SHAPE MEETS FUNCTION: STRUCTURAL MODELS IN PRIMATOLOGY

Edited by Emiliano Bruner

Proceedings of the 20th Congress of the International Primatological Society Torino, Italy, 22-28 August 2004

MORPHOLOGY AND MORPHOMETRICS JASs Journal of Anthropological Sciences Vol. 82 (2004), pp. 47-66

A geometric morphometric approach to airorhynchy and functional cranial morphology in Alouatta (Atelidae, )

Emiliano Bruner1,2, Simone Mantini1 & Giorgio Manzi1,2

1) Dipartimento di Biologia Animale e dell’Uomo, Università La Sapienza, Piazzale Aldo Moro 5, 00185 Roma. e-mail: [email protected]

2) Istituto Italiano di Paleontologia Umana, Piazza Mincio 2, 00198 Roma

Summary – The skull of the howler monkeys (Alouatta spp., Atelidae) is characterised by a generalised rotation of the splanchnocranium with respect to the neurocranial antero-posterior axis. This process, referred to as airorhynchy, is the result of a derived structural relationship between basicranium, vault, and facial districts. A number of variables – such as diet and social behaviour – probably co-evolved with the remodelling of the cranial functional matrix. We used a landmark-based analysis to explore the geometri- cal model of the skull in the genus Alouatta. Shape comparisons were performed by using superimposition procedures and the Euclidean distance matrix. In the latter analysis, a method is proposed in order to visu- alise variations of form through chromatic maps and interpolant functions. The comparison with other gen- era of Atelidae shows a marked neurocranial flattening in Alouatta as well as muzzle projection and enlargement, nuchal flattening, relative basicranial lengthening, and tilting of the occipital foramen. Only minor differences were visible in relation to facial shape, suggesting that significant changes depend on the relationship between splanchnocranium and neurocranium, rather than on localised anatomical variations. The limited vault development constrained by the basicranial structures probably involved the extreme retroflexion of the basal angle. Airorhynchy can be interpreted as an additional adjustment to fit this struc- tural network beyond the biomechanical range of the cranial base hypoflexion. This cranial functional matrix is directly related to feeding and social changes, representing an interesting evolutionary “package”. In Pongo pygmaeus a similar process is associated with a different structural pattern, mostly related to the flattening of the upper facial structures, maxillary midsagittal enlargement, and palatal tilting.

Keywords – Alouatta, functional craniology, airorhynchy, Pongo, geometric morphometrics, Euclidean distance matrix analysis.

Introduction The genus Alouatta includes about 10 and up to 19 subspecies distributed between The genera Alouatta, Ateles, Brachyteles, and Mexico and Argentina (see Groves, 2001). They Lagothrix, are the living representatives of the are generally divided into a palliata group (main- family Atelidae Gray 1825, forming a mono- ly Central America) and a seniculus group (South phyletic taxon. Extant howlers are grouped into America, including A. caraya). Despite the con- the Mycetinae subfamily, which includes the fos- troversies on the phylogenetic relationships with- sil genus Stirtonia (Szalay & Delson, 1979) and in the subfamily Atelinae, there is a general possibly the subfossil genus Paralouatta (Rivero agreement about the phyletic independence of & Arredondo, 1991; but see Horovitz & Alouatta with respect to all the other genera McPhee, 1999). Extant howler monkeys (genus (Rosenberg, 1981; Horovitz & Mayer, 1995; Alouatta) represent one of the most distinctive Schneider et al., 1996). According to a mt-DNA taxa of the Neotropical primatological fauna. analysis, the two subfamilies show a divergence 48 Alouatta skull morphology

time close to 16 million years before present latter is associated with the social structure of the (Ma), while the Trans-Andean and Cis-Andean howlers. Although the development of the Alouatta groups may diverged about 7 Ma mandible was hypothesised to have the principal (Cortés-Ortiz et al., 2002). The evolutionary role within this structural network (Osman Hill, patterns are not as clear, because of the mosaic 1962), it is not possible to split its causes and variability expressed in the atelids, which proba- consequences in an evolutionary perspective, as bly includes a large percentage of plesiomorphic these processes depend on a general rearrange- characters as well as parallelisms and homoplastic ment of the whole functional matrix of the skull. features. Alouatta and Brachyteles share many Interestingly, airorhynchy has also been dental traits because of the comparable size-relat- described in Pongo pygmaeus, associated –- as in ed dietary specialisation. Alouatta and Lagothrix Alouatta – with the enlargement of the masseter share many postcranial traits, which were influ- and the development of vocal sacs (Shea, 1985). enced by the locomotion patterns (see Szalay & Alouatta also shares some morphological affini- Delson, 1979), and probably some plesiomor- ties with the cranium of the fossil phic features like the enlarged zygomatico-faciale genus Aegyptopithecus, mostly in the frontal areas foramen displayed by the Oligocenic fossil genus (Simons, 1987). It has also been hypothesised Parapithecus (Simons, 2004) and occasionally by that the two genera share a similar body size and Aotus (quoted in Osman Hill, 1962: p. 22). locomotory system (see Rasmussen, 2002). A Alouatta displays a set of rather interesting comparison between Alouatta and Aegyptopithecus features when compared to the other Atelids, could thus be useful to suggest comments on the leading early authors to describe this taxon as evolutionary history of the howlers’ morphology. “the most derived [genus] of the recent Cebidae”, Although Alouatta represents a unique and “unmistakable … on account of its peculiar extremely specialised taxon within the evolution- form”, and with a “bestial appearance” (see ary radiation of extant primates, the literature on Osman Hill, 1962). A trichromatic vision its morphology and anatomy is rather scarce and evolved in howlers independently from the mostly based on general anatomical descriptions. catarrhine evolution (see Jacobs, 2004; Heesy This explorative analysis is aimed at creating a and Ross, 2004). The cranial anatomy is very basic framework for future investigations on this specialised. The hyoid bone is extremely devel- genus, through a “geometric dissection” of its oped, forming large vocal sacs (Schön, 1971) cranial morphology. Different superimposition that strongly characterise the howlers communi- procedures were used to compare the cranial cation system accounting for the intense loud shape in Alouatta with the skull of the other calls produced by isolated or grouped individu- atelids. The process of airorhynchy in this mor- als. The cranial structure is rather different from photype was also compared with the analogous the basic morphology of the extant platyrrhini, changes described in Pongo, in order to consider because of the extreme airorhynchy. Airorhynchy differences and affinities between these two line- can be defined as a dorsoventral rotation of max- ages. Finally, we performed a comparison with illary structures, or the upward rotation of the cranial shape in Aegyptopithecus in order to splanchnocranial functional axis onto the neuro- propose some general considerations on the evo- cranial one. As a consequence of to the subse- lution of the Mycetinae cranial anatomy. quent separation between face and braincase, the mandibular ramus enlarges to fill the structural Materials and methods gap between the occlusal plane and the mandibular articulation. In Alouatta, the Sample enlargement of the ramus is associated with the In this explorative analysis, Euclidean three- development of the masseter, and the develop- dimensional coordinates were sampled from ment of the hyoid bone. The former morpholo- eight adult Alouatta belonging to the seniculus gy is associated with a folivourous diet, while the group, and averaged in order to compute a mean E. Bruner et al. 49

configuration of landmarks. Only one adult with MORPHEUS ET AL. (Slice, 2000) by align- specimen represents each of the other atelids ment of the whole sample according to the genus. Unfortunately, the rear vault of the only Principal Component eigenvectors, and subse- Lagothrix presently available is damaged, and quently projecting the resulting coordinates on a only the facial structures were compared. One lateral plane (only the left side was used in the Cebus was included as non-Atelids morphologi- shape comparisons). The coordinate systems cal outgroup, as well as a cast of the 1966 com- were superimposed by Procrustes (both 3D and plete skull of Aegyptopithecus (see Rasmussen, 2D data) and Bookstein Superimposition (2D 2002). We also considered one adult male Pongo data) using MORPHEUS ET AL. (Slice, 2000), and and one adult male Pan in order to compare the shape differences were visualised using geometri- airorhynchy in howlers and orangs. Although cal comparison and thin-plate spline (Bookstein, this descriptive analysis cannot check the within- 1991). The first procedure operates a translation genus and between-sexes variability, it is assumed of the coordinate systems to a common centroid, that the intergeneric differences should be larger scaling to unitary centroid size, and a rotation to allow a generic and explorative morphological between corresponding landmarks using a least- comparison. Data were sampled at the Museum squares approach. The second procedure super- of Anthropology Giuseppe Sergi, in Rome imposes the configuration onto a common base- (Bruner & Manzi, 2001). line. Bookstein superimposition was used to compare 2D configurations relatively to the Shape comparison neurocranial length (nasion-opisthocranion). We used landmark coordinates and superim- The information available from these two position to compare the cranial geometry in dif- superimpositions is rather complementary: ferent taxa, both in two and three dimensions. A while the first normalises the effect of size dif- three-dimensional bilateral configuration of 42 ferences (i.e. approaching “shape”) the second is landmarks was selected to describe the cranial useful to compare phenotypic variations with morphology (Tab. 1; Fig. 1). The entire skull was respect to a specific functional/structural refer- considered, as well as a sub-configuration includ- ence (the neurocranial axis). ing only the facial morphology. Landmarks were Procrustes distances were computed by sampled with a Microscribe 3DX (Immersion TPSSMALL 1.19 (Rohlf, 1998) and visualised Corporation). Using the three-dimensional according to the unweighted pair-group method dataset, two-dimensional data were computed using arithmetic averages (UPGMA) cluster pro-

Fig. 1 - Complete three-dimensional configuration of landmarks (42), shown in lateral view (basicranial landmarks are not visible). Landmarks are plotted on the skull of Alouatta (a) and linked to produce a geometric reference model (b). See Table 1 for landmarks description. 50 Alouatta skull morphology

cedure by SYNTAX 2000 (Podani, 1997), in order ogy, and this is the case with the atelids skull to check the main phenotypic affinities. (Fig. 2). The inter-orbital area represents a very Euclidean Distance Matrix Analysis important source of information, because of its (EDMA) was used to perform a further compar- role in convergence, encephalisation, and in the ison, in order to check form differences regard- relationship between face and vault. Landmarks less of scaling procedures (see Richtsmeier et al., from this district were not included in the pres- 2002). A subset of 17 lateral-projected land- ent study, because of the lack of correspondence marks was used to limit the number of inter- between the different taxa. Usually, the maxillary landmark distances. EDMA were performed processes reach the frontal suture separating the using WINEDMA (Cole 2002). Interlandmark ipsilateral nasal and lacrimal bones. This pattern distances exceeding 2 standard deviations from allows the recognition of two osteometric points, the mean value of the whole Form Difference namely maxillofrontale and dacryon (see White & Matrix (FDM) were considered as determinant Folkens, 2000). This is also the pattern found in interlandmark distances. Landmarks with a medi- Cebus. In Lagothrix, the maxillary processes meet an distance value exceeding 5% of the average above the nasal bones, whose lateral borders con- median value were considered as influential land- verge at nasion, while the two maxillofrontale marks. The median variation of each landmark converge toward the midsagittal plane. from the FDM was used to compute and map an Brachyteles shows a marked superior thinning of interpolating function in order to visualise the the maxillary processes, and a tendency to over- pattern of form differences (Form Difference lap dacryon with maxillofrontale. In Ateles the Maps). Maps were computed by SURFER 7.0 maxillary processes do not reach the frontal (Golden Software, Inc.), using a radial basis suture, being separated by the lacrimal bones. function through multiquadric interpolation Alouatta displays a pattern similar to Ateles, with (Carlson & Foley, 1991), suggested for small an almost constant lacrimo-nasal suture that, samples and scattered data. Interpolations with however, is limited in young specimens (Osman- other common functions gave overall similar Hill, 1962). Furthermore, the nasal bones are results. These maps allow a visualisation of the extremely different in the four genera: broad and form differences, describing patterns of increas- short in Lagothrix, curved and converging ing/decreasing interlandmark distances. The upward in Brachyteles, thin and converging in minimum represents areas of relative shortening, Ateles, and almost parallel in Alouatta. Other and the maximum value represents areas of rela- functional areas such as the temporal fossa tive lengthening. The intermediate value repre- (Bruner et al., nd) display a similar condition sents a mean size difference between the two among the Atelids skull. Clearly, further studies forms, expressed as a ratio between each inter- are needed to describe the intra-generic and landmark value (ratio = 1 means no difference). intra-specific variability of these features. It is worth noting the difficulties in considering bio- Anatomical notes on the cranial structure of the logical homology in these variable structures. Atelids Future analyses must be necessary developed so The configuration used in this explorative as to carefully consider the biological correspon- analysis is based almost entirely on type I land- dence. marks (Bookstein, 1991), according to the bio- logical principle of homology. In general, in geo- Results metric morphometrics the operational homology concept can offer a more useful reference (Smith, Three-dimensional shape comparisons 1990), in considering the structural parameters Figure 3 shows the skull shape 3D compar- of the anatomical system. The inter-specific vari- isons between the Alouatta morphology and the ability often makes it difficult to rely on homol- other genera considered. In Alouatta there is a E. Bruner et al. 51

Tab. 1 - Landmarks and definitions.

clear midsagittal enlargement of the premaxillary pared to Ateles, which in contrast has narrower structures and lower face, with Brachyteles also zygomatic arches. Malar bones are relatively showing a more depressed nasal area compared widened only in Cebus. The whole nuchal plane to the other specimens. The maxillary complex is rotated upward, and the foramen magnum is widens (mostly when compared with Ateles) consequently tilted backward. except when compared with Brachyteles. A The differences are less marked in consider- marked frontal bone flattening is associated ing the superimposition of the facial shape (Fig. with a loss of frontation, i.e. a marked inclina- 4). Alouatta shows a light premaxillar enlarge- tion of the orbital surface. Orbits are also rela- ment and upper facial flattening. The midface is tively narrower. This configuration of the upper also depressed, except when compared with face is extreme when compared to the Cebus Brachyteles that displays an even more flattened morphology. The cranial base of Alouatta is profile. The maxilla is relatively narrower in lengthened, because of the backward shifting of Ateles and wider in Cebus. The orbital surfaces are the spheno-occipital suture. In Cebus, the base no longer inclined, and the frontation is also is further shortened because of the backward improved because of a backward shifting of the infe- position of the posterior palatal spine. The vault rior orbitale with respect to the muzzle area. Orbits in Alouatta is extremely flattened, with the are narrower, particularly in comparison to Ateles bregma in a relatively forward position. The and Brachyteles. Because of the maxillary develop- neurocranium is relatively narrow, mostly com- ment, malars are relatively shifted backwards. 52 Alouatta skull morphology

Fig. 2 - The interorbital area in the atelids. Alouatta and Ateles show a contact between lacrimal and nasal bones. In Brachyteles the contact between the maxillary processes and the frontal bone is almost limited to a single point. In Lagothrix, the maxillary processes meet above the nasal bones, forming a intermaxillary suture. Labels: m: maxilla; n: nasal bone; l: lacrimal bone; f: frontal bone.

When the overall phenotypic affinity is con- Brachyteles. Orbits are more verticalised in Cebus. sidered as Procrustes distances, the skull of In Alouatta the entire vault (including the frontal Alouatta differs from the other Atelids (Fig. 5a) and orbital areas) is extremely flattened, the more than Cebus. In contrast, the facial shape is nuchal structures are tilted backward and highly comparable with the Brachyteles and Ateles upward, and the cranial base (posterior palatal configurations (Fig. 5b). spine to spheno-occipital suture) is lengthened. The malar area varies in each species. Brachyteles Two-dimensional shape comparisons shows the least amount of shape differences com- In figure 6 the pairwise comparisons between pared to Alouatta. The distortion grids stress the the skull shapes are computed on the 2D pro- neurocranial flattening, nuchal rotation, muzzle jected data (lateral projection, left side). The enlargement, and basicranial lengthening. Procrustes superimposition shows the maxillary Compared to Cebus, Alouatta shows a vertical enlargement and premaxillary projection in facial compression related to the premaxillary Alouatta, and the midfacial flattening in rotation, and a coronal structural compression at E. Bruner et al. 53

Fig. 3 - Pairwise comparisons of the complete configurations of landmarks after Procrustes super- imposition (left lateral and superior views). 54 Alouatta skull morphology

the level of the porion. Compared to Ateles, the ing the neurocranial axis, the orientation of the face is rotated on the neurocranial axis but not foramen magnum shows a very limited rota- vertically compressed. The coronal structural tion. It is worth noting that while the compression is localised at the level of the pteri- Procrustes superimposition better describes the on, and the maxilla is significantly enlarged. overall phenetic affinity, the Bookstein super- Compared to Brachyteles, there is a marked mid- imposition is suitable to promote structural facial enlargement, and a minor structural com- hypotheses (when the baseline is assumed to rep- pression at the level of the porion. In all cases the resent a functional reference). flexion of the neurocranial structures onto the facial one is well expressed. Pongo vs Pan Through the baseline superimposition, it is Following the Procrustes superimposition, possible to compare shapes in relation to the the main differences identified between the Pan same neurocranial length. Brachyteles is more and Pongo cranial shapes (Fig. 7) concern a similar to Alouatta than the other two taxa. The marked frontal flattening and maxillary enlarge- differences concern the upper and midfacial ment in the latter. The maxilla-premaxilla com- enlargement and neurocranial flattening. Ateles plex undergoes a sagittal development without and more specifically Cebus show a further widening, showing a relative inclination of the reduction of splanchnocranium and verticalisa- alveolar border. The orbits are smaller, but there tion of the orbits. Interestingly, by superimpos- is no loss of frontation. The neurocranium

Fig. 4 - Pairwise comparisons of the facial configurations of landmarks after Procrustes superimpo- sition (left lateral view). E. Bruner et al. 55

widens posteriorly, and the nuchal area becomes The neurocranial superimposition emphasis- shorter and taller. The foramen magnum is es the neurocranial/basicranial vertical develop- slightly tilted backwards. The zygomatic arches ment in Pongo mainly in relation to the lower enlarge. The cranial base does not lengthen, and areas, and the forward enlargement of the middle the vault does not flatten except at the level of and lower face associated to the zygomatic move- the frontal squama. ment. There is a palatal tilting and loss of fronta- Comparing the projected two-dimensional tion, without rotation or displacement of the cranial shapes in Pongo and Pan in lateral view by foramen magnum. Procrustes superimposition, the former shows a marked frontal flattening with only a minor rel- Euclidean Distance Matrix Analysis ative vertical shortening of the vault. In contrast The Euclidean distance matrix comparison the nuchal area is relatively taller. The develop- was computed between Alouatta and Brachyteles ment of the maxilla-premaxilla complex is asso- and between Pongo and Pan, to compare the sim- ciated to a rotation of the alveolar border, with- ilar airorhynchy patterns. out changes in the orientation of the lower and Alouatta/Brachyteles - The FDM shows a similar midfacial profile. The cranial base is scarcely size for the two specimens (mean and standard lengthened, and the foramen magnum is rather deviation for all the interlandmark distances: tilted backward without a posterior displace- 0.99 ± 0.13). The influential landmarks show a ment. There is a strong forward shifting of the lengthening at the rhinion (median ratio = 1.10), inferior border of the zygomatic arch, but the and shortening at superior orbital rim (0.94), malar/pteric areas are quite comparable. There is opisthocranion (0.95), and bregma (0.87). The no clear loss of frontation. The distortion grids determinant interlandmark distances show a are mainly characterised by the zygomatic shift- shift of the rhinion from the inferior orbital rim ing and by the neurocranial tilting associated to (ratio = 1.73) and a shortening of the distance the vault and frontal flattening, and nuchal between porion and the spheno-occipital suture enlargement. (0.68). Furthermore, there is a general involve-

Fig. 5 - UPGMA phenograms computed on the Procrustes distances, considering the complete (a) and facial (b) configurations (Alo: Alouatta; Ate: Ateles; Bra: Brachyteles; Ceb: Cebus; Lag: Lagothrix). 56 Alouatta skull morphology

Fig. 6 - Pairwise comparisons of the complete configuration of landmarks after two-dimension- al lateral projection (26 landmarks). Comparisons are visualised by Procrustes super- imposition of the geometrical model (left), distortion grids and thin-plate spline (mid- dle), and baseline superimposition on the neurocranial length (right).

ment of landmarks in relation to vault and molar (1.23) areas, and basion (1.21), with rela- nuchal flattening. The map of the form differ- tive shortening at the orbital rim (1.01-1.06), ence interpolation stresses the flattening of the nasion (1.00), and bregma (1.07). The determi- neurocranium and the muzzle forward develop- nant interlandmark distances again show a shift ment (Fig. 8a). of the rhinion from the inferior orbital rim (ratio Pongo/Pan - The FDM shows a larger size for = 2.10), and a relative approaching between the Pongo (1.14 ± 0.17). The influential landmarks lateral orbital border and nasion (0.65). There is show a relative lengthening at the prosthion a general involvement of landmarks related to (median ratio = 1.21), premolar (1.26) and nuchal, vault, and frontal flattening. The map of E. Bruner et al. 57

Fig. 7 - Pairwise comparison between Pan and Pongo after Procrustes superimposition (left lat- eral and superior views).

the form difference interpolation stresses a short- Cranial structure and evolution in Alouatta ening of the supraorbital landmarks and the The shape comparisons between the Atelids downward enlargement of the maxilla (Fig. 8b). genera provide a general description of the struc- Aegyptopithecus ture of the skull in Alouatta. The entire neuro- Only some landmarks are available on the cranium is extremely flattened, and at the same Aegyptopithecus cast, mostly because of the time narrower than that of the other atelids. It is preservation of the specimen. Comparing the noteworthy that Alouatta shows an early cessa- Alouatta and Aegyptopithecus cranial shape by tion of neurocranial growth related to an early superimposition, a structural affinity is easily obliteration of all the cranial sutures (quoted in recognised mostly because of the muzzle Osman Hill, 1962: p.16), and it is clear that the enlargement and because of the similar rela- timing of craniosynostosis had a major role in tionship between facial and neurocranial dis- the evolution of the cranial morphology in this tricts (Fig. 9). The maxillary complex is more morphotype. The flattening of the frontal bone developed in the Oligocenic taxon, even though is associated to a relative posterior shortening of it is relatively narrower compared to Alouatta. the frontal squama and inclination of the orbits. The neurocranium shows similar proportions. The subsequent orbital morphology characteris- The most important difference is observed at es Alouatta as probably the least frontated the foramen magnum: in both cases it is placed anthropoid (Bruner et al., 2002). It is interesting in a posterior position, but in Aegyptopithecus is to note that orbits are inclined with respect to not tilted as in the howler. The distortion grid the neurocranial axis, but not considering the from the lateral projection stresses the relative facial morphology alone. The morphogenetic splanchnocranial development and the absence processes leading to a generalised flattening of of a clear inclination of the occipital foramen in the neurocranial and basicranial structures occur the former. almost entirely during the post-natal growth, with the early ontogenetic stages showing a more Discussion typical basikyphosis that is partially preserved in adults along the brainstem (Osman Hill, 1962). In view of the explorative nature of these Anyway, Jeffery described interesting differences results, some comments are provided on the between Alouatta and Macaca during prenatal structural craniology in Alouatta, on the phylo- ontogeny (Jeffery, 2003). In both species the bas- genetic patterns of the Atelids, and on the paral- icranial retroflexion (i.e. hypoflexion) is correlat- lel evolution of airorhynchy in Pongo pygmaeus. ed with brain growth and development, but in 58 Alouatta skull morphology

Fig. 8 - Form Difference Maps comparing Alouatta versus Brachyteles (a) and Pongo versus Pan (b). The configuration is based on 17 lateral-projected landmarks. The median values from the Form Difference Matrix of each landmark are interpolated using a multiquadric radial basis function (other similar functions do not change the overall pattern). The scale provides a quantification of the processes, referring to the median ratio between the interlandmark distances of the numerator and that of the denominator. In Alouatta, a vault vertical flattening associated to a forward muzzle development is observed, while in Pongo there is a supraorbital backward flattening associated to a downward premaxilla/maxilla development. E. Bruner et al. 59

Fig. 9 - Pairwise comparisons between Alouatta and Aegyptopithecus through Procrustes superimposition of two-dimensional left lateral projections.

Alouatta this process is faster, associated with Considering the importance of the basicrani- lower rates of total and posterior cranial base um and in particular of the spheno-occipital lenghtening, and associated with the infratentor- suture in the development of the cranial func- ial absolute and relative development. tional matrix (e.g. Lieberman et al., 2000), this Furthermore, in Alouatta the volumetric brain process has probably a major role within the evo- expansion is generally slower, but the cranial base lutionary changes described in Alouatta. angle is always much higher than in macaques of Actually, the pressure along the anterior cranial similar maturation quotient, and the infratento- base can easily involve the rotation of the rial (i.e. cerebellar) structures are relatively more splanchnocranium and be related to the flatten- developed (see details in Jeffery, 2003). All these ing of the vault. Ross et al. (2004) have recently features must be involved in the final neurocra- published an extensive analysis on the basicranial nial morphology of the howler monkeys, and flexion in Primates. Species-specific averaged should be carefully considered in terms of struc- data from Ceboidea and Cercopithecoidea tural correlations between cranial parts. The shown the outstanding position of Alouatta faster retroflexion stresses neurocranial flattening because of its basicranial length, cranial capacity, and airorhynchy, which seem to be anyway evi- and cranial base angle. The basicranium is very dent since the earliest fetal stages (at least at the lengthened when compared with the small cra- 40% of the total prenatal period). The scarce nial capacity (approximately 50 cc.), probably lenghtening of the clivus is probably involved in with a large contribution of its anterior parts the basioccipital shortening and nuchal rotation, (Fig. 10a). Consequently, the encephalisation also necessarily affected by the structural adjuste- index relative to the basicranial length is ments related to the infratentorial development. extremely small. The spatial packing hypothesis The slower rate of cerebral growth leads to a gen- states that in primates an increase in the relative eralised neurocranial limited enlargement. encephalisation index is associated with a flexion The basicranium flattens and lengthens of the cranial base (see Ross et al., 2004). through the stretching of the sphenoid body, dis- Alouatta does not fit the relationship described placing the spheno-occipital suture and the fora- for the Ceboidea-Cercopithecoidea variation, men magnum backward. The pneumatisation of showing even a less flexed base compared with its the sphenoidal body in Alouatta (Osman Hill, low relative encephalisation or, conversely, a 1962) is probably secondarily related to this small encephalisation index for such cranial base sagittal development between the vomer and the angle (Fig. 10b). Some structural constraints spheno-occipital suture. must be involved in this configuration, where the 60 Alouatta skull morphology

vault flattening and sphenoid lengthening act alone. Brachyteles shows an interesting phenotyp- directly onto the cranial base flexion and ic similarity with Alouatta, despite a species-spe- endocranial volume. The howler’s base angle is cific flattening of the midfacial areas. In addi- theoretically compatible with a larger encephali- tion, the shared interorbital anatomy between sation index, within the Ceboidea- Alouatta and Ateles should be further explored. Cercopithecoidea variation. In contrast, the These results and the comparison with scarce encephalisation index would require an Aegyptopithecus can suggest some comments on extreme retroflexion of the base, probably out of the evolution of the howlers’ cranial morphology. the available structural range. It is therefore pos- In the latter, the splanchnocranium is tilted onto sible to hypothesise that this extremely low the neurocranial axis because of the scarce cranial encephalisation index has involved the maxi- development and the marked maxilla/premaxilla mum possible retroflexion, plus an additional relative enlargement. The similar morphology compensatory adjustment represented by the displayed by Aegyptopithecus and Alouatta is rotation of the splanchnocranium (namely, probably related to a similar relationship airorhynchy). The low encephalisation index between face and neurocranium, but there is no must be related more to neurocranial flattening evidence of shared structural processes. The and narrowing (i.e. scarce cranial capacity, relat- antero-posterior muzzle enlargement in the spec- ed to slow rate of cerebral growth, suture fusion, imen of the Oligocene is not associated with a etc.) than to an actual and absolute lengthening general maxillary widening, and the foramen of the sphenoid. The Euclidean Distance Matrix magnum - although shifted to a posterior posi- Analysis (based onto the ratios between absolute tion - is not tilted as in Alouatta. In considering values) largely supports this hypothesis, stressing the muzzle shortening described along the the neurocranial flattening and midface enlarge- anthropoids evolution and the peculiar nuchal ment, without any localised changes at the cra- shape in Alouatta, it is reasonable to hypothesise nial base (see Fig. 8). a parallelism between a derived condition in the The same process tilts the occipital foramen howlers and a plesiomorphic morphology for backwards, promoting a flattening of the nuchal Aegyptopithecus. In the latter, the skull morphol- area and characterising the in vivo posture of the ogy can be easily interpreted as an archaic expres- howlers. It is worth noting that the foramen sion of the volumetric ratio between neurocrani- magnum is much less inclined with respect to the um and splanchnocranium. In Alouatta, the dif- neurocranial axis. ferent relationship between muzzle and cranial The premaxilla-maxilla complex undergoes a base (sphenoidal lengthening and associated relevant development, through a general enlarge- shifting of the foramen magnum) suggests that ment and forward projection of the midface. the rotation of the facial axis is probably related All these features increase the airorhynchy, to a secondary (i.e. derived) splanchnocranial producing a widening of the enlargement associated to with a specific adap- neurocranial/splanchnocranial axis and a mor- tive context (Osman-Hill, 1962). phological “crease” (Bookstein, 2000) between These inferences do not involve comments pterion and porion, which suggests that these on the ancestral morphology of the Atelids them- surfaces may have a major structural role. The selves, and on the within-group polarity of their relative position of the occipital foramen onto characters, including airorhynchy. The probable the neurocranial length and the relative position phyletic relationships within the Atelids of the orbits onto the facial morphology stress (Horovitz & Mayer, 1995) is largely charac- the hypothesis of different relationships between terised by mosaic patterns and a certain amount functional parts (face, base, and vault) without of homoplasy, as can be possibly hypothesised major local rearrangements. The peculiar mor- comparing postcranial and dental traits (Szalay phology of the skull in Alouatta is actually rather & Delson, 1979). Furthermore, the fossil record “atelid-like” when the facial shape is considered is rather scattered (Hartwig & Meldrum, 2002; E. Bruner et al. 61

Fig. 10 - Relationship between brain volume (BV; in cubic centimeters), basicranial length (BL: fora- men caecum - sella - basion; in centimeters), cranial base angle (CBA; degree), and index of relative encephalisation (IRE; cube root of endocranial volume/BL), in Ceboidea (white circles), Cercopithecoidea (black circles), and Alouatta (A. belzebul, A. palliata; crosses). Data from Ross et al., 2004 (IRE = IRE5; BL = BL2; CBA = CBA4). 62 Alouatta skull morphology

MacPhee and Horovitz, 2002) and the group Pongo, it was also hypothesised that the orbital seems scarcely derived with respect to the morphology is independent upon airorhynchy, Oligocene primates, in relation to size, habits, which was associated to palatal tilting and devel- and locomotion (Conroy, 1990). All these con- opment of the temporal muscle (Penin & Baylac, siderations suggest that their interesting evolu- 1999). The comparison between Pan and Pongo tionary history is far from being understood. shows different superimposition results in the present analysis and the work published by Penin Airorhynchy in Alouatta and Pongo & Baylac (compare Figure 2a in Penin & Baylac, Compared to Pan, Pongo is mainly charac- 1999). In both studies there is a frontal and terised by a lower splanchnocranial enlargement, nuchal flattening associated with a palatal tilting. braincase vertical stretching, and supraorbital In contrast, Penin & Baylac found a maxillary flattening. The flattening at the supraorbital rotation and relative neurocranial flattening. structures is very stressed, involving loss of These differences can be related to the different frontation with respect to the neurocranial axis landmarks considered, or to the use of unilateral but not in relation to the entire skull shape. A vs. bilateral configurations. Anyway, these differ- minor upper vault flattening is associated with a ences once more stress the need of caution when relevant vertical stretching of the braincase. The considering superimposed forms (Richtsmeier et nuchal area undergoes a subsequent flattening al., 2002), suggesting a complementary use of and relative widening. The foramen magnum is multiple techniques. Different superimpositions slightly tilted, but not displaced backwards. must be compared, and integrated using coordi- Interestingly, once more the occipital foramen is nate-free approaches to “Form” such as the angled compared to the whole skull shape, but Euclidean Distance Matrix Analysis. This analy- not considering the neurocranial axis. Similarly, sis is extremely useful to describe the actual geo- the cranial base does not show a marked length- metrical differences between systems of coordi- ening. The premaxilla-maxilla complex shows a nates, even if results are more difficult to display sagittal development, mainly in relation to a (Cole & Richtsmeier, 1998). The visualisation downward shifting and inclination of the alveo- method proposed in this paper can improve the lar border. The zygomatic arches are very interpretation of the interlandmark data, synthe- enlarged and flared. It must be noted that once sising the information available from the Form more the polarity of the structural condition in Difference Matrix, and allowing a direct com- chimps and orangs has not been discussed, most- parison of morphological patterns. ly taking into account the phenotypic affinity When comparing the airorhynchy in Pongo between Pongo and the extinct Dryopithecinae. and Alouatta, some observations can be synthe- In Pongo, Shea (1985) hypothesised that sised to consider this interesting parallel evolu- airorhynchy can be associated to simognathism, tionary process linking anatomy, ecology, and supraorbital flattening, orbital shape, mandible behaviour (Fig. 11). In Pongo, Shea (1985) sug- enlargement, nasal floor and ethmoid structure, gested that airorhynchy is probably unrelated to and morphology of the anterior cranial fossa. major local rearrangements, but rather depends According to both superimposition and on a repositioning of the face compared to the Euclidean distance matrix analysis, there is no neurocranial structure. Interestingly, this also evidence of a clear midfacial flattening. The seems to be the case for Alouatta. They share a lower and middle facial profile in Pongo seems marked flattening of the orbital/supraorbital not to differ from the shape expressed in Pan, structures, the orbital shape (tall and narrow), except for a generalised enlargement. Rhinion and a consequent mandibular enlargement. In shows even a further projection considering the contrast, the maxillary growth is characterised by overall size increase in Pongo. The actual flatten- a general development and forward projection in ing involves the upper face, including the nasion. the howlers, and by a sagittal downward stretch On the contrary, in comparing Pan and in the orangs. In Alouatta and Pongo there is a E. Bruner et al. 63

Fig. 11 - Differences between the cranial shape in Alouatta and Pongo. Vectors display the differences of the three-dimensional configuration between Alouatta and a con- sensus computed from all the other Atelids (a), and between Pongo and Pan (b) after Procrustes superimposition (solid line: Alouatta and Pongo shapes). The superimposed configurations (c) show a Procrustes comparison between Alouatta (solid line) and Pongo (dashed line).

nuchal flattening, associated with a rotation of The complementary role of different the occipital foramen. In any case, the braincase approaches should be stressed once again. flattening and narrowing in the previous taxon is Geometric morphometrics and EDMA are use- not comparable with the braincase vertical ful to consider shape and form respectively. stretching and widening of the latter. Although the theoretical background of these Furthermore, Pongo does not display the relative methodologies can be discussed to improve the cranial base lengthening with the consequent resolution and power of the morphological nuchal morphology. The frontal flattening itself analysis (Richtsmeier et al. 2002; Rohlf 2003), is oriented downward in Alouatta and backward they can be used simultaneously to increase the in Pongo, with respect to to the cranial shape. range of available models. Differences between In synthesis, in both species airorhynchy is their respective results must not be considered possibly related to changes in the relationship necessarily as a bias, but in terms of complemen- between face and vault, the consequences of tary information. In geometric morphometrics, which are mainly found at their structural joint. different superimposition procedures can also Actually, the supraorbital structures were contribute to improve the structural knowledge hypothesised to represent this architectural inter- of the morphological systems, optimising the face (Lieberman, 2000). Taking into account the shape comparison (General Procrustes Analysis) anatomical variations of the temporal fossa in the or considering specific functional networks atelids, also the pteric area must be carefully con- (Bookstein Superimposition). sidered. Anyway, the different cranial structure in All these explorative and descriptive data on howlers and orangs required different rearrange- the cranial morphology in Alouatta must be fur- ments of the whole system. ther developed through the consideration of the 64 Alouatta skull morphology

intra-specific, inter-specific, and inter-generic the skull of Atelids: notes on the cranial variability, especially by analysing the possible anatomy in Alouatta. Manuscript. role of allometry and ontogenetic scaling with Carlson, R.E. and Foley, T.A. 1991 - The respect to airorhynchy. Some species of howlers Parameter R2 in Multiquadric Interpolation. show a sexual dimorphism based on the per- Computers Mat. App., 21: 29-42. amorphic trajectory in males (Zingeser, 1966), Cole T.M. III 2002 - WinEDMA: Software for probably related to time-dependent differential Euclidean distance matrix analysis. Version growth (Couette, 2002). Furthermore, in 1.0.1 beta. Kansas City: University of Alouatta there are some indications of a possible Missouri - Kansas City School of Medicine. female size-related selection, and of a relevant Cortes-Ortiz L., Bermingham E., Rico C., male growth after maturity (Jones et al., 2000). Rodriguez-Luna E., Sampaio I. & Ruiz- Even more localised differences as the inclination Garcia M. 2002 - Molecular systematics and of the foramen magnum show individual varia- biogeography of the Neotropical monkey tions (Osman-Hill, 1962). All these data make genus, Alouatta. Mol. Phyl. Evol., 26: 64-81. the study of structural cranial morphology in Cole T.M. & Richtsmeier J.T. 1998 - A simple Alouatta a quite complex and promising issue. method for visualisation of influential land- marks when using Euclidean Distance Matrix Analysis. Am. J. Phys. Anthropol., Acknowledgments 107: 273-283. Couette S. 2002 - Quantitative Charatterisation of We are grateful to Cristina Giacoma and Marco the Cranial Development in Cebus apella and Gamba, and to all the staff of the 20th Congress of Alouatta seniculus (Primates, Platyrrhini): The the International Primatological Society, in which Contribution of Geometrical Morphometric the symposium on geometric morphometrics and Metods. Folia Primatol., 73: 149-164. Groves C. 2001 - . primatology was included. Luca Fiorenza provided Smithsonian Institution Press. Washington useful anatomical comments. Anna Loy improved and London. this paper with constructive comments and advices. Hartwig W.C. & Meldrum D.J. 2002 – platyrrhines of the northern Neotropics. In W.C.Hartwig (Ed): The Primate Fossil References Record, pp. 175-188. Cambridge, Cambridge University Press. Bookstein F.L. 1991 - Morphometrics Tools for Heesy C.P. & Ross C.F. 2004 - Mosaic Evolution Landmark Data: Geometry and biology. of Activity Pattern, Diet, and Color Vision in Cambridge, Cambridge University Press. Haplorhine Primates. In C.F. Ross & F. Kay Bookstein F.L. 2000 - Creases as local features of (eds): Anthropoid Origins: New Visions, pp. deformation grids. Med. Im. An., 4: 93-110. 665-700. Kluwer Academic/Plenum Bruner E. & Manzi G. 2001 - The Primates Publishers. collection of the Museum of Anthropology Horovitz I. & Meyer A. 1995 - Systematics of “G. Sergi”, Rome (with some notes on tax- the New World monkeys (Platyrrhini, onomy and systematics). Riv. Antropol., 79: Primates) based on 16S mitochondrial DNA 229-242. sequences: A comparative analysis of differ- Bruner E., Fiorenza L., Manzi G. 2002 - Orbital ent weghting methods in cladistic analysis. Frontalization and Morphological Cranial Mol. Phyl. Evol., 4: 448-456. Variability in Anthropoidea. Folia Primatol., Horovitz I. & MacPhee R.D.E. 1999 - The qua- 73: 297-337. ternary Cuban platyrrhinaae Paralouatta Bruner E., Mantini S., Fiorenza L. & Manzi G. varonai and the origin of Antillean monkeys. nd - Phylogeny and structural morphology in J. Hum. Evol., 36: 33-68. E. Bruner et al. 65

Jacobs G.H. 2004 - Photopigment Variation Morphometrics. Year. Phys. Anthropol., 45: and the Evolution of Anthropoid Vision. In 63-91. C.F. Ross & R.F. Kay (eds): Anthropoid Rivero M. & Arredondo O. 1991 - Paralouatta Origins: New Visions, pp. 645-664. Kluwer varonai, a new platyrrhinae from Academic/Plenum Publishers. Cuba. J. Hum. Evol., 21: 1-11. Jeffery N. 2003 - Brain expansion and compar- Rohlf F.J. 1998 - tpsSmall 1.19. Ecology and ative prenatal ontogeny of the non-homi- Evolution. SUNY at Stony Brook, New York. noid primate cranial base. J. Hum. Evol., Rohlf F.J 2003 - Bias and error in estimates of 45: 263-284. mean shape in geometric morphometrics. J. Jones A.L., Degusta D., Turner S.P., Campbell Hum. Evol., 44: 665-683. C.J. & Milton K. 2000 - Craniometric Rosenberger A.L. 1981 - Systematics: the higer Variation in Population of Mantled Howler taxa. In A.F. Coimbra-Filho & R.A. Monkeys (Alouatta palliata): Evidence of Mittermeier (eds): Ecology and Behavior of size Selection in Females and Growth in neotropical Primates, vol 1, pp.9-27. Rio de Dentally Mature Males. Am. J. Phys. Janeiro: Academia Brasileira de Ciencias. Anthropol., 113: 411-434. Ross C., Henneberg M., Ravosa M.J. & Richard Lieberman D.E. 2000 - Ontogeny, homology, and S. 2004 - Curvilinear, geometric and phylo- phylogeny in the hominid craniofacial skeleton: genetic modeling of basicranial flexion: is it the problem of the browridge In P. O’Higgins adaptive, is it constrained? J. Hum. Evol., & M. Cohn (eds): Development, growth and 46: 185-213. evolution. Implication for the study of the Schon M.A. 1971 - The anatomy of the hominid skeleton, pp. 85-122. Academic Press. Resonating Mechanism in Howling Lieberman D.E., Ross C. & Ravosa M. 2000 - Monkeys. Folia Primatol., 15: 117-132. The primate cranial base: ontogeny, func- Shea B.T. 1985 - On Aspects of Skull Form in tion, and integration. Yrb. Phys. Antrhopol., African Apes and Orangutans, With 43: 117-169. Implications for Hominoid Evolution. Am. J. MacPhee R.D.E. & Hovitz I. 2002 - Extinct Phys. Anthropol., 68: 329-342. Quaternary platyrrhinae of Greater Antilles Simons E.L. 1987 - New faces of Aegyptopithecus and Brazil.. In W.C. Hartwig (Ed): The from the Oligocene of Egypt. J. Hum. Evol., Primate Fossil Record, pp. 189-200. 16: 273-290. Cambridge University Press. Simons E.L. 2004 - The Cranium and Osman Hill W.C. 1962 - Primates. Comparative Adaptations of Parapithecus grangeri, a anatomy and taxonomy. Vol. V. Edinburgh Stem Anthropoid from the Fayum University Press, Edinburgh. Oligocene of Egypt. In C.F. Ross & R.F. Penin X. & Baylac M. 1999 - Comparaison Kay (eds) : Anthropoid Origins : New tridimentsionelle des cranes de Pan et Visions, pp. 183-204. Kluwer Academic/ Pongo par superimpositions procrusti- Plenum Publisher. ennes. C.R. Acad.Sci. 1999. (Ser III) Slice D. 2000 - Morpheus et al.; Ecology and 322 (12): 1099-1104. Evolution, SUNY at Stony Brook, New York. Podani J. 1997 - Syn-Tax 5.1: a new version for Szalay F. & Delson E. 1979 - Evolutionary PC and Macintosh computers. Coenoses, 12: History of the Primates. Academic Press: 149-152. New York. Rasmussen D.T. 2002 - Early Catarrhines of the White T.D. & Folkens P.A. 2000 - Human African Eocene and Oligicene,. In W.C. osteology. 2nd Edition. San Diego: Hartwig (Ed): The Primate Fossil Record, pp. Academic Press. 203-220. Cambridge University Press. Zingeser M.R. 1966 - Occlusofacial Relationships Richtsmeier J.T., Burke Deleon V. & Lele S.R. in the Mature (Alouatta 2002 - The Promise of Geometric caraya). Am. J. Phys. Anthropol. 24: 171-180 66