Paleodiet of Extinct Platyrrhines With Emphasis on the Caribbean Forms: Three-Dimensional Geometric Morphometrics of Mandibular Second Molars


  • Siobhán B. Cooke

    Corresponding author
    1. Department of Anthropology, New York Consortium on Evolutionary Anthropology (NYCEP), The Graduate Center, The City University of New York, New York, New York
    2. Department of Evolutionary Anthropology, Duke University, Durham, North Carolina
    • Department of Evolutionary Anthropology, Duke University, 130 Science Drive, Room 108, Box 90383, Durham NC 27708
    Search for more papers by this author
    • Tel: 919-660-7386


A three-dimensional geometric morphometric approach was employed to examine shape variation in laser-scan generated models of lower second molars and its relationship to diet in a sample of 9 extant and 16 extinct platyrrhine genera. Principal component analysis of twenty-three x,y,z landmarks describing the occlusal table and sidewalls showed that dental relief was the main contributing factor to variation along the first axis. Discriminant function analysis (DFA) of PC 1 scores and centroid size accurately classified extant platyrrhines according to dietary preference; however, without centroid size, the DFA was less successful. Within this framework, most of the fossil platyrrhines, including specimens from Patagonia, Colombia, Brazil, and the Caribbean, were predicted to have had a frugivorous diet, but several taxa were classified as having a frugivorous/insectivorous diet, the middle Miocene Neosaimiri, Patasola, and Laventiana, all from La Venta. Alouattins, including the La Ventan Stirtonia and the Cuban Paralouatta, showed variable classification as either frugivores or folivore/frugivores. Xenothrix, from Jamaica, was classified either as a frugivore or frugivore/omnivore. Dietary profiles across different extinct platyrrhine communities are compared and discussed in a paleoecological context. Anat Rec, , 2011. © 2011 Wiley Periodicals, Inc.

The extant platyrrhines show a great deal of dietary diversity with most species exploiting a variety of food sources seasonally to meet their nutritional needs (Table 1). This makes the strict classification of diet and the concomitant identification of clear morphological correlates problematic. This is particularly vexing when attempting to classify the dietary profiles of extinct species, but a number of techniques for dietary profiling have been developed and applied to the reconstruction of paleodiets. At their core, all dietary profiling methodologies attempt to identify the morphological elements essential for processing different types of foods, as different food types require different “tools” to break them down. In recent years, a large body of research on the dental morphology of the living platyrrhines in combination with work on the material properties of leaves, fruits, seeds, and insects has greatly advanced our understanding of the functional capacities of primate dentition. Through analogy with these living forms, paleodietary patterns can be inferred for extinct primates.

Table 1. Diet in extant platyrrhines
inline image

Beginning in the 1970s, researchers identified and quantified aspects of molar tooth morphology thought to be adaptations to different diets. Initially, there were several lines of thought on how diet might influence morphology. Kay (1975) suggested that a food type would only exert selective pressure on molar shape if that food were consumed habitually, though he noted that some less frequently consumed, yet important foods might have a “special selective influence.” Rosenberger and Kinzey (1976) introduced the idea of “critical function” whereby the major selective pressures on molar morphology would largely be determined by the mechanical properties of foods that were both essential for adequate nutrition and required specialized adaptations to process. For example, Alouatta consumes significant quantities of easily processed soft ripe fruit in addition to leaves. While soft ripe fruit can be broken down without difficulty, leaves must be finely sheared to gain maximal nutritional value. The primary molar adaptation or “critical function” is to the leaves—not to the easily processed fruit.

The idea of critical function was echoed in subsequent studies. Anapol and Lee (1994) found a significant correlation between the type of protein consumed and the morphology of the dentition, such that two primates that consumed approximately equal quantities of fruit, but relied on very different protein sources, had very different dental morphology. Additionally, fallback foods (foods that are readily available when preferred foods are scarce; see Marshall and Wrangham, 2007) also play a significant role in the evolution of dental morphology (Rosenberger and Kinzey, 1976; Kinzey, 1978; Altman, 1998; Marshall and Wrangham, 2007; Lucas et al., 2009). For many primate species, fallback foods require morphological specializations to process, that is, they may be tough or hard. Fallback foods may make up a relatively small percentage of the annual diet, but during periods of preferred food scarcity, fallback foods are relied upon for a major portion of the diet making them important for survival (Marshall and Wrangham, 2007).

A variety of methodologies have been developed to identify the morphological correlates to these dietary patterns. Perhaps the most influential has been Kay's shearing quotient (SQ) (Kay, 1978, 1984), a measure of the relative length of a tooth's shearing crests. SQ is calculated by first regressing the total length of shearing crests on a molar against a proxy for body size—usually molar length—in a sample of frugivores. The SQ is then calculated from the residuals between the expected length of shearing crests on a tooth if a primate were a frugivore and the actual shear length. Kay found primates that are heavily reliant on leaves or insects have relatively longer shearing crests than do more frugivorous species; frugivores can be further broken down by SQ into ripe fruit specialists and those specializing in hard objects. Hard object specialists are distinguishable from ripe fruit specialists by their extremely low degree of molar shear. However, SQ alone was not successful in distinguishing between the molars of folivorous and insectivorous primates in the absence of information on body size. In addition to work on living forms, SQ has been employed in the reconstruction of paleodiet in many different taxa including platyrrhines (e.g., Meldrum and Kay, 1997), early anthropoids (e.g., Kay and Simons, 1980), cercopithecoids (e.g., Benefit, 2000), and Miocene apes (e.g., Ungar and Kay, 1995).

Attempts to quantify morphological adaptation to diet have now extended beyond SQ to include a number of three-dimensional measures relevant to molar crown function including characterizations of dental topography (e.g., Reed, 1997; Zuccotti et al., 1998; Dennis et al., 2004), measures of relief (e.g., M'Kirera and Ungar, 2003; Boyer, 2008), and analyses of geometric complexity using orientation patch count (OPC) (Evans et al., 2007). OPC has been employed successfully to make fine distinctions in diet; animals that include large amounts of structural carbohydrate tend to have more complex tooth crowns than those that do not. Additionally, Boyer's (2008) relief index (RFI) (a ratio of the three-dimensional area of a molar's enamel cap divided by the area of the tooth projected onto a plane) has had some success in differentiating insectivorous primates from more folivorous primates without using body size information. He found primates that include more structural carbohydrates, or insects in their diets tend to have a high RFI. RFI equates well with the relative ability of a tooth to shear, but also contains information about overall crown height. These methodologies have some advantages over SQ, but may not necessarily provide additional data. All are less reliant on homologous structures for taking measures, such that animals with radically different morphology (e.g., carnivorans and rodents—see Evans et al., 2007) can be compared. This is advantageous when analyzing long extinct groups, which may not have morphologies similar to living forms.

This article further contributes to this line of inquiry through the use of three-dimensional geometric morphometric (3DGM) techniques in the analysis of platyrrhine molars. While 3DGM has been extensively and successfully used in the analysis of cranial, and to a lesser extent, postcranial morphology (e.g., Frost et al., 2003; Wiley et al., 2005; Harvati et al., 2004; Nicholson and Harvati, 2006; Baab, 2008; Tallman, 2010), few studies have utilized landmark-based 3DGM as a tool for exploring primate molar shape and functional anatomy. This is partially because until recently few data collection techniques were available to handle dentition. Most 3DGM data are collected using a microscribe digitizer or similar measurement tool. These types of tools cannot accurately measure very small dentitions, and researchers interested in applying 3DGM techniques to the study of dental morphology mainly employed a stereophotogrametric system to collect x,y,z coordinate landmarks on dentition (e.g., Savara, 1965; Teaford, 1982; Hartman, 1989). The combination of the use of laser scan or CT scan generated models of teeth with data collection programs such as Landmark Editor (Wiley et al., 2005) have, for the first time, made collecting detailed 3DGM data possible on small specimens with relative ease. However, because of the newness of the techniques only a few authors have begun to publish these types of dental analyses.

In a series of papers, Skinner et al. (2008, 2009a, b) collected three-dimensional data on CT scans to examine shape differences in the enamel–dentin junction (EDJ) among different species and subspecies of Pan and between Australopithecus africanus and Paranthropus robustus. The authors were able to distinguish different species and different tooth positions. Their work has particular use in paleontological contexts where dental wear may completely obliterate crown surface morphology.

Additional research on tooth crowns using three-dimensional point data was conduced by Singleton et al. (2011). The authors quantified the metameric shape variation in the mandibular molars of Pan and found that different Pan species share patterns of allometry and metameric variation. They were also able to detect differences between m1 and m2 morphology that allowed them to sort molars by tooth position, which would certainly be valuable in a paleontological context.

To date, no research has examined functional correlates of platyrrhine molar shape in the context of a landmark based three-dimensional analysis. Here, x,y,z coordinate landmarks are used to quantify shape in extinct and extant platyrrhine molars. Three main research questions are addressed:

  • 1Can coordinate-based landmark data be used to accurately classify living forms by diet?
  • 2What does shape variation in platyrrhine mandibular m2s tell us about diet?
  • 3What are the hypothetical diets of the fossil platyrrhines included in this study?


The sample was drawn from 208 extant platyrrhine mandibular tooth rows (Table 2) and 22 extinct platyrrhine tooth rows or isolated m2s (Table 3). Taxa were chosen to represent a variety of dietary profiles and taxonomic groups across the platyrrhine primates, but sampling was not exhaustive. All of the extant primate teeth had negligible wear. Teeth were rejected if cusp tips showed dentin or significant flattening. The extant sample was collected from museum specimens in the American Museum of Natural History, Smithsonian Institution National Museum of Natural History, Museu Nacional, Rio de Janeiro, Brazil, and the Museu de Zoologia de Universidade de São Paulo, Brazil. Each tooth row was cleaned with acetone to remove debris and was molded using President Jet Microsystem ployvinylsiloxane medium body (Coltène/Whaledent). The molds were reinforced with President Putty Soft (Coltène/Whaledent) and then cast using F-82 epoxy with a TP-41 hardener (Eastpoint Fiberglass, Eastpoint Michigan). To prepare for laser scanning, all casts were painted with a thin layer of matte water-based acrylic paint to correct for glare. The sample of extinct primates was composed of casts made from original specimens in the same manner detailed above or from casts obtained from museums or other researchers. The extinct sample includes representatives from the three major geographic areas where platyrrhine fossils have been found: Patagonia, La Venta, Colombia, and the Greater Antilles. Specimens chosen for inclusion in this study exhibit a variety of dental morphologies and phylogenetic affinities.

Table 2. Extant platyrrhine sample
Cebus capucinus25
Saimiri boliviensis27
Aotus sp.29
 A. brumbacki1
 A. lemurinus4
 A. trivirgatus4
 A. vociferans9
 A. azarae1
 A. infulatus6
 A. nigriceps4
Callicebus sp.27
 C. caligatus2
 C. cupreus23
 C. personatus2
Pithecia sp.19
 P. aequatorialis2
 P. irrorata10
 P. pithecia7
Alouatta sp.22
 A. palliata14
 A. seniculus8
Lagothrix sp.23
 L. cana7
 L. lagotricha12
 L. lugens1
 L. poepigii2
 L. sp.1
Ateles geoffroyi19
Brachyteles arachnoides17
Table 3. Extinct platyrrhine sample
Graph symbolSpecies AgeSites, localities, or formationsFamily relationship
  • a


AAntillothrix bernensisMDH 01HoloceneLa Jeringa, Parque del Este, Dominican RepublicCebidae ?
BInsulacebus toussaintianaUF 11417HoloceneTrouing Jérémie no. 5, HaitiPitheciidae
CXenothrix mcgregoriAMNHM 148198 (type)HoloceneLong Mile Cave, JamaicaPitheciidae
D1Paralouatta varonaiMHNH Cueva Alta 1996PleistoceneCueva Alta, CubaAtelidae
D2Paralouatta varonaiMNNH V123PleistoceneCueva del Mono Fósil, CubaAtelidae
ECaipora bambuiorumIGC-UFMG 05 (type)PleistoceneToca da Boa Vista, Bahia, BrazilAtelidae
F1Neosaimiri fieldsiUCMP 39205 (type)MioceneLa Venta, ColombiaCebidae
F2Neosaimiri fieldsiIGM-KU 89002MioceneLa Venta, ColombiaCebidae
F3Neosaimiri fieldsiIGM-KU 89034MioceneLa Venta, ColombiaCebidae
GLaventiana annectensIGM-KU 8801aMioceneLa Venta, ColombiaCebidae
HMohanimico hershkovitziIGM 181500MioceneLa Venta, ColombiaCebidae
IPatasola magdalenaeIGM 184332MioceneLa Venta, ColombiaCebidae
JAotus dindensisIGM-KU 8601 (type)MioceneLa Venta, ColombiaCebidae
KCebupithecia sarmientoiaUCMP 38762MioceneLa Venta, ColombiaPitheciidae
L1Stirtonia tatacoensisIGM-KU 8102MioceneLa Venta, ColombiaAtelidae
L2Stirtonia tatacoensisIGM-KU 8215MioceneLa Venta, ColombiaAtelidae
MDolichocebus gaimanesisMPEF 5146MioceneChubut, ArgentinaCebidae ?
N1Carlocebus carmenensisMACN-SC 43MioceneSanta Cruz, ArgentinaPitheciidae ?
N2Carlocebus carmenensisMACN-SC 250MioceneSanta Cruz, ArgentinaPitheciidae ?
N3Carlocebus carmenensisMACN-SC 63MioceneSanta Cruz, ArgentinaPitheciidae ?
P1Soriacebus ameghinorumMACN-SC 2 (type)MioceneSanta Cruz, ArgentinaPitheciidae ?
P2Soriacebus ameghinorumMACN-SC 379MioceneSanta Cruz, ArgentinaPitheciidae ?

The cast tooth rows and m2s were laser scanned using an LDI Surveyor AM-66RR laser scanner with an RPS 120 sensor at 25 μm interpoint distances. This process created three dimensional point clouds, which were then surface rendered in Geomagic Studio 11 (Geomagic, Inc.) to create virtual three-dimensional dental models. In instances where the entire tooth row was scanned, the m2 was isolated using the editing functions in Geomagic Studio 11 (Geomagic, Inc.). Because these cropped specimens lacked the mesial and distal tooth surfaces in the interstitial region, no landmarks were located in this area.

Landmark Editor (Wiley et al., 2005) was used to place 23 x,y,z coordinate landmarks on the occlusal surface and sidewalls of m2. Landmarks were chosen to outline major dental features on the occlusal surface including cusp apices, basin low points, points of intersection between two crests, and the distal and mesial-most points on the occlusal surface. Landmarks placed on the sidewall of the tooth included points of maximum curvature and the cemento-enamel junction (CEJ; Table 4, Fig. 1). Landmarks were chosen based on their repeatability and functional significance. Error tests performed indicated a low error rate (Supporting Information).

Figure 1.

A: Occlusal view of m2 of Callicebus cupreus (AMNH 98370) with landmarks, (B) buccal view, (C) lingual view. Points are connected with wireframes generated in morphologika2 (O'Higgins and Jones, 2006).

Table 4. Mandibular m2 landmarks
Occlusal surface landmarks
1Metaconid apex
2Protoconid apex
3Hypoconid apex
4Entoconid apex
5Mesial-most point on occlusal surface
6Distal-most point on occlusal surface
7Lowest point on the protocristid - usually at the midline
8Lowest point on the cristid obliquid
9Point at which the preentocristid and postmetacristid meet
10Lowest point in the trigonid basin
11Lowest point in the talonid basin
Sidewall landmarks
12Point of maximum curvature directly below the protoconid
13Point of intersection of the ectoflexid with the buccal wall
14Point of maximum curvature directly below the hypoconid
15The cemento-enamel junction (CEJ) directly below the protoconid
16The CEJ directly below the intersection of the ectoflexid with the buccal wall
17The CEJ directly below the hypoconid
18Point of maximum curvature directly below the entoconid
19Point of maximum curvature directly below where the preentocristid and postmetacristid meet
20Point of maximum curvature directly below the metaconid
21The CEJ directly below the entoconid
22The CEJ directly below the below where the preentocristid and postmetacristid meet
23The CEJ directly below the metaconid

After landmark placement, the x,y,z coordinate points were exported from Landmark editor (Wiley et al., 2005). The points were aligned using generalized Procrustes analysis (GPA) (morphologika2, O'Higgins and Jones, 2006). GPA is a statistical technique used to minimize the least squares distance between sets of landmark points by scaling, rotation, transposition, and translation (Gower, 1975; Rohlf and Slice, 1990). Principal components analysis (PCA) was performed in PAST (Hammer et al., 2001) on the GPA-aligned landmark points to explore overall shape variation in the sample (Figs. 2–4).

Figure 2.

Plot of PC 1 versus PC 2 with 95% equal frequency ellipses. Extinct species are lettered (Table 3). PC 1 represents 21.6% of the total variance, and PC 2 accounts for 14.0% of the total variance. Wireframe deformations (from a distolingual view) show shape changes at the extremes of the axes.

Figure 3.

Plot of PC 2 versus PC 3 with 95% equal frequency ellipses. PC 2 represents 14.0% of the total variance, and PC 3 accounts for 9.5% of the total variance. Extinct species are lettered (Table 3). Wireframes deformations (from a distolingual view) show shape changes at the extremes of the axes.

Figure 4.

Plot of PC 1 versus PC 3. PC 1 represents 21.6% of the total variance, and PC 3 accounts for 9.5% of the total variance. Extinct species are lettered (Table 3). Wireframes deformations (from a distolingual view) show shape changes at the extremes of the axes.

Finally, the living species were assigned dietary categories based on annual dietary information culled from the literature (Table 1). Dietary categories include both the major food source contributing 40% to the annual diet and the major protein source or fallback food if a clear one exists (following Anapol and Lee, 1994). Alouatta and Brachyteles were classified as folivore/frugivores, Saimiri as a frugivore/insectivore, Callicebus, Aotus, Ateles, and Lagothrix as frugivores, Pithecia as a frugivore/seed predator and Cebus as a frugivore/omnivore. Here, the frugivore category is fairly broad and encompasses the highly frugivorous Ateles and Lagothrix as well as the seasonally frugivorous Callicebus and Aotus.

To test the efficacy of landmark data for diet prediction in fossil taxa, discriminant function analysis (DFA) (SPSS, Rel. 11.01) (SPSS Inc., 2001) (Tables 5 and 6) was performed on PC 1, PC 1 and centroid size, and the GPA aligned landmarks. The analysis was conducted entering the independent variables together, and with prior probabilities calculated from the group size. Extant forms were grouped by diet, and fossils were left ungrouped. Cross-validation of the models was performed in SPSS. To do this, one sample was removed from the dataset to determine the new discriminant rule. The removed sample was then classified using the new discriminant rule to determine the rate of correct classification by diet.

Table 5. Discriminant function analyses (DFA) of PC 1 score, (B) PC 1 score and centroid size, and (C) GPA aligned landmarks
  Predicted Group Membership
Group Folivore/ FrugivoreFrugivore/ InsectivoreFrugivoreFrugivore/ Seed predatorFrugivore/ Omnivore
Frugivore/Seed predator001126
 67.6% of original grouped cases correctly classified. 46.9% of cross-validated cases were correctly classified.
Frugivore/Seed predator000172
 96.6% of original grouped cases correctly classified. 87.4% of cross-validated cases were correctly classified.
Frugivore/Seed predator000190
 100.0% of original grouped cases correctly classified. 46.9% of cross-validated cases were correctly classified.
Table 6. Discriminant function analysis (DFA) dietary classification of extinct platyrrhines
  PC 1 ScorePC 1 Score and Centroid SizeGPA aligned Landmarks
Greater Antilles
 Antillothrix bernensisMDH 01FrugivoreFrugivoreFrugivore
 Insulacebus toussaintianaUF 11417FrugivoreFrugivoreFrugivore
 Xenothrix mcgregoriAMNHM 148198 (type sp.)FrugivoreFrugivoreFrugivore/Omnivore
 Paralouatta varonaiMHNH Cueva Alta 1996FrugivoreFrugivoreFolivore/Frugivore
 P. varonaiMNNH V123FrugivoreFrugivoreFolivore/Frugivore
 Caipora bambuiorumIGC-UFMG 05 (type sp.)FrugivoreFrugivoreFrugivore
La Venta, Colombia
 Neosaimiri fieldsiUCMP 39205 (type sp.)Frugivore/InsectivoreFrugivoreFrugivore
 N. fieldsiIGM-KU 89002Frugivore/InsectivoreFrugivore/InsectivoreFrugivore
 N. fieldsiIGM-KU 89034Folivore/FrugivoreFrugivore/InsectivoreFrugivore
 Laventiana annectensIGM-KU 8801aFolivore/FrugivoreFrugivore/InsectivoreFrugivore
 Mohanimico hershkovitziIGM 181500FrugivoreFrugivoreFrugivore
 Patasola magdalenaeIGM 184332Folivore/FrugivoreFrugivore/InsectivoreFrugivore
 Aotus dindensisIGM-KU 8601 (type sp.)FrugivoreFrugivoreFrugivore
 Cebupithecia sarmientoiUCMP 38762FrugivoreFrugivoreFrugivore
 Stirtonia tatcoensisIGM-KU 8102FrugivoreFrugivoreFolivore/Frugivore
 S. tatcoensisIGM-KU 8215FrugivoreFrugivoreFolivore/Frugivore
Patagonia, Argentina
 Dolichocebus gaimanesisMPEF5146FrugivoreFrugivoreFrugivore
 Carlocebus carmenensisMACN-SC 43FrugivoreFrugivoreFrugivore/ Seed predator
 C. carmenensisMACN-SC 250FrugivoreFrugivoreFrugivore
 C. carmenensisMACN-SC 63FrugivoreFrugivoreFrugivore
 Soriacebus ameghinorumMACN-SC 2 (type sp.)Folivore/FrugivoreFrugivoreFrugivore
 S. ameghinorumMACN-SC 379FrugivoreFrugivoreFrugivore


Principal Components Analyses

Figure 2 shows a PCA plot of landmarks 1–23. Variation along PC 1 (21.6% of total variance) is primarily driven by cusp height in combination with basin depth. The species showing the most extreme version of high dental relief (relative height of the cusps above the tooth basin) are the folivorous/frugivorous Alouatta and Brachyteles and the frugivorous/insectivorous Saimiri. The largely frugivorous species Ateles, Lagothrix, Aotus, and Callicebus have less dental relief and lower flatter crowns. The most extreme variations in this latter pattern are found in the frugivorous/seed predacious Pithecia and frugivorous/omnivorous Cebus, which have nearly flat crowns with little delineation between the trigonid and talonid basins, abbreviated crests, and overall low cusp height.

Among the fossils forms, Paralouatta varonai, Patasola magdalenae, Laventiana annectens, and the three Neosaimiri fieldsi individuals fall within the range of frugivorous/insectivorous Saimiri and the folivore/frugivores along PC 1. The remainder of the extinct platyrrhines fall within the large region of overlap along PC 1, but several alignments are worth noting: Caipora bambuiorum, a Pleistocene ateline, is remarkable for its similarity to Ateles, and Aotus dindensis consistently groups with extant Aotus, though the Aotus distribution along PC 1 does primarily fall within a large region of overlap. Cebupithecia sarmientoi, considered a Miocene pitheciin, falls outside of the range of Pithecia, but within the range of Callicebus. The alouattin, Stirtonia tatacoensis, falls just outside the range of modern Alouatta. Carlocebus carmenensis specimens show a wide degree of variability along PC 1, with MACN SC 205 (N2) within the range of Alouatta, Brachyteles, and Saimiri, and MACN SC 43 (N1) falling with the platyrrhines having far less dental relief.

PC 2 (Figs. 2 and 3) shows significant overlap for most taxa, and there is a weak allometric component to the shape. In a regression of PC 2 against centroid size, the r2 value is 0.39. Variation along this axis is largely governed by the relative position of the entoconid, the relative length of the trigonid, and the degree of wasting at the CEJ. While there is substantial overlap for most taxa, PC 2 does differentiate the large-bodied folivore/frugivores Alouatta and Brachyteles from the frugivore/insectivore Saimiri. The fossil alouattin, Stirtonia, demonstrates similar morphology to Alouatta as does Paralouatta along PC 2, though Stirtonia is separated from Alouatta along PC 3. Xenothrix mcgregori also falls near the position of Alouatta at the edge of the range of modern variation, but is distinct when comparing PC 2 and PC 3 (Fig. 3). Again, Aotus dindensis clusters within the Aotus group, and Caipora groups with Ateles. Carlocebus shows a wide range of variation along this axis, also.

Variation along PC 3 (Figs. 3 and 4) is accounted for by the relative length of the m2, the degree of waisting at the CEJ, and height of the point of maximum curvature on the tooth sidewall in combination with the degree of separation of the cusp tips on the crown surface. There is considerable overlap among the living genera with only Cebus falling outside the confidence limits of Aotus and Ateles along this axis. Most of the fossil platyrrhines fall within the large area of overlap, but on the whole tend to have a much lower point of maximum curvature and greater sidewall flare than do the living forms. The Caribbean platyrrhines, Xenothrix and Insulacebus, are extreme in this morphology, a trait that has been noted previously as one of the distinctive features of these taxa (Rosenberger, 1977; Cooke et al., 2011).

Discriminant Function Analyses

DFA successfully grouped living taxa of known dietary profile into dietary groups at a rate better than chance. A DFA using only PC 1 scores classified 67.6% of platyrrhines into the correct dietary category [Table 5(A)]. The majority of misclassifications occurred between folivorous/frugivorous Alouatta and Brachyteles and frugivorous/insectivorous Saimiri, and between the frugivorous/seed predacious Pithecia and frugivorous/omnivorous Cebus. The addition of centroid size yielded much better results with 96.6% correctly classified [Table 5(B)], and made distinguishing between the large bodied folivore/frugivores and the smaller frugivorous/ insectivorous Saimiri possible. A DFA of the GPA aligned complete landmark set yielded 100% correct classification [Table 5(C)]. In cross-validation studies, PC 1 and centroid size fared best with 87.4% of classifications correctly classified. The other two approaches cross-validated poorly with PC 1 and the complete landmark set, both classifying 46.9% of cases correctly.

To predict the hypothetical diet, the fossil forms were left ungrouped in the DFA. Results of predicted dietary groups are shown in Table 6. Most fossil platyrrhines were classified with the frugivores in all analyses, but classification did vary across analyses for several taxa. The alouattins Stirtonia and Paralouatta were classified as frugivores when using PC 1 or PC 1 and centroid size, but were classified as folivore/frugivores when using the landmark set. The position of Neosaimiri was also equivocal. Using only PC 1, Neosaimiri specimens were classified as folivore/frugivores or frugivore/insectivores, but the addition of centroid size resulted in their classification as frugivore/insectivore with the exception of UCMP 39205, which also had the largest centroid size. The landmark data placed Neosaimiri with the frugivores. Both Patasola and Laventiana also were classified as folivore/frugivore (PC 1) or frugivore/insectivore (PC 1 and centroid size), but like Neosaimiri were grouped as frugivores using the complete landmark set. One species, Xenothrix, was categorized as a frugivore/omnivore in the complete landmark analysis, but elsewhere fell with the frugivores. Differences in specimen allocation were also encountered with Carlocebus. One (MACN SC 43) was classified as a frugivore/seed predator using the landmark set and as a frugivore in the other analysis; the other two Carlocebus specimens were always grouped with the frugivores.


Molar Morphology and Diet in the Extant Platyrrhines

Principal component 1 largely differentiated genera based on diet (Fig. 5). The greatest separation existed between clusters containing: (1) the folivore/frugivores, Alouatta and Brachyteles and the frugivore/insectivore, Saimiri, (2) the largely frugivorous primates, Ateles, Lagothrix, Aotus, and Callicebus, and (3) the frugivore/seed predator, Pithecia, and the frugivore/omnivore Cebus. This distribution largely mirrors the abilities of Kay's SQ (Kay 1978, 1984) to distinguish platyrrhines of different dietary guilds. The landmark-based data do provide additional information about the exact shape changes that occur across this distribution, however. This method was least successful at differentiating primates specializing in frugivory, frugivory/seed predation, and frugivory/omnivory (Table 5).

Figure 5.

PC 1 scores of extant taxa. Average PC 1 score for each genus is shown below the form's wireframe. Diet is represented by the color blocks: light blue represents folivore/frugivores and frugivore/insectivores, yellow represents frugivores, and pink represents frugivore/seed predator and frugivore/omnivores. The regions of dietary overlap are shown in green and orange.

The folivorous/frugivorous platyrrhines, Brachyteles and Alouatta, have a distinct morphology marked by high crowns, deep basins, deep intercuspal notches on the buccal and lingual sides of the crown, and a long cristid obliquid. While both have similarities in molar form, genetic (e.g., Schneider et al., 2001; Opazo et al., 2006; Herke et al., 2007; Wildman et al., 2009) and postcranial evidence (e.g., Rosenberger and Strier, 1989), as well as subtle morphological differences in dentition (Rosenberger, 1992; Anthony and Kay, 1993) indicate parallel adaptation to a leafy diet. The diets of both of these species contain a significant quantity of leaves (Table 1), which have specific processing demands. The toughness or resistance to crack propagation of cellulose is considerable and requires long blades to process (Kay, 1975; Rosenberger and Kinzey, 1976; Lucas, 2004). Additionally, some leaves consumed by Alouatta, and by extension other leaf-eating platyrrhines, contain mechanical defenses such as silica (Teaford et al., 2006), which will wear teeth down very quickly, necessitating high crowns to prolong the life of the tooth.

The frugivorous/insectivorous Saimiri boliviensis falls within the range of the folivorous/frugivorous primates along PC 1. Mischaracterizations in the DFA included the classification of Saimiri within the folivore/frugivore group and vice versa. These are unsurprising results that have been supported in other functional studies. The processing of insects requires biomechanical adaptations similar to those necessary for leaf eating, and specialists in both soft-bodied and hard-bodied insects have been found to have significant shearing crest development. Hard-bodied insects tend to be tough and brittle, while soft-bodied insects are subject to ductile deformations and must be sliced through completely in order to break them down (Strait, 1993; Lucas, 2004). Consequently, primates that derive a significant portion of their diet from insects have been difficult to distinguish from leaf-eating primates using functional measures of shearing crest length. Primarily, researchers have relied on body size as a way of separating the folivorous from the insectivorous, as the energy requirements of small-bodied primates (<500 g) preclude reliance on leaves as a major food source (Kay, 1975). While Saimiri was statistically indistinguishable from Alouatta along PC 1 (ANOVA, P = 0.99), it could be differentiated from the folivore/frugivores along PC 2, a principal component that does have a weak correlation with size. Additionally, the DFA of PC 1 scores and centroid size had much more success in classifying these groups than did PC 1 scores alone. Morphologically, the landmarks along PC 2 that separate Saimiri from Alouatta and Brachyteles include those that outline its relatively shorter trigonid, and more distally placed hypoconid. This likely maximizes the length of the cristid obliquid as each of its endpoints are moved further away from each other.

In contrast, the molars of primates that do not rely on leaves or insects for an important portion of their diets are marked by relatively larger basin areas, lower crown relief, and an overall squarer molar outline, which largely correlates with reduced shearing ability, lower resistance to wear, and increased surface area for crushing. This pattern exists across platyrrhines ranging in body size from Aotus (800–900 g) to the very large Lagothrix and Ateles (6,000–10,000 g) (average weights compiled by Ford and Davis, 1992). Of this group, the least frugivorous are Callicebus and Aotus, which exploit a greater variety of food types than do the most committed soft-fruit frugivores, Ateles and Lagothrix (Table 1). Most of this dietary variation appears to be seasonal (Wright, 1989), and when availability allows it, fruit can be consumed in proportions similar to those observed in Ateles and Lagothrix (Kinzey, 1978; Terborgh, 1983; Wright, 1989; Herrera and Heymann, 2004). Along PC 1, Aotus and Callicebus are statistically indistinguishable from each other (ANOVA, P = 0.99); both genera maintain a greater degree of dental relief than the committed soft-fruit frugivores—a likely adaptation to this more flexible dietary profile. Lagothrix and Ateles also show significant overlap along PC 1 (ANOVA, P = 0.99).

The most extreme in the trend toward low levels of relief, and consequently the most positive PC 1 scores, are evident in the frugivorous/seed predacious Pithecia and the frugivorous/omnivorous Cebus. Along PC 1 they were statistically indistinguishable (ANOVA, P = 0.19) and were misclassified in the DFA of PC 1 most frequently; Cebus was classified as Pithecia 36% of the time and Pithecia was classified as Cebus 32% of the time [Table 5(A), Fig. 5]. Their molars have a wide basin, low cusps, and are square in outline, resulting in enlarged area for crushing food items. While Pithecia and Cebus have different dietary profiles, they both incorporate hard objects into their feeding behavior, though the species of Cebus included in this analysis, C. capucinus is a less committed hard object consumer than Cebus apella (Table 1). Pithecia initially breaches the hard outer covering of seeds with its anterior teeth before crushing them with the cheek teeth. Its molars have a unique enamel structure that makes them very resistant to crack propagation when crushing hard seeds (Martin et al., 2003). Very little of their diet is composed of leaves or insects.

Like Saimiri, Cebus includes invertebrate prey (including insects) as a significant portion of its diet during certain times of the year (Tomblin and Cranford, 1994; Janson and Boinski, 1992), though the types of invertebrate prey exploited by the two cebines differ. At localities surveyed by Janson and Boinski (1992), Cebus tended to consume social insects and snails as their primary invertebrate prey, while Saimiri specialized in grasshoppers and caterpillars. Functionally, the delicate cusps and crests that are usually the hallmark of a committed insectivore are incompatible with the processing of hard fruit and nuts, which Cebus also does. The great diversity of the Cebus diet likely explains its lack of specialization for insectivory despite its inclusion of this resource in its diet.

Dietary Prediction in the Continental Extinct Platyrrhines

The platyrrhine fossil record begins with the 26 Ma Bolivian platyrrhines Branisella boliviana (Hoffstetter, 1969) and Szalatavus attricuspis (Rosenberger et al., 1991a), though this article is largely concerned with the later platyrrhines of the Miocene and does not include these primates in the analysis. The Bolivan fossils provide a brief glimpse into the earliest period of platyrrhine evolution, but these small-bodied monkeys remain phylogenetically enigmatic (Fleagle and Tejedor, 2002), though some researchers have argued for a special relationship with callitrichines (Horovitz, 1999; Takai et al., 2000). Between the Oligocene Bolivian deposits and the rich fossil localities of the Patagonian Colhuehuapian and Santacrucian South American Land Mammal Ages there is only one other primate known: Chilicebus carrascoensis from the early Miocene Abanico Formation of Chile. It is represented by a well-preserved skull (Flynn et al., 1995).

From the Argentine Patagonian Miocene several primates are known including Dolichocebus gaimanensis, Mazzonicebus aimendrae dating to approximately 20 Ma (Kay et al.,1999; Kay 2010) Soriacebus ameghinorum, S. adrianae, Carlocebus carmenensis, C. intermedius, Killikaike blakei, Tremacebus harringtoni, and Homunculus patagonicus. These have been alternately hypothesized to be stem platyrrhines not monophyletically related to the modern radiation (Kay, 1990; Fleagle et al., 1997; Kay et al., 2008; Hodgson et al., 2009; Kay and Fleagle, 2010) or allied with members of the modern radiation (Rosenberger, 1979; Rosenberger, 2002; Tejedor et al., 2006; Rosenbegrer et al., 2009; Rosenberger, 2010). Within the latter phylogenetic scheme, Dolichocebus and Killikaike are considered as cebids, Soriacebus, Carlocebus, and Homunculus fall with the pitheciids, and Tremacebus is considered a relative of owl monkeys. One additional form, Proteopithecia neuquenensis, is 15.7 ± 0.07 Ma old and has been described as a pitheciid (Kay et al., 1998).

The fossils included in this study, Soriacebus, Carlocebus, and Dolichocebus, fell within the large cluster of frugivorous platyrrhines in the PCA and were mostly classified as frugivores in the DFA (Table 6, Fig. 6). They are morphologically similar to each other with elevated trigonids, distinct ectoflexids, cusps of moderate height—not as extreme in dental relief as the committed folivore/frugivores and frugivore/insectivores nor as flat as Pithecia and Cebus. Soriacebus and Carlocebus also shared a lower point of maximum curvature on the tooth sidewall. Dentally, Soriacebus is particularly intriguing; its anterior dentition consists of elongate incisors in combination with large canines, which favorably compare with the modern pitheciins Pithecia, Chiropotes, and Cacajao (Fleagle et al.,1987; Fleagle, 1990). The molars, however, are not at all similar to any modern forms, and both Soriacebus specimens fall within the range of overlap of the frugivores and folivore/frugivores along PC 1. This mosaicism may indicate that the anterior dentition evolved a more modern morphology well before the typically low relief molars seen in modern pitheciins (Rosenberger, 1992).

Figure 6.

PC 1 scores of extinct taxa. The key is as follows: Cebus red cross, Saimiri blue square, Aotus purple circle, Callicebus green diamond, Pithecia blue star, Alouatta pink filled square, Lagothrix yellow filled triangle, Ateles light blue triangle, Brachyteles grey filled diamond. For extinct species for which there is more than one individual an average shape and average PC 1 score was generated. PC 1 score for each genus is shown below the form's wireframe. Diet is represented by the color blocks: light blue represents folivore/frugivores and frugivore/insectivores, yellow represents frugivores, and pink represents frugivore/seed predator and frugivore/omnivores. The regions of dietary overlap are shown in green and orange.

Carlocebus shows a larger degree of variation across its distribution than in other fossil species for which 2–3 specimens were included. This calls into question the inclusion of one specimen (MACN SC 43) within this species. While two of the fossils cluster closely together in all analyses and are classified by DFA as frugivores, MACN SC 43 exhibits lower dental relief and falls with the frugivore/seed predators in one analysis (Table 6). The range of the Carlocebus distribution does mirror the extremes of the distributions for Callicebus cupreus, but it is worth considering the possibility that a mixed assemblage of multiple species might be included within the Carlocebus material. As additional specimens are found, a better understanding of the natural distribution of this group will likely develop.

Dolichocebus, here tentatively considered a cebid (but see Kay et al., 2008 and Kay and Fleagle, 2010 for an alternate view), falls within the region of overlap between the frugivores and Saimiri and the folivore/frugivores. In the DFA, it was classified as a frugivore, indicating that this form, if closely related to Saimiri or the cebines, as Rosenberger holds (1979, 2002, 2010), is more primitive and not as committed to partial insectivory. Compared with Saimiri, Dolichocebus has larger basins, an overall lower dental profile, and like the other Patagonian primates more closely approximated cusps in the trigonid, such that the width of the tooth anteriorly is largely a result of a buccal flaring. A diet of mixed resources is predicted for this form, which is in keeping with the paleoecology of the Patagonian region (Fleagle et al., 1997).

Patagonia represents the furthest reaches of the platyrrhine radiation at latitudes around 50 degrees south, and the environment would have been subject to significant seasonal fluctuations in temperature and light (Fleagle et al., 1997; Rosenberger et al., 2009). Isotopic and faunal evidence suggest that grasslands had already begun to establish themselves over much of South America with grazers and browsers present in the same localities as the primates (Flynn and Wyss, 1998; Jacobs et al., 1999; MacFadden, 2000). The primates of this region likely inhabited marginal environments of dry or gallery forests (Fleagle et al., 1997; Rosenberger et al., 2009; Perry et al., 2010), and many fossils show significant dental wear (Perry et al., 2010). The fact that this region exhibits much lower primate species diversity than is found in later fossil assemblages from more humid tropical environments equates well with this paleoecological reconstruction. The primates included in this study from this period fall within the range of Callicebus and Aotus—two primates, which while frugivorous, maintain a certain amount of dietary flexibility. In the seasonal marginal environments of Miocene Patagonia, dietary flexibility would have been likely with Soriacebus, Carlocebus, and Dolichocebus exploiting a variety of food types (see Rosenberger et al., 2009). These primates maintain a certain degree of shearing ability but clearly are neither committed insectivores nor folivores.

A large gap in the platyrrhine fossil record occurs between the Patagonian deposits and the site of La Venta in Colombia, dated to 13.5–11.8 Ma (Guerrero, 1997). The fauna includes 12 platyrrhine species, many of which can be definitively associated with modern groups. La Venta represents a faunal community with similar levels of ecological diversity as modern platyrrhine communities (Wheeler, 2010). Seven species representing all three platyrrhine families are included in this analysis: the alouattin Stirtonia tatacoensis, the cebids Neosaimiri fieldsi, Laventiana annectens, Mohanimico hershkovitzi, Patasola magdalenae, Aotus dindensis, and the pitheciid Cebupithecia sarmientoi. Stirtonia, Neosaimiri, and Cebupithecia have been recognized as being phylogenetically linked with Alouatta, Saimiri, and the pitheciins, respectively, but the relationships of the other taxa have been a matter of some debate. Kay (1990) has argued that Mohanimico hershkovitzi and Aotus dindensis are conspecific while Rosenberger et al. (1990) have advocated for their separate generic status and a different classification. They suggested that Mohanimico may be a callitrichine and Aotus dindensis a pitheciid, but most importantly they hold that Aotus dindensis is a representative of the modern genus. Here, the inclusion of A. dindensis within the modern genus is accepted, but Aotus is classified in the Family Cebidae following the genetic evidence (Schneider et al., 2001; Opazo et al., 2006; Herke et al., 2007; Wildman et al., 2009). Laventiana was originally described as a primitive cebid possibly located at the point of divergence between the calltrichines and the cebines (Rosenberger et al., 1991b). There is general agreement on the placement of Laventiana within Cebidae, but some have advocated its inclusion either within Neosaimiri fieldsi (Takai, 1994) or as a separate Neosaimiri species (Kay and Meldrum, 1997).

Within the analyses conducted here, several interesting dietary trends emerge. First, Stirtonia is morphologically similar to modern Alouatta, so much so that an ancestor–descendent relationship has been argued as a possibility for the two forms (Delson and Rosenberger, 1984), but it displays a less extreme version of the alouattin dentition. Along PC 1, Stirtonia falls outside of the range of Alouatta, but does group with it along PC 2 and PC 3. This is largely as a result of the relative position of the cusps on the crown surface, particularly the entoconid, the degree of waisting at the CEJ, and the location of maximum curvature on the tooth sidewall. In the DFA including only PC 1 scores or PC 1 scores and centroid size, Stirtonia was classified with the frugivores, a result that mirrors previous analyses on the shearing crest lengths of this species (Kay et al., 1987; Fleagle et al., 1997). Interestingly, in the DFA of the GPA aligned landmarks, Stirtonia was classified with the folivore/frugivores. These mixed results seem to indicate that Stirtonia had evolved the cusp placement and overall dental shape of the alouattins—as captured by the complete landmark set—but did not possess the extreme degree of dental relief which accounts for variation along PC 1. Arguably, the complete landmark set contains much more phylogenetic information that is not functional than does PC 1, which is perhaps driving this result. On the other hand, as there are interspecific differences in feeding within modern genera, and we should not expect Stirtonia to precisely mirror any modern Alouatta in its feeding behavior.

Several extinct La Venta platyrrhines have been classified as having a diet that is at least partially insectivorous. The small-bodied Patasola magdalenae falls within the range of the frugivore/insectivore and folivore/frugivore cluster along PC 1 and within the range of Alouatta and Brachyteles along PC 2 (Fig. 2). Patasola has been reconstructed as having a body size of around 350–550 g (Cooke et al., 2011) and is phylogenetically linked with callitrichines (Kay and Meldrum, 1997). Metabolic constraints would have made a predominantly leafy diet unlikely at that size, and an at least partially insectivorous diet would best explain the morphology. This indicates that the separation along PC 2 between the modern folivorous/frugivorous and frugivorous/insectivorous platyrrhines is likely not a result of a functional adaptation to processing insects. In the DFA, Patasola is classified as a folivore/frugivore when only PC 1 scores are employed, but with the addition of centroid size it is falls with the frugivore/insectivores. This result is at odds with another analysis utilizing shearing crests, which suggested Patasola had a frugivorous diet (Fleagle et al., 1997). While landmark analyses show Patasola has dental relief within the range of the frugivore/insectivore Saimiri, these analyses do not incorporate callitrichines. Their addition would include morphological variation not analyzed here and might refine the placement of Patasola.

Two other cebines, Neosaimiri and Laventiana, fall in close proximity to each other in the PCA and within the broad cluster of folivore/frugivores and frugivore/insectivores along PC 1 (Fig. 2). They too were classified as frugivore/insectivores in the DFA. Along PC 2, both species fall at the edge or outside of the distribution of modern Saimiri, indicating that while there is evidence that Neosaimiri and Laventiana may have had a similar dietary profile as the modern genus, other morphological differences exist. Along PC 2, Neosaimiri and Laventiana differ from Saimiri in the relative length of the trigonid (shorter in Saimiri), and their more mesially positioned entoconid. In studies of relative shearing crest lengths, Neosaimiri and Laventiana were found to have somewhat shorter crests than modern Saimiri indicating a more mixed diet (Fleagle et al., 1997), a result replicated here.

Aotus dindensis falls within the distribution of modern Aotus in plots of the first three principal components, though it also falls within the range of Lagothrix in several analysis—highlighting some of the morphological similarities of molar form between the two frugivorous species if size is discarded. While there are other extinct platyrrhines which also fall in the range of Aotus along some axes (notably Mohanimico), none is as consistent in its grouping with extant Aotus. Aotus-like morphology has been argued to be present at a very early date in the fossil record based on cranial (Fleagle and Rosenberger, 1983) and cranio-dental (Setoguchi and Rosenberger, 1987; Takai, 2009) fossil material. It seems that fully Aotus-like molar morphology was present by the mid-Miocene. In all DFA, Aotus dindensis is classified as a frugivore, and may have had a similar ecological niche as modern Aotus based on these analyses as well as research on a facial fragment and tali (Setoguchi and Rosenberger, 1987; Gebo et al.,1990). Mohanimico, which exhibits similar molar morphology to Aotus dindensis, falls within this range as well and is consistently classified as a frugivore.

The pitheciine Cebupithecia sarmientoi has dental morphology that compares favorably with the modern pitheciins including molars with low relief and canines showing a triangular cross-sectional outline. It does maintain some primitive characteristics including smooth enamel (Fleagle et al., 1997) and smaller unmolarized premolars. Previous analyses have hypothesized a diet of fruit with the potential for some seed predation (Fleagle et al., 1997). These analyses support that conclusion. Cebupithecia falls within the range of Callicebus along PC 1. Within the DFA, Cebupithecia groups with the frugivores and not Pithecia, though two of the pitheciin genera, Chiropotes, and Cacajao, were not included. It should be noted that a Cebupithecia m1 was used. While pitheciin m1s and m2s are morphologically quite similar some differences exist—particularly greater differentiation between the trigonid and talonid and a somewhat more closely spaced metaconid and protoconid. The possibility that these results may be driven by the subtle differences in m1 and m2 morphology should be considered.

The La Venta platyrrhines included in this analysis exhibit a much larger range of dietary profiles than do the Patagonian forms (Fig. 6), indicating more dietary differentiation and an increase in available ecological niches. Overall, the variation observed in the La Venta sample mirrors the extant sample, but the La Venta primates do not show the extremes of dental relief (very low and very high), though that certainly could be due to sampling error. Paleoecological studies have reconstructed La Venta as a riparian mosaic with significant forest cover (Kay and Madden, 1997a, b) that may have followed meandering rivers (Guerrero, 1997; Kay and Madden, 1997a, b). The region received a substantial amount of rain (1500–2000 mm per year), but much less than the amount received by the closed canopy forests of modern Amazonia (Kay and Madden, 1997a, b). It would have been broadly similar to the western Amazonian lowlands of today and the primates had a similar ecological diversity (Wheeler, 2010).

Very little fossil evidence for the platyrrhines exists between the middle Miocene forms from La Venta and the Pleistocene and Holocene subfossils from Brazil and the Caribbean. Two species, a cebid, Acrecebus fraileyi, and an atelid, Solimoea acrensis, have been described from the Huyaquerian (9–6.8 Ma) of Brazil (Kay and Cozzuol, 2006), but these are not included here.

In the 1990s, two large-bodied Pleistocene platyrrhines were recovered from the Brazilian cave site Toca da Boa Vista in Bahia, Caipora bambuiorum and Protopithecus brasiliensis. Caipora bore a striking resemblance to Ateles, despite being nearly twice its size. The second primate, Protopithecus, resembled Alouatta in cranial, though not dental morphology and was also very large-bodied, leading researchers to suggest that these two primates represented Pleistocene megafauna (Cartelle and Hartwig, 1996; Hartwig and Cartelle, 1996; Cooke et al., 2007). The Protopithecus skull has no preserved lower molars so it was not included in this study. Caipora, however, has a well-preserved dentition and in analyses it falls well within the Ateles distribution along PC 1 and PC 2 and clusters with the frugivores in all DFAs. It has the low dental relief, wide basins, and low crowns that are the hallmark of frugivorous forms. Except for its much larger size, the molars of Caipora fit comfortably within the variation of Ateles. Based on the molar morphology, this platyrrhine exploited a similar feeding niche to the modern frugivorous atelines, indicating that the large body size radiation of atelid primates that exists today also once included large-bodied forms that survived well into the Plesitocene.

Dietary Prediction in the Extinct Caribbean Platyrrhines

The final group of platyrrhines includes the subfossil forms from the Greater Antillean islands Jamaica, Hispaniola, and Cuba. These five species, Xenothrix mcgregori, Antillothrix bernensis, Insulacebus toussaintiana, and Paralouatta varonai and P. marianae (the latter represented by a single talus) have been considered to be a monophyletic group most closely related to modern Callicebus (MacPhee et al., 1995; Horovitz and MacPhee, 1999; MacPhee and Horovitz, 2004), members of different extant clades (Rosenberger, 1977; Rímoli, 1977; Ford, 1990; Rivero and Arredondo, 1991; Rosenberger, 2002; Rosenberger et al., 2011; Cooke et al., 2011), or most recently, Kay et al. (2011) have hypothesized that Antillothrix might be a stem platyrrhine. Overall, these primates are marked by unique dental morphologies that more closely resemble the Miocene Patagonian forms than the modern mainland radiation (Rosenberger et al., 2011; Cooke et al., 2011). Nearly all of the Greater Antillean material is Pleistocene or Holocene in age, but it is likely that these primates entered the Caribbean during the Miocene or perhaps before (Rosenberger, 1978; MacPhee et al., 1995; Iturralde-Vinent and MacPhee, 1999). The earliest evidence of endemic Greater Antillean platyrrhines comes from Cuban Miocene deposits dating to 17.5–18.5 Ma, in which the type specimen for P. marianae was found (MacPhee et al., 2003).

The endemic Cuban primate, Paralouatta varonai, is a hypothesized alouattin (Rivero and Arredondo, 1991; Rosenberger et al., 2009, but see MacPhee et al., 1995 and Horovitz and MacPhee, 1999 for an alternate view), but is dentally distinctive from the other alouattins in some aspects of its morphology. The species falls within the range of Alouatta and Brachyteles along PC 1 and PC 2. Since this analysis only considers lower molar morphology, it is worth noting that the maxillary molars of Paralouatta are morphologically quite distinct from Alouatta and have a prominent cingulum from which the hypocone arises (Fig. 7). Cranially, it has the long, low, small-brained airorhynchous skull of the other alouattins (Rivero and Arredondo, 1991).

Figure 7.

Maxillary (above) and mandibular (below) left second molar of Alouatta seniculus (left) and Paralouatta varonai (right). Scale bar indicates 1 mm. Mesial is to the left.

Dietarily, Paralouatta is classified with the frugivorous primates except in the DFA of the landmark set where it fell with the folivorous/frugivorous forms. While the overall shape outlined by this landmark set resembles the modern folivorous/frugivorous primates, the m2 is far more bunodont than in Alouatta or Brachyteles. Paralouatta likely represents an early branch of the alouattin lineage prior to the evolution of committed folivorous/frugivorous habits. Additional evidence for frugivorous or partially frugivorous alouttins exists in the Brazilian Protopithecus (not included in this analysis), which has the characteristic cranial morphology of the alouattins, but like Paralouatta does not exhibit a lower dentition adapted to a folivorous/frugivorous diet (Cooke et al., 2007). While the possibility of adaptation to a unique island environment and re-evolution of bunodonty theoretically exists, given the other primitive aspects of the Paralouatta dentition, it seems very unlikely.

The two Hispaniolan species, Antillothrix and Insulacebus, bear a close morphological resemblance to each other. Insulacebus is considered a pitheciid (Cooke et al., 2011), and new fossil evidence is causing a re-evaluation of the phylogenetic placement of Antillothrix. Like the Patagonian platyrrhines, these species exhibit distinct flaring ectoflexids, elevated trigonids, and a moderate degree of dental relief. Insulacebus has the bucco-lingually compressed mesial cusps present in several Patagonian forms (Cooke et al., 2011). Insulacebus and Antillothrix differ from the Patagonian primates in size; they are both much larger, estimated to be around 4,000 g (Rosenberger et al., 2011; Cooke et al., 2011). Along PC 1 and PC 2, both Insulacebus and Antillothrix fall broadly within the range of Callicebus and both are classified as frugivores in the DFA.

The Jamaican platyrrhine Xenothrix has been variously linked with Callicebus (Rosenberger, 1977; MacPhee and Horovitz, 2004) and Aotus (Rosenberger, 2002), but demonstrates a unique morphology unknown elsewhere in the platyrrhine radiation. It lacks a third molar, but analyses have shown that it likely lost this tooth independently of the callitrichine primates (Rosenberger, 1977). Additionally, it is fairly large-bodied—around 5,000–6,000 g (MacPhee and Meldrum, 2006; Cooke et al., 2011). It has closely approximated cusps, a high degree of buccal flare, a low point of maximum curvature on the tooth sidewall, and a relatively long trigonid. Its molars are somewhat polycuspate. While additional analyses including other two-molared forms might be warranted, it seems most likely that Xenothrix occupied an ecological niche unique to its island habitat. In the DFA, it is variously reconstructed as a frugivore or frugivore/omnivore—a reflection of its low dental relief—but the possibility that it exploited resources untapped by the comparative sample is certainly worth considering. Additional evidence for this may be found in its unusual postcranial adaptations (MacPhee and Fleagle, 1991; MacPhee and Meldrum, 2006). Isotopic studies of this form might be particularly enlightening as they could potentially provide a different line of evidence for its dietary profile.

Little is known of the paleoecology of the Caribbean prior to the Pleistocene when primates entered the region and became isolated from the mainland radiation of platyrrhines. Today, the Greater Antilles show a remarkable degree of ecological diversity. Hispaniola alone has rainforest, dry lowlands dominated by cacti and scrub, montane cloud forest, and high mountain pine forest. While it is unknown whether this ecological diversity existed in the past, faunal distributions on Hispaniola do indicate some areas were highly endemic (Cooke et al., 2011).

To date, we do not have more than one species of primate from any one site or larger ecological region on any of the Greater Antillean islands. Two species of primate are known from different regions on Cuba and Hispaniola, and the two primate skulls (Rosenberger et al., 2011; Kay et al., 2011) from the southeastern Dominican Republic, have yet to be compared with each other to confirm that they are conspecifics. Thus far, in the Caribbean, we do not have evidence of a “primate community” in the sense that we do in La Venta or even (to a lesser extent) Patagonia. It remains to be seen whether or not the Greater Antillean islands will each harbor a community of primates jointly sharing habitats as with the continental fauna, or if the monkeys will be organized as narrowly adapted individual species living in separate ecological zones.


Landmark-based 3DGM is successful in differentiating extant platyrrhines with different dietary profiles, and can be employed as one of the methodologies used in reconstructing paleodiet in extinct forms. In discriminant function analyses, PC 1 scores combined with centroid size, can differentiate primates with different dietary profiles and can be used to reconstruct hypothetical diets of extinct platyrrhine primates. Without centroid size, less definition between folivores/frugivores and frugivore/insectivores is obtained, as studies of shearing crest length have also concluded. The predominantly frugivorous, but somewhat eclectic feeders, Aotus and Callicebus, bridge the gap between the frugivorous/insectivorous and folivorous/frugivorous forms and the committed frugivores (Ateles and Lagothrix), frugivore/seed predators (Pithecia) and frugivore/omnivores (Cebus).

The extinct forms of Miocene Patagonia cluster close to the distribution of Aotus and Callicebus—two platyrrhines who, while predominantly frugivorous, do exploit other food sources seasonally. This feeding niche would have been compatible with the paleoecological reconstructions from this region. In such marginal environments, maintenance of some degree of shearing ability would be advantageous in order to allow dietary flexibility in medium-sized platyrrhines. However, one form, Proteropithecia neuquenensis, not analyzed here, may have occupied a more specialized dietary niche. In the more humid tropical environment of La Venta, Colombia where there was a wider array of dietary niches available, the fossil platyrrhines show a greater range of dental morphology from frugivore/insectivores to primates who may have been exploiting fruit and hard seeds. The dental morphologies of the primates of La Venta more closely resemble the variation seen in the Amazonian platyrrhine communities of today than do either of the other two regions explored in this analysis. The Greater Antilles harbored a unique fauna, which more closely matched the morphologies seen in the Patagonian forms than the continental forms from La Venta or the modern period. All the Caribbean forms, but Xenothrix, maintained a modern degree of dental relief also falling within the range of Aotus or Callicebus. While paleoecological information on this region is scant, like the primates of Patagonia the Caribbean forms likely inhabited marginal niches where dietary flexibility was essential.


Author thanks Melissa Tallman and Alfred Rosenberger for helpful criticism, which greatly improved this manuscript. A big thanks to great number of researchers, museum curators, and museum staff for allowing access to specimens, including Castor Cartelle (Instituto de Geociencias, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil), Linda Gordon (National Museum of Natural History, Smithsonian Institution), Manuel Iturralde-Vinent (Museo Nacional de Historia Natural, La Habana, Cuba) Richard Hulbert (Florida Museum of Natural History), Ross MacPhee (American Museum of Natural History) Christian Martinez (Museo del Hombre Dominicano, Santo Domingo, Dominican Republic), Renato Rímoli (Museo del Hombre Dominicano, Santo Domingo, Dominican Republic), Marcelo Tejedor (Universidad Nacional de la Patagonia “San Juan Bosco” Esquel, Argentina) , Leandro Salles (Museu Nacional, Rio de Janeiro, Brazil), and last, but certainly not least, the tireless Eileen Westwig, who has helped me navigate the mammalogy collections of the American Museum of Natural History more times than I can count. Finally, thanks are also due to Clare Davidson for her constant support and encouragement.