Exploring Phylogenetic and Functional Signals in Complex Morphologies: The Hamate of Extant...

Exploring Phylogenetic and FunctionalSignals in Complex Morphologies:

The Hamate of Extant Anthropoids as aTest-Case Study

SERGIO ALM�ECIJA,1,2,3* CALEY M. ORR,4 MATTHEW W. TOCHERI,5,6

BIREN A. PATEL,7,8AND WILLIAM L. JUNGERS1

1Department of Anatomical Sciences, Stony Brook University School of Medicine, StonyBrook, New York

2Institut Catal�a de Paleontologia Miquel Crusafont, Universitat Aut�onoma de Barcelona,Edifici Z (ICTA-ICP), campus de la UAB, c/ de les Columnes, s/n., 08193 Cerdanyola del

Valles, Barcelona, Spain3NYCEP Morphometrics Group

4Department of Anatomy, Midwestern University, Downers Grove, Illinois5Human Origins Program, Department of Anthropology, National Museum of NaturalHistory, Smithsonian Institution, 10th and Constitution Avenue NW, Washington, DC

6Department of Anthropology, Center for the Advanced Study of Hominid Paleobiology,The George Washington University, Washington, DC

7Cell and Neurobiology, Keck School of Medicine, University of Southern California,Los Angeles, California

8Human and Evolutionary Biology Section, Department of Biological Sciences, Universityof Southern California, Los Angeles, California

ABSTRACTThree-dimensional geometric morphometrics (3DGM) is a powerful

tool for capturing and visualizing the “pure” shape of complex structures.However, these shape differences are sometimes difficult to interpretfrom a functional viewpoint, unless specific approaches (mostly based onbiomechanical modeling) are employed. Here, we use 3DGM to explorethe complex shape variation of the hamate, the disto-ulnar wrist bone, inanthropoid primates. Major trends of shape variation are explored usingprincipal components analysis along with analyses of shape and size cova-riation. We also evaluate the phylogenetic patterning of hamate shape byplotting an anthropoid phylogenetic tree onto the shape space (i.e., phylo-morphospace) and test against complete absence of phylogenetic signalusing posterior permutation. Finally, the covariation of hamate shape andlocomotor categories is explored by means of 2-block partial least squares(PLS) using shape coordinates and a matrix of data on arboreal locomotorbehavior. Our results show that 3DGM is a valuable and versatile tool forcharacterizing the shape of complex structures such as wrist bones inanthropoids. For the hamate, a significant phylogenetic pattern is foundin both hamate shape and size, indicating that closely related taxa aretypically the most similar in hamate form. Our allometric analyses showthat major differences in hamate shape among taxa are not a direct con-

Grant sponsors: Fulbright Commission and the Generalitatde Catalunya (S.A.), the Spanish Ministerio de Econom�ıa yCompetitividad (S.A.), the AAPA Professional DevelopmentGrant (S.A.); the Smithsonian Scholarly Studies Grant Program(M.W.T.); Leakey Foundation (B.A.P.); Wenner-Gren Foundation(C.M.O.); National Science Foundation; Grant numbers: 2009BFUL 00049, 2009 BP-A 00226, CGL2011-27343, NSF-BCS1316947, NSF-BCS-1317047, NSF-BCS 1317029.

*Correspondence to: Sergio Alm�ecija; Department of Anatomi-cal Sciences, Stony Brook University School of Medicine, StonyBrook, NY 11794-8081. E-mail: [email protected]

Received 3 October 2014; Accepted 11 October 2014.

DOI 10.1002/ar.23079Published online in Wiley Online Library (wileyonlinelibrary.com).

THE ANATOMICAL RECORD 298:212–229 (2015)

VVC 2014 WILEY PERIODICALS, INC.

sequence of differences in hamate size. Finally, our PLS indicates a sig-nificant covariation of hamate shape and different types of arboreal loco-motion, highlighting the relevance of this approach in future 3DGMstudies seeking to capture a functional signal from complex biologicalstructures. Anat Rec, 298:212–229, 2015. VC 2014 Wiley Periodicals, Inc.

Key words: wrist morphology; 3D geometric morphometrics;shape–function complex

In primates, the hamate forms the distal ulnar aspectof the carpus (Fig. 1). It articulates proximally with thetriquetrum (sometimes with the lunate as well), radiallywith the capitate, and distally with the fourth and fifthmetacarpal bases. It has a hook-shaped projection, thehamulus, on its distopalmar side, which is the attachmentsite for two intrinsic muscles of the fifth digit (flexor digitiminimi and opponens digiti minimi). Variation in hamateshape has been used in studies of the evolution of modernhuman manipulative capabilities because human hamatemorphology appears to favor the ability to form a varietyof grips that are employed to construct and manipulatetools (Marzke et al., 1992, 1998; Marzke and Marzke,2000; Marzke, 2013; Orr et al., 2013). The functional dif-ferentiation of monkey and ape wrists has also been atopic of particular interest for primate morphologists(Lewis, 1965, 1969, 1972; McHenry and Corrucini, 1975;O’Connor, 1975; Cartmill and Milton, 1977; Sarmiento,1988; Lewis, 1989; Richmond, 2006). Specifically, hamatemorphology has been important in discussions of homi-noid locomotor evolution because of its role in wrist mobil-ity (i.e., ranges of motion), which is necessary forbehaviors such as vertical climbing and below-branch sus-pension used by extant apes (e.g., Tuttle, 1967; Lewis,1972; Jenkins and Fleagle, 1975; O’Connor, 1975; Lewis,1977; Beard et al., 1986; Sarmiento, 1988; Spoor et al.,1991; Sarmiento, 1994; Fleagle, 1999).

Monkeys and extant hominoid primates differ consid-erably in terms of ulnar wrist morphology. Most of thesedifferentiating features are related to the derived com-plex in crown hominoids in which the ulnar styloid pro-cess is retracted from the proximal carpal row (Lewis,1989; Sarmiento, 1988). Although the expression variesacross apes, the imposition of fibrocartilage between theulnar styloid process and the triquetropisiform complex(and reduction of the triquetrum) is thought to dramati-cally alter the load transmission regime through thewrist. This reorganization has been traditionally relatedto the capacity for increased supination at the distalradioulnar joint necessary for below-branch locomotion(Lewis, 1972; Sarmiento, 1985, 1988; Lewis, 1989). How-ever, the loss of ulnocarpal articulation in the 12million-year-old fossil great ape Pierolapithecus catalau-nicus, in combination with its inferred orthograde bodyplan (Moy�a-Sol�a et al., 2004) but moderate hand length(Moy�a-Sol�a et al., 2005; Alm�ecija et al., 2009; Alba et al.,2010), suggests that changes in the ulnar side of thewrist might have been initially related to extant great-ape-like vertical climbing and not specifically suspen-sion. It is likely that differences in hamate shapebetween apes and monkeys at least in part reflect thatreorganization of the ulnar side of the carpus.

Lewis (1972) described the wrist morphology of cercopi-thecoids as being largely similar across major taxonomicgroups with colobines and cercopithecines exhibiting littlevariation in hamate form. He described both groups asexhibiting a wedge-shaped body set obliquely within thewrist, with a proximally oriented triquetral facet thatforms a stable platform. The orientation of the articularfacets and the corresponding blocky triquetrum and well-developed styloid process of the ulna probably facilitateload transfer on the ulnar side of the wrist while restrict-ing ulnar deviation (Lewis, 1989; Sarmiento, 1988).O’Connor (1975) explained the relative uniformity ofhamate morphology among cercopithecoid monkeys asresulting from their use of quadrupedal locomotion involv-ing dorsiflexion of the wrist either in palmigrade or digiti-grade postures (irrespective of habitat/substratedifferences; see also Patel, 2009). Hyperextenison at thefifth carpometacarpal joint (and the fourth to a lesserdegree) appears to facilitate such hand postures in pro-nograde quadrupeds. This is permitted by the lack of adeveloped “hook” on the hamate (i.e., hamulus) togetherwith a proximal articular surface on the fourth and fifthmetacarpals that extends onto the dorsal surface of theshaft (Corruccini, 1975; O’Connor, 1975).

In contrast to the minimal morphological variation inthe hamate documented thus far in quadrupedal mon-keys, considerable diversity in hamate shape has beenreported for extant apes and humans. In both Pongo andthe hylobatids (that highly rely heavily on below-branchsuspension), the hamate is described as composing alarger proportion of the midcarpal articular surface andcontributing to a quasi-ball-and-socket joint that allowsconsiderable mobility (Jenkins and Fleagle, 1975; Jen-kins, 1981; Richmond et al., 2001). Lewis (1972, 1989)also described hylobatid hamates as retaining a some-what oblique-set within the overall carpus (similar tomonkeys). However, the hylobatid hamate has anulnarly facing triquetral facet and a rounded proximal“head” that articulates with the lunate and nearlyexcludes this latter bone from articulating with the capi-tate. The proximal aspects of the hamate and capitatetogether contribute to a globular distal midcarpal rowthat is engulfed by the lunate, scaphoid, and centrale(e.g., Jenkins and Fleagle, 1975; Lewis, 1989). Conse-quently, it does not present a large, flat, platform-likearticulation. Thus, the hylobatid configuration is instark contrast to the stable weight-bearing platform forthe triquetrum in monkeys. Lewis (1972, 1989) consid-ered the hylobatid midcarpus to be a less well-developedversion of the great ape wrist, which he viewed asideally adapted to suspensory locomotion. However, thehylobatid geometry is better considered as a uniquely

SHAPE ANALYSIS OF EXTANT ANTHROPOID HAMATES 213

derived modification that facilitates midcarpal rotationduring their highly specialized style of ricochetal bra-chiation (e.g., Jenkins, 1981; Kivell et al., 2013).

Wrist morphometric analyses, such as individualshape ratios based on traditional linear measurementsand angles have been used in the past to describe differ-

ences amongst different locomotor categories in primates(e.g., Sarmiento, 1988; Spoor et al., 1991). Other studieshave employed multivariate statistical analyses toexplore morphometric variation in the wrist in compari-son to a priori assigned locomotor groups (Begun andKivell, 2011). However, the hamate, like other wrist

Fig. 1. Line drawing depicting a left human hand skeleton in dorsal view. The hamate is highlighted incolor and its ulnar (upper) and distal (bottom) views are shown. The hamulus, as well as the triquetral,fourth and fifth metacarpal (mc) facets are indicated.

214 ALM�ECIJA ET AL

bones with complex shapes and multiple articular surfa-ces for surrounding bones, has proved difficult to charac-terize numerically at the species and even genus level.Alternative approaches based on three-dimensional (3D)quantification of carpal and tarsal anatomy have beenparticularly useful at discriminating among closelyrelated taxa (Tocheri et al., 2003, 2005, 2007; Marzkeet al., 2010; Tocheri et al., 2011; Dunn et al., 2014). Forexample, this approach was used to compare thehamates of fossil hominin taxa within the context ofmodern great ape-human variation (Orr et al., 2013).

Since the 1990s, the ongoing development of three-dimensional geometric morphometrics (3DGM) hasenabled the simultaneous analysis and visualization ofcomplex morphology (Bookstein, 1991; Rohlf and Mar-cus, 1993; Dryden and Mardia, 1998; Hammer andHarper, 2006). This method quantifies shape more com-prehensively than two-dimensional geometric morpho-metrics (2DGM) and other methods, and may oftencapture subtle shape differences between specimenswithin a species or among different taxa. Subsequently,the shape differences can be intuitively visualized as“warpings” of 3D surfaces generated from thin-platespline algorithms (Wiley et al., 2005). Several recentexamples exist in anthropology using 3DGM to charac-terize human and primate postcranial elements (e.g.,Turley et al., 2011; Alm�ecija et al., 2013; Bastir et al.,2013; Sylvester, 2013; Tallman et al., 2013; Turley andFrost, 2013; Knigge et al., this volume).

Geometric morphometric analyses characterize com-plex shapes and reveal overall phenetic affinities effec-tively. However, the functional significance of shapedifferences captured by 3DGM is often difficult to inter-pret. For instance, the observed shape differences oftenlack the explicitly biomechanical foundation that typicallyunderlies studies based on linear or other 3D measure-ments taken to represent specific elements of a particularmechanical model (e.g., load and lever arms, torques, etc.)(Terhune, 2013; Knigge et al., this volume). Nevertheless,3DGM analyses can use coordinate data that contain bio-mechanically relevant dimensions to help interpret thevisual results from a functional point of view (e.g.,Alm�ecija et al. 2013; Figueirido et al., 2013). A commonprocedure is to rely on multivariate statistical analyses(e.g., discriminant function analysis, MANOVA, etc.) totest for differences among “functional” groups defined apriori (e.g., Polly, 2008). This is similar to the way thatmultivariate analyses are conducted using more tradi-tional measurements. Another approach is to perform amultivariate regression or multivariate multiple regres-sion of shape coordinate data and, for example, locomotorcategories, taxonomic groups, substrate use, body masses,etc. and inspect how dependent variables of interest relateto each other and to the shape coordinates. Degree andpattern of covariation can be assessed by comparingangles between the projected vectors of different regres-sion analyses (e.g., Turley et al., 2011; Rein and Harvati,2013; Turley and Frost, 2013). However, results of regres-sion approaches are still difficult to interpret when dif-ferent locomotor categories (or covariates) are correlatedwith each other. For example, hypothetical shape changescorrelated with “terrestriality” may not be clearlyseparated from those correlated with general“quadrupedalism” (whether it is arboreal or terrestrial).

In this study, we use 3D coordinate shape dataderived from 3D surface scans to accomplish four mainobjectives:

(1) Characterize the complex morphology of thehamate among extant anthropoid primates in an effortto distinguish among taxa relying solely on their majorpatterns of shape variation. Based on previous studies,we hypothesize that hominoids will show a larger degreeof shape diversity in comparison to platyrrhine and cer-copithecoid monkeys. In comparison to hominoids, pla-tyrrhine and cercopithecoid monkeys are expected toshare larger metacarpal articular surfaces extendingonto smaller hamuli facilitating dorsiflexion at the car-pometacarpal joints.

(2) Test for the presence of phylogenetic structure inhamate shape and size (i.e., are closely related taxabeing more similar to one another than more distantlyrelated taxa?). We expect hamate form (i.e., shape plussize) in hominoids to be different from catarrhine andplatyrrhine monkeys, with hylobatids being the mostdivergent of all apes in terms of shape.

(3) Explore the predictable covariation of hamateshape and size among taxa to assess if shape differencesacross taxa are related to size differences. Although weexpect hylobatids and monkeys to differ in both shapeand size relative to great apes, it is unlikely that differ-ences in overall hamate shape will be caused by differen-ces in size alone. To our knowledge, predictability of thatcovariation has not been inspected in a comparable data-set of anthropoid primates (but see Orr et al., 2013 foran analysis of specific univariate features and hamatesize in great apes and humans).

(4) Test if (and how) hamate shape covaries with amatrix of known locomotor behaviors to examinewhether shape differences (overall) can be reasonablyinterpreted in a functional context. We hypothesize thatshape differences between monkey, hylobatid and greatape taxa are related to both phylogenetic divergence andassociated differences in locomotor behaviors. Hence, weexpect a significant portion of total shape variation tosignificantly covary with predominant locomotor behav-iors. For example, given previous observations in thespecialized type of locomotion and wrist morphology ofhylobatids, we expect a large amount of the total anthro-poid hamate shape variance to significantly covary withlocomotor behaviors (i.e., due at least to brachiation inhylobatids).

MATERIALS AND METHODS

Sample and Data Collection

We sampled 253 hamates representing 18 extant gen-era, including platyrrhines, cercopithecoids, hylobatids,and hominids (great apes and humans) (Fig. 2; Table 1).Hominoid taxa (i.e., hylobatids and hominids) weresampled more extensively to capture their patterns ofvariation, whereas monkey taxa (especially platyrrhines)were included as comparisons to help better contextual-ize the hominoid groups. To capture the overall form(i.e., shape plus size) of the anthropoid hamate, we col-lected 23 landmarks (Fig. 3; Table 2) from the surfacesof 3D digital hamate models using Landmark Editorsoftware (ver. 3.6) (Wiley et al., 2005). The 3D modelswere obtained using laser scanning (Nextengine, Scan-studio HD Pro), with the exception of three small


platyrrhine hamates (Aotus, Saguinus, Saimiri) thatwere acquired using micro-CT imaging with surfacesextracted using the segmentation tool of Avizo software(ver. 7.0). All raw scans were edited with Geomagic(vers. 12–2013) to clean the meshes, fill holes, andstandardize all hamate models to a similar number ofvertices. It has been shown that scale-free data collectedfrom 3D surface models obtained from different sourcesare reasonably compatible (Tocheri et al., 2011). All final3D models were opened in Landmark Editor for datacollection.

The phylogenetic relationships of the taxa examined(Fig. 2) were recovered from a chronometric consensustree based on molecular data and downloaded from “The10KTrees Project” (ver. 3; http://www.10ktrees.fas.har-vard.edu) (Arnold et al., 2010). Some nomenclature, suchas the new generic name (Hoolock) for hoolock gibbons(Mootnick and Groves, 2005), was updated to reflectrecent taxonomic revisions.

Analyses

Shape data were obtained from raw coordinatesthrough a full (generalized) Procrustes fit analysis—which rotates, translates and size-scales the landmarkconfigurations to unit of centroid size—and posteriororthogonal projection on to the tangent space (Drydenand Mardia, 1998). Major patterns of hamate shape vari-ation were explored through a principal componentsanalysis (PCA). The original coordinate system of theshape space was transformed to the principal compo-

nents of the covariance matrix of the original variables,such that the new shape space was centered on the aver-age shape configuration. Thus, the first principal compo-nent axis runs through the major axis of variation in thedata with subsequent axes running at orthogonal anglesthrough the minor axes of variation, implying a rigidrotation of the original variables (Polly, 2008; Pollyet al., 2013). Shape differences among our extant sampleof anthropoid primates were tested by means of multi-variate analysis of variance (MANOVA) of the first threeprincipal components (which concentrate 53.4% of thevariance). Post hoc significant differences among platyr-rhines, cercopithecoids, hylobatids, the three great apegenera and modern humans was assessed throughHotelling’s P values (Bonferroni’s corrected).

To assess the phylogenetic patterning in the shapespace, we mapped a phylogenetic tree onto the shapespace defined by the two first axes of a PCA of the covar-iance matrix of extant taxa means. Hypothetical ances-tral states (internal nodes in the tree) werereconstructed using squared-change parsimony (Maddi-son, 1991) with a method developed for geometric datain which shape is treated as a single multidimensionalcharacter and weighed by branch length (Klingenbergand Gidaszewski, 2010). Subsequently, the estimatedancestral node configurations were plotted on the origi-nal shape space, and the branches of the tree were con-nected to obtain a phylomorphospace (Rohlf, 2002;Sidlauskas, 2008; Monteiro, 2013). The previous stepswere necessary to subsequently test for the presence of aphylogenetic signal in hamate shape space by means of

Fig. 2. Time-calibrated phylogenetic tree showing the relationships among the modern anthropoid pri-mates sampled in this study. Representative hamates are presented in dorsoulnar view for selected taxa.


http://www.10ktrees.fas.harvard.edu

http://www.10ktrees.fas.harvard.edu

a permutation test approach (Laurin, 2004) extended formultivariate analysis of coordinate data (Klingenbergand Gidaszewski, 2010). This technique simulates thenull hypothesis of complete absence of phylogeneticstructure among hamate shapes in extant anthropoids.The species mean shape configurations were randomlydistributed as the TIPS of the phylogeny in 10,000 per-mutations. For each permutation, tree length (i.e., thesum of the squared Procrustes distances between ances-tral and descendant shapes for all branches) was com-puted. If more than 5% of the resulting tree lengthscomputed in the permutation test were greater than theone obtained with the original data, the null hypothesisof absence of phylogenetic structure in the data wasrejected.

Covariation of hamate shape and size (i.e., allometry)was explored using multivariate regression of all theProcrustes coordinates on log-transformed centroid size(logCS), using both species mean configurations (i.e.,TIPS of the phylogeny) and phylogenetic independentcontrasts (PICs; Felsenstein, 1985). To assess the reli-ability of these multivariate regressions, a permutationtest (10,000 iterations) against the null hypothesis ofindependence between the dependent (hamate shape)and independent (logCS) variables was performed.

We also performed a two-block partial least squares(PLS) analysis to test for covariation between hamate

shape and locomotor categories in selected taxa (Rohlfand Corti, 2000). This method differs from multivariateregression in that both the shape and covariate blocksare treated symmetrically rather than as one set of vari-ables (X axis) being used to predict variation in theother set of variables (Y axis). These new pairs of varia-bles (the different PLSes) represent linear combinationsof variables within the original two sets (the blocks).These linear combinations are constructed so that thenew variables account for as much of the covariation aspossible between the two original sets with the goal tofind relationships between them without assuming thatone is the cause of the variation in the other. In ourcase, one block was constituted by hamate shape (Pro-crustes coordinates) and the other by a matrix of thefrequencies of use for several types of arboreal locomo-tion by extant hominoid and selected anthropoid species(Table 3). Data on arboreal locomotion was used becauseit is available for more taxa (Thorpe and Crompton,2006; their Table 7 and references therein). The RVcoefficient, a multivariate analogue of the squared corre-lation (Escoufier, 1973), was used as an overall measureof association between the two blocks, and a permuta-tion test (10,000 iterations) was employed against thenull hypothesis of complete independence. In addition, asecond PLS analysis was carried out on the PICs forboth blocks of variables to test if covariation between

TABLE 1. Sample of hamates analyzed in this study

Genus Species/subspecies Female Male Unknown Total

Homo Homo sapiens 8 33 14 55Pan Pan paniscus 10 11 1 22

P. troglodytes troyglodytes 5 5 1 11P. troglodytes schweinfurthii 0 2 0 2P. troglodytes verus 1 2 0 3P. troglodytes ssp. 2 2 1 5

Gorilla Gorilla gorilla 7 15 1 23G. beringei beringei 4 4 3 11G. beringei graueri 6 11 3 20

Pongo Pongo pygmaeus 7 7 0 14Pongo abelii 1 4 0 5Pongo sp. 0 1 0 1

Symphalangus Symphalangus syndactylus 4 6 0 10Hoolock Hoolock hoolock 2 3 0 5Hylobates Hylobates klossi 5 0 0 5

Hylobates lar 2 1 0 3Nomascus Nomascus leucogenys 0 1 0 1Papio Papio anubis 4 5 0 9

Papio cynocephalus 1 1 0 2Macaca Macaca mulatta 5 5 0 10Nasalis Nasalis larvatus 3 5 0 8Alouatta Alouatta seniculus 4 4 4 12Ateles Ateles belzebuth 1 0 0 1

Ateles fuscieps 0 2 0 2Ateles geoffroyi 2 0 1 3Ateles paniscus 1 2 0 3Ateles sp. 1 0 0 1

Brachyteles Brachyteles arachnoides 0 0 1 1Lagothrix Lagothrix lagothrica 0 0 1 1

Lagothrix sp. 0 0 1 1Aotus Aotus trivirgatus 0 0 1 1Saguinus Saguinus oedipus 0 0 1 1Saimiri Saimiri sciureus 0 0 1 1

Total sample 253


hamate shape and arboreal locomotion was detectedwhen considering phylogenetic structure. The amount ofshape variance explained by each PLS axis was com-puted by dividing the “singular value” (or “singularwarp”) of each PLS by total variance within the shapeblock (sum of eigenvalues, computed after a PCA of theshape data).

To visualize morphological variation associated witheach axis in each analysis, coordinates of shape changealong the axes were imported into Landmark Editor andused to warp (using thin-plate splines) a 3D hamatemodel of Pan troglodytes (shown in Fig. 3) into desiredscore values. All the statistical analyses were carried outusing MorphoJ (ver. 1.05; Klingenberg, 2011), PAST

Fig. 3. Left hamate of Pan troglodytes illustrating the 23 surface landmarks employed in this study.Warped versions of the same three-dimensional hamate model are used in subsequent figures.


(ver. 3.01; Hammer et al., 2001) and SPSS softwarepackages (ver. 17).

RESULTS

Hamate Shape Variation Among ExtantAnthropoids

Overall, humans, gorillas, hylobatids, and monkeys(and to a lesser degree chimps and orangutans) canbe differentiated from one another based on theirhamate shapes when the three first PC axes (account-ing for 53.4% of the variance, Fig. 4D) are inspectedtogether (Fig. 4A). Post hoc comparisons reveal thatmonkeys, hylobatids, Pongo, Gorilla, Pan, and H. sapi-ens are statistically different from each other based onthe variation along the three first axes (P< 0.0001).

Our analyses also show significant differences betweenhamate shape in platyrrhines and cercopithecoids(P< 0.05).

PC1 (25.8% of total variance) is related to overall pro-portions of the hamate body and hamulus, as well as tothe relative depth of the fourth and fifth metacarpalarticulations. This axis generates four relatively distinctclusters of hylobatids, monkeys, Pan/Pongo, andGorilla/Homo from left (negative) to right (positive)(Fig. 4B,C). Most Ateles specimens are near the negativeextreme of the monkey cluster, overlapping with hyloba-tids. Hylobatids have hamate bodies and hamuli thatare proximodistally long, and radioulnarly and dorsopal-marly narrow (Fig. 4D). The fifth metacarpal facet proj-ects distally over that of the fourth metacarpal whilethe medial side of the hamate is oriented obliquely. Con-versely, human and gorilla hamates are proximodistallyshort, and radioulnarly and dorsopalmarly wide, with asigmoid medial side and a more vertical overall orienta-tion of the bone.

PC2 (15.3% of variance) is also related to the radioul-nar width of the hamate body, along with the size andorientation of the hamulus and the articular surfacesfor the triquetrum and metacarpals. Along this axis,hylobatids are maximally separated from monkeys, withhominid taxa falling in between these two extremes,although humans overlap more with monkeys than dothe other hominids (Fig. 4B). Negative values corre-spond to radioulnarly narrow hamates with largehamuli that project distally, triquetral facets that aredorsopalmarly wide proximally and dorsopalmarly nar-row distally, and fifth metacarpal facets that are radio-ulnarly narrow relative to those of the fourthmetacarpal. In contrast, positive values on this axisdescribe hamates with a radioulnarly wide body, a smalland less distally projecting hamulus, a triquetral facetthat is dorsopalmarly narrow proximally and distallybut widest at its midpoint, and a fifth metacarpal facetthat that is larger than that of the fourth metacarpaland extends onto the hamulus. Together, PC1 and PC2separate hominids from both hylobatids and anthropoidmonkeys (Fig. 4B).

PC3 (12.3% of variance) is related to hamate bodyproportions, the shape and distal projection of thehamulus, and the outline shape of the metacarpalarticulations. This axis clearly separates gorillas frommodern humans (negative and positive values respec-tively), while also acting to separate Pan and monkeys(negative values) from Pongo and hylobatids (positivevalues). Negative values correspond to radioulnarlywider and proximodistally shorter hamates (but to asmaller extent than seen in PC 1 and 2), distally pro-jecting hamuli and a trapezoidal outline for the meta-carpal facets with its larger dimension located dorsally.In addition, the fifth metacarpal facet extends distallyonto the hamulus. Positive values correspond withhamate bodies that are slightly longer proximodistalyand slightly narrower radioulnarly, hamuli that extendpalmarly and a square-shaped articular outline for themetacarpals. The fifth metacarpal facet does notextend on to the hamulus. Overall, the combination ofPC1 and PC3 (Fig. 4C) separates humans, hylobatidsand gorillas (with the exception of two specimens)from the remaining taxa.

TABLE 2. Landmarks employed in this study tocapture hamate external morphology

# Type Description

1 II Junction of triquetral and capitatesurfaces on the proximopalmar side

2 II Point of maximum inflexion on thepalmar border of capitate surface

3 II Junction on the palmar border of thecapitate and metacarpal IV surfaces

4 II Junction of the triquetral/lunate andcapitate surfaces on the dorsal side

5 III Midpoint between 4 and 6 on the dorsalborder of the capitate surface

6 II Junction of the dorsal border of thecapitate and metacarpal IV surfaces

7 II Proximal root of the hamuluson the palmar side

8 II Most palmarly protruding pointof the hamulus

9 II Most distally protruding pointof the hamulus

10 II Junction of metacarpals IV and Von the palmar side

11 II On the triquetral surface,distodorsal border

12 III Midpoint between 11 and 1313 II On the triquetral surface,

distopalmar border14 II On the triquetral surface, most

convex point of the dorsal side15 II Midpoint of the triquetral surface16 II On the triquetral surface,

point of maximum inflexionon the palmar side

17 II Most proximal point on the junctionof triquetral and capitate surfaces

18 II Radiopalmar border on themetacarpal IV surface

19 II Ulnopalmar border on themetacarpal V surface

20 III Midpoint between 10 and 22,following the border betweenthe surfaces for metacarpals IV and V

21 III Midpoint between 19 and 2322 II Junction of metacarpals IV and V

on the dorsal side23 II Ulnodorsal border on the

metacarpal V surface

The description of the landmark type is after Bookstein(1991).


Phylogenetic Structure in the ExtantAnthropoid Hamate Shape and Size

Shape space patterns based on species means reason-ably display a phylogenetic structure (Fig. 5). This isconfirmed by the results of our permutation test(P< 0.0001), allowing us to reject the null hypothesis ofcomplete lack of phylogenetic signal. However, the maxi-mum axis of variance in shape space (i.e., PC1) isdefined by monkeys (especially some platyrrhine species)and hylobatids, respectively, with great apes in an inter-mediate position. Branch lengths indicate that, not onlywithin hominoids but also within the whole sample ofanthropoid species examined here, most of the shapechanges occurred in the branch (Fig. 5, marked in red)connecting the hypothetical ancestor of all hominoids(LCAH) and that of hylobatids. Modern hominids exhibitfewer shape differences in their hamates than do hyloba-tids (i.e., they have changed less) since their divergencefrom the LCAH. This phylomorphospace (reconstructedbased solely on the extant taxa included) suggests thatsince the LCAH, the hamates of modern humans andgorillas have “evolved more” than that of chimps (i.e.,common chimpanzees and bonobos), which are recon-structed by this analysis as having secondarily evolvedinto a morphology similar to that of orangutans (thusindicating homoplasy). Orangutan hamate shapeappears more conservative among great apes. Similarlyas with hamate shape, when hamate centroid size ismapped onto our anthropoid phylogeny (Fig. 6), differen-ces in size are evident among taxa and phylogeneticsignal is further detected (P< 0.0001).

Predicable Covariation of Hamate Shape andSize

The multivariate regression of shape and log-transformed centroid size (logCS) based on speciesmeans (TIPS) shows a significant relationship (P< 0.05),although accounting only for minimal portion (8.4%) oftotal shape variation (Fig. 7A). An inspection of the plot

suggests that the relationship is mostly artifactual,being caused by the presence of three platyrrhine spe-cies that are comparatively small (Saimiri, Saguinus,Aotus). This is confirmed by the regression of the phylo-genetically independent contrasts (PICs) of hamateshape and logCS, revealing no significant relationship(P 5 0.8; Fig. 7B). This result indicates that differencesin hamate shape among extant hominoids and otheranthropoids are not merely “explained” or predicted bytheir differences in size.

Locomotor Signatures in the Hamate

The results of PLS show that there is a strong andsignificant overall association between hamate shapeand the types of arboreal locomotion (RV 5 0.73; P <0.0001). The relationship stands (although not asstrongly) when considering phylogenetic structurethrough the use of PICs (RV 5 0.57; P < 0.05). Figure 8shows the results for the first two PLS that account for98% of the total covariance.

Along PLS1 (79.5% of total covariation and 99% of thetotal shape variation), negative values in the arboreallocomotion block are related to brachiation and arm-swinging, and covary most with proximodistally longand radioulnarly narrow hamate bodies, as well as withlarge and distally extending hamuli and convex trique-tral facets. Conversely, positive values in the locomotorblock are mostly related to “quadrupedal/tripedal walk”behaviors and to a lesser degree to “vertical climbingand descent” modes of locomotion. These latter arboreallocomotor types covary most in the hamate shape blockwith proximodistally shorter and radioulnarly broaderhamates, with less distally protruding hamuli and“spiral” shaped triquetral facets.

Along PLS2 (18.2% of total covariation and 47.3% ofthe total shape variation), negative values covary mostwith “orthograde clamber and transfer” behaviors and toa lesser extent with “quadrupedal/tripedal walk”,whereas in the hamate shape block they relate to proxi-modistally short and radioulnarly wide hamates, with

TABLE 3. Proportion of arboreal locomotion used in the two-block partial least squares (PLS) analysis

Taxon QTW VCD BW OCT BFS DL

Pan paniscus 0.32 0.53 0.01 0.255a 0.09 0.04Pan troglodytes schweinfurthii 0.36 0.49 0.07 0.05 0.05 0.00Pan troglodytes verus 0.22 0.68 0.03 0.08a 0.07 0.01Gorilla beringei beringei 0.53 0.4 0.02 0.06a 0.05 0.00Gorilla gorilla gorilla 0.19 0.48 0.05 0.17 0.03 0.01b

Pongo pygmaeusc 0.18 0.26 0.07 0.22 0.13 0.01Symphalangus syndactylus 0.00 0.32 0.08 0.00 0.59 0.02Hylobates lard 0.01 0.16 0.02 0.00 0.67 0.14Papio Anubis 0.68 0.21 0.00 0.00 0.00 0.10Ateles belzebuth 0.21 0.13 0.01 0.28 0.22 0.02Lagothrix lagotricha 0.29 0.14 0.00 0.30 0.09 0.04

QTW, quadrupedal and tripedal walk; VCD, vertical climb and descent; BW, bipedal walk; OCT, orthograde clamber andtransfer; BFS, brachiation and forelimb swing; DL, drop and leap.Data were obtained from Table 7 in Thorpe and Crompton (2006). Their percentages are presented here as a proportion.aAverage data for males and females, obtained from Carlson (2005).bUnknown, but because drop and leap probably constitute a very small component of western gorillas locomotor repertoire,it was approximated at 1%.cThorpe and Crompton (2006) study, omitting juveniles.dData reported in Thorpe and Crompton (2006) for gibbons corresponds to pooled data for H. agilis, H. lar and H. pileatusfrom different sources. Because data come from different sources, and not all arboreal behaviors reported in the originalstudies were included, percentages often do not add to 100%.


Fig. 4. Hamate shape variation among extant anthropoid primates.A: Major taxonomic groups are differentiated along the first three prin-cipal components (PCs). B: Plot showing PC1 vs. PC2 (25.8% and15.3% of variance explained, respectively). C: Plot showing PC1 vs.PC3 (12.3% of variance explained). D: Percentages of the total var-

iance accounted for each PC and warped surfaces (based on a thin-plate spline) representing shape changes associated with each axis (inulnar, dorsal, and distal views, respectively). PC1 represents values of20.20 and 10.20; PC2 represents values of 20.20 and 10.20; PC3represents values of 20.15 and 10.15.

large facets for the metacarpals (especially the fifth),small and nonprojecting hamuli and dorsopalmarly shorttriquetral facets on the proximal portion. In contrast,positive values that mostly covary in the arboreal loco-motor block are related to “vertical climb and descent”behavior. In the hamate shape block, this corresponds toslightly proximodistally longer and radioulnarly nar-rower hamates, with larger and distally projectinghamuli, and larger and more globular proximal portionsof the triquetral facets (thus forming a true spiral facet).

When the distribution of taxa in the PLS bivariateplots is inspected (Fig. 8), overall, closely related taxaoccupy a similar position in the scatter, similarly to theanthropoid hamate shape space (Figs. 4, 5).

DISCUSSION

Characterization of Hamate Morphology AmongExtant Anthropoids

Our 3DGM results largely agree with previous obser-vations on the relative uniformity of hamate shapewithin cercopithecoids, as well as the extensive overlapwith platyrrhine taxa (Fig. 4). At the same time, whencompared at the species level, platyrrhines exhibit con-sistent shape differences vis-�a-vis cercopithecoids (Fig.5), although more species and specimens per taxa arenecessary to formally confirm this observation. However,overall, major anthropoid taxonomic groups can be char-

acterized by means of their major trends of hamateshape variation (Fig. 4A). By relying on the two majoraxes of variation (Fig. 4B), extant hominoid hamates arereadily distinguishable from those of extant monkeys,and in turn, those of hylobatids can be distinguishedfrom those of great apes and humans. Within hominidae,human and gorilla hamates are more easily differenti-ated compared with those of chimps and orangutans(Fig. 4A,C). This observation might be explained by sym-plesiomorphy of the latter two in comparison to themore derived morphologies of humans and gorillas (Fig.5) as similarly observed by others (Kivell et al., 2013).

Inspection of the Phylogenetic Structure inAnthropoid Hamate Shape

As explained above, major shape differences existamong major taxonomic groups, and thus a significantphylogenetic pattern is detected (Fig. 5). Not only is thecolobine Nasalis similar to the papionin taxa, in accordwith prior work (Lewis, 1972; O’Connor, 1975), but pla-tyrrhines also overlap almost completely in shape spacewith cercopithecoids despite their early divergence time(Fig. 2). In contrast, extant hominoids—which are moreclosely related to each other than to platyrrhines or cer-copithecoids—occupy a larger portion of the shape space(i.e., higher shape disparity). This is remarkable consid-ering that this analysis incorporates fewer terminal taxafrom the hominoids than from monkeys (12 vs. 14

Fig. 5. Phylomorphospace of the anthropoid hamate. A: The phylog-eny presented in Fig. 1 is projected onto the shape space defined bythe two first principal components (PCs) of an analysis performed onthe covariance matrix amongst extant species means. Internal nodemorphologies were reconstructed using squared-change parsimonyand branch lengths were derived from estimated divergence times. Tofacilitate the morphospace interpretation, 3D surfaces representing

extant species average configurations (in dorsoulnar view) were plot-ted. Phylogenetic structure is detected in this shape space (treelength: 0.22; P< 0.0001). LCAH indicates the reconstructed positionfor the last common ancestor of all modern hominoids. The longestbranch (red marked) is leading from the LCAH to the hylobatids. B:Percentages of the total variance accounted for each PC.


respectively). These results suggest that hominoids, as agroup, have experienced increased selective pressures onwrist morphology in comparison to other anthropoids.The hominoid’s larger hamate shape disparity is in largepart due to the highly distinctive hylobatids; the longestbranch in the hominoid part of the tree occurs betweenthe last common ancestor of all hominoids (LCAH) andthat of all hylobatids (Fig. 5, marked in red), suggestinga large amount of evolutionary change in the lineageleading to gibbons and siamangs. Similar results havebeen found recently when relying on a Mosimann shapespace and a phylogenetic PCA (Kivell et al., 2013). Theseresults support the hypothesis that hylobatids are highlyautapomorphic in their hamate morphology (more sothan great apes relative to monkeys). Derived wrist mor-phology in hylobatids is probably related to their veryspecialized ricochetal brachiation (e.g., Jenkins, 1981),which could be considered part of an early hominoidradiation in a new adaptive zone.

Fossil evidence also strengthens the view that hyloba-tids are too derived in their wrist joints to be consideredgood ancestral models for hominoids. Based on the studyof the wrist bones of Proconsul, Lewis (1972) observedmorphological traits enhanced for greater mobility incomparison to monkeys and hypothesized that an earlyfeature defining hominoids in their new ecological nichedifferent from other catarrhine primates was related to“brachiation”. However, he admitted that the term bra-chiation, with the connotations of a highly derivedgibbon-like behavior, was probably not the mostadequate term to describe the entire repertoire of arbo-real activities displayed by early hominoids. In fact, lat-ter studies have found that, on the basis of their totalmorphological patterns, early and middle Miocene apesrelied heavily on arboreal palmigrade quadrupedalism

and some type of climbing, but not below-branch suspen-sion (e.g., Ward et al., 1993; Madar et al., 2002; Ishidaet al., 2004; Moy�a-Sol�a et al., 2004; Ward, 2007),although some workers have inferred knuckle-walkingfor Sivapithecus (Begun and Kivell, 2011).

The shape space defined by anthropoid primatemeans (Fig. 5) is similar to that of individual speci-mens (Fig. 4B) although it shows an inversion of thetwo major axes of variance (i.e., PC1 in the first anal-ysis is PC2 in the second and vice versa). There is ashape shift in modern hominoids relative to monkeys.This shift involves an enlargement of the hamuluswith a concomitant reduction of the relative size ofthe metacarpal facets, as well as a dorsopalmarenlargement of the proximal portion of the triquetralfacet, which also happens to be more globular (seechanges in PC2 of Fig. 4). This latter morphologymay be related to the “concavo-convex spiral complex”described in extant great apes by Lewis (1972; seediscussion below). As for the enlargement of thehamulus in hominoids, the small (and likely plesio-morphic) hamulus described for Proconsul (Beardet al., 1986) and Sivapithecus (Spoor et al., 1991) sug-gests that its size increased independently in hyloba-tids and extant great apes. The results of our PLS(Fig. 8) reinforce the idea that the small hamuli offossil apes are related to quadrupedalism (PLS1) and/or even orthograde clambering (PLS2).

Extant great apes (especially gorillas) and humansoccupy a region of the hamate shape space that is char-acterized by dorsopalmarly extended hamates (andhamuli) that are also less obliquely oriented. This is alsoevidenced by the more nearly coplanar positioning of themetacarpal facets (i.e., the fifth metacarpal facet doesnot extends as far distally beyond the fourth when the

Fig. 6. Anthropoid phylogeny mapped on hamate centroid size. A significant phylogenetic signal isdetected (tree length: 848.88; P < 0.0001).


Fig. 7. Predictable covariance of anthropoid hamate shape and size. A: Anthropoid species averageshapes regressed on log-transformed centroid size (TIPS) reveals a significant relationship (P < 0.05)although only accounting for 8.4% of the total shape variation. B: Regression of phylogenetic independ-ent contrasts (PICs) of shape and log-transformed centroid size reveals no significant relationship(P 5 0.8).

radial side of the bone is aligned vertically as it does inmonkeys and hylobatids; see PC1 in Fig. 4).

Since the reconstructed LCAH, orangutans (followedby chimps) seem to possess the most plesiomorphichamate shape, while gorillas and humans with stockierhamate bodies and palmarly extending hamuli displaythe most derived hamates within extant Hominidae, inagreement with recent findings using similar methods(Kivell et al., 2013). While the hamate morphology ofgorillas can be intuitively associated with weight bearingin these large-bodied terrestrial hominoids (Sarmiento,1994), the morphology of the human hamate (freed fromselective locomotor pressures) should be explored underthe scope of enhanced manipulation. Future analysesincluding fossil species near major evolutionary splittingevents (e.g., Alm�ecija et al., 2013) and employing phylo-genetically weighted PCA (e.g., Kivell et al., 2013; Pollyet al., 2013) may allow a better determination of thepolarity and degree of shape change within each lineage,and thus properly allow us to assess and resolve the pat-terns and tempo of hominoid hamate evolution.

Is There Predictable Covariation betweenHamate Shape and Size?

Although significant phylogenetic patterning exists inboth hamate shape and size (Figs. 5, 6), a very weak cor-relation of hamate shape and size exists when speciesmeans are compared to each other (Fig. 7A), and even

becomes statistically nonsignificant when studying theirphylogenetic independent contrasts (Fig. 7B). This indi-cates that the observed shape differences between bigtaxonomic groups are not simply correlated with overallsize. However, it could be argued that some of the loco-motor behaviors displayed by different anthropoid taxaanalyzed here are more or less compatible with a certainbody mass. For example, among other factors, brachia-tion is facilitated in hylobatids because they are small(e.g., Preuschoft and Demes, 1985; Swartz, 1989). Inthat sense, certain behaviors could show some amount ofsignificant covariation with size (predictable or not) andin turn with specific aspects of shape (see following sec-tion). Future analyses (Alm�ecija et al., in preparation)will inspect if and how patterns of hamate shape varia-tion within hominoid species are predictably correlatedwith size.

Exploring Covariation of Hamate Shape andLocomotion

Most of the hamate functional morphology discussionhas focused on the shape of the triquetral articular sur-face and the length and orientation of the hamulus. Inthis section, we discuss how previous observations onthe anthropoid hamate morphology relate to our results.

Lewis (1972) considered African apes to share a com-mon hamate morphology: a narrower version of thewedge-shaped monkey hamate that is oriented with the

Fig. 8. Two-block partial least squares (PLS) analysis of hamateshape and arboreal locomotor categories revealing a significant asso-ciation (RV 5 0.73; P < 0.0001). A: PLS1 scores of arboreal locomo-tion vs. hamate shape. B: PLS2 scores of arboreal locomotion vs.hamate shape. C: Total amount of covariance contained in each PLS

axis as well as a biplot with the PLS coefficients for the arboreal loco-motion block. D: PLS coefficients for the hamate shape block are dis-played as shape changes corresponding to PLS scores of (20.2 and10.1) in each axis.


long axis of the body directed proximodistally instead ofobliquely as in monkeys and hylobatids. This morphologyreduces the articulation with the lunate (increasing artic-ulation with the capitate) and causes the triquetral facetto be aligned more obliquely. This arrangement causesthe triquetral facet to be more ulnarly oriented overall,but distally “inflected” toward the hamulus, resulting inthe so-called “spirally concavo-convex” facet. Lewis (1972)contended that the “spiral” character of the triquetralfacet contributed to the “screw mechanism” of the midcar-pal complex, working as a “close-packing” system limitingextension, which he argued is an adaptation to resistingthe tensile forces associated with regular suspension.Such screw-clamp kinematics of the carpus has beenfound in humans and chimpanzees (MacConaill, 1941;Orr et al., 2010), but the exact role of triquetrohamateform in that mechanism is not clear (Orr, 2010). Further-more, Jenkins and Fleagle (1975) argued that the pres-ence of the spiral concavo-convex complex is common inquadrupeds such as Macaca, thus falsifying the connec-tion between this trait and below-branch suspension. Sar-miento (1994) pointed out that the spiral complexeffectively increases the radius of curvature of the carpalrow proximally and increases the surface area of the dis-tal triquetrohamate joint that is directed orthogonaltransarticular loads when the midcarpus is extended. Assuch, he suggests that this character is related to quadru-pedalism rather than suspension.

It has been hypothesized that the overall degree of con-vexity of the hamate triquetral facet reflects the potentialrange of flexion-extension at the midcarpus (Spoor et al.,1991). This facet is more curved (or globular) in apes andflatter proximally in terrestrial monkeys, with the latterostensibly preventing extension during terrestrial locomo-tion in nonprimate mammals (Yalden, 1971). However,cadaver-based experiments indicate that although Papiohas limited mobility at the triquetrohamate joint duringdorsiflexion, palmigrade-capable monkeys exhibit muchlarger ranges of motion at this articulation than do chim-panzees or orangutans (Orr, 2010).

Sarmiento (1988) provided the first quantitative studyrelating the orientation of the triquetral facet (comput-ing angles between the capitate, triquetral and metacar-pal facets) with the capacity of weight transfer throughthe triquetrohamate joint. As with Lewis, Sarmientorelated the more ulnarly oriented triquetral facets ofhominoids such as Pan and Pongo to less effectiveweight transmission and wider range of movement atthe midcarpal joint (in comparison to that of monkeys).However, he noted that the triquetral facet in gorillas isnearly parallel to the articular surfaces for the metacar-pals, thus favoring weight support (Sarmiento, 1988,1994). Other studies using traditional goniometric meas-urements (Richmond, 2006) and least-squares planes fitto segmented joint surfaces of 3D polygon models of thehamate (Orr et al., 2013) had similar results with goril-las displaying the most proximally oriented triquetralsurfaces.

Our PLS results based on arboreal locomotor behav-iors and with PLS1 accounting for 79.5% of shape-arboreal locomotion covariation (and up to 99% of thetotal hamate shape variation) also indicate that quadru-pedal primates exhibit wider hamate bodies with lessglobular and more proximally oriented triquetral facetsand smaller hamuli than do suspensory taxa (Fig. 8). In

terms of the “spiral” complex on the triquetral facet—assuming that its morphology is related to the long axisof the hamate body being more proximodistally aligned,with a globular (more convex) proximal end and a moreconcave distal portion—our PC1 of the entire anthropoidsample (Fig. 4) captures a gradient (negative to positive)from hylobatids to monkeys, followed by Pongo/Pan andfinally Gorilla/Homo. Therefore, our results agree withthose of Lewis (1972, 1989) in indicating better develop-ment of this complex in extant great apes (and humans)than in monkeys. However, hylobatids show the mostdisparate (autapomorphic) morphology, with no trace of“spiralization”. Instead, they exhibit an oblique proximo-distal axis of the hamate with a large and very globulartriquetral facet that lacks the distal concavity. Thisshape is most likely related to the “ball-and-socket” mid-carpal morphology that has been widely interpreted asan adaptation for their very specialized ricochetal bra-chiation (e.g., Tuttle, 1969; Lewis, 1972; Usherwoodet al., 2003). Hylobatids also exhibit relatively proximo-distal long hamates, which Kivell et al. (2013) hypothe-size is related to their overall long forelimbs (Drapeauand Ward, 2007) in relation to enhanced suspension.Hylobatids are also the smallest of the modern homi-noids (Smith and Jungers, 1997), which is probablytightly related to their unique capability among homi-noids to perform a true “ricochetal brachiation” involvinga period of free flight (e.g., Conroy and Fleagle, 1972;Preuschoft and Demes, 1985; Swartz, 1989; Hunt, 1991).Hylobatids therefore define the shape that is related tobrachiation along the first axis of the PLS analysis (Fig.8A,C).

Although there is some overlap, Pan and Gorilla dis-play a more globular proximal triquetral facet (i.e., con-cavoconvex) than does Pongo (PC3 in Fig. 4C). This mayreflect a wider weight-bearing midcarpal surface in theAfrican apes. The complex morphology of the spiral con-cavoconvex morphology found in some monkeys (Jenkinsand Fleagle, 1975), Proconsul (Lewis, 1972), orangutans,and especially some African ape taxa might be explainedin terms of the functional demands of arboreal locomo-tion in these forms. These locomotor behaviors requireboth enhanced mobility (i.e., rotation and ulnar devia-tion) and stability. The close-packing mechanism pro-vided by the spiral facet has been hypothesized to beadvantageous during knuckle-walking (e.g., Richmondet al., 2001), but the distribution of this feature mayinstead indicate its importance during use of the handin arboreal settings. High positive loadings of verticalclimbing associated with a highly convex proximal tri-quetral facet in PLS2 further suggest that this is thecase (Fig. 8).

In reference to the greater development of the hamu-lus in extant hominoids in comparison to monkeys, ithas been related to the action of the flexor carpi ulnaris(which primarly inserts on the pisiform) but acts on thehamulus by means of the pisohamate ligament (Sch€onand Ziemer, 1973; O’Connor, 1975; Lewis, 1977). A pal-marly protruding hamulus favors powerful flexion of afully extended wrist (Sarmiento, 1988). Conversely, thehamulus is more distally oriented in hominoid taxa thatregularly use arboreal supports. A distally orientedhamulus improves the lever arm of the flexor carpi ulna-ris for ulnar deviation, flexion of an already flexed wrist,or preventing dorsiflexion (Sarmiento, 1988, 1994; Patel


et al., 2012); thus, it is probably related to increasingwrist torque potential during climbing or suspension.Sarmiento (1994) pointed out that the hamulus of goril-las is better developed than in chimpanzees, with amore palmar extension similar to that of modernhumans. He argued that this provides an advantageouslever arm for flexion of an extended wrist and preventsextension at the hamate-metacarpal fifth joint. Such anarrangement may be related to use of the hand for pro-pulsion during terrestrial quadrupedalism (Sarmiento,1994; see also Patel et al., 2012).

As for the palmar extension of the modern humanhamate, it should probably be interpreted as a modifica-tion for enhanced manipulation (Orr et al., 2013). Twointrinsic hypothenar muscles (flexor digiti minimi andopponens digiti minimi) originate on the hamulus andinsert on the ulnar side of the fifth metacarpal shaft andthe base of the fifth proximal phalanx. In humans, thesemuscles act in flexion and opposition of the fifth digit tothe lateral fingers, thus facilitating the palmar cuppingand cylindrical grips important for a variety of humangrips (Marzke et al., 1992, 1998; Reece, 2005; Niewoeh-ner, 2006; Marzke, 2013). Our PLS results using thewhole hamate landmark configuration as a multidimen-sional shape variable (Fig. 8) indicate that a distally pro-truding hamulus is associated with both suspension(negative values along PLS1) and vertical climbing (posi-tive values along PLS2), thus reinforcing the previouslystated arguments. However, the hominoid suspensorytraits are not observed in Ateles, as previously reported(e.g., Lewis, 1989; Kivell et al., 2013) which has beenexplained by the use of a prehensile tail during suspen-sory behaviors and their substantial amount of timeengaging in quadrupedal locomotion (Cant et al., 2001;see Table 3). Other studies have shown convergencesbetween Ateles and hylobatids concentrated in theshoulder, humeral shaft, and elbow (e.g., Rose, 1996;Larson, 1998; Young, 2003; Patel et al., 2013).

Although a PLS based on the PICs of shape and arbo-real locomotion (not shown) was still found to be statisti-cally significant (although accounting for a lesserportion of total covariance), the proximity of closelyrelated taxa in the PCA (Fig. 4, 5), size (Fig. 6), and PLSplots (Fig. 8) highlights the fact that closely related taxanot only look similar but they also do similar things.Although differences in shape exist between closelyrelated taxa (e.g., Pan is more similar to Pongo than toHomo), overall, large taxonomic groups (i.e., platyrrhineand cercopithecoids, hylobatids and great apes) sharesimilar morphology (Fig. 5) and locomotion (Fig. 8).Therefore, our results on hamate shape and its covaria-tion with locomotion support the idea that some strictlyadaptive morphological changes (i.e., associated alsowith a change in function; Gould and Vrba, 1982) origi-nate with phylogenetic events, strengthening the viewthat phylogeny and adaptation are part of the same com-plex (Polly et al., 2013 and references therein).

SUMMARY AND CONCLUSIONS

In this work, and with a deep exploratory spirit, wesought to quantitatively characterize a biologically com-plex morphology, such is the case of the anthropoidhamate, using three-dimensional geometric morphomet-rics. Major trends of shape variation were inspected in

253 anthropoid primates using principal componentsanalysis on the covariance matrix of 23 Procrustes-aligned surface landmarks. This approach allowed us tocharacterize and distinguish the hamate of hominoidsfrom those of platyrrhine and cercopithecoid monkeys.Furthermore, differences between hylobatids and greatapes and even among great ape genera were capturedwithout relying on a priori grouping methods like canon-ical discriminant functions. This indicates that 3DGM isa reliable tool for accurately capturing subtle shape dif-ferences in complex biological structures.

The anthropoid primate phylogenetic tree was mappedonto hamate shape and size spaces by reconstructinginternal nodes using squared-change parsimony. Thisallowed us to perform permutation tests that identifiedthe presence of phylogenetic structure in both hamateshape and size. This indicates that, overall, more closelyrelated taxa tend to exhibit more similar hamates (prob-ably also due to phylogenetic clustering of similar loco-motor behaviors). Although different anthropoidtaxonomic groups exhibit both different hamate shapeand size, allometric regressions reveal very small to non-significant (TIPS vs. PICs respectively) predictable cova-riation between hamate shape and size. Thus, ourresults show that differences in hamate size alone do notexplain differences in hamate shape among taxa. We fur-ther hypothesized about the functional evolution of thehominoid wrist on the basis of the hamate.

We also used two-block partial least squares of shapedata and a matrix of arboreal locomotor behaviors totest the covariance of hamate shape and observed loco-motion without relying on the predictive models implicitin regression analyses. This analysis revealed that,among anthropoids, some aspects of hamate shapecovary significantly with the frequency of brachiation,arboreal quadrupedalism and vertical climbing. Futurework incorporating fossils will focus on further elucidat-ing the evolutionary histories of hamate phenotypic evo-lution and functional morphology.

ACKNOWLEDGEMENTS

We are indebted to the following researchers and cura-tors for granting access to collections under their care:Emmanuel Gilissen and Wim Wendelen, Royal Museumof Central Africa; Patrick Semal and Georges Lenglet,Royal Belgian Institute of Natural Sciences; Judy Chu-pasko, Museum of Comparative Zoology; Eileen Westwigand Neil Duncan, American Museum of Natural History;Richard Thorington and Linda Gordon, United StatesNational Museum (Smithsonian). Ashley Hammond gen-erously provided most of the Symphalangus hamatescans used in this study from the Bavarian State Zoolog-ical Collections. Soledad de Esteban-Trivigno offeredinsightful methodological discussion. Paul O’Higgins andKristi Lewton educated one of us (S.A.) in the applica-tion and use of PLS to coordinate shape data and matri-ces of locomotor behaviors. A “Transmitting Science”course on GM and phylogenetics imparted by Chris Klin-genberg during September 2013 near Barcelona inspiredpart of the analyses performed in this work. We areespecially grateful to Siobh�an Cooke and Claire Terhunefor inviting us to provide a manuscript to this specialissue of The Anatomical Record. This is NYCEP morpho-metrics contribution number 86.


LITERATURE CITED

Alba DM, Alm�ecija S, Moy�a-Sol�a S. 2010. Locomotor inferences inPierolapithecus and Hispanopithecus: reply to Deane and Begun(2008). J Hum Evol 59:143–149.

Alm�ecija S, Alba DM, Moy�a-Sol�a S. 2009. Pierolapithecus and thefunctional morphology of Miocene ape hand phalanges: paleobio-logical and evolutionary implications. J Hum Evol 57:284–297.

Alm�ecija S, Tallman M, Alba DM, Pina M, Moy�a-Sol�a S, JungersWL. 2013. The femur of Orrorin tugenensis exhibits morphomet-ric affinities with both Miocene apes and later hominins. NatCommun 4:2888.

Arnold C, Matthews LJ, Nunn CL. 2010. The 10kTrees website: anew online resource for primate phylogeny. Evol Anthropol 19:114–118.

Bastir M, Garc�ıa Mart�ınez D, Recheis W, Barash A, Coquerelle M,Rios L, Pe~na-Meli�an �A, Garc�ıa R�ıo F, O’Higgins P. 2013. Differen-tial growth and development of the upper and lower humanthorax. PLoS ONE 8:e75128.

Beard KC, Teaford MF, Walker A. 1986. New wrist bones of Procon-sul africanus and P. nyanzae from Rusinga Island, Kenya. FoliaPrimatol 47:97–118.

Begun DR, Kivell TL. 2011. Knuckle-walking in Sivapithecus? Thecombined effects of homology and homoplasy with possible impli-cations for pongine dispersals. J Hum Evol 60:158–170.

Bookstein FL. 1991. Morphometric tools for landmark data: geome-try and biology. Cambridge: University Press.

Cant JG, Youlatos D, Rose MD. 2001. Locomotor behavior of Lago-thrix lagothricha and Ateles belzebuth in Yasun�ı National Park,Ecuador: general patterns and nonsuspensory modes. J Hum Evol41:141–166.

Carlson KJ. 2005. Investigating the form–function interface in Afri-can apes: relationships between principal moments of area andpositional behaviors in femoral and humeral diaphyses. Am JPhys Anthropol 127:312–334.

Cartmill M, Milton K. 1977. The lorisiform wrist joint and the evo-lution of "brachiating" adaptations in the Hominoidea. Am J PhysAnthropol 47:249–272.

Conroy GC, Fleagle JG. 1972. Locomotor behavior in living and fos-sil pongids. Nature 237:103–104.

Dryden I, Mardia K. 1998. Statistical shape analysis. New York:Wiley.

Dunn RH, Tocheri MW, Orr CM, Jungers WL. 2014. Ecologicaldivergence and talar morphology in gorillas. Am J Phys Anthro-pol 153:526–541.

Escoufier Y. 1973. Le traitement des variables vectorielles. Biomet-rics 29:751–760.

Felsenstein J. 1985. Phylogenies and the comparative method. AmNat 125:1–15.

Figueirido B, Tseng ZJ, Mart�ın-Serra A. 2013. Skull shape evolutionin durophagus carnivorans. Evolution 67:1975–1993.

Fleagle JG. 1999. Primate adaptation and evolution, 2nd ed.Amsterdam: Elsevier.

Gould SJ, Vrba ES. 1982. Exaptation—a missing term in the sci-ence of form. Paleobiology 8:4–15.

Hammer O, Harper D. 2006. Paleontological data analysis. Malden:Blackwell Publishing.

Hammer Ø, Harper DAT, Ryan PD. 2001. PAST: paleontological sta-tistics software package for education and data analysis. PalaeoElectron 4:1–9.

Hunt KD. 1991. Positional behavior in the Hominoidea. Int J Pri-matol 12:95–118.

Ishida H, Kunimatsu Y, Takano T, Nakano Y, Nakatsukasa M.2004. Nacholapithecus skeleton from the Middle Miocene ofKenya. J Hum Evol 46:69–103.

Jenkins, FA. 1981. Wrist rotation in primates: a critical adaptationfor brachiators. Symp Zool Soc Lond 48:429–451.

Jenkins FA, Fleagle JG. 1975. Knuckle-walking and the functionalanatomy of the wrist in living apes. In: Tuttle RH, editor. Primatefunctional morphology and evolution. The Hague: Mouton, p 213–227.

Kivell T, Barros A, Smaers J. 2013. Different evolutionary pathwaysunderlie the morphology of wrist bones in hominoids. BMC EvolBiol 13:229.

Klingenberg CP. 2011. MorphoJ: an integrated software package forgeometric morphometrics. Mol Ecol Resour 11:353–357.

Klingenberg CP, Gidaszewski NA. 2010. Testing and quantifyingphylogenetic signals and homoplasy in morphometric data. SystBiol 59:245–261.

Laurin M. 2004. The evolution of body size, Cope’s rule and the ori-gin of amniotes. Syst Biol 53:594–622.

Larson SG. 1998. Parallel evolution in the hominoid trunk and fore-limb. Evol Anthropol 6:87–99.

Lewis OJ. 1965. Evolutionary change in the primate wrist and infe-rior radio-ulnar joints. Anat Rec 151:275–286.

Lewis OJ. 1969. The hominoid wrist joint. Am J Phys Anthropol 30:251–268.

Lewis OJ. 1972. Osteological features characterizing the wrists ofmonkeys and apes, with a reconsideration of this region in Dryo-pithecus (Proconsul) africanus. Am J Phys Anthropol 36:45–58.

Lewis OJ. 1977. Joint remodelling and the evolution of the humanhand. J Anat 123:157–201.

Lewis OJ. 1989. Functional morphology of the evolving hand andfoot. Oxford: Clarendon Press.

MacConaill MA. 1941. The mechanical anatomy of the carpus andits bearing on some surgical problems. J Anat 75:166–175.

Madar SI, Rose MD, Kelley J, MacLatchy L, Pilbeam D. 2002. NewSivapithecus postcranial specimens from the Siwaliks of Pakistan.J Hum Evol 42:705–752.

Maddison WP. 1991. Squared-change parsimony reconstructions ofancestral states for continuous-valued characters on a phyloge-netic tree. Syst Zool 40:304–314.

Marzke MW. 2013. Tool making, hand morphology and fossil homi-nins. Phil Trans R Soc B 368:20120414.

Marzke MW, Marzke RF. 2000. Evolution of the human hand:approaches to acquiring, analyzing and interpreting the anatomi-cal evidence. J Anat 197:121–140.

Marzke MW, Tocheri MW, Steinberg B, Femiani JD, Reece SP,Linscheid RL, Orr CM, Marzke RF. 2010. Comparative 3D quanti-tative analyses of trapeziometacarpal joint surface curvaturesamong living catarrhines and fossil hominins. Am J Phys Anthro-pol 141:38–41.

Marzke MW, Toth N, Schick K, Reece S, Steinberg B, Hunt K,Lindscheid RL, An KN. 1998. EMG study of hand muscle recruit-ment during hard mammer percussion maufacture of Oldowantools. Am J Phys Anthropol 105:315–332.

Marzke MW, Wullstein KL, Viegas SF. 1992. Evolution of the power(“squeeze”) grip and its morphological correlates in hominids. AmJ Phys Anthropol 89:283–298.

McHenry HM, Corruccini RS. 1975. The wrist of Proconsul africa-nus and the origin of hominoid postcranial adaptations. In Cio-chon RL, Corruccini RS, editors. New interpretations of ape andhuman ancestry. New York: Plenum Press, p 353–367.

Monteiro LR. 2013. Morphometrics and the comparative method:studying the evolution of biological shape. Hystrix 24:12.

Mootnick A, Groves C. 2005. A new generic name for the hoolockgibbon (Hylobatidae). Int J Primatol 26:971–976.

Moy�a-Sol�a S, K€ohler M, Alba DM, Casanovas-Vilar I, Galindo J.2004. Pierolapithecus catalaunicus, a new Middle Miocene greatape from Spain. Science 306:1339–1344.

Niewoehner WA. 2006. Neanderthal hands in their proper perspec-tive. In: Harvati, K., Harrison, T, editors. Neanderthals revisited:new approaches and perspectives. Dordrecht: Springer, p 157–190.

O’Connor BL. 1975. The functional morphology of the cercopithecoidwrist and inferior radioulnar joints, and their bearing on someproblems in the evolution of the Hominoidea. Am J Phys Anthro-pol 43:113–122.

Orr CM. 2010. Adaptations to knuckle-walking and digitigrady: athree-dimensional kinematic and morphometric analysis of theanthropoid wrist. Doctoral dissertation, Arizona State University,Tempe, AZ.


Orr CM, Leventhal EL, Chivers SF, Marzke MW, Wolfe SW, CriscoJJ. 2010. Studying primate carpal kinematics in three-dimensionsusing a computed-tomography-based markerless registrationmethod. Anat Rec 293:692–709.

Orr CM, Tocheri MW, Burnett SE, Awe RD, Saptomo EW, SutiknaT, Jatmiko, Wasisto S, Morwood MJ, Jungers WL. 2013. Newwrist bones of Homo floresiensis from Liang Bua (Flores, Indone-sia). J Hum Evol 64:109–129.

Patel BA. 2009. Not so fast: speed effects on forelimb kinematics incercopithecine monkeys and implications for digitigrade posturesin primates. Am J Phys Anthropol 140:92–112.

Patel BA, Larson SG, Stern, JT. 2012. Electromyography of wristand finger flexor muscles in olive baboons (Papio anubis). J ExpBiol 215:115–123.

Patel BA, Ruff CB, Simons ELR, Organ JM. 2013. Humeral cross-sectional shape in suspensory primates and sloths. Anat Rec 296:545–556.

Polly PD. 2008. Adaptive zones and the pinniped ankle: a three-dimensional quantitative analysis of carnivoran tarsal evolution.In: Sargis E, Dagosto M, editors. Mammalian evolutionary mor-phology: a tribute to Frederick S. Szalay. Dordrecht: Springer. p167–196.

Polly PD, Lawing AM, Fabre A-C, Goswami A. 2013. Phylogeneticprincipal components analysis and geometric morphometrics.Hystrix 24:9.

Prado-Martinez J, Sudmant PH, Kidd JM, Li H, Kelley JL,Lorente-Galdos B, Veeramah KR, Woerner AE, O’Connor TD,Santpere G, Cagan A, Theunert C, Casals F, Laayouni H, MunchK, Hobolth A, Halager AE, Malig M, Hernandez-Rodriguez J,Hernando-Herraez I, Prufer K, Pybus M, Johnstone L, LachmannM, Alkan C, Twigg D, Petit N, Baker C, Hormozdiari F,Fernandez-Callejo M, Dabad M, Wilson ML, Stevison L,Camprubi C, Carvalho T, Ruiz-Herrera A, Vives L, Mele M,Abello T, Kondova I, Bontrop RE, Pusey A, Lankester F, KiyangJA, Bergl RA, Lonsdorf E, Myers S, Ventura M, Gagneux P,Comas D, Siegismund H, Blanc J, Agueda-Calpena L, Gut M,Fulton L, Tishkoff SA, Mullikin JC, Wilson RK, Gut IG, GonderMK, Ryder OA, Hahn BH, Navarro A, Akey JM, Bertranpetit J,Reich D, Mailund T, Schierup MH, Hvilsom C, Andres AM, WallJD, Bustamante CD, Hammer MF, Eichler EE, Marques-Bonet T.2013. Great ape genetic diversity and population history. Nature499:471–475.

Preuschoft H, Demes B. 1985. Influence of size and proportions onthe biomechanics of brachiation. In: Jungers WL, editor. Size andscaling in primate biology. New York: Plenum. p 383–399.

Reece S. 2005. Getting a grip: functional morphology of the fifth rayin eight catarrhine species. PhD Dissertation, Arizona StateUniversity.

Rein TR, Harvati K. 2013. Exploring third metacarpal capitate facetshape in early hominins. Anat Rec 296:240–249.

Richmond BG. 2006. Functional morphology of the midcarpal jointin knuckle-walkers and terrestrial quadrupeds. In: Ishida H, Tut-tle R, Pickford M, Nakatsukasa M, editors. Human origins andenvironmental backgrounds. New York: Springer. p 105–122.

Richmond BG, Begun DR, Strait DS. 2001. Origin of human biped-alism: the knuckle-walking hypothesis revisited. Am J PhysAnthropol Suppl 33:70–105.

Rohlf FJ. 2002. Geometric morphometrics and phylogeny. In:MacLeod N, Forey PL, editors. Morphology, shape, and phylogeny.London: Taylor and Francis. p 175–193.

Rohlf FJ, Corti M. 2000. Use of two-block partial least-squares tostudy covariation in shape. Syst Biol 49:740–753.

Rohlf FJ, Marcus LF. 1993. A revolution in morphometrics. TrendsEcol Evol 8:129–132.

Rose M. 1996. Functional morphological similarities in the locomo-tor skeleton of Miocene catarrhines and platyrrhine monkeys.Folia Primatol 66:7–14.

Sarmiento EE 1985 Functional differences in the skeleton of wildand captive orangutans and their adaptive significance. PhD dis-sertation, New York University.

Sarmiento EE. 1988. Anatomy of the hominoid wrist joint: its evolu-tionary and functional implications. Int J Primatol 9:281–345.

Sarmiento EE. 1994. Terrestrial traits in the hands and feet ofgorillas. Am Mus Novitates 3091:1–56.

Sch€on MA, Ziemer LK. 1973. Wrist mechanism and locomotorbehavior of Dryopithecus (Proconsul) africanus. Folia Primatol 20:1–11.

Sidlauskas B. 2008. Continuous and arrested morphological diversi-fication in sister clades of characiform fishes: a phylomorphospaceapproach. Evolution 62:3135–3156.

Smith RJ, Jungers WL. 1997. Body mass in comparative primatol-ogy. J Hum Evol 32:523–559.

Spoor CF, Sondaar PY, Hussain ST. 1991. A new hominoid hamateand first metacarpal from the Late Miocene Nagri Formation ofPakistan. J Hum Evol 21:413–424.

Swartz SM. 1989. Pendular mechanics and the kinematics andenergetics of brachiating locomotion. Int J Primatol 10:387–418.

Sylvester AD. 2013. A geometric morphometric analysis of themedial tibial condyle of African hominids. Anat Rec 296:1518–1525.

Tallman M, Alm�ecija S, Reber SL, Alba DM, Moy�a-Sol�a S. 2013.The distal tibia of Hispanopithecus laietanus: More evidence formosaic evolution in Miocene apes. J Hum Evol 64:319–327.

Terhune CE. 2013. How effective are geometric morphometric tech-niques for assessing functional shape variation? An example fromthe great ape temporomandibular joint. Anat Rec 296:1264–1282.

Thorpe SKD, Crompton RH. 2006. Orangutan positional behaviorand the nature of arboreal locomotion in Hominidea. Am J PhysAnthropol 131:384–401.

Tocheri MW, Orr CM, Larson SG, Sutikna T, Jatmiko, SaptomoWW, Due RA, Djubiantono T, Morwood MJ, Jungers WL. 2007.The primitive wrist of Homo floresiensis and its implications forhominin evolution. Science 317:1743–1745.

Tocheri MW, Razdan A, Williams RC, Marzke MW. 2005. A 3Dquantitative comparison of trapezium and trapezoid relative artic-ular and nonarticular surface areas in modern humans and greatapes. J Hum Evol 49:570–586.

Tocheri MW, Solhan CR, Orr CM, Femiani J, Frohlich B, GrovesCP, Harcourt-Smith WE, Richmond BG, Shoelson B, Jungers WL.2011. Ecological divergence and medial cuneiform morphology ingorillas. J Hum Evol 60:171–184.

Turley K, Frost SR. 2013. The shape and presentation of the Catar-rhine talus: a geometric morphometric analysis. Anat Rec 296:877–890.

Turley K, Guthrie EH, Frost SR. 2011. Geometric morphometricanalysis of tibial shape and presentation among catarrhine taxa.Anat Rec 294:217–230.

Tuttle RH. 1967. Knuckle-walking and the evolution of hominoidhands. Am J Phys Anthropol 26:171–206.

Ward C. 2007. Postcranial and locomotor adaptations of hominoids.In: Henke W, Tattersall I, editors. Handbook of paleoanthropology.Heidelberg: Springer. p 1011–1030.

Ward CV, Walker A, Teaford MF, Odhiambo I. 1993. Partial skeletonof Proconsul nyanzae from Mfangano Island, Kenya. Am J PhysAnthropol 90:77–111.

Wiley DF, Amenta N, Alcantara DA, Ghosh D, Kil YJ, Delson E,Harcourt-Smith W, Rohlf FJ, John KS, Hamann B. 2005. Evolu-tionary morphing [Abstract]. In: IEEE Visualization Conference.Minneapolis, MN. p 431–438.

Yalden DW. 1971. The functional morphology of the carpus in ungu-late mammals. Acta Anat 78:461–487.

Young NM. 2003. A reassessment of living hominioid postcranialvariability: implications for ape evolution. J Hum Evol 45:441–464.


Exploring Phylogenetic and Functional Signals in Complex Morphologies: The Hamate of Extant...

Documents

Transcript of Exploring Phylogenetic and Functional Signals in Complex Morphologies: The Hamate of Extant...