Craniofacial Adaptations to Tree-Gouging Among Marmosets

Authors

  • Elliott C. Forsythe,

    Corresponding author
    1. Department of Anthropology, Southern Illinois University, Carbondale, Illinois
    • Department of Anthropology, Southern Illinois University, 3525 Faner Hall, Carbondale, IL 62901-4502
    Search for more papers by this author
    • Tel.: (847) 354-1514; Fax: (618) 453-5037

  • Susan M. Ford

    1. Department of Anthropology, Southern Illinois University, Carbondale, Illinois
    Search for more papers by this author

Abstract

Many primates rely on exudates as dietary items, but comparatively few elicit exudates via tree-gouging. Marmosets are the only haplorhines to extensively utilize this behavior during feeding. Several studies have explored craniofacial adaptations to this behavior, but its morphological correlates are a matter of debate. Various studies suggest that gougers exhibit bite-force maximizing adaptations, load resistance adaptations, and/or jaw-gape maximizing characteristics. All of these seemingly incompatible biomechanical adaptations have previously been argued for marmosets. This study utilizes multivariate and univariate analyses to compare gouging and non-gouging callitrichids for 25 biomechanically relevant craniofacial variables to address this form–function debate. We show that marmosets differ from non-gouging callitrichids in few craniofacial characteristics. Specifically, three craniofacial features differentiate them from non-gougers: relatively longer basicrania, narrower palates, and shorter coronoid processes. We suggest that these characteristics are consistent with a mosaic model for gouging adaptations. In particular, we argue that: (a) shortening the coronoid processes facilitates relatively larger maximum jaw-gapes, (b) basicranial elongation facilitates the extended neck/head posture utilized by marmosets during gouging to maximize gapes, and (c) narrowing the palate serves to more effectively dissipate forces through the maxillary canines during substrate anchoring. Previous studies have documented some of these characters as typical of marmosets, but this combination has not been interpreted as core elements of the marmoset adaptive complex. Marmosets exhibit a unique anatomical repertoire that biomechanically adapts them to both increased jaw-gape and the force dissipation regime associated with tree-gouging. Comparisons among marmoset taxa may enlighten the evolutionary history of the features reported here. Anat Rec,, 2011. © 2011 Wiley Periodicals, Inc.

Although mammals rarely utilize exudates (here defined as gums, saps, resins, and latexes) as a food resource, many primate species rely on them extensively (Hershkovitz,1977; Nash,1986; Power,1996,2010; Peres,2000; Porter et al.,2009; Vinyard et al.,2009; Garber and Porter,2010). As an order, Primates have an atypical predisposition to exudate feeding (Power,1996,2010) and seem uniquely able to incorporate exudates into their diets in significant quantities. At least 37 primate species, plus Paleolithic and contemporary humans, have been documented to feed on exudates (Nash,1986; Johns et al.,2000; Peres,2000; Porter et al.,2009; Vinyard et al.,2009; Garber and Porter,2010), while only a handful of other mammals, mostly marsupials, consume exudates in notable quantities (e.g., Lee and Cockburn,1985; Power,2010; Rosenberger,2010). Exudates can be both difficult to acquire in quantity and difficult to digest (Power,2010), and therefore exudate dependent feeders can be expected to have adaptations to cope with these challenges. This study explores possible mechanical adaptations to exudate acquisition in a group of noted primate exudate specialists, the marmosets (Mico, Callithrix, and Cebuella) of the Neotropics.

Because of the low nutritional quality and difficulty of actively harvesting exudates (Power,2010), few exudate-consuming primate species actively elicit exudate production from plants. Most exudate-consuming species only coincidently ingest exudates by exploiting foods that contain them or opportunistically exploit exudates from naturally occurring flows (Nash,1986; Kelly,1993; Power,1996,2010; Smith2010). A more limited set of exudativores, primarily the marmosets, actively elicit exudate production by stimulating plants to generate exudate flows. In both active and opportunistic exudativores (Nash,1986), gut adaptations to exudate processing have been reported in the literature (Nash,1986,1989; Power,2010; Smith,2010). For example, exudativores such as Phaner furcifer, Euoticus elegantus, and marmosets have specialized gut morphologies (e.g., an elongated caecum or proximal hindgut), which may aid in more efficient exudate fermentation (Ferrari and Martins,1992; Power,2010; Smith,2010). In combination with this specialized gut morphology, several studies have suggested that some exudate-consuming primates (e.g., marmosets and Galago moholi) have a specialized set of physiological mechanisms in their hind-gut which facilitate exudate digestion (Caton et al.,1996,2000; Power and Oftedal,1996; Power,2010). While limited, the available studies indicate that exudate consumption, regardless of the process by which the exudates are acquired, can serve as a selective pressure for specialized anatomical and physiological adaptations.

A limited set of exudate specialists exhibit, in addition to the processing adaptations discussed above, a set of characters linked to active exudate acquisition (Nash,1986; Vinyard et al.,2009; Ravosa et al.,2010; Hogg et al., 2011). Some exudate-feeding taxa, such as Cebuella, Callithrix, and Mico, use their anterior dentition to open wounds in trees (a behavior known as tree-gouging), stimulating exudate flows and increasing the abundance of exudates available for consumption. Dental adaptations to this style of exudate harvesting are well documented among marmosets, and include elongated and narrow lower incisors, incisiform mandibular canines that do not project above the incisor tips, tall incisor crowns, loss (or extreme thinning) of enamel on the lingual surface of the lower incisors, and extreme enamel decussation on the labial surface of the incisors and canines (Rosenberger,1978,2010; Nash,1986; Nogami and Natori,1986; Ravosa et al.,2010; Hogg et al., 2011). These dental characters are well expressed in all marmosets studied to date. Craniomandibular adaptations to exudate harvesting amongst marmosets have been explored in several studies (Dumont,1997; Vinyard et al.,2003; Taylor and Vinyard,2004; Vinyard and Ryan,2006; Aguiar and Lacker,2009; Taylor et al.,2009; Vinyard et al.,2009; Mork et al.,2010; Ravosa et al.,2010; Rosenberger,2010), but these studies disagree on how marmosets functionally differ from non-gouging taxa; little consensus has emerged as to how marmosets are structured to deal with the difficulties associated with tree-gouging (Dumont,1997; Williams et al.,2002; Vinyard et al.,2003; Taylor and Vinyard,2004; Vinyard and Ryan,2006; Aguiar and Lacher,2009; Taylor et al.,2009; Vinyard et al.,2009; Mork et al.,2010; Ravosa et al.,2010; Rosenberger,2010).

Dumont (1997) suggested that mammalian gougers, including marmosets, exhibit morphologies consistent with increased force generation (e.g., elongated in-levers) and resisting mechanical loading (wide crania) in comparison to non-gouging species. This finding seems inconsistent with the in vivo data on bite-force in Callithrix jacchus during gouging reported by Vinyard et al. (2001), who show that marmosets use lower bite forces during gouging than during other biting activities. According to Vinyard et al. (2001), the most notable feature of marmoset gouging is their use of increased jaw-gape during gouging in comparison to their other feeding behaviors. Williams et al. (2002) and Vinyard et al. (2003) suggest that the incongruence between Vinyard's experimental work and Dumont's morphological findings is likely due to sampling bias. Dumont (1997) utilized a very broad, phylogenetically disparate sample for her analyses that may have made interpretation difficult. In contrast, Williams et al. (2002) and Vinyard et al. (2003) show that in narrower primate-only analyses, gougers tend to exhibit morphologies consistent with increased maximum jaw-gape, but not increased force production or resistance. Jaw-gape maximizing characters have also been reported among marmosets by Taylor and Vinyard (2004,2008), Vinyard and Ryan (2006), Vinyard et al. (2009), Aguiar and Lacher (2009), and Taylor et al. (2009), suggesting that jaw-gape maximization may have been a particularly important selective pressure during marmoset evolution. More recent work suggests that while jaw-gape maximization was an important component of marmoset biomechanical evolution, marmoset jaws likely evolved in a biomechanical context that favored increased force transmission, resistance and jaw-gape maximization (Mork et al.,2010; Ravosa et al.,2010; Hogg et al., 2011).

This study adds to the growing body of data on marmoset craniofacial biomechanics by further addressing morphological correlates of tree-gouging among marmosets by comparing them with other, non-gouging members of the Callitrichidae (the tamarins, callimicos, and marmosets). This sampling strategy follows Vinyard et al. (2003), who suggest that craniofacial adaptations to a feeding behavior are best understood in terms of changes from morphologies in closely related taxa; morphological comparisons with distantly related gum harvesters are unlikely to clarify the biomechanical evolution of traits within a particular clade. However, unlike Vinyard et al. (2003), we include a more diverse sample of callitrichid species to better understand how marmosets as a group are morphologically adapted to exudate harvesting (following Aguiar and Lacher,2009).

MATERIALS AND METHODS

Twenty-five linear measurements (Table 1; Fig. 1) were taken on the craniofacial skeleton of a sample of callitrichid crania (N = 305; Table 2) by E.F. using Mitutoyo digital calipers accurate to the nearest 0.01 mm. These measures largely follow Dumont (1997) and especially Vinyard et al. (2003), but include some additional measurements unique to this study. All metrics were collected three times and averaged to minimize measurement error. Each linear measurement was standardized using both the geometric mean (GM) for size and a biomechanical standard (BM; standard is total mandible length), techniques chosen following Anapol and Lee (1994), Dumont (1997), and Vinyard et al. (2003). After standardization, the two data sets (GM and BM corrected) were treated separately. All subsequent analyses were conducted only on the standardized data sets, since variation in size is fairly well documented amongst the callitrichids (Hershkovitz,1977; Ford and Davis,1992). Both data sets resulted in very similar findings, so only the GM results are reported here except where explicitly noted.

Figure 1.

Images of C. jacchus depicting the measures used here. SW and TPL not shown. (Abbreviations given in Table 1).

Table 1. Measurement names, definitions, acronyms, and defining landmarks
MeasurementAcronymDefinitionLandmarks
Condylocanine lengthCCLMax. distance between condylion and the mesial margin of the canine at the alveolusCondylioin to indet.
Condylo-M1 lengthCMLMax. distance between condylion and the mesial margin of M1 at the alveolusCondylion to indet.
Total mandibular lengthTMLMax. distance between gonion and symphysionGonion to symphysion
Maximum condyle lengthMCLMax. AP distance across the mandibular condyleNone
Maximum condylar widthMCWMax. ML distance across the mandibular condyleNone
Symphysis lengthSLMax. IS distance between gnathion and symphysionGnathion to symphysion
Symphysis widthSWLabiolingual distance from gnathion to symphysion perpendicular to symphysis lengthGnathion to symphysion (PERP)
Maximum temporal fossa widthMFWMax. internal ML width of temporal fossaNone
Biglenoid breadthBGBMax. ML distance between lateral margins of the glenoid fossaeNone
Interorbital breadthIOBMinimum ML distance between the medial borders of the eye orbitNONE
Biorbital breadthBOBMax. ML distance between lateral margins of eye orbitEctoconchion to ectoconchion
Lower face heightLFHMax. IS distance between Prosthion and RhinionProsthion to rhinion.
Upper face heightUFHMax. IS distance between the most superior point on the nasal aperture to GlabellaBregma to glabella
Table 2. Sample defined by sex and by percent wild caught
GenusSpeciesMaleFemaleUnknownTotal%Wild
Non-Gougers
 Callimicogoeldii881170%
 LeontopithecusTOTAL   23 
 Leontopithecussp.22150%
 Leontopithecuschrysomelas10010%
 Leontopithecuschrysopygus1001100%
 Leontopithecusrosalia664160%
 SaguinusTOTAL   172 
 Saguinussp.95014100%
 Saguinusfuscicollis12972868%
 Saguinusgeoffroyi1724344100%
 Saguinusilligeri22040%
 Saguinuslabiatus651120%
 Saguinusleucopus67013100%
 Saguinusmidas2305100%
 Saguinusmystax7651844%
 Saguinusnigricollis0014140%
 Saguinusoedipus11722040%
Gougers
 CallithrixTOTAL   28 
 Callithrixsp.  11?
 CallithrixJacchus151112748%
 Cebuellapygmaea12171303%
 MicoTOTAL   35 
 Micoargentatus91012080%
 Micohumeralifer105015100%

The sample was assembled from the Field Museum of Natural History, Chicago, IL, and National Museum of Natural History, Washington D.C. (Table 2). Both captive and wild-caught specimens were measured due to the scarcity of available wild-caught specimens. To control for the effect of captivity on craniofacial form, Student's t-tests were used to compare every variable for wild versus captive specimens at the species level; if a significant difference was found for even one variable, the captive specimens were completely excluded from any further analyses for that species. This strict criterion was used to eliminate artifacts that may have been brought on by captive dietary habits.

Mico, Callithrix, and Cebuella are here defined as active gougers, and the remaining taxa, Leontopithecus, Callimico, and Saguinus, are defined as non-gougers, based on an extensive review of the available literature (Kinzey et al.,1975; Hernandez-Camacho and Cooper,1976; Hershkovitz,1977; Terborgh,1983; Nash,1986; Rylands,1989; Garber,1984,1988,1992; Emmons and Feer,1990; Egler,1992; Rosenberger,1992; Rylands1993a, b; Harrison and Tardif,1994; Veracini1997; Martins and Setz,2000; Correa et al.,2000,2002; Porter,2001; Kierulff et al.,2002; Knogge and Heymann,2003; Kostrub,2003; Vilela and de Faria,2004; Digby et al.,2007; Vinyard et al.,2009; Taylor et al.,2009; Smith,2010). Specimens of Callibella (van Roosmalen and van Roosmalen,2003) were not available, and therefore this gouger was not included in our analysis. Unfortunately, it is not possible at present to assign a quantitative value to the amount of gouging engaged in by many of the taxa included in this analysis, so the broad categories of gouger and non-gouger are used, though we acknowledge that some marmosets (Cebuella and Callithrix) likely rely on exudates much more extensively than others (Mico).

A variety of statistical methods were used to analyze the data set. First, the data were analyzed using discriminant function analysis, with the intent of determining how effectively the measured characters can discriminate between the taxa based on their harvesting behavior. Each species was coded as either a gouger or non-gouger, resulting in two populations for discrimination. Both complete and step-wise functions were calculated. The accuracy of the discriminant functions was determined using jack-knife co-verification.

This study also compares gougers and non-gougers using univariate comparisons. Previous studies have relied extensively on pattern recognition through multivariate morphometrics (Dumont,1997; Viguier,2004; Aguiar and Lacher,2009; but see Vinyard et al.,2003), and have often failed to determine any statistical significance for the observed patterns of craniofacial variation. In this study, the pooled gouger data set was compared to the non-gouger group for each character as an initial assessment of the relationship of each trait to harvesting behavior. In addition, each gouging taxon was independently compared to each non-gouging taxon for every character to elucidate more specific relationships between feeding behaviors and the morphological traits: this step resulted in nine univariate pairwise comparisons for each character in addition to the pooled group comparisons. Shapiro–Wilk tests were used to assess normality: as all data fit the normal distribution, we opted to use the parametric Student's t-test for all univariate comparisons. Due to the large number of tests, we applied a sequential Bonferroni correction to protect against type 1 error. A character was considered a “gouging character” only if both the grouped gougers differed from grouped non-gougers and each gouger exhibited a particular state for that character to the exclusion of each non-gouger. All of these analyses were conducted in JMP 8.0 (SAS Institute Inc., Cary, NC).

RESULTS

Wild-Caught Versus Captive Specimens

The comparisons of wild-caught and captive callitrichid primates showed little difference between them. However, a few significant differences were observed, which caused a number of captive specimens to be removed from further analysis. The observed differences are briefly discussed below.

For the GM standardized data, Student's t-tests only found significant differences for two species of tamarins, which led to the removal of 17 captive specimens: 9 captive Saguinus fuscicollis and 8 captive S. oedipus. Wild S. fuscicollis specimens were found to have significantly narrower crania directly posterior to the orbits (MSW: P = 0.023), significantly shorter distances from the condyle to M1 and the canine (CCL: P < 0.000, CML: P = 0.004), and a significantly larger span between the lateral edges of the glenoid fossae (BGB: P = 0.004) compared to captive specimens. The wild S. oedipus specimens were found to have significantly wider symphyses (SW: P = 0.001) than the captive specimens. The removal of these specimens resulted in the reduction of the sample to 288 crania for the GM analysis.

An even greater number of specimens, 21, were removed from the BM standardized data set. Significant differences were found between captive and wild specimens for three species, leading to the removal of four captive Mico argentatus, nine captive Saguinus fuscicollis, and eight captive S. oedipus specimens. Wild Mico argentatus specimens were found to have significantly narrower crania directly posterior to the orbits (MSW: P = 0.006), narrower symphyses (SW: P = 0.01), and shorter distances between the later margins of the orbits (BOB: P = 0.021) compared to captive individuals. Wild S. fuscicollis specimens were found to have significantly larger dimensions from the condyle to M1 and the canine (CML: P = 0.012, CCL: P = 0.002), and wild S. oedipus specimens have significantly larger distances across the zygomatic arches (BZB: P = 0.022), significantly narrower symphyses (SW: P = 0.002), and significantly greater distances between the lateral margins of the glenoid fossae (BGB: P = 0.019). These exclusions resulted in a reduction of the sample size to 284 callitrichid crania for the BM analyses.

Multivariate Analyses

The discriminant functions on size corrected data are extremely effective at categorizing the taxa analyzed into either gouging or non-gouging behavioral categories. The original data set gave a classification accuracy of 99.7%, with only a single specimen being misclassified (a captive adult Callithrix specimen of unknown sex and species affiliation was classified as non-gouging). The cross-validated (jack-knifed) analysis showed slightly decreased accuracy, with a function accuracy of 98.3%. In the latter analysis, five specimens were misclassified (the aforementioned Callithrix specimen, plus among non-gougers, two captive, adult Callimico specimens, one male and the other female, one adult female Saguinus geoffroyi, and one adult male Saguinus midas specimen). The two largest function coefficients are associated with palate width at canine (PWC: 1.019) and coronoid process height (CPH: 0.943), suggesting these traits are crucial in the discrimination between gougers and non-gougers. Bizygomatic breadth has the next largest coefficient (BZB: 0.798). All other variables have function coefficients ≤ 0.436.

The step-wise discriminant functions have an un-validated function accuracy of 99.3%, and a cross-validated (jack-knifed) accuracy as high as the total discriminant function (98.3%). Analysis shows that after 13 steps, the functions do not significantly increase in accuracy. Again, the two largest function coefficients PWC (−0.867) and CPH (−0.796), followed by BZB (−0.628). The misclassified specimens in the un-validated functions are the same aforementioned adult Callithrix and a captive female Callimico. In the cross-validated function, the Callithrix and Callimico specimens are again misclassified, as well as an adult male Leontopithecus specimen, one adult female Saguinus geoffroyi, and one adult male Saguinus midas specimen.

Univariate Analyses

Very few individual characters exhibit differences that are consistently significant both in comparing the grouped gougers versus non-gougers and in all pair-wise contrasts between gouging and non-gouging callitrichids. In fact, for the GM data set there are only three variables that are consistently significantly different in all relevant contrasts: PWC (P < 0.0001 for grouped comparison), CPH (P < 0.0001 for grouped comparison), and anterior skull length (ASL, P < 0.0001 for grouped comparison). Gougers invariably have narrower PWC (see Fig. 2), shorter CPH (see Fig. 3), and longer anterior basicrania (ASL, see Fig. 4) than do non-gougers. Using a BM (mandibular length, in the BM data set), while the same three characters differ between pooled gougers and non-gougers (all P < 0.0001), only PWC and CPH significantly differ between gougers and non-gougers for all paired comparisons of individual genera.

Figure 2.

Boxplots depicting variation in PWC. The group comparison between gougers and non-gougers is significant (P < 0.0001), as are the individual pair-wise comparisons.

Figure 3.

Boxplots depicting variation in CPH. The group comparison between gougers and non-gougers is significant (P < 0.0001), as are the individual pair-wise comparisons.

Figure 4.

Boxplots depicting variation in ASL. The group comparison between gougers and non-gougers is significant (P < 0.0001), as are the individual pair-wise comparisons.

It is notable that BZB, which is important in the discriminant function analysis for differentiating gougers from non-gougers, does differ significantly between gougers and non-gougers for the pooled group comparisons (e.g., P = 0.007 with the GM data), but fails to differentiate specific gouging from non-gouging taxa. There is substantial overlap between many of the gougers and non-gougers, suggesting that while there is a tendency for gougers to differ from non-gougers, the distribution of this character fails to reach the strict criterion we set for qualifying as a “gouging character”: not all gougers are characterized by a particular state to the exclusion of all non-gougers. We suspect that the discriminant function and pooled gouging versus non-gouging results are driven by the very distinctive cranium of Leontopithecus (a non-gouger with a very narrow skull), as well as the highly distinctive cranium of Cebuella (whose broad, short cranium is significantly different from all three non-gougers in BZB). This finding emphasizes the need for caution when interpreting multivariate comparisons of composite groups, such as “gouger” and “non-gouger.”

DISCUSSION

These analyses show that there are relatively few consistent differences between gouging and non-gouging callitrichids; however, there are a few differences that likely have important biomechanical implications. Marmosets consistently differ from their non-gouging sister taxa in three cranial/mandibular characters: shortened coronoid processes, elongated basicrania, and narrower palates. Below we offer some possible interpretations of these character states.

The shorter coronoid process (CPH) in gougers compared to non-gougers reported here confirms findings of Vinyard et al. (2003), Viguier (2004), Vinyard and Ryan (2006), Vinyard et al. (2009), and Aguiar and Lacher (2009). However, these results are in contrast to Dumont (1997), who suggested that gouging mammals have taller ascending rami with higher condyles than non-gougers. It is noteworthy that Dumont (1997) included a galago (Euoticus) in her analysis, and strong evidence indicates that some galagos (e.g., Otolemur) have evolved a set of adaptations for exudate harvesting that differ markedly from those exhibited by marmosets (see Burrows and Smith,2005). The balance of the evidence suggests that, contra Dumont (1997), all marmoset gougers have shortened ascending rami and especially coronoid processes. Ravosa (1990) has persuasively argued that decreasing the length of the coronoid process is a mechanism by which mammals can increase gape, especially if this reduction is associated with an increase in muscle fiber length. Marmosets fit this biomechanical model quite well, with the combination of the shortened coronoid process demonstrated here and increased masticatory muscle fiber lengths in comparison to non-gouging callitrichids (Taylor et al.,2009; see also Taylor and Vinyard,2004,2008). Thus, the shortening of the coronoid process among gougers reported here, in combination with muscle architectural features, is consistent with the conclusion that gouging marmosets are mechanically capable of producing larger gapes than non-gouging callitrichids.

While this interpretation is compelling and is consistent with previous form-function models (see Vinyard et al.,2009), we are unable to rule out the possibility that the shorter coronoid process among gougers may also relate to a decreased utilization of the m. temporalis during feeding behavior, a possibility that is not in contradiction to the proposed selection for wide gape. Dumont (1997) argues that nectar feeding mammals exhibit decreased coronoid processes as a result of their poorly developed temporalis musculature. A variety of experimental studies have shown that decreasing the size of the m. temporalis results in decreased size of the coronoid process (Washburn,1947; Boyd et al.,1967). In fact, muscle architecture data presented by Taylor et al. (2009) suggest that, when compared to tamarins, marmosets have a decreased reliance on the m. temporalis for force generation. Future work on relative masticatory muscle size and activity in these taxa is needed to address this hypothesis.

Alternatively, it is also possible that shortening the coronoid process is a mechanism by which marmosets actually increase force production, at least in some regions of the mandible. The coronoid process of non-gouging callitrichids is both tall and hooked posteriorly; the shortened coronoid process in marmosets also lacks most of the posterior hook, thus moving the insertion of the m.temporalis anteriorly relative to the origin, compared to non-gougers. While we did not measure the location of the m. temporalis insertion relative to the temporomandibular joint in this study, qualitative observations of the coronoid processes location suggests that it is relatively more anterior in marmosets. Spencer and Demes (1993) suggested that anterior migration of the m. temporalis insertion in some hominins results in increased leverage of the m. temporalis, which results in increased force production efficiency at the anterior dentition. Wright (2005) found that this model was supported within the platyrrhines, since species that utilize their anterior dentition in difficult harvesting strategies (Cebus) exhibit some of the greatest anterior migration of the m. temporalis, while other taxa, which emphasize the posterior dentition during mastication (Aotus, Callicebus), have more posteriorly located m. temporalis insertions. This suggests that the shortening of the coronoid processes among gouging callitrichids might effectively increase the ability of gougers to produce a particular level of bite-force at the anterior dentition. More work is needed to test this hypothesis.

The basicranium of gouging callitrichids shows a significant elongation relative to non-gougers (character ASL). This is particularly true for the GM dataset, where every contrast shows a significant difference between gougers and non-gougers. This elongation is apparently caused by an expansion of the sphenoidal region of the basicranium among gougers, as palate length and total skull length do not differ between gougers and non-gougers. Basicranial elongation in marmosets results in anterior displacement of the incisors and mandibular symphysis away from the neck, increasing the amount of space between the mandibular symphysis and cervical anatomy. This space can limit jaw-gape when the mouth is open to near maximum levels due to contact between cervical structures and the inferior aspects of the mandible: Smith (1984) emphasized the importance of avoiding contact between cervical structures and the mandible during jaw-gapes among primates that utilize wide gapes for feeding and display. This explanation seems consistent with in vivo observations of gouging amongst marmosets: when gouging, marmosets utilize near maximum jaw-gapes (Vinyard et al.,2001) and actively expanding the space between their cervical anatomy and the mandible by using an extended neck posture (our observation). The elongated basicranium of marmosets accentuates the extended neck posture via anatomical mechanisms and mitigates possible limitations on jaw-gapes imposed by cervical structures.

Last, both the univariate and multivariate analyses suggest that gouging callitrichids have a significantly narrower palate across the canines (PWC) than non-gougers. In fact, the discriminant functions for both data sets show that the single best variable for discriminating gougers and non-gougers is PWC. Aguiar and Lacher (2009) have also pointed out narrow palate width as a diagnostic feature of marmosets. The narrowed palate of marmosets is largely brought on via repositioning of the upper canines medially compared to non-gougers, and in addition to this repositioning, the upper canines of gouging marmosets are more vertically implanted than the upper canines of non-gougers, which project laterally from the tooth-row; this was noted by, who suggested that Callithrix (then including all gougers) have a more vertically implanted and a less “criss-crossed” canine occlusal pattern than non-gougers. More recently, Rosenberger (2010) has emphasized the importance of the more medial and vertically lower incisors of marmosets in reducing bending and strain during gouging bouts, an observation analogous to our observations of the palate.

The narrow anterior palate at the canines does not have an obvious association with jaw-gape maximization like the aforementioned characters; we hypothesize that this character may relate to the biomechanics of anchoring during a gouging bout. When anchoring to a substrate, forces are transferred between the dentition/face and the substrate (Vinyard et al.,2001; Rosenberger,2010). The more medial positioning and more vertical implantation of the upper canines of gougers may be conducive to more direct and buttressed transfer of the forces through the canines to the maxilla, in contrast to a more indirect transfer of forces through the splayed and more widely spaced canines on non-gougers (see Rosenberger,2010 for discussion of similar traits in the mandible). If this is a correct interpretation, we might expect other maxillary adaptations to force dissipation (e.g., alveolar bone density, larger tooth root surface area; see Hogg et al., 2011), in addition to the narrow palate and more closely placed vertical canines. Further research is needed to clarify the role of palate width, and especially canine positioning, in tree-gouging and other harvesting behaviors.

Forsythe and Ford (2010) have begun exploring variation among marmoset genera in the characters identified in this study as possible correlates of tree-gouging. Preliminary findings suggest that all of these traits (CPH, PWC, and ASL) are best expressed in Cebuella, as is evident in Figs. 2 and 3 here. Cebuella consistently stands out for its extreme expression of particular gouging-related traits, while the larger marmosets are much more similar morphologically. Aguiar and Lacher (2009) have suggested that the extreme expression of cranial traits in Cebuella is approached in the morphology of Callibella. We are currently testing the relationship between body size, exudate reliance and the morphological characters identified in this study as gouging-related traits.

CONCLUSIONS

Our biometric investigation of the craniofacial morphology of callitrichid primates identified several biomechanically important features that are probably adaptations to tree-gouging among marmosets. Gougers have shorter coronoid processes, elongated anterior basicrania, and narrower anterior palates than non-gouging tamarins and callimicos. We suggest that shortening of the coronoid process is related to increasing potential maximum jaw-gape, a finding consistent with previous anatomical research (Vinyard et al.,2003,2009) and in vivo studies of jaw-gape (Vinyard et al.,2001), but caution that alternations in the anatomy and line of action of the temporal muscles may also influence coronoid morphology. Elongation of the marmoset basicranium has not been previously reported, and we propose it augments the use of an extended neck posture by marmosets during gouging, thus facilitating an increase in jaw-gape abilities compared to non-gougers. Finally, narrowing of the palate is proposed to act as a force dissipation mechanism during substrate anchoring by marmosets, in association with more vertically implanted upper canines, a finding analogous to the conclusions of Rosenberger (2010) based on mandibular characters. These findings are broadly consistent with the biomechanical model recently suggested by Ravosa et al. (2010) and Hogg et al. (2011), which predicts that marmosets should exhibit a mosaic of characteristics linked to jaw-gape maximization and force resistance/transmission.

A biomechanical model of callitrichid evolution based upon these findings would suggest that at the very base of the marmoset clade a series of derived facial characters originated that adapted stem marmosets to an exudativorous foraging niche. These characters included shortening of the coronoid processes, increased length of the anterior basicranium, and narrowing of the anterior palate bringing the canines medially compared to the primitive, tamarin-like condition. This character suite seems to relate largely to increasing jaw-gape. When combined with observations from the anterior dentition, symphysis and temporomandibular joint (Mork et al.,2010; Ravosa et al.,2010; Rosenberger,2010; Hogg et al., 2011), a consistent pattern emerges: marmosets exhibit a pattern of characters linked to the divergent biomechanical demands associated with both jaw-gape maximization and force production, transmission, and dissipation. Further research comparing marmoset taxa known to vary in size and exudate dependence, which may reasonably be expected to have experienced the selective pressures guiding morphological evolution of the masticatory apparatus differently, may help clarify the processes resulting in this morphological pattern.

Acknowledgements

Authors would like to thank Alfred Rosenberger for inviting us to contribute to this special issue of Anatomical Record. Also, authors thank Bill Stanley (FMNH) and Linda Gordon (NMNH) for access to the collections under their care, and Christopher Vinyard for providing us with video of gouging in captive common marmosets. EF also thanks Lauren Forsythe for her critical insights and help with data collection. Robert Corruccini provided extensive statistical advice, and three anonymous reviewers offered comments that significantly improved this manuscript.

Ancillary