Three-Dimensional GeometricMorphometric Analysis of Talar

Morphology in Extant Gorilla Taxafrom Highland and Lowland Habitats

RYAN P. KNIGGE,1* MATTHEW W. TOCHERI,2,3 CALEY M. ORR,4

AND KIERAN P. MCNULTY1

1Evolutionary Anthropology Lab, Department of Anthropology, University of Minnesota,Minneapolis, Minnesota

2Human Origins Program, Department of Anthropology, National Museum of NaturalHistory, Smithsonian Institution, Washington, DC

3Center for the Advanced Study of Hominid Paleobiology, Department of Anthropology,The George Washington University, Washington, DC

4Department of Anatomy, Midwestern University, Downers Grove, Illinois

ABSTRACTWestern gorillas (Gorilla gorilla) are known to climb significantly

more often than eastern gorillas (Gorilla beringei), a behavioral distinc-tion attributable to major differences in their respective habitats (i.e.,highland vs. lowland). Genetic evidence suggests that the lineages lead-ing to these taxa began diverging from one another between approxi-mately 1 and 3 million years ago. Thus, gorillas offer a specialopportunity to examine the degree to which morphology of recentlydiverged taxa may be “fine-tuned” to differing ecological requirements.Using three-dimensional (3D) geometric morphometrics, we comparedtalar morphology in a sample of 87 specimens including western (low-land), mountain (highland), and grauer gorillas (lowland and highlandpopulations). Talar shape was captured with a series of landmarks andsemilandmarks superimposed by generalized Procrustes analysis. Abetween-group principal components analysis of overall talar shape sepa-rates gorillas by ecological habitat and by taxon. An analysis of only thetrochlea and lateral malleolar facet identifies subtle variations in troch-lear shape between western lowland and lowland grauer gorillas, poten-tially indicative of convergent evolution of arboreal adaptations in thetalus. Lastly, talar shape scales differently with centroid size for highlandand lowland gorillas, suggesting that ankle morphology may track body-size mediated variation in arboreal behaviors differently depending onecological setting. Several of the observed shape differences are linkedbiomechanically to the facilitation of climbing in lowland gorillas and tostability and load-bearing on terrestrial substrates in the highland taxa,providing an important comparative model for studying morphologicalvariation in groups known only from fossils (e.g., early hominins). AnatRec, 298:277–290, 2015. VC 2014 Wiley Periodicals, Inc.

Key words: foot; arboreality; terrestriality; talus; tarsals

Grant sponsor: Wenner-Gren Foundation; Grant number:7822.

*Correspondence to: Ryan P. Knigge, Department of Anthro-pology, University of Minnesota, Minneapolis, MN 55454, USA.E-mail: knigg008@umn.edu

Received 3 October 2014; Accepted 11 October 2014.

DOI 10.1002/ar.23069Published online in Wiley Online Library (wileyonlinelibrary.com).

THE ANATOMICAL RECORD 298:277–290 (2015)

VVC 2014 WILEY PERIODICALS, INC.

The talus serves as the connection between the bonesof the lower limb and those of the rest of the foot. Proxi-mally, the talus articulates with the tibia and fibula toform the talocrural joint complex, and distally it articu-lates with the calcaneus to form the subtalar joint andthe navicular as a part of the midtarsal complex. Amongprimates, the talus is centrally involved in plantarflexion-dorsiflexion and abduction-adduction of the foot at thetalocrural joint, supination-pronation at the subtalarjoint, and plantarflexion-dorsiflexion at the midtarsaljoint (Levangie and Norkin, 2011). Thus, primate talarmorphology is a reasonably strong indicator of foot func-tion in living taxa, and is useful for interpreting locomo-tor adaptations in fossil hominins (Latimer et al., 1987;Latimer, 1991; Gebo and Schwartz, 2006; DeSilva, 2008,2009; DeSilva and Devlin, 2012; Su et al., 2013). Multiplestudies have included analyses of primate talar shapeand functional morphology using a wide variety of meth-odological approaches (Lisowski et al., 1974; Wood,1974a,b; Langdon, 1986; Latimer et al., 1987; Lewis,1989; Harcourt-Smith, 2002; Berillon, 2004; Gebo andSchwartz, 2006; Jungers et al., 2009; Marivaux et al.,2010, 2011; Turley and Frost, 2013). Regardless ofwhether these studies used a more traditional morpho-metric approach or three-dimensional (3D) geometricmorphometrics, the overall results have been similar:there is substantial talar shape variation among primatetaxa, and this variation is partitioned according to differ-ences in ankle function, substrate use/preference, bodymass, and phylogeny. Here, we specifically evaluate therelationship between morphological variation in the talusand the ecology and locomotor substrate use within thegenus Gorilla.

Three-dimensional talar shape has recently been stud-ied by Turley and Frost (2013), who used geometric mor-phometrics to study variation in a sample of extantcatarrhines. In their study, a set of 30 landmarks (modi-fied from Harcourt-Smith, 2002) was used to quantifythe morphology and size of each talus, and subsets ofthese were analyzed to quantify and compare the proxi-mal (trochlea, lateral and medial malleolar facets) anddistal (talar head and posterior calcaneal facet) articularsurfaces separately. They found that overall talar mor-phology was greatly influenced by size and by substratepreference (Turley and Frost, 2013). Looking separatelyat the proximal and distal articular surfaces, however,they demonstrated that proximal facet shapes covarywith the frequency of arboreal and terrestrial behaviorswhile distal facet shapes reflect phylogeny as well assubstrate preference (Turley and Frost, 2013). Theyshow that components of talar shape associated with ter-restrial substrates have proximal articulations thatreflect stability for dorsiflexion (e.g., smaller, flattermedial and lateral malleolar facets and a higher, flattertrochlea) in addition to distal articulations that exhibitmedial axis stability (e.g., a flatter, laterally placed pos-terior calcaneal facet and a flat talar head). Conversely,shape components coinciding with arboreality demon-strate flexibility in movement while remaining stable onunsteady substrates (e.g., asymmetric trochlear rims, adeep trochlear groove, a concave and posteriorly placedposterior calcaneal facet) (Turley and Frost, 2013).

Another recent study analyzed talar shape using non-geometric morphometric methods (joint angles, surface

areas, and curvatures), focused solely on variationamong gorillas (Dunn et al., 2014). This study employedtwo specific measurements to compare gorilla taxa: themediolateral curvature of the trochlear surface (meas-ured using quadric surface fitting; see Marzke et al.,2010 for methodological details) and the surface area ofthe lateral malleolar facet relative to the entire talararea (see Tocheri et al., 2005 for methodological details).These articular surfaces are of interest based on theexpectation that strong mediolateral curvature of thetrochlea and a larger relative lateral malleolar facetarea (resulting in a higher lateral trochlear rim) resultin a more inverted foot posture, which is biomechani-cally advantageous for movement on arboreal substrates(Latimer et al., 1987). Conversely, a flatter trochlear sur-face and smaller relative lateral malleolar facet arearesult in a more neutral foot position favorable for ter-restrial locomotion and limiting mediolateral excursionof the foot (Latimer et al., 1987; Harcourt-Smith andAiello, 2004; DeSilva, 2008). The results showed thattalar shape falls along a morphocline that tracks func-tion in terms of differing frequencies of arboreal behav-iors, which vary according to ecological habitat (Dunnet al., 2014). Lowland gorillas have tali characterized byrelatively higher lateral trochlear rims and more medio-laterally curved trochleae, similar to other arboreally-adapted primates (Gebo and Schwartz, 2006; Turley andFrost, 2013), whereas highland gorillas have relativelyeven trochlear rims and mediolaterally flatter troch-leae—more broadly similar to modern human tali (Dunnet al., 2014).

Ecological Differences Among Gorillas

Gorillas living in lowland areas, which comprise equa-torial rainforest habitats typically below 900 m abovesea level (ASL) (Mayaux et al., 2004), climb and eat fruitmore often than those living in highland areas (Casimir,1975; Tuttle, 1970; Goodall and Groves, 1977; Tuttle andWatts, 1985; Tutin et al., 1991; Yamagiwa et al., 1992;Tutin and Fernandez, 1993; Remis, 1994; Yamagiwa andMwanza, 1994; Doran, 1996; Goldsmith, 1996; Yamagiwaet al., 1996; Doran and McNeilage, 1998; Remis, 1998;Goldsmith, 1999; Goldsmith, 2003; Robbins and McNei-lage, 2003; Ferriss, 2005). These lowland habitats ofteninclude high-canopied, continuously-distributed foreststhat provide ample opportunities for gorillas to exploitseasonal fruits and build nests atop large supports (Tut-tle, 1970; Tuttle and Watts, 1985; Tutin et al., 1991;Tutin and Fernandez, 1993; Remis, 1994, 1998; Gold-smith, 1996, 1999; Doran and McNeilage, 1998; Ferriss,2005). In contrast, highland gorilla habitats are com-posed of montane rainforest that are typically above1,500 m ASL (Mayaux et al., 2004) and provide fewerarboreal opportunities or incentives for gorillas. Forinstance, trees are smaller and less continuously distrib-uted, and seasonal fruits are scarce (Schaller, 1963;Groves, 1970, 1971; Tuttle, 1970; Fossey and Harcourt,1977; Tuttle and Watts, 1985; Doran, 1996; Doran andMcNeilage, 1998, 2001; Stewart et al., 2001; Robbinsand McNeilage, 2003).

Although direct quantitative comparisons of climbingfrequency among gorilla taxa are lacking, a reasonableproxy derives from examinations of the number of fruitspecies in the diet of different gorilla taxa, which reflects

278 KNIGGE ET AL.

the relative climbing effort needed to incorporate fruitand other arboreal resources into the diet (Remis, 1994,1998). Robbins and McNeilage (2003) compared thenumber of fruit species they observed in Bwindi moun-tain gorillas with those of multiple studies of other goril-las. Their comparisons show a clear relationshipbetween the number of fruit species in gorilla diet andaltitude: Virunga (Karisoke) mountain gorillas (2,700–3,400 m ASL) had 1 fruit species in their diet (Watts,1984; Vedder, 1984); Bwindi mountain gorillas (1,300–2,400 m ASL) had 16 different species (Robbins andMcNeilage, 2003); highland populations of grauer goril-las (1,800–3,300 m ASL) had 20 fruit species whereaslowland grauer gorillas (600–1,300 m ASL) had 48(Yamagiwa et al., 1992, 1994, 1996); western lowlandgorillas (<900 m ASL) exploited the greatest diversity offruit species, having between 77 and 115 in their diets(Williamson et al., 1990; Nishihara, 1995; Tutin, 1996;Remis, 1997a,b).

Study Goals

Even though it has been argued that talar morphologyoften exhibits a large amount of intraspecific variation(see Lovejoy, 1978), the aforementioned studies under-score the important knowledge that can be derived fromconsiderations of talar shape. To further explore thefunctional implications of talar variation, we undertook3D geometric morphometric analyses of landmarks andsemilandmark patches from samples of western lowland(G. g. gorilla), mountain (G. b. beringei), and grauer (G.b. graueri) gorillas (following the taxonomy proposed byGroves, 2001, 2003). Unlike previous work, this narrowtaxonomic study focuses on talar shape amongst closelyrelated populations that differ in their proportional useof locomotor substrates. Dunn et al. (2014) provided asimilarly narrow taxonomic study of talar shape, and webuild on those results here by quantifying the morphol-ogy of the entire talus and of only the proximal articula-tions, using a method that is effective for identifyingsubtle but potentially important differences in morphol-ogy. We further evaluate the hypothesis that gorillas liv-ing in broadly similar ecological conditions (i.e.,highland versus lowland habitats) share common aspectsof talar morphology (particularly in the trochlea and lat-eral malleolar facet) that are plausibly related to foot setand positioning on arboreal versus terrestrial substrates.Alternatively, it is possible that talar shapes may moreclosely approximate the phylogenetic relationships ofgorillas (i.e., eastern versus western gorillas) instead ofreflecting substrate use. Lastly, we address whether par-ticular shape components associated with substrate usein catarrhines generally (Turley and Frost, 2013) aresimilar to the morphological differences documentedamong gorillas.

MATERIALS AND METHODS

The sample for this study included tali from 87 adultor nearly-adult male and female gorillas: 31 western (G.gorilla) and 56 eastern (G. beringei). The eastern gorillasample included 27 mountain (G. b. beringei) and 29grauer gorillas (G. b. graueri). All 31 of the westerngorillas derive from lowland localities, as do 14 of thegrauer gorillas. The remaining 15 grauer and 27 moun-

tain gorillas derive from highland localities. Museumrecords were used to determine collection localities andaltitude (Fig. 1; Table 1). Nearly all specimens (N 5 84)were scanned using a NextEngine 3D Scanner HD laserscanner (macro setting, �16 scans per orientation, mini-mum two orientations per bone). ScanStudio HD PROsoftware was used to align and merge each set of scans,and the resulting surface was subsequently exported asa triangular mesh. The triangular meshes of each bonewere then aligned, merged, and digitally cleaned usingGeomagic Studio software, and then exported as a final3D model. The remaining three specimens were digitizedusing a SIEMENS Somatom Emotion CT scanner (110kV, 70 mA, 1 mm slice thickness, 0.1 mm reconstructionincrement, H50 moderately sharp kernel). Final 3D mod-els of these specimens were generated using MaterialiseMimics software and then digitally cleaned using Geo-magic Studio software. The different methods used forscanning and creating the 3D talus models were notfound to affect the results of this study.

A total of 189 three dimensional landmarks and semi-landmarks were collected for each talus by placingpoints and patches on the 3D models using LandmarkEditor software (Wiley, 2006) (Fig. 2). All left tali werereflected to create a full right-side sample prior to plac-ing the landmarks. The following six articular surfaceswere quantified using semilandmark patches (where kequals the number of semilandmarks per patch): troch-lea (k 5 36), lateral malleolar facet (k 5 30), medial mal-leolar facet (k 5 24), navicular facet (i.e., talar head)(k 5 28), posterior calcaneal facet (k 5 36), and anteriorcalcaneal facet (k 5 30). These semilandmarks were ini-tially constructed as equally spaced points along thetalar articular surfaces and were anchored using a modi-fied version of the landmark set described by Harcourt-Smith (2002) (landmarks 1–30 in Fig. 2). Five additionallandmarks were placed at non-articular locations on thetalus and were used during the analysis (landmarks 31–35 in Fig. 2).

Raw configurations of landmarks and semilandmarkscontain information regarding position, size, and orien-tation. Thus, these configurations were transformed intoshape variables using a generalized Procrustes analysis(Gower, 1975; Rohlf and Slice, 1990) performed in theGeomorph package for R (Adams and Otarola-Castillo,2013). During superimposition, semilandmarks wereallowed to slide along tangents to these curves underthe criterion of minimizing Procrustes distance acrossall specimens (see Gunz and Mitteroecker, 2013) andprojected back on to the surface. The alternative crite-rion, minimizing bending energy (Bookstein, 1997; Gunzet al., 2005), was also utilized to slide semilandmarksbut yielded results congruent with those reported here.

Multivariate ordinations were used to summarizeshape distances in low-dimensional subspaces. Becauseour research questions are aimed at understanding thedifferences between four gorilla populations (western,mountain, highland grauer, and lowland grauer), weused a between-group principal component analysis(BGPCA) which projects the full dataset onto theeigenvectors of the group mean configurations (Boules-teix, 2005; Mitteroecker and Bookstein, 2011). Thisapproach has the benefit of emphasizing group differ-ences rather than variation across all specimens. How-ever, it does not involve matrix inversion nor does it

3D GEOMETRIC MORPHOMETRIC ANALYSIS OF GORILLA TALAR SHAPE 279

warp the shape space as do approaches based oncanonical variates.

An initial BGPCA was performed on the full set of 189landmarks and semilandmarks to assess variation inoverall talar morphology. A separate analysis was alsocarried out using only the 66 semilandmarks of the troch-lear and lateral malleolar surfaces. Data for the latteranalysis were extracted following specimen superimposi-tion of the full set of landmarks and semilandmarks in

order to preserve the articular shapes and their anatomicpositions/orientations within the talus. This approach pro-vides a further test of previously published results, whichhighlighted the shape variation of the trochlea and lateralmalleolar articular surfaces related to ecological con-straints and substrate preferences both within gorillasand among various catarrhine taxa (Dunn et al., 2014;Turley and Frost, 2013). Both BGPCAs were performedusing SAS 9.3 software (SAS Institute, Cary NC).

TABLE 1. Sample sizes by taxon, habitat, sex, and museum collectiona

Western (G. g. gorilla) Grauer (G. b. graueri) Mountain (G. b. beringei)

Lowland Lowland Highland Highland

Collection Male Female Unknown Male Female Unknown Male Female Male Female Unknown

AMNH 3 5 1USNM 5 3 1 2 2RMCA 1 2 1 8 6 2 4ANSP 4 2MCZ 2 2ASU 1RBINS 1 1 11MSGP 5 9 1KNM 1 1Total by sex 16 13 2 3 1 11 8 6 11 15 1Total by taxon 31 15 14 27

aMuseum abbreviations: AMNH: American Museum of Natural History, New York; USNM: United States NationalMuseum; RMCA: Royal Museum for Central Africa; ANSP: Academy of Natural Sciences, Philadelphia; MCZ: Museum ofComparative Zoology; ASU: Arizona State University; RBINS: Royal Belgian Institute of Natural Sciences; MSGP: Moun-tain Gorilla Skeletal Project; KNM: National Museums of Kenya.

Fig. 1. Map of central Africa showing the current distribution ofextant gorilla taxa as described by the IUCN (International Union forConservation of Nature) red list of threatened species. Only westernlowland (G. g. gorilla), mountain (G. b. beringei), and grauer (G. b.graueri) gorilla subspecies were used in this study. All of the mountain

gorillas sampled are from the southern range shown (i.e., Virungalocalities only). The insert provides a closer view of the eastern gorilladistribution (G. beringei) which also specifies the ranges of mountaingorillas and highland versus lowland grauer populations.

280 KNIGGE ET AL.

Lastly, to explore the effects of overall talar size onmorphology, we computed a multivariate regression ofthe shape variables on log centroid size and display thesize/shape relationship by plotting multivariate regres-sion scores against log centroid size for each group.Lacking a direct measure of body mass for these speci-mens, log centroid size of the talus was used as a goodproxy for overall body size (cf., Parr et al., 2011). Themultivariate regression scores were computed andexported using MorphoJ (Klingenberg, 2011).

RESULTS

Overall Talar Morphology

Results of the BGPCA on overall talar morphology areillustrated in Figures 3 and 4. The first between-groupPC axis explains 47.4% of among-group variance androughly separates highland (mountain and highland

grauer) from lowland (western and lowland grauer)gorilla populations rather than distinguishing the gorillaspecies, G. gorilla and G. beringei. The shape compo-nents that coincide with separating highland from low-land gorillas are visualized in Figure 4a. For example,low values for BGPC 1 represent a higher lateral troch-lear rim that extends more dorsally than the medialrim, an anteriorly extended and mediolaterallyrestricted trochlear head, and flatter lateral malleolarand posterior calcaneal facets. Conversely, higher valuesfor BGPC 1 depict a flatter trochlear surface and talarhead, and more concave lateral malleolar and posteriorcalcaneal facets (Fig. 4a).

Taxonomic differences become apparent on BGPC 2(34.3%), where grauer gorillas separate from the othertwo populations, in particular highland grauer gorillasseparate from western gorillas (Fig. 3a). The shape com-ponents driving the variation along this axis depict a

Fig. 2. Landmarks visualized on a western lowland gorilla right talus in (a) dorsal, (b) plantar, (c) poste-rior, and (d) anterior views. Semilandmarks patches were placed on the 6 articular surfaces (outlined inblue) using a set of 30 landmarks (white circles) as anchors, and 5 landmarks (red circles) were selectedfrom nonarticular areas.

3D GEOMETRIC MORPHOMETRIC ANALYSIS OF GORILLA TALAR SHAPE 281

flatter or slightly convex trochlear surface, an anteropos-teriorly shorter posterior calcaneal facet, and mediolater-ally narrower flexor hallucis longus groove as highervalues are attained along this vector (Fig. 4b). The finalcomponent (BGPC 3) accounts for the residual variance(18.3%) and reflects the variation of lowland grauer andmountain gorillas from highland grauer and westerngorillas (Fig. 3b). The morphology described along thisvector includes variation in the curvature of the anteriortrochlear region and medial malleolar facet and theanteroposterior length of the anterior calcaneal facet(Fig. 4c).

Trochlea and Lateral Malleolar Facet

Analysis of the trochlea and lateral malleolar facethighlighted more subtle variations in surface morphol-ogy that are partitioned across the three between-groupPC shape vectors. Figure 5a shows a continuum alongBGPC 1 with mountain gorillas and lowland grauergorillas occupying opposite ends of the distribution. In avisualization of this vector (Fig. 6a), higher values repre-sent an expansion of the posterior, lateral trochlear rimin the dorsal/posterior direction and conversely a reduc-tion of this aspect for lower values. The variation alongBGPC 2 highlights the distinct nearly convex trochleafor low values (coinciding with the range for highlandgrauer gorillas) resulting from a dorsal expansion of thecentral aspect of the trochlear surface (Fig. 6b). Alongthis axis, western, mountain, and lowland grauer goril-las retain slightly positive values while highland grauergorillas are markedly negative. The shape differencesreflected along BGPC 3 are again associated with the

shape of the lateral trochlear rim. Positive values dem-onstrate a more dorsally expanded central aspect of thelateral trochlear rim (Fig. 6c).

Interestingly, the shape changes associated with sepa-rating highland and lowland gorillas in the full datasetare distributed across the first and third axes of thetrochlea and lateral malleolar facet analysis, but in amore subtle manner. For example, the positive ends ofBGPC 1 and 3 of the reduced dataset (occupied by low-land grauer and western gorillas, respectively in Figs.5a,b) are both associated with increased height (i.e., dor-sally extended) of the lateral trochlear rim, which is alsovisualized along BGPC 1 of the full dataset (Figs. 3a,4a). However, this configuration is achieved differentlyas each vector primarily influences a different aspect ofthe trochlear rim. Lowland grauer gorillas have anexpanded posterior trochlear rim (Fig. 6a) while westerngorillas have a higher central trochlear rim (Fig. 6c),and these shape characteristics are evident in examplesof actual lowland grauer and western gorilla tali ratherthan only these components of shape variation (Fig. 7).

Size and Shape

Results comparing talar shape with talar size show asignificant correlation between the regression scores andlog centroid size for all four gorilla groups (Table 2).This is not surprising given the substantial differencebetween males and females in body size. However, thereis a clear difference between highland and lowland popu-lations in the way talar shape scales with size (Fig. 8).As tali from both highland and lowland populationsincrease in size, their size-correlated components of

Fig. 3. Plots of the between-group PCA using the full set of 189landmarks and semilandmarks with 95% concentration ellipses: (a)BGPC 1 vs. 2 and (b) BGPC 1 vs. 3. BGPC 1, 2, and 3 account for47.4, 34.3, and 18.3% of the total among-group variance, respectively.BGPC 1 roughly separates highland from lowland gorillas and thushighlights shape components associated with ecological variation(e.g., height of the lateral trochlear rim, shape of the talar head, andcurvature of the lateral and posterior calcaneal facets). The shape

components that separate highland grauer gorillas from western goril-las at opposite ends of BGPC 2 include the depth of the trochleargroove, the anteroposterior length of the posterior calcaneal facet,and width of the flexor hallucis longus groove. Variation driving BGPC3 highlights the curvature of the anterior trochlear region and medialmalleolar facet. The four gorilla groups are represented as: western-5 green triangles; lowland grauer 5 orange squares; highlandgrauer 5 purple squares; and mountain 5 red diamonds.

282 KNIGGE ET AL.

shape converge on a morphology that is similar amongadult males of all groups.

DISCUSSION

Previous studies of primate tali that included gorillashave primarily sampled western lowland gorillas(Gorilla gorilla gorilla) (Harcourt-Smith, 2002; Berillon,2004; Gebo and Schwartz, 2006; Jungers et al., 2009;Parr et al., 2011). However, an analysis focused solely ongorilla talar morphology demonstrated dramatic varia-tion related to ecological differences among populations(Dunn et al., 2014). Figure 7 provides a comparison ofactual gorilla tali from each group demonstrating thediversity of talar shape among extant gorillas. Other stud-ies of gorilla anatomy have documented morphological

differences plausibly related to variation in locomotorbehavior, such as differences in hand and foot segmentalproportions (Sarmiento, 1994; Jabbour, 2008), hallucalabduction (Tocheri et al., 2011), cross-sectional geometrythrough ontogeny (Ruff et al., 2013), and vertebral for-mula (Williams, 2012).

Our results also suggest that ecology plays a signifi-cant role in shaping gorilla talar morphology, corroborat-ing and extending the results of previous work. Whenconsidering morphological differences among groupmeans, the largest component of variation distinguishesthose populations that live in low-elevation equatorialrainforests from those who inhabit high-elevation mon-tane rainforests. This distinction runs counter to estab-lished genetic relationships that have identifiedmountain and grauer gorillas as sister taxa to the

Fig. 4. Visualization of shape change along (a) BGPC 1, (b) BGPC 2,and (c) BGPC 3 for the full talar landmark set. The models representidealized shapes at the ends of the corresponding BGPC axes. Thefollowing aspects of morphology contain the major shape componentsfor the specified BGPC axes: BGPC 1—lateral trochlear rim and cur-

vature of the talar head and posterior calcaneal facet; BGPC 2—cur-vature of the trochlear groove, width of the flexor hallucis longusgroove, and anteroposterior length of the posterior calcaneal facet;BGPC 3—anterior extension of the anterior calcaneal facet and curva-ture of the medial malleolar facet and anterior trochlear surface.

3D GEOMETRIC MORPHOMETRIC ANALYSIS OF GORILLA TALAR SHAPE 283

exclusion of western taxa (Ruvolo, 1997; Saltonstallet al., 1998; Jensen-Seaman and Kidd, 2001; Jensen-Seaman et al., 2003; Jensen-Seaman et al., 2004; Thal-mann et al., 2005; Anthony et al., 2007a,b; Thalmannet al., 2007; Scally et al., 2012). Rather than reflectingphylogenetic relationships, as one might expect in alandmark-based study (see Bookstein, this volume),major aspects of talar morphology of lowland grauergorillas resemble that of the more distantly related west-ern gorillas, suggesting that these shape data provide afunctional rather than phylogenetic signal.

Western gorillas are known to engage in substantiallymore arboreal activities than mountain gorillas as aresult of residing in more densely forested lowlandregions (Tutin et al., 1991; Remis, 1994; Doran, 1996;Goldsmith, 1996; Remis, 1998; Goldsmith, 1999). Corre-spondingly, field studies have shown that mountaingorillas are primarily terrestrial and have a less variedlocomotor repertoire than western lowland gorillas(Schaller, 1963; Groves, 1970, 1971; Tuttle, 1970; Fosseyand Harcourt, 1977; Tuttle and Watts, 1985; Doran,1996; Doran and McNeilage, 1998, 2001; Stewart et al.,2001; Robbins and McNeilage, 2003). Grauer gorillas areinteresting in this regard because, though more closelyrelated to mountain gorillas, they inhabit both highlandand lowland habitats (Mehlman, 2008). Comparing theshapes delineated along BGPC 1 of the full talar dataset(Fig. 4a), one can see that differences between highlandand lowland gorillas bear out the functional predictionsmade by Dunn et al. (2014).

Understanding the exact functional significance of thedifferences among gorilla taxa warrants further detailedbiomechanical study. However, the most likely possibilityis that differences in ankle morphology impact “foot set”(the orientation of the plantar surface of the foot relative

to the long axis of the tibia) during habitual foot pos-tures used on arboreal versus terrestrial supports. Inparticular, the asymmetry in height (i.e., dorsal exten-sion) of the medial and lateral trochlear rims exhibitedby lowland gorillas results in a cone-shaped articularsurface with a supratalar joint space that is oblique tothe plantarflexion-dorsiflexion axis of the ankle. Thisarrangement is thought to impart concomitant adductionof the foot with plantarflexion and abduction with dorsi-flexion, which in turn results in an abducted knee andlateral travel of the shank over the ankle when the footis planted on a substrate (Latimer et al., 1987). Such aconfiguration probably allows for habitually abductedlower limb postures and inverted foot positioning (sole ofthe foot turned medially toward the substrate) as usedin vertical climbing and above-branch quadrupedalbehaviors (Meldrum, 1991; Isler, 2005).

In contrast to the more asymmetrical medial and lat-eral talar trochlear rims and more curved trochlear sur-face seen in lowland gorillas, the more equal rim heightsand flatter trochlea in highland gorillas are broadly sim-ilar to the condition exhibited by humans (Dunn et al.,2014). This likely facilitates the use of plantigrade footpostures in which the sole of the foot is oriented approxi-mately perpendicular to the long axis of the tibia whenthe sole is placed downward (e.g., Latimer et al., 1987;Harcourt-Smith and Aiello, 2004; DeSilva, 2008). Inhumans, the tibial platform is also slightly cone-shaped,and this results in plantar flexion being coupled withabduction of the foot and dorsiflexion being coupled withadduction (Inman, 1976). However, because the suprata-lar joint space is more approximately parallel to theinferred plantarflexion-dorsiflexion axis of the ankle,concomitant mediolateral deviations of the foot mayoccur to a lesser degree than in highland gorillas.

Fig. 5. Results from the between-group PCA of only the trochleaand lateral malleolar facet with 95% concentration ellipses: (a) BGPC1 vs. 2 and (b) BGPC 1 vs. 3. BGPC 1, 2, and 3 account for 49.8,36.4, and 13.8% of the total among-group variance, respectively.BGPC 1 and 3 are associated with changes in the height of the poste-

rior and central portions of the lateral trochlear rim, respectively.BGPC 2 reflects variation in the curvature of the trochlear groove. Thefour gorilla groups are represented as: western 5 green triangles; low-land grauer 5 orange squares; highland grauer 5 purple squares; andmountain 5 red diamonds.

284 KNIGGE ET AL.

Consequently, the knee should travel over the plantedfoot in a path that more closely approximates a parasag-ittal plane. This anatomical arrangement combines afoot position that should be more stable on relativelyflat, terrestrial substrates with efficient lower leg kine-matics while walking on the ground (Latimer et al.,1987).

Whether or not Latimer et al.’s (1987) model forlower limb kinematics applies to gorillas requires fur-ther experimental validation as well as complementarymorphometric work on the rest of the lower limb. Alter-natively (or as a complement to the kinematic model),the more mediolaterally curved talocrural surface oflowland gorillas (associated with the asymmetricaltrochlear rims) may better accommodate a varied load-ing regime as might be expected on irregular arborealsupports, while the overall flatter joint of highlandgorillas may maximize load transmission efficacy in a

restricted range of joint positions (c.f. Hamrick, 1996).Specifically, DeSilva (2008) suggests that the highlykeeled talar trochlea (of lowland gorillas and most non-human anthropoids according to DeSilva’s data) mayhelp to maintain joint congruence during forceful inver-sion of the foot because the lateral trochlear lip abutsthe fibula closely in such positions. With a less pro-nounced lateral lip, the talus can “tilt” away from thedistal tibia and fibula, decreasing overall joint congru-ence, possibly damaging ligaments and cartilage duringhigh impact loading. As with the current study, DeSilva(2008) found that humans and mountain gorillas havemuch flatter (non-keeled) talocrural articular surfaces.Regardless of the exact mechanism involved, the appa-rent convergence of talocrural joint shape betweenbipedal humans and highland gorillas suggests thatthis shared condition is the result of adaptation to ter-restrial substrates.

Fig. 6. Visualization of shape change along BGPC 1 (a), BGPC 2 (b),and BGPC 3 (c) for the trochlea and lateral malleolar facet only. Simi-lar to Fig. 4, the models represent idealized shapes at the ends of thecorresponding BGPC axes. The following aspects of morphology con-

tain the major shape components for the specified BGPC axes: BGPC1—height of the posterior aspect of the lateral trochlear rim; BGPC2—curvature of the trochlear groove; BGPC 3—height of the centralaspect of the lateral trochlear rim.

3D GEOMETRIC MORPHOMETRIC ANALYSIS OF GORILLA TALAR SHAPE 285

Differences in the scaling relationship of talar sizeand shape among gorillas also support the hypothesisthat certain aspects of the morphology may be associatedwith substrate use. Behavioral data indicate that maleand female mountain gorillas are similar in their degreeof terrestriality (Tuttle and Watts, 1985; Doran, 1996,1997; Remis, 1998). In contrast, western lowland gorillasexhibit sex differences in the proportion of time spent onarboreal substrates (Remis, 1995, 1999). Remis (1995)has suggested that the large body size of male western

gorillas may constrain their abilities to frequently climband engage in other arboreal positional behaviors. Suchsex differences in arboreal behavior may be driven bythe seasonal variation in fruit availability with femalewestern lowland gorillas maintaining a consistent levelof arboreality regardless of fruit distribution, whilemales become more terrestrial and less frugivorouswhen fruit is scarce or only accessible on smaller, termi-nal branches (Remis, 1999). The scaling relationshipbetween log centroid size and talar shape in westernand lowland grauer gorillas (Fig. 8) suggests that larger-bodied individuals (inferred by talar centroid size) attaina talar morphology more closely resembling that of thelarger-bodied terrestrial highland gorillas. This is con-sistent with a model by which morphological variationtracks body-size mediated differences in substrate use inlowland taxa (which largely manifest as male/female dif-ferences due to the pronounced size dimorphism ingorillas).

Although a significant relationship between talar sizeand shape also exists for both mountain and highland

Fig. 7. Examples of actual tali from each gorilla group in posterior, dorsal, lateral, and plantar views.Each talus is a 3D model of an actual specimen and falls near the mean values for each particular groupin the analysis so as to approximate the mean morphology for that group.

TABLE 2. Multivariate regression of shape variableson log centroid size (CS)

Regression scores vs. log(CS)

P valueR2

Western (lowland) 0.49 0.00001Grauer (lowland) 0.77 0.00002Grauer (highland) 0.53 0.00300Mountain (highland) 0.26 0.00610

286 KNIGGE ET AL.

grauer gorillas (Fig. 8), it may be a consequence of sexdifferences or allometric scaling that is unrelated to dif-ferences in substrate use, as behavioral observationssuggest that mountain gorilla males and females aremore similar in substrate use than are lowland gorillas(Remis, 1994; Doran-Sheehy et al., 2009). This relation-ship between body size (inferred by talar size) and sub-strate use could be further tested with ontogeneticsamples. For example, Ruff et al. (2013) found signifi-cant changes in inter-limb strength proportions occurredabruptly in infant mountain gorillas at around 2 yearsof age. Their result corresponds with ontogenetic behav-ioral data that show mountain gorillas become signifi-cantly less arboreal/more terrestrial at this age (Doran,1997; Ruff et al., 2013). Unfortunately, the gorilla talusis not completely ossified until well after this age so itwould be difficult to study in terms of external shape.However, one might predict that trabecular structure ororientation within the gorilla talus may vary accordingto ontogenetic locomotor behavioral patterns, althoughsuch an approach has thus far produced mixed resultsfor adult hominoid tali (DeSilva and Devlin, 2012; Suet al., 2013).

The research presented here focuses specifically ongorilla tali, but certain aspects of the results substanti-ate the conclusions of Turley and Frost (2013) regardingoverall catarrhine talar morphology, and in particularthose related to proximal facet shape and substrate use.

In both our study and that of Turley and Frost (2013),the variations in trochlear and lateral malleolar facetshape reflect the degree to which catarrhine primates(including gorillas) engage in arboreal versus terrestrialbehaviors. Furthermore, Turley and Frost (2013) suggestthat the shapes of the distal facets are greatly influencedby phylogeny and, to a lesser degree, substrate prefer-ence. In our study, we did give specific attention to onlythe shapes of the distal facets, but the results yieldedplots and shape vectors similar to the analysis using thewhole talus (Figs. 3, 4). In this regard, the main aspectsof distal facet shape variation are related to ecologicaldifferences (highland versus lowland) rather than phy-logeny (eastern versus western). The distal facet shapesfound to be related to substrate preference in catar-rhines demonstrate that arboreal forms have a moreconcave posterior calcaneal facet and a rounder, moreconvex talar head relative to terrestrial forms (Turleyand Frost, 2013). In our analysis, the more arboreal low-land gorillas share a similarly rounded and convex talarhead, but conversely have a flatter posterior calcanealfacet (Figs. 4a, 7) in comparison to the more terrestrialhighland gorillas. This may suggest that the gorilla talusis uniquely modified in relation to terrestrial and arbo-real behaviors in at least some ways that deviate fromthe general catarrhine pattern.

One of the benefits of 3D geometric morphometrics isthe ability to detect more subtle variations in shape thatmay be overlooked or difficult to quantify using othermethods; this is particularly evident in the shape of thelateral trochlear rim. Dunn et al. (2014) used the rela-tive area of the lateral malleolar facet to indirectlyapproximate the degree to which the lateral trochlearrim extends dorsally. They found that western gorillashave the largest relative lateral malleolar area; however,among eastern gorillas, relative lateral malleolar areadid not differ significantly (Dunn et al., 2014). In thisanalysis, we found higher lateral trochlear rims in bothwestern and lowland grauer gorillas, but, importantly,each group achieves this configuration in a differentway, expanding different aspects of the lateral trochlearrim. The central portion of the lateral trochlear rim isdorsally expanded in western gorillas (Fig. 6c) while theposterior portion is expanded posteriorly and dorsally inlowland grauer gorillas (Fig. 6a). These specific charac-teristics of the lateral trochlear rim are also evident inactual western and lowland grauer gorilla tali displayedin Figure 7. Although these minor variations are detect-able through analysis of 3D shape, they likely achievesimilar functional results, and may be evidence for inde-pendently evolved adaptations to arboreality in each lin-eage, rather than representing the possible primitivecondition for all living gorillas.

The results from this study have implications forinterpreting the paleobiology and evolution of extincttaxa including fossil hominins. Indeed, although allknown fossil hominin feet show clear primary adapta-tions to bipedality, there is a surprising amount of diver-sity in early hominin talocrural morphology that doesnot necessarily follow a temporal trend or to be consist-ent within lineages (Harcourt-Smith and Aiello, 2004;Gebo and Schwartz, 2006). For example, tali attributedto Australopithecus afarensis (Hadar specimens AL288-1as and AL333-147) have platform-like talocrural jointsurfaces with medial and lateral rims of more equal

Fig. 8. Plot of the multivariate regression scores of the full landmarkdataset against log(centroid size) to illustrate the relationship betweentalar shape variation and talar size across the four gorilla groups (seeTable 2 for regression statistics). All gorilla groups exhibit a significantrelationship between the size and shape of the talus, although thelowland groups appear to scale differently in comparison with thehighland gorillas which may reflect the variation in substrate use.Western 5 green; lowland grauer 5 orange; highland grauer 5 purple;mountain 5 red; males 5 plus signs; females 5 squares; unknown 5 -triangles. Blue boxes indicate specimens belonging to captiveindividuals.

3D GEOMETRIC MORPHOMETRIC ANALYSIS OF GORILLA TALAR SHAPE 287

height similar to the condition in humans (Latimeret al., 1987; DeSilva, 2008; Ward et al., 2012) and nowdocumented for highland gorillas (present study andDunn et al., 2014). In contrast, tali commonly attributedto either a species of early Homo or to one of the robustaustralopiths (e.g., OH8, KNM-ER 813, KNM-ER 1464)exhibit talocrural articular surfaces with markedlyasymmetrical rims and midline grooving (Harcourt-Smith and Aiello, 2004; Gebo and Schwartz, 2006;DeSilva, 2008) that is somewhat similar to the conditiondocumented here in lowland gorillas. Using gorilla diver-sity as a model suggests that variation in early hominintalar morphology may reflect similar adaptive tinkeringto meet the demands of the local habitus and differinglevels of arboreality, likely with correlative functionaleffects that may have led to biomechanically distinctforms of bipedalism in different hominin clades (Har-court-Smith and Aiello, 2004; Gebo and Schwartz, 2006).Future work is necessary to further test thesehypotheses.

CONCLUSIONS

The anatomical variation in talar morphology quanti-fied here using geometric morphometric techniques isconsistent with the hypothesis that the foot of westerngorillas, and to some degree in lowland grauer gorillas,exhibits specific adaptations to arboreal locomotion. Incontrast, mountain and highland grauer gorillas exhibittalar morphology suited to load transmission and stabil-ity of the foot on terrestrial substrates. Although someof these results have previously been demonstratedusing other quantitative metrics, 3D geometric morpho-metrics sheds new light on the subtle ways in whichthese varying anatomical configurations are exhibited indifferent gorilla taxa. Given the relatively recent diver-gence of these taxa, the results reflect the degree towhich talar morphology may be fine-tuned to local envi-ronments even within a clade of closely-related primatesor even intraspecifically in response to body-size medi-ated differences in substrate use. The association of talarsize and shape with substrate use in gorillas provides animportant comparative model for interpreting morpho-logical variation and the paleobiology of fossil taxaincluding early hominins.

ACKNOWLEDGEMENTS

Curatorial assistance and access to collections providedby Richard Thorington and Linda Gordon (USNM),Emmanuel Gilissen and Wim Wendelen (RMCA), NedGilmore (ANSP), Patrick Semal and Georges Lenglet(RBINS), Shannon McFarlin and Tony Mudakikwa(MGSP), Eileen Westwig (AMNH), Judy Chupasko(MCZ), Ogeto Mwebi (KNM), and Diane Hawkey (ASU)is gratefully acknowledged. We thank the Rwandan gov-ernment for permission to study skeletal remains cura-ted by the MGSP, an effort made possible by fundingsupport from the National Science Foundation (BCS-0852866, BCS-0964944), National Geographic Society’sCommittee for Research and Exploration, and the Lea-key Foundation, and infrastructure support from theDian Fossey Gorilla Fund International’s KarisokeResearch Center. We also thank Samantha Porter forassistance with creating the figures.

LITERATURE CITED

Adams DC, Otarola-Castillo E. 2013. Geomorph: an R package forthe collection and analysis of geometric morphometric shape data.Method Ecol Evol 4:393–399.

Anthony NM, Clifford SL, Bawe-Johnson M, Abernethy KA,Bruford MW, Wickings EJ. 2007a. Distinguishing gorilla mito-chondrial sequences from nuclear integrations and PCR recombi-nants: guidelines for their diagnosis in complex sequencedatabases. Mol Phylogenet Evol 43:553–566.

Anthony NM, Johnson-Bawe M, Jeffery K, Clifford SL, AbernethyKA, Tutin CE, Lahm SA, White LJT, Utley JF, Wickings EJ,Bruford MW. 2007b. The role of Pleistocene refugia and rivers inshaping gorilla genetic diversity in central Africa. Proc Natl AcadSci USA 104:20432–20436.

Berillon G. 2004. In what manner did they walk on two legs? Anarchitectural perspective for the functional diagnostics of theearly hominid foot. In: Meldrum DJ, Hilton CE, editors. Frombiped to strider: the emergence of modern human walking, run-ning, and resource transport. New York: Kluwer Academic. p 85–100.

Bookstein FL. 1997. Landmark methods for forms without land-marks: morphometrics of group differences in outline shape. MedImage Anal 1:225–243.

Boulesteix A-L. 2005. A note on between-group PCA. Int J PureAppl Math 19:359–366.

Casimir MJ. 1975. Feeding ecology and nutrition of an easterngorilla group in the Mt. Kahuzi region (Republic of Zaire). FoliaPrimatol 24:81–136.

DeSilva JM. 2008. Vertical climbing adaptations in the anthropoidankle and midfoot: implications for locomotion in Miocene catar-rhines and Plio-Pleistocene hominins. PhD Dissertation, The Uni-versity of Michigan.

DeSilva JM. 2009. Functional morphology of the ankle and the like-lihood of climbing in early hominins. Proc Natl Acad Sci USA106:6567–6572.

DeSilva JM, Devlin MJ. 2012. A comparative study of the trabecu-lar bony architecture of the talus in humans, non-human prima-tes, and Australopithecus. J Hum Evol 63:536–551.

Doran DM. 1996. Comparative positional behavior of the Africanapes. In: McGrew MC, Marchant LF, Nishida T, editors. Greatape societies. Cambridge: University Press. p 213–224.

Doran DM. 1997. Ontogeny of locomotion in mountain gorillas andchimpanzees. J Hum Evol 32:323–344.

Doran DM, McNeilage A. 1998. Gorilla ecology and behavior. EvolAnthropol 6:120–131.

Doran DM, McNeilage A. 2001. Subspecific variation in gorillabehavior: the influence of ecological and social factors. In: RobbinsMM, Sicotte P, Stewart KJ, editors. Mountain gorillas, three deca-des of research at Karisoke. Cambridge: Cambridge UniversityPress. p 123–149.

Doran-Sheehy D, Andrianady M, Lodwick J. 2009. Sex differencesin western gorilla arboreality. Am J Phys Anthropol Suppl 48:120.

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

Ferriss S. 2005. Western gorilla (Gorilla gorilla). In: Caldecott J,Miles L, editors. World atlas of great apes and their conservation.Berkeley: University of California Press. p 104–127.

Ferriss S, Robbins MM, Williamson EA. 2005. Eastern gorilla(Gorilla beringei). In: Caldecott J, Miles L, editors. World atlas ofgreat apes and their conservation. Berkeley: University of Califor-nia Press. p 128–152.

Fossey D, Harcourt AH. 1977. Feeding ecology of the free-rangingmountain gorillas. In: Clutton-Brock TH, editor. Primate ecology.London: Academic Press. p 415–449.

Gebo DL, Schwartz GT. 2006. Foot bones from Omo: implicationsfor hominid evolution. Am J Phys Anthropol 129:499–511.

Goldsmith ML. 1996. Ecological influences on the ranging andgrouping behavior of western lowland gorillas at Bai Hok€ou, Cen-tral African Republic [PhD Dissertation]. Stony Brook, NY: StateUniversity of New York.

288 KNIGGE ET AL.

Goldsmith ML. 1999. Ecological constraints on the foraging effort ofwestern gorillas (Gorilla gorilla gorilla) at Bai Hok€ou, CentralAfrican Republic. Int J Primatol 20:1–24.

Goldsmith ML. 2003. Comparative behavioral ecology of a lowlandand highland gorilla population: where do Bwindi gorillas fit? In:Taylor AB, Goldsmith ML, editors. Gorilla biology: a multidiscipli-nary perspective. Cambridge, UK: Cambridge University Press. p358–384.

Goodall AJ, Groves CP. 1977. The conservation of the eastern goril-las. In: Prince Rainier HRH, III, Bourne GH, editors. Primateconservation. New York: Academic Press. p 599–637.

Gower JC. 1975. Generalized procrustes analysis. Psychometrika40:33–51.

Groves CP. 1970. Population systematics of the gorilla. J Zool 161:287–300.

Groves CP. 1971. Distribution and place of origin of the gorilla.Man 6:44–51.

Groves CP. 2001. Primate taxonomy. Washington, DC: SmithonianInstitution Press.

Groves CP. 2003. A history of gorilla taxonomy. In: Taylor AB, Gold-smith ML, editors. Gorilla biology: a multidisciplinary perspec-tive. Cambridge, United Kingdom: Cambridge University Press. p15–34.

Gunz P, Mitteroecker P. 2013. Semi-landmarks: a method for quan-tifying curves and surfaces. Hystrix 24:103–109.

Gunz P, Mitteroecker P, Bookstein FL. 2005.Semi-landmarks inthree dimensions. In: Slice DE, editor. Modern morphometrics inphysical anthropology. New York: Kluwer Academic. p 73–98.

Hamrick MW. 1996. Articular size and curvature as determinantsof carpal joint mobility and stability in strepsirhine primates.J Morphol 230:113–127.

Harcourt-Smith WEH. 2002. Form and function in the hominoidtarsal skeletal structure, PhD Dissertation, London: UniversityCollege London.

Harcourt-Smith WEH, Aiello LC. 2004. Fossils, feet and the evolu-tion of human bipedal locomotion. J Anat 204:403–416.

Inman VT. 1976. The joints of the ankle. Baltimore: Williams andWilkins.

Isler K. 2005. 3D-kinematics of climbing in hominoids. Am J PhysAnthropol 126:66–81.

Jabbour RS. 2008. Geographic variation in the forelimb and hind-limb skeletons of African apes, PhD Dissertation. New York: TheCity University of New York.

Jensen-Seaman MI, Deinard AS, Kidd KK. 2003. Mitochondrial andnuclear DNA estimates of divergence between western and east-ern gorillas. In: Taylor AB, Goldsmith ML, editors. Gorilla biol-ogy: a multidisciplinary perspective. Cambridge, UK: CambridgeUniversity Press. p 247–268.

Jensen-Seaman MI, Kidd KK. 2001. Mitochondrial DNA variationand biogeography of eastern gorillas. Mol Ecol 10:2241–2247.

Jensen-Seaman MI, Sarmiento EE, Deinard AS, Kidd KK. 2004.Nuclear integrations of mitochondrial DNA in gorillas. Am J Pri-matol 63:139–147.

Jungers WL, Larson SG, Harcourt-Smith W, Morwood MJ, SutiknaT, Rokhus Due Awe, Djubiantono T. 2009. Descriptions of thelower limb skeleton of Homo floresiensis. J Hum Evol 57:538–554.

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

Langdon JH. 1986. Functional morphology of the Miocene hominoidfoot. New York: Karger.

Latimer B. 1991. Locomotor adaptations in Australopithecus afaren-sis: the issue of arboreality. In: Senut B, Coppens Y, editors. Ori-gine(s) de la Bip�edie chez les Hominid�es. Paris: CNRS. p 169–170.

Latimer B, Ohman JC, Lovejoy CO. 1987. Talocrural joint in Afri-can hominoids: implications for Australopithecus afarensis. Am JPhys Anthropol 74:155–175.

Levangie PK, Norkin CC. 2011. Joint structure and function. Phila-delphia: F.A. Davis.

Lewis OJ. 1989. Functional morphology of the evolving hand andfoot. New York: Oxford University Press.

Lisowski FP, Albrecht GH, Oxnard CE. 1974. The form of the talusin some higher primates: a multivariate study. Am J PhysAnthropol 41:191–215.

Lovejoy CO. 1978. A biomechanical review of the locomotor diver-sity of early hominids. In: Jolly CJ, editor. London: Duckworth. p403–429.

Marivaux L, Beard KC, Chaimanee Y, Dagosto M, Gebo DL, Guy F,Marandat B, Khaing K, Kyaw AA, Oo M, Soe AN, Jaeger JJ.2010. Talar morphology, phylogenetic affinities and locomotoradaptation of a large-bodied amphipithecid primate from the latemiddle Eocene of Myanmar. Am J Phys Anthropol 143:208–222.

Marivaux L, Tabuce R, Lebrun R, Adaci M, Mahboubi M, BensalahM. 2011. Talar morphology of azibiids, strepsirhine-related prima-tes from the Eocene of Algeria: phylogenetic affinities and locomo-tor adaptation. J Hum Evol 61:447–457.

Marzke MW, Tocheri MW, Steinburg B, Femiani JD, Linscheid RL,Orr CM, Marzke RF. 2010. Comparative 3D quantitative analysesof trapeziometacarpal joint surface curvatures among living catar-rhines and fossil hominins. Am J Phys Anthropol 141:38–51.

Mayaux P, Bartholome E, Fritz S, Belward A. 2004. A new land-cover map of Africa for the year 2000. J Biogeogr 31:861–877.

Mehlman PT. 2008. Current status of wild gorilla populations andstrategies for their conservation. In: Stoinski TS, Steklis HD,Mehlman PT, editors. Conservation in the 21st century: gorillasas a case study. New York: Springer. p 3–54.

Meldrum DJ. 1991. Kinematics of the cercopithecine foot on arbo-real and terrestrial substrates with implications for the interpre-tation of hominid terrestrial adaptations. Am J Phys Anthropol84:273–289.

Mitteroecker P, Bookstein FL. 2011. Linear discrimination, ordina-tion, and the visualization of selection gradients in modern mor-phometrics. Evol Biol 38:100–114.

Nishihara T. 1995. Feeding ecology of western lowland gorillas inthe Nouabale-Ndoki National Park, Congo. Primates 36:151–168.

Parr CH, Chatterjee HJ, Soligo C. 2011. Inter- and intra-specificscaling of articular surface areas in the hominoid talus. J Anat281:386–401.

Remis MJ. 1994. Feeding ecology and positional behavior of westernlowland gorillas (Gorilla gorilla gorilla) in the Central AfricanRepublic (PhD Dissertation), Yale University.

Remis MJ. 1995. Effects of body size and social context on the arbo-real activities of lowland gorillas in the Central African Republic.Am J Phys Anthropol 97:413–433.

Remis MJ. 1998. The gorilla paradox: effects of habitat and bodysize on the positional behavior of lowland and mountain gorillas.In: Strasser E, Fleagle JGH, Rosenberger A, McHenry HM, edi-tors. Primate locomotion. New York: Plenum Press. p 95–106.

Remis MJ. 1999. Tree structure and sex differences in arborealityamong western lowland gorillas (Gorilla gorilla gorilla) at BaiHokou, Central African Republic. Primates 40:383–396.

Robbins MM, McNeilage A. 2003. Home range and frugivory pat-terns of mountain gorillas in Bwindi Impenetrable National Park,Uganda. Int J Primatol 24:467–491.

Rohlf FJ, Slice D. 1990. Extensions of the procrustes method for theoptimal superimposition of landmarks. Syst Biol 39:40–59.

Ruff CB, Loring Burgess M, Bromage TG, Mudakikwa A, McFarlinSC. 2013. Ontogenetic changes in limb bone structural propor-tions in mountain gorillas (Gorilla beringei beringei). J Hum Evol65:693–703.

Ruvolo M. 1997. Genetic diversity in hominoid primates. Annu RevAnthropol 26:515–540.

Saltonstall K, Amato G, Powell J. 1998. Mitochondrial DNA vari-ability in Grauer’s gorillas of Kahuzi-Biega National Park.J Hered 89:129–135.

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

Scally A, Dutheil JY, Hillier LW, Jordan GE, Goodhead I, HerreroJ, Hobolth A, Lappalainen T, Mailund T, Marques-Bonet T, et al.2012. Insights into hominid evolution from the gorilla genomesequence. Nature 483:169–175.

Schaller GB. 1963. The mountain gorilla: ecology and behavior. Chi-cago: The University of Chicago Press.

3D GEOMETRIC MORPHOMETRIC ANALYSIS OF GORILLA TALAR SHAPE 289

Stewart KJ, Sicotte P, Robbins MM. 2001. Mountain gorillas of theVirungas: a short history. In: Robbins MM, Sicotte P, Stewart KJ,editors. Mountain gorillas: three decades of research at Karisoke.Cambridge: Cambridge University Press. p 2–26.

Su A, Wallace IJ, Nakatsukasa M. 2013. Trabecular bone anisotropyand orientation in an Early Pleistocene hominin talus from EastTurkana, Kenya. J Hum Evol 64:667–677.

Thalmann O, Fischer A, Lankester F, Paabo S, Vigilant L. 2007.The complex evolutionary history of gorillas: insights fromgenomic data. Mol Biol Evol 24:146–158.

Thalmann O, Serre D, Hofreiter M, Lukas D, Eriksson J, VigilantL. 2005. Nuclear insertions help and hinder inference of the evo-lutionary history of gorilla mtDNA. Mol Ecol 14:179–188.

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.

Tutin CEG, Fernandez M. 1993. Composition of the diet of chimpan-zees and comparisons with that of sympatric lowland gorillas inthe Lop�e Reserve, Gabon. Am J Primatol 30:195–211.

Tutin CEG, Fernandez M, Rogers ME, Williamson EA, McGrewMC. 1991. Foraging profiles of sympatric lowland gorillas andchimpanzees in the Lop�e Reserve, Gabon. Philos Trans R SocLond B Biol Sci 334:179–186.

Tuttle RH. 1970. Postural, propulsive, and prehensile capabilities inthe cheiridia of chimpanzees and other great apes. In: BourneGH, editor. The chimpanzee. Vol. II. New York: Karger, Basel. p167–253.

Tuttle RH, Watts DP. 1985. The positional behavior and adaptivecomplexes of Pan gorilla. In: Kondo S, editor. Primate morpho-

physiology, locomotor analyses and human bipedalism. Tokyo:University of Tokyo Press. p 261–288.

Vedder AL. 1984. Movement patterns of a group of free-rangingmountain gorillas (Gorilla gorilla beringei) and their relation tofood availability. Am J Primatol 7:73–88.

Ward CV, Kimbel WH, Harmon EH, Johanson DC. 2012. New post-cranial fossils of Australopithecus afarensis from Hadar, Ethiopia(1990–2007). J Hum Evol 63:1–51.

Watts DP. 1984. Composition and variability of mountain gorilladiets in the central Virungas. Am J Primatol 7:323–356.

Wiley D. 2006. Landmark editor 3.0. Institute for Data Analysisand Visualization, University of California, Davis. Available at:http://graphics.idav.ucdavis.edu/research/EvoMorph.

Williams SA. 2012. Variation in anthropoid vertebral formulae:implications for homology and homoplasy in hominoid evolution.J Exp Zool B Mol Dev Evol 318:134–147.

Williamson EA, Tutin CEG, Rogers ME, Fernandez M. 1990. Com-position of the diet of lowland gorillas at Lope in Gabon. Am JPrimatol 21:265–277.

Wood BA. 1974a. A Homo talus from East Rudolf, Kenya. J Anat117:203–204.

Wood BA. 1974b. Evidence on the locomotor pattern of Homo fromearly Pleistocene of Kenya. Nature 251:135–136.

Yamagiwa J, Maruhashi T, Yumoto T, Mwanza N. 1996. Dietaryand ranging overlap in sympatric gorillas and chimpanzees inKahuzi-Biega National Park, Zaire. In: McGrew MC, MarchantLF, Nishida T, editors. Great ape societies. Cambridge UK: Cam-bridge University Press. p 82–98.

Yamagiwa J, Mwanza N. 1994. Day-journey length and daily diet ofsolitary male gorillas in lowland and highland habitats. Int J Pri-matol 15:207–224.

Yamagiwa J, Mwanza N, Yumoto T, Maruhashi I. 1992. Travel dis-tances and food habits of eastern lowland gorillas: a comparativeanalysis. In: Itoigawa N, Sugiyama Y, Sackett GP, ThompsonRKR, editors. Topics in primatology. Tokyo: Tokyo UniversityPress. p 267–281.

290 KNIGGE ET AL.