mm2. a Nerve site that was unable to be used to obtain an axon count or cross-sectional area. the most distal portion of the main branch of the man- juvenile nerves had an average density of 583 axons per dibular nerve in the mandible that supplies the symphy- ROI (626). seal region. Site V2-5 was the most proximal portion of Mean axon count was 31,115 (64,480 SD) at site V3-1; the maxillary nerve and was sampled to provide a com- 20,248 (64,160) at site V3-2; 11,705 (63,093) at site V3- parison to the number of sensory fibers along the upper 3; 3,249 (61,387) at site V3-4; and 18,815 (64,452) at parts of the face. A Dremel rotary tool was used to ex- site V2-5. Mean cross-sectional area was 9.5 mm2 (63.6 pose the inferior alveolar canal and gain access to the mm2) at site V3-1; 5.6 mm2 (62.7 mm2) at site V3-2; 3.3 distal portions of the mandibular nerve. A section of mm2 (61.6 mm2) at site V3-3; 1.1 mm2 (60.5 mm2) at approximately one cm was excised from each nerve. site V3-4; and 5.66 mm2 (2.25 mm2) at site V2-5. Full Harvested nerves were fixed in 3% glutaraldehyde for nerve cross-sectional area and axon count for each sam- 48 hr immediately following dissection. Nerve samples ple site are recorded in Table 2. were processed, embedded in paraffin, and sectioned as In the six specimens that had all four sites along the 5-mm slices on a Leica rotary microtome, mounted, and mandibular nerve well represented, about 28.59% of the stained with hematoxylin and eosin (H&E) stain. Nerve total number of axons branched away from the main slides were photographed on an Olympus BX41TF trunk before site V3-2. These fibers include virtually all microscope with an Olympus DP71 at 43 and 403. Mul- of the motor rami to the jaw muscles as well as several tiple photos of each nerve at 43 magnification were significant sensory rami. Another 32.71% of the total taken and collaged to image the entire nerve cross-sec- number of axons branched between V3-2 and V3-3. This tion. These images were combined in Adobe Photoshop means that about 40% of the mandibular nerve solely CS2 and imported into ImageJ to obtain the cross-sec- innervates the mandible proper. Within the mandible, tional area of each nerve section. Four 403 images were 26.46% of the total number of axons branched between taken at random sites within each nerve section for indi- V3-3 and V3-4 and 12.24% of the total number of axons vidual axon counting. Each of the 403 images was terminated after site V3-4. counted manually within ImageJ’s Cell Counting mod- ule. Total axon number in each nerve cross-section was Endocast and Trigeminal Fossa Volume calculated from the axon number per 403 region of in- terest and then multiplied by the total cross-sectional Mean endocast volume for the sample was 8,753 mm3 area of the nerve to estimate the total number of axons (63,665 mm3). Further separating the sample into juve- in each cross-section. Scaling relationships among varia- nile, subadult, and adult ages we find endocranial vol- bles were determined via reduced major axis (RMA) ume averages of 3304 mm3 (6124 mm3), 7,616 mm3 regression analysis conducted in NCSS 2007 statistical (6208 mm3), and 11,387 mm3 (61,641 mm3), respec- software using an adjusted R (Rbar) for small sample tively. Mean trigeminal fossa volume for the sample was size, R-squared, a post hoc Bonferroni adjusted P value, 210 mm3 (6118 mm3). Separation by approximate aged and slope analysis using confidence intervals to test the yielded 50 mm3 trigeminal fossa volume (612 mm3) for hypothesis that the experimentally derived slope differed juvenile, 163 mm3 (62 mm3) for subadult, and 293 mm3 from that expected for particular equations for isometry. (674 mm3) for adult alligators. Full results for volumes are shown in Table 1. RESULTS Scaling Analysis Axon Count The volume of the trigeminal fossa correlated signifi- After sectioning and mounting all nerve samples, cantly with endocranial volume (r 5 0.98) and skull three sites out of 45 were of unsuitable quality for cross- length (r 5 0.98) (Table 3, Fig. 5). Endocast volume cor- sectional area or axon counting. It was noted that axon related significantly with skull length (r 5 0.99). The density was uniform across the sample except for the vertical diameter of the maxillomandibular foramen two youngest specimens. The sub adult and adult nerves strongly correlates with the volume of the trigeminal had an average density of 285 axons per ROI (670). The fossa (r 5 0.93). The maxillomandibular foramen 8 GEORGE AND HOLLIDAY TABLE 3. Results of regression analysis of skeletal variables including variables (y vs. x), regression equation, adjusted Pearson correlation coefficient (r), R2, P value, expected slope of isometry, and confidence interval of regression slope (CI) Variables (Y v X) Equation r R2 P value Isometry CI Alligator Log mmf 3 log sl y 5 (20.5686) 1 (0.6102)x 0.96 0.90 0.0001 1.0 0.44–0.78 Log ev 3 log sl y 5 (0.4246) 1 (1.4909)x 0.99 0.97 0.0001 3.0 1.27–1.71 Log Vfv 3 log sl Y 5 (22.7741) 1 (2.1494x 0.98 0.95 0.0001 3.0 1.73–2.57 Log mmf 3 log ev Y 5 (20.7298) 1 (0.4060)x 0.96 0.91 0.0001 0.33 0.30–0.51 Log Vfv 3 log ev Y 5 (23.3506) 1 (1.4326)x 0.98 0.96 0.0001 1.0 1.20–1.67 Log mmf 3 log Vfv Y 5 (0.2465) 1 (0.2715)x 0.93 0.86 0.0002 0.33 0.18–0.36 Log V3.1 axon count 3 log sl Y 5 (0.2854) 1 (3.8249)x 0.74 0.49 NS 1.0 0.06–0.52 Log V2.5 axon count 3 log sl Y 5 (0.3318) 1 (3.4901)x 0.46 0.09 NS 1.0 20.25–0.91 Log mmf axon count 3 log sl Y 5 (0.301) 1 (3.9928)x 0.67 0.37 NS 1.0 0.01–0.60 Log V3.1 axon count 3 log mmf Y 5 (0.4149) 1 (4.136)x 0.89 0.74 0.0018 1.0 0.03–0.80 Log V2.5 axon count 3 log mmf Y 5 (0.5745) 1 (3.7731)x 0.91 0.79 0.0008 1.0 0.31–1.45 Log mmf axon count 3 log mmf Y 5 (0.4695) 1 (4.2936)x 0.67 0.37 NS 1.0 0.01–0.94 Pooled Alligator (AL015) and six additional crocodyliforms Log mmf 3 log sl Y 5 (0.7395) 1 (5.0727)x 0.06 0.03 NS Log ev 3 log sl Y 5 (1.1629) 1 (1.2089)x 0.62 0.27 NS Log Vfv 3 log sl Y 5 (3.3214) 1 (20.3752)x 0.16 0.17 NS Log Vfv 3 log ev Y 5 (2.2418) 1 (3.1176)x 0.03 0.20 NS Log mmf 3 log ev Y 5 (20.1239) 1 (0.2351)x 0.50 0.10 NS Log mmf 3 log Vfv Y 5 (0.2336) 1 (0.2672)x 0.70 0.39 NS Bonferroni adjusted P for Alligator-only data is P 5 0.0042; for pooled data is P 5 0.0083. NS, not significant. Abbreviations: ev, endocast volume; mmf, maxillomandibular foramen diameter; sl, skull length; Vfv, trigeminal fossa volume. significantly correlates with skull length (r 5 0.96) and nerves. Because densely packed mechanoreceptors affect endocast volume (r 5 0.96). the size of a nerve (Muchlinski, 2010), their presence, as The number of axons and nerve cross sectional area well as absence, in a somatic region is reflected in axon at each sample site show positive allometry with skull count and nerve fiber size. These results show axon length and trigeminal fossa volume except for at site V3- count and nerve cross-sectional area correlates with 2. At V3-2, the number of axons correlates with skull both trigeminal fossa volume and skull length in Alliga- size and trigeminal fossa volume but scales with nega- tor. Assuming that other extant crocodylians follow the tive allometry. Axon density at each sample site also same pattern, the trigeminal fossa volume and diameter shows a pattern of significant correlation and negative of the maxillomandibular foramen are informative allometry with skull length and trigeminal fossa volume. metrics for inferring trigeminal nerve size, and therefore informative proxies for facial sensitivity, mechanorecep- tion, and other sensory input. DISCUSSION The results illustrated here indicate that head length, Comparative Utility in Fossils brain size, and trigeminal nerve size are consistently related to each other in Alligator mississippiensis. De- These results indicate that foramen size may be an spite the small sample size, the distribution of axons is accurate predictor of nerve size and axon number in fos- largely uniform across all individuals. The larger, adult sil crocodyliforms. These findings are also important for individuals appear to have more nerves branching inferring sensory receptor density in soft tissue from within the adductor chamber (i.e., between sites V3-1 only skeletal sources. Previous investigations on cranial and V3-2) compared to smaller individuals (Figs. 2 and nerves and their branches with respect to foramen size 4). However, it remains unclear to what degree these (DeGusta, 1999; Muchlinski, 2008) and sensory mecha- changes are due to increases in motor or sensory func- noreceptors (Soares, 2002; Muchlinski, 2010) agree with tion. The decrease in axon density as individuals grow our results. More sensory receptors innervated by the older is most likely due to the additional space occupied maxillary and mandibular nerves will require more by the myelin sheath around each axon. Axon diameter axons within each nerve (Kandel et al., 2000), thus as well as the conduction velocity of peripheral nerves larger nerves, and a larger trigeminal ganglion size. relates to the thickness of the myelin sheath (Kiernan To illustrate the use of the Alligator data in a small et al., 1996; Kandel et al., 2000; Michailov et al., 2004; case study, we collected relevant data from: three extant Moldovan et al., 2006). Although the thickness of the crocodyliform species, Crocodylus niloticus, Crocodylus myelin sheath does not continue to grow throughout on- johnstoni, and Melanosuchus niger; and three extinct togeny, the youngest alligators sampled in our study crocodyliforms, the basal eusuchian Leidyosuchus cana- may not have completed myelination of their axons. densis, the sebecid neosuchian Hamadasuchus rebouli, The sizes of the maxillary and mandibular nerves and the dyrosaurid neosuchian cf. Rhabdognathus (Fig. relate to the magnitude of sensory coverage as well as 2). These fossil taxa have similar skull lengths and occu- different sensory modalities that are carried by these pied different ecological niches. Leidyosuchus canadensis  CROCODYLIFORM FACIAL SENSATION AND EVOLUTION 9 Fig. 5. Scaling analysis of alligators with supplemental extant and fossil crocodyliforms. (A–F) Blue line is the regression of Alligator-only sample with R2 as reported in Table 3. Additional extant and fossil croc- odyliforms identified in Key. 10 GEORGE AND HOLLIDAY (ROM 1903) is a basal brevirostrine crocodylian, an supports the hypothesis that the DPR system was an early representative of the clade containing crocodiles, emergent system among early eusuchians and not a alligators and caimans (Brochu and McEachran, 2000; primitive feature of neosuchian crocodyliforms. However, Brochu, 2003), from Late Cretaceous of North America. it is equally possible that dyrosaurs secondarily lost an It bears numerous features similar to modern alligators enhanced DPR system for reasons still unclear. Regard- including a platyrostral, triangular skull, mediolaterally less, this study demonstrates that trigeminal ganglion broad occiput, and a dense array of facial neurovascular size is an informative metric for analyzing sensory adap- foramina (Soares, 2002). Rhabdognathus (CNRST- tations in crocodyliforms and potentially other fossil SUNY-190; Brochu et al., 2002) is a short-snouted dyro- archosaurs. Further investigation into the rich crocodyli- saur from the Paleocene of Northern Africa. Dyrosaurs form fossil record may elucidate how different taxa were marine neosuchian crocodyliforms characterized by responded to environmental cues and when the neuro- very long, slender snouts, enlarged dorsotemporal fossae, logic osteological correlates of the DPR system first rectangular occiputs, and some facial neurovascular pit- appeared. ting. Hamadasuchus rebouli (ROM 52620) has a tall, Our investigation of the alligator trigeminal fossa and oreinirostral skull with some facial neurovascular pit- peripheral branches of the trigeminal nerve shows a ting, and is a terrestrial predator related to sebecid neo- relationship between trigeminal fossa size, and thus tri- suchian crocodyliforms (Larsson and Sues, 2007). Like geminal ganglion, and skull length. These findings sup- the Alligator sample, each specimen was CT-scanned at port the idea that crocodyliforms of a given head size, 0.625-mm slice thickness, had its cranial and trigeminal should have a predictable sensory sensitivity based on endocasts reconstructed and other measurements skeletal data. With the future addition of other species collected. to such investigations, it will be possible to make better No significant correlations were found among fossil inferences about the sensory potential in species where and extant species data pooled with an alligator of simi- only fossil data are available. More in-depth analysis of lar skull length (AL015). The relationships between cra- alligator soft tissues, specifically the most terminal nial endocast volume and skull length (r 5 0.62) and branches of the trigeminal nerves as well as DPRs will maxillomandibular foramen diameter and trigeminal help infer not only the sensory spread in an extinct spe- fossa volume (r 5 0.70) are strong. However, the relation- cies, but also which sensory modalities may have been ships between trigeminal fossa volume and skull length present. With application of similar anatomical data on (r 5 0.16), and cranial endocast volume (r 5 0.03) are extant species as a baseline, these findings suggest that weak. These findings suggest that differences in trigemi- neurologic osteological correlates of the trigeminal sys- nal fossa volume relative to brain or skull size may be tem are informative features useful for investigating due to differences in sensory magnitude rather than rel- crocodyliform as well as archosaur somatosensory ative differences with nervous tissue as a whole (Table evolution. 3, Fig. 5). Although a larger sample size is necessary to understand the relationships between the skull, brain, ACKNOWLEDGEMENTS and trigeminal nerve in crocodyliforms, we interpret the lack of clear signal among these variables to be result of The authors thank undergraduates J. Kim, R.J. Skiljan, phylogenetic or behavioral effects associated with adap- and C. Gant for assistance in the lab. They thank Ruth tations of the cranial sensory system rather than size Elsey and Rockefeller State Refuge, LA for supplying al- alone. ligator material. They thank JR Hutchinson (Royal Vet- Leidyosuchus had endocranial and trigeminal fossa erinary College), La Ferme Aux Crocodiles, France and volumes similar to an alligator of similar skull length. St. Augustine Alligator Farm, US for sharing data of C. Rhabdognathus had a large endocranial volume com- niloticus and M. niger and LM Witmer (Ohio University) pared to that expected for an alligator of similar skull for sharing CT data of cf. Rhabdognathus and C. john- length and a trigeminal fossa volume that is similar to stoni (via JRH). that found in a similar-sized alligator. Hamadasuchus had small endocranial and trigeminal fossa volumes compared to an alligator of similar size and a relatively LITERATURE CITED small trigeminal fossa in relation to its endocranial vol- Allen D. 2005. The inferred evolutionary history of integumentary ume. Thus, Hamadasuchus and Rhabdognathus had sense organs in crocodylomorphs. J Vertebr Paleontol 25(Suppl to smaller trigeminal fossae, and thus likely decreased fa- 3):31A. cial sensation compared to Leidyosuchus and Alligator. Brochu CA. 1999. Phylogenetics, taxonomy, and historical biogeog- Several hypotheses could explain this pattern, though raphy of Alligatoroidea. J Vertebr Paleontol 19:9–100. they remain poorly supported given our small sample Brochu CA. 2003. Phylogenetic approaches toward crocodylian his- size. First, the larger trigeminal ganglia found in Lei- tory. Annu Rev Earth Pl Sci 31:357–397. dyosuchus and Alligator may be derived compared to Brochu CA, McEachran JD. 2000. Phylogenetic relationships and divergence timing of Crocodylus based on morphology and the fos- other crocodyliforms and reflect eusuchian origins of the sil record. Copeia 2000:657–673. derived DPR system. Second, the smaller trigeminal Brochu CA, Bouar e ML, Sissoko F, Roberts EM, O’Leary MA. 2002. ganglia in Hamadasuchus and Rhabdognathus may A dyrosaurid crocodyliform braincase from Mali. J Paleontol reflect an ecologically relevant, diminished sense of face 76:1060–1071. touch compared to other crocodyliforms. One might DeGusta D, Gilbert W, Turner S. 1999. Hypoglossal canal size and expect that terrestrial crocodyliforms such as Hamada- hominid speech. Proc Natl Acad Sci USA 96:1800–1804. suchus had less need for DPR-based facial sensation, Dehnhardt G, Kaminski A. 1995. Sensitivity of the mystacial vibris- and thus possessed smaller trigeminal ganglia. However, sae of harbor seals (Phoca vitulina) for size differences of actively the decreased ganglion size in the marine dyrosaur touched objects. J Exp Biol 198:2317–2323.  CROCODYLIFORM FACIAL SENSATION AND EVOLUTION 11 Ebara S, Kumamoto K, Matsuura T, Mazurkiewicz J, Rice F. 2002. Molenaar GJ. 1978a. The sensory trigeminal system of a snake in Similarities and differences in the innervation of mystacial vibris- the possession of infrared receptors. I. The sensory trigeminal sal follicle-sinus complexes in the rat and cat: a confocal micro- nuclei. J Comp Neurol 179:123–135. scopic study. J Comp Neurol 449:103–119. Molenaar GJ. 1978b. The sensory trigeminal system of a snake in Gregory JE, Iggo A, Mcintyre AK, Proske U. 1987. Electroreceptors the possession of infrared receptors. II. The central projections of in the platypus. Nature 326:386–387. the trigeminal nerve. J Comp Neurol 179:137–151. Holliday CM, Witmer LM. 2007. Archosaur adductor chamber evo- Morhardt A. 2009. Dinosaur smiles: do the texture and morphology lution: integration of musculoskeletal and topological criteria in of the premaxilla, maxilla, and dentary bones of sauropsids pro- jaw muscle homology. J Morphol 268:457–484. vide osteological correlates for inferring extra-oral structures reli- Holliday CM, Witmer LM. 2009. The epipterygoid of crocodyliforms ably in dinosaurs? MS Thesis, Western Illinois State University, and its significance for the evolution of the orbitotemporal region Macomb, IL. of eusuchians. J Vertebr Paleontol 29:715–733. Morhardt A, Bonnan MF, Keiler T. 2009. Dinosaur smiles: correlat- Hopson JA. 1979. Paleoneurology; pp. 39–146 in C. Gans (ed.), ing premaxilla, maxilla, and dentary foramina counts with extra- Biology of the Reptilia Academic Press, New York. oral structures in amniotes and its implications for dinosaurs. J Jungers WL, Pokempner AA, Kay RF, Cartmill M. 2003. Hypoglos- Vert Paleont, 29 (Suppl to 3): 152A. sal canal size in living hominoids and the evolution of human Muchlinski MN. 2008. The relationship between the infraorbital fo- speech. Hum Biol 75:473–484. ramen, infraorbital nerve, and maxillary mechanoreception: Kandel E, Schwartz J, Jessell T. 2000. Principals of neural science. implications for interpreting the paleoecology of fossil mammals 4th ed. New York: McGraw-Hill. based on infraorbital foramen size. Anat Rec 291:1221–1226. Kay RF, Cartmill M, Balow M. 1998. The hypoglossal canal and the Muchlinski MN. 2010. A comparative analysis of vibrissa count and origin of human vocal behavior. Proc Natl Acad Sci USA 95:5417– infraorbital foramen area in primates and other mammals. J 5419. Hum Evol 58:447–473. Larsson HCE, Sues H-D, 2007. Cranial osteology and phylogenetic Nicolelis M, Lin R, Chapin J. 1997. Neonatal whisker removal relationships of Hamadasuchus rebouli (Crocodyliformes: Mesoeu- reduces the discrimination of tactile stimuli by thalamic ensem- crocodylia) from the Cretaceous of Morocco. Zool J Linn Soc bles in adult rats. J Neurophysiol 78:1691–1706. 149:533–567. Soares D. 2002. An ancient sensory organ in crocodilians. Nature Leitch DB, Catania KC. 2012. Structure, innervation and response 417:241–242. properties of integumentary sensory organs in crocodilians. Jour- Tsai HP, Holliday CM. 2011. Ontogeny of the alligator cartilage nal of Experimental Biology 215:4217–4230. transiliens and its significance for sauropsid jaw muscle evolu- Kiernan MC, Mogyoros I, Burke D. 1996. Differences in the recov- tion. PLoS One 6:e24935. ery of excitability in sensory and motor axons of human median von During M. 1973. The ultrastructure of lamellated mechanore- nerve. Brain 119:1099–1105. ceptors in the skin of reptiles. Anat Embryol 143:81–94. Manger PR, Perrigrew JD. 1996. Ultrastructure, number, distribu- von During M. 1974. The ultrastructure of the cutaneous receptors tion and innervation of electroreceptors and mechanoreceptors in in the skin of Caiman crocodilus. Abhandlungen Rhein.-Westfal. the bill skin of the platypus, Ornithorhynchus anatinus (part 1 of Akad 53:123–134. 2). Brain Behav Evol 48:27–40. von During M, Miller MR. 1979. Sensory nerve endings in the skin Michailov GV, Sereda MW, Brinkmann BG, Fischer TM, Haug B, and deeper structures. In: Gans C, Northcutt RG, Ulinski P, edi- Birchmeier C, Role L, Lai C, Schwab MH, Nave K. 2004. Axonal tors. Biology of the reptilia, 9. Neurology A. New York: Academic neuregulin-1 regulates myelin sheath thickness. Science 304:700– Press. p 407–442. 703. Wineski L. 1983. Movement of the cranial vibrissae in the golden Moldovan M, Sorensen J, Krarup C. 2006. Comparison of the fastest hamster (Mesocritus auratus). J Zool 200:261–280. regenerating motor and sensory myelinated axons in the same pe- Witmer LM, Ridgely RC. 2009. New insights into the brain, brain- ripheral nerve. Brain 129:2471–2483. case, and ear region of tyrannosaurs (Dinosauria, Theropoda), Molenaar GJ. 1974. An addition trigeminal system in certain with implications for sensory organization and behavior. Anat Rec snakes possessing infrared receptors. Brain Res 78:340–344. 292:1266–1296.