Allometry and Interspecific Differences in the Facial Cranium of Two Closely Related Macaque Species

Interpreting evolutionary history of macaque monkeys from fossil evidence is difficult, because their evolutionary fluctuations in body size might have removed or formed important morphological features differently in each lineage. We employed geometric morphometrics to explore allometric trajectories of craniofacial shape in two closely related species, Macaca fascicularis and M. fuscata. These two species exhibit a single shared allometric trajectory in superoinferior deflection of the anterior face, indicating that the differences in this feature can be explained by size variation. In contrast, two parallel trajectories are demonstrated in craniofacial protrusion, indicating that even if they are comparable in size, M. fuscata has a higher and shorter face than M. fascicularis. The degree of facial protrusion is most likely a critical feature for phyletic evaluation in the fascicularis group. Such analyses in various macaques would help to resolve controversies regarding phyletic interpretations of fossil macaques.


Introduction
Macaques are medium-sized cercopithecine monkeys, which are currently distributed widely in the southern, southeastern, and eastern regions of Asia, and in a restricted area of northern Africa [1]. It is considered that this group probably arose as early as the Late Miocene in northern Africa and spread to Eurasia by the beginning of the Pliocene [2,3]. Dispersal to eastern Eurasia occurred by the Late Pliocene, and macaques subsequently accomplished successful adaptive radiation in the southern, southeastern, and eastern regions of Asia [2]. The living species are usually classified into four species-groups, that is, sylvanus, silenus, sinica, and fascicularis groups ( [2,4,5], and see also [6]).
The fascicularis group, which is continuously distributed over a vast area extending in its broadest area from western to eastern Asia, includes four species: Macaca fascicularis in the tropical area from Indochina to Indonesia, M. mulatta in the subtropical and temperate areas from eastern Afghanistan to China, M. cyclopis in the subtropical area of Taiwan, and M. fuscata in the subtropical to subfrigid areas of Japan [7,8]. Evolutionary and dispersal histories of the fascicularis group have been proposed based on their morphological variations, fossil records [7,9], and evidence of the mitochondrial and Y-chromosome genes [10][11][12]. According to their studies, ancestral M. fascicularis might have originated in the equatorial region [7,9] and subsequently dispersed to the islands on the Sunda Shelf during periods of marine regression due to glacio-eustasy and northward to the mainland of Southeast Asia [7,9]. M. mulatta may have diverged from a stock of M. fascicularis around 2.5 Ma and subsequently, were widely distributed in western, southern, and eastern Asia [10][11][12]. The eastern populations of M. mulatta colonized separately in Taiwan and Japan during the period of marine regression around or before 0.4 Ma and became to M. cyclopis and M. fuscata, respectively [7,12]. The eastern populations of M. mulatta then retreated southward to its present latitudinal zone during a glacial episode [7,9,12]. Thus, the current distribution has been formed by intricate dispersal events of the four species with the recent speciation during the Pleistocene period, during which at least six major glacial episodes occurred with significant climate fluctuations [6,13].
The four living species of the fascicularis group are often discriminated by size. The length of the cranium decreases in following order: M. fuscata, M. cyclopis, M. mulatta, and M. fascicularis [7]. Size differences are usually attributed to Bergmann's rule [7], which postulates that animals living in colder regions tend to have larger body sizes than close relatives residing in warmer regions. This adaptation effectively maintains body temperature in colder conditions. In addition to size measurements, the four species are identified by specific anatomical features of the cranium. The lateral orbital rim is relatively thicker, and the orbits, interorbital region, and foramen magnum are relatively narrower in M. fascicularis than in the other species. Furthermore, the orbits and neurocranium are relatively wider, and the facial cranium is relatively higher in M. fuscata than in the other species [14]. Such interspecific differences in cranial morphology could, however, be explained by the differences in body size among the four species, that is, an allometric trend common to the fascicularis group. In the studies on Chinese macaques [15], baboons [16], and the entire Papionini tribe [17], such common allometric trends have been commonly observed across species, for example, larger individuals usually have a protruding, inferiorly deflecting face and relatively small orbits as compared to those of the smaller individuals. Because of several intrinsic changes in the distribution and climatic conditions, body sizes have most likely been unstable throughout the evolution of each species of the fascicularis group, for example, dental remains of M. fascicularis discovered from archaeological sites in Borneo dating back to the Late Pleistocene era are larger than those of the living conspecifics [18]. Such evolutionary fluctuations in the body size might make a fossil specimen look different in some morphological features from its true living relatives, which can be explained by a common allometric trend. Such features reflect climate conditions rather than phylogeny, posing a potential risk of wrongly identifying the phyletic position of a given fossil specimen.
In the present investigation, we employed geometric morphometrics to explore the allometric trajectories of craniofacial shape in M. fascicularis and M. fuscata, which are quite distinct with respect to morphology, distribution, and climate of habitats. We demonstrated a common allometric trend for the fascicularis group, and then, attempted to identify the features, which are independent of size destabilization, in order to differentiate between the two species even in fossil remains.

Materials and Methods
The Three-dimensional (3D) coordinates were measured according to the anatomical landmarks of facial crania by using a 3D digitizer (Microscribe MX; Immersion Corporation, USA). The 9 landmarks measured are shown in Figure 1 and listed in Table 1. The measurements on the left side are typically used for analysis to avoid redundancy. The specimen with the broken left side was analyzed with the measurements of the right side after horizontal reversal.
The following analyses were performed using the geometric morphometrics software Morphologika version 2.5 [19,20]. Generalized Procrustes analysis (GPA; [19,21]) was carried out to register landmark configurations by eliminating the translational and rotational differences and scaling them to the best fit. GPA was achieved by scaling and optimally superimposing all landmark coordinates to minimize the sum of squared distances between homologous landmarks. The registered landmark configurations were then represented as points in the non-Euclidean shape space of Kendall. The centroid size was computed for each specimen at the same time as the square root of the sum of squared Euclidean distances from each landmark to the centroid [19].
Principal component analysis (PCA) was carried out to identify the major axes of variation in the shape of the facial crania. PCA was performed in the tangent plane to Kendall's shape space by using the vector of the tangent space coordinates [19].
Shape variability represented by each principal component (PC) was visualized by reconstructing hypothetical forms of the wireframe along each PC. The visualizations were further interpreted using Cartesian transformation grids calculated from the triplets of thin-plate splines (TPS; [19,22]). The grids derived from TPS indicate the deformation of the space surrounding a reference shape into that surrounding a target shape, wherein the deformation involves minimum bending.
To evaluate our interpretations of shape differences represented by the PCs, we calculated indices and an angular measurement, including Relative Snout Length, Relative Palatal Length, and Facial Deflection on the basis of the digital landmarks (Table 2). Then, correlations between the measurements and the PC scores were examined using R version 2.12.0. [23].
Allometric trajectories for the two species are indicated by PC scores against centroid size, and their differences are examined by analysis of covariance (ANCOVA) and Welch's t-test by using R version 2.12.0.

Results
The centroid size of M. fuscata is significantly larger than that of M. fascicularis (t = −27.7, P < .001). PC1 and PC2 account for 31.4% and 22.6% of the total variance, respectively, and they are distinct with respect to the PCs enlisted in Table 3. PC1 is inversely correlated with the centroid size for all the samples of the two species (R 2 = 0.42, P < .001); M. fascicularis samples have significantly higher PC1 score than M. fuscata samples (t = 26.8, P < .001; Figure 2(a)). PC1 indicates anteroposterior shearing of the Cartesian transformation grid, indicating that the anterior portion of the face is deflected more superiorly to produce airorhynchy in a subject with a higher PC1 score (Figure 3(a)), and klinorhynchy in a subject with a lower PC1 score (Figure 3(b)). PC1 is inversely correlated with the angular measurement of Facial Deflection (r = −0.88, P < .001).

Discussion
The shape variations represented by the first two PCs depend on the size in M. fascicularis and M. fuscata. These two allometric scaling patterns found in our study are commonly observed across varied clades of the papionin monkeys, for example, in the studies on Chinese macaques [15], baboons [16], and the entire Papionini tribe [17]. However, the two shape variations observed in this study have distinct allometric trajectories with respect to each other.
PC1 exhibits a single allometric trajectory shared by the two species. This suggests that the differences in body size principally produce interspecific differences in the superoinferior deflection of the anterior face between M. fascicularis and M. fuscata. This indicates that if the hypothetically ancestral M. fuscata had been comparable in size to the current M. fascicularis, the fossil M. fuscata could have a deflected face similar to that of the living M. fascicularis. This finding suggests that the deflection in the anterior face should be carefully used for evaluating the phyletic relationships of fossil species to living taxa, at least among the fascicularis group.
In contrast, PC2 exhibits two parallel allometric trajectories; the trajectory is vertically lower in M. fuscata relative to M. fascicularis. This finding means that M. fuscata has a higher and shorter face than M. fascicularis, even if they are comparable in size. Such a difference is supported by the findings of Fooden [6] and Mouri [24], which demonstrated that M. fascicularis has long facial dimensions relative to skull size compared to M. fuscata at a comparative age or size. Thus, the differences in the longitudinal-to-vertical proportion of the face against skull size are probably more critical for the phyletic evaluation of a given fossil specimen relative to the living taxa in the fascicularis group.
The present findings suggest that for the fascicularis group, the degree of facial deflection represented by PC1 likely follows the latitudinal cline in body size, which would lead to klinorhynchy in cold climate. The inferior deflection of the anterior face inevitably extends the nasal profile and cavity; such a feature might have selective advantage for effectively warming the inspired air in the nasal cavity under cold conditions. However, the interspecies differences in this feature per se are rather regarded as simply a consequence of adaptive modifications in body size as predicted by Bergmann's rule. The variation in facial protrusion represented by PC2 possibly reflects critical modifications occurring during speciation under any selection pressures among the members of the fascicularis group. The vertical transposition of the trajectories in the two species could be explained by evolutionary fluctuations in the general scaling. The relatively short face of M. fuscata might reflect functional modifications to the masticatory apparatus. Antón [25] also demonstrated such features in M. fuscata during the examination of the masticatory apparatus and suggested that the short and high face is desirable for dissipating greater occlusal loads. Indeed, M. fuscata may have a tougher diet than the other macaques [26,27]. On the other hand, Mouri [24] suggested that the long face of M. fascicularis contributes to the large gape, which permitted the presence of long canines in the members of this species. The longer 6 Anatomy Research International canines of M. fascicularis, which could be used as a weapon, are probably related to the severe male-to-male competition within this species as compared to that in M. fuscata [24]. Such severe competition may have originated because of the weak seasonality of reproduction by M. fascicularis, wherein there are usually only small numbers of females involved in the estrus cycle [14]. Thus, differences in facial protrusions among the members of the fascicularis group could be generated by selective processes, which are different for each species, for example, feeding functions and social systems. This would reflect critical morphological modifications occurring during each speciation.
The macaque fossil specimens from the Pleistocene period found so far are mostly isolated teeth and fragmentary jaws. Nevertheless, several fossil specimens retain facial regions [2,3,28], for example, M. anderssoni from the Early Pleistocene of Honan, China [29]; M. robusta from the Middle Pleistocene of Choukoutien, China [28,30]; Macaca cf. robusta from the Middle or Late Pleistocene of Turupong, South Korea [31]; M. speciosa subfossilis from the Late Pleistocene of Thung-Lang, Vietnam [32]; and M. fuscata from the Late? Pleistocene of Shikimizu, Japan [33]. The phyletic positions of these specimens are usually evaluated based on geological approximations or morphological similarities to current animals. Nevertheless, fossil specimens occasionally do not possess the stereotypical combinations of cranial features seen in current animals. This tends to confound the process of phyletic reconstruction. For instance, a fossil cranium of M. speciosa subfossilis was regarded to be closely related to the extant M. arctoides based on geological approximation and morphological similarities in the nasal cavity [34] and malars [35]. However, this specimen does not possess a prominent preorbital concavity, which is a distinct feature of M. arctoides [34]. Such ambiguity can be cleared by future studies on allometric trajectories of cranial shape in various macaque subgroups. This approach, as well as the indices and angular measurements correlated with PCs, can be applied to the aforementioned fossil cranial specimens, some of which are facial fragments. Future findings about the allometry and interspecific differences in the dentognathic features will contribute to identifying the phyletic position of more fragmentary fossil materials.

Conclusions
We demonstrated that two closely related macaque species, M. fascicularis and M. fuscata, exhibit a single shared allometric trajectory in the supero-inferior deflection of the anterior face, and two parallel trajectories with respect to craniofacial protrusion. The interspecific difference in the latter feature is not explained by the evolutionary modification in body size. The craniofacial protrusion is one of the most reliable characteristics for phyletic evaluation of a given fossil specimen relative to living taxa, at least within the members of the fascicularis group. Future efforts with respect to allometric analyses in various macaques are expected to provide critical references for solving continuing controversies about the phyletic relationships between fossil macaques and living taxa.