Tribe Papionini is well-represented in the Plio-Pleistocene African fossil record with extensive fossil samples from eastern and southern Africa (reviewed in Jablonski and Frost, 2010). Papionins begin appearing in the later Miocene of northern Africa with Macaca, the sister taxon to all other extant papionins ( Figure 1 ), a genus that is especially well-represented in the later Plio-Pleistocene records of Europe and Asia.

Genus Parapapio appears in the later Miocene and early Pliocene record of eastern Africa, followed by larger samples in the Plio-Pleistocene of southern Africa. The exact phylogenetic position of this genus remains unclear, but current data support a basal phylogenetic position for Parapapio relative to other extant and fossil genera. Parapapio appears to have regionally diversified in the Pleistocene of southern Africa, with some earlier population potentially giving rise to the extant African papionin lineages (Jablonski, 2002; Jablonski and Frost, 2010; Pugh and Gilbert, 2018; Williams et al., 2007).

Following the presumed divergence of Theropithecus from Papio and Lophocebus around 4 million years ago ( Figure 1 ; Cronin and Meikle, 1979; Perelman et al., 2011), Theropithecus dominates eastern African assemblages between ~3.5 to 1.5 million years ago, with increases in body size eventually leading to one of the largest monkey species ever known (Jablonski, 2002). During this time, southern African Pleistocene assemblages include representatives of Parapapio, Theropithecus, and notably, early Papio (e.g., Freedman, 1957, 1960, 1961, 1965; Heaton, 2006; Jones, 1937; Pugh and Gilbert, 2018; Szalay and Delson, 1979).

Starting in the Early Pleistocene, Papio hamadryas is well documented in southern Africa, and the species has been recognized in the Middle Pleistocene of the eastern African fossil record (Jablonski and Frost, 2010; Frost and Alemseged, 2007; Brasil et al., 2023). While the fossil records for Theropithecus and Papio are relatively rich among the extant papionin genera, the record of Lophocebus is limited to fragmentary remains from a few eastern African sites; Cercocebus is even more poorly represented, found only in the Early Pleistocene of southern Africa, and Mandrillus is entirely absent from the fossil record (Jablonski, 2002; Jablonski and Frost, 2010). Despite the uneven representation of papionin taxa in the fossil record, the relatively rich records of Papio, Theropithecus, and their candidate ancestor Parapapio, with many relatively complete cranial specimens, provide valuable data for assessing the evolution of craniodental anatomy in several papionin lineages.

Significant and pervasive craniofacial covariation characterizes many, but not all, cercopithecid species, generating species-specific patterns of covariation in this family (Monson, 2020). It has been hypothesized that the evolution of craniofacial morphology in more prognathic species like Papio is under greater constraint because of shared underlying genetic and developmental factors (Monson, 2020; Porto et al., 2009). The extent to which dental anatomy covaries with craniofacial anatomy in primates has been less well-studied (but see Polychronis and Halazonetis, 2014) despite the well-documented covariation between many areas of the cranium, including the face, braincase, and basicranium (e.g., Ackermann, 2002, 2005, 2009; Cheverud, 1982; Dayan et al., 2002; Esteve-Altava et al., 2015; Hallgrímsson et al., 2004, 2007; Lieberman, 2011; Neaux, 2017; Pérez-Claros et al., 2015; Polanski and Franciscus, 2006; von-Cramon-Taubadel and Smith, 2012).

Papio and Theropithecus have longer postcanine tooth rows relative to body mass than other cercopithecids, hypothesized to be related to their prognathic faces (Martin, 1993). Facial length is correlated with RUNX2 gene expression in anthropoid primates and linked to craniodental disorders in humans, commonly with the presence of extra teeth (Ritzman et al., 2017). Multivariate studies of tooth length also reveal a significant correlation with mandibular length, supporting that larger teeth are associated with larger jaws (Hikita et al., 2018).

Previous work has posed the hypothesis that the long third molars of papionins (relative to the other molars) are phenotypically associated with longer faces among the papionin clade (Monson et al., 2022b). This hypothesis builds from earlier studies that identified a strong correlation between average postcanine tooth area and facial size in a sample of cercopithecids (Scott, 2011). The aim of the present study is to investigate whether proportions of postcanine tooth length (relative to other postcanine teeth) are correlated with measures of facial length in a comparative phylogenetic framework. We used paired data from individuals to analyze both inter- and intraspecific variation in facial length and postcanine tooth proportions. Quantifying how these phenotypes covary has significant implications for our understanding of craniodental evolution and development; limited covariation suggests that traits may have been evolving independently, while significant covariation suggests shared genetic, developmental, and/or functional networks between traits, often resulting in linked evolutionary histories.

Additionally, this investigation stands to bear on our interpretations of taxonomic diversity in the fossil record. As the majority of primate fossil remains are craniodental, morphological traits of the cranium are commonly used to differentiate fossil species (e.g., Hartwig, 2002; Jablonski and Frost, 2010). In addition to cranial morphology, relative proportions of the third molar, as well as the first molar and fourth premolar, have been cited extensively in taxonomic classification schemes for extant and fossil papionins (e.g., Fleagle and McGraw, 1999; Gilbert, 2013), making an understanding of the allometric relationship between these traits fundamental to refining taxonomic and phylogenetic hypotheses.

Moreover, the last two million years of Homo evolution are characterized by a general reduction in jaw length, dental arcade length, tooth size, and facial length, with variation in these traits used to distinguish among hypothesized Plio-Pleistocene fossil species (e.g., Spoor et al., 2005). Investigating whether these changes are phenotypically linked is important for our understanding of craniodental variation and the selective pressures that shaped human evolution. Extreme facial length has evolved convergently in several cercopithecid genera, making them an ideal clade in which to investigate phenotypic covariation among craniodental anatomy (Collard and O’Higgins, 2001; Leigh, 2007; Monson, 2020; Singleton, 2005).

Our key goal for this work was to test whether facial length, as captured by palatal length and prosthion - glabella length, is phenotypically correlated with postcanine tooth proportions. A thorough assessment of craniofacial and dental anatomy in extant primates, where traits are paired and measured from the same individual, is essential for interpreting patterns of covariation in these traits. To meet this goal, we measured facial length and postcanine tooth lengths for N=314 extant cercopithecid individuals spanning 10 species across 10 genera, and addressed the following questions:

Our sample includes cranial and dental measurements for N=314 extant cercopithecid individuals, held in osteological collections at the American Museum of Natural History (AMNH) in New York, NY, USA, the Human Evolution Research Center (HERC) and the Museum of Vertebrate Zoology (MVZ) in Berkeley, CA, USA, the National Museum of Natural History (NMNH) in Washington, DC, USA, the Center for Evolutionary Origins of Human Behavior at Kyoto University (PRISK) in Inuyama, Japan, and the Zurich Natural History Museum (UZH) in Zurich, Switzerland.

All measurements were taken using analog DialMax calipers. Cranial and dental data were matched, measured from the same individual. Matching data requires that specimens are complete and in good condition, and grants the statistical power needed to assess both inter- and intraspecific correlations between traits. All individuals included in this study had a complete cranium and face, with complete maxillary and mandibular postcanine dentitions.

We only measured adult individuals as assessed by the eruption of the third molars. We attempted to measure sex-balanced samples whenever possible, but many papionins (e.g., Theropithecus, Mandrillus) are extremely rare in museum collections. Our sample includes n=190 papionins, n=57 cercopithecins, and n=67 colobines. See Table 1 for sample size details.

The extant and fossil data used in this study represent years of data collection in many different museum collections around the world. Given the complexity of gathering a dataset of this size, and in favor of maximizing taxonomic breadth and sample sizes, we focused on linear metrics for this study, collecting two cranial and eight dental measurements from all individuals sampled. The two cranial measurements provide different measures of facial length: (1) facial length (measured from prosthion to glabella) and (2) palatal length (measured from orale to staphylion). We did not focus on cranial length as a metric in this study because: 1) cranial length is a proxy for body size, and larger individuals often possess robust muscle attachments on the occipital bone, introducing measurement error (e.g., resulting from variation in external occipital crest morphology), and 2) very few fossil specimens preserve enough anatomy to confidently measure total cranial length; palatal and facial length, however, are preserved more frequently. Given these caveats, we focused on facial and palatal length, and the phenotypic association between these cranial traits and postcanine dental proportions, for this study. However, we did calculate relative facial length for a subsample (as noted by minimum sample sizes in Table 1 ) of extant primates (prosthion - glabella length divided by maximum cranial length) to investigate the effects of body size variation on the relationship between facial length and dental proportions.

We also measured the lengths of the fourth premolar and all three molars (P4-M3) for both the maxilla and the mandible to calculate two dental ratios: MMC (molar module component, measured as M3 length divided by M1 length), and PMM (premolar-molar module, measured as M2 length divided by P4 length; Hlusko et al., 2016). Quantitative genetic analyses demonstrate that these dental ratios are heritable (e.g., h² mandibular MMC=0.69, h² mandibular PMM=0.45) and do not covary with sex or body size in baboons (genus Papio; Hlusko et al., 2016; Monson et al., 2022a). Because molar lengths are guided by iterative developmental processes (e.g., Kavanagh et al., 2007), ratios of relative third molar length are more informative than absolute tooth lengths or widths, which are frequently linked to body size (e.g., Hlusko et al., 2016). The MMC and PMM dental ratios also have strong phylogenetic signal in mammals (e.g., Monson et al., 2019a; Zuercher et al., 2021). Additionally, because these dental ratios carry taxonomic signal, and teeth are relatively abundant in the hominid fossil record, MMC and PMM have already been employed in paleoanthropological investigations (e.g., Brasil et al., 2020; Monson et al., 2018, 2022a).

As part of our analysis, we also collected data on molar, fourth premolar, and facial/palatal lengths for n=50 fossil papionins from genera Parapapio, Procercocebus, Papio, and Theropithecus. Data for Victoriapithecus, a putative early ancestor of the cercopithecid primate family, were taken from the literature (Benefit, 1993; Benefit and McCrossin, 1997) and represent compiled data for multiple specimens. All other cranial and dental fossil data were measured by the authors and are matched, measured from individual fossils that retain all relevant morphology. The fossils are held in repositories at the Ditsong Museum of Natural History and the Evolutionary Studies Institute at the University of the Witwatersrand (South Africa), the National Museum of Ethiopia, and the University of California Museum of Paleontology (USA). See SI for raw data used in this study.

We log-corrected all cranial measurements and tested that all traits were distributed normally prior to analysis. We used the describeBy function in the psych package (Revelle and Revelle, 2015) to generate descriptive statistics and characterize the data. Average values for all traits are reported in Table 1 (extant cercopithecids). Average values for each trait were used to run a series of phylogenetic generalized least squares (PGLS) analyses, which test for covariation between traits while accounting for phylogenetic relationships. These analyses examine whether traits are correlated more than would be expected from neutral evolution and phylogenetic relatedness. All phylogenetic analyses were conducted using a published molecular supertree trimmed to species of interest (Faurby and Svenning, 2015). We chose this molecular phylogeny because it retains Papio, Lophocebus, and Theropithecus as sister taxa in an unresolved polytomy (in contrast to 10ktrees [Arnold et al., 2010] which places Lophocebus as sister to Theropithecus in a monophyletic clade with Papio placed basally, and more recent mammalian supertrees [e.g., Upham et al., 2019] that continue to place Lophocebus and Cercocebus in a monophyletic mangabey clade in contrast to current consensus that these species are within different clades of the papionin phylogeny [P/L/T vs. C/M]).

We ran PGLS analyses comparing trait variation between palatal length, facial length, relative facial length, and the two dental ratios (PMM and MMC, for both the maxillary and mandibular dentitions), using the pgls function in caper (Orme et al., 2013). All PGLS analyses are interspecific comparisons of quantitative trait values. We also compared trait values across clades with phylogenetic ANOVA using aov.phylo in geiger (Harmon et al., 2008). We performed a preliminary assessment of the impact of sexual dimorphism on trait covariation in our extant sample using ANOVA.

To assess intraspecific variation in craniodental traits, we calculated correlations between trait values within a single species using the rcorr function in the Hmisc package (Harrell and Harrell, 2019). We aimed for sample sizes of n=30 for each species included in the analyses. Seven species had adequate sample sizes for correlation analyses: Cercopithecus mitis, Cercocebus torquatus, Lophocebus albigena, Macaca mulatta, Papio hamadryas, Colobus guereza, and Presbytis rubicunda. We also included Mandrillus sphinx, with a minimum sample size of n=28, in the correlation analyses.

Craniodental trait variation was visualized at species and subfamily/tribe levels for both extant and fossil taxa using bivariate regressions, plotted with ggplot2 (Wickham, 2016). We plotted dental ratios, as well as one dental length (which increased fossil sample size), against palatal and facial length. All analyses were conducted in the statistical programming software R version 4.3.2 (R Core Team, 2021).

The strong taxonomic discrimination provided by measures of facial length and relative premolar size supports that these traits are heritable (as has already been demonstrated for relative premolar size; Hlusko et al., 2016; Monson et al., 2022a) and may be important for facilitating occupation of distinctive morphospaces associated with ecological and dietary niches.

Extant cercopithecids occupy distinctive morphospaces of postcanine dental proportions and facial/palatal lengths. While PMM does not appear to be associated with diet broadly across mammals (Monson et al., 2019a), this study clearly demonstrates the diversity of second molar and fourth premolar lengths relative to facial/palatal length in cercopithecid primates ( Figure 3 ). These craniodental traits also partition the Cercocebus/Mandrillus and Papio/Lophocebus/Theropithecus clades ( Figure 2D ), aligning with previous work by Fleagle and McGraw (1999, 2002) demonstrating the relatively larger fourth premolar in Cercocebus and Mandrillus compared to Papio, Lophocebus, and Theropithecus, which they attribute to feeding behaviors and hard-object processing. Fleagle and McGraw (2002) also demonstrate that Macaca, which they use as an analogue for the papionin ancestor, has relatively larger fourth premolars, and they hypothesize that the enlarged first molars of Papio/Lophocebus/Theropithecus evolved independently in that clade.

A growing body of work has identified a strong allometric component to primate craniofacial variation with important implications for fossil taxonomy and paleontological studies (e.g., Collard and O’Higgins, 2001; Gilbert, 2013; Monson et al., 2017; Plavcan, 2002, 2003; Shirai and Marroig, 2010). Our work supports these previous studies and emphasizes that dental variation, and particularly molar proportions, should also be considered in assessments of primate craniofacial allometry and evolutionary relationships.

Most researchers now agree that prognathism is derived and evolved convergently in papionins (Benefit and McCrossin, 1991, 1993; Collard and O’Higgins, 2001; Cronin and Sarich, 1975; Delson, 1975; Disotell, 1994, 1996; Harris, 2000, 2002; Jolly, 1970; Strasser and Delson, 1987). The position of Victoriapithecus, a basal cercopithecoid, in trait morphospace suggests that the ancestral papionin morphology of dental proportions and facial lengths is likely similar to that seen in extant Macaca, Cercocebus, and Lophocebus. Fossil Papio, Parapapio and Procercocebus all look similar to extant baboons, although they tend to have relatively shorter third molars and shorter faces than extant taxa. These findings support that Parapapio and Procercocebus may represent an intermediate fossil morphology between the ancestral phenotypes typified by Victoriapithecus and other short-faced papionins, and the derived longer face in extant Papio and Mandrillus (see Brasil et al. [2025, this same issue] for a detailed history of the taxonomy and phylogenetic hypotheses for Parapapio). Further assessment of the relationships between postcanine tooth length proportions and facial/palatal length may help provide some indication of ancestor-descendent relationships, although additional fossils will be required to more thoroughly test these hypotheses.

If some early population of Parapapio is ancestral to other papionins, this would support that long faces evolved convergently in Papio/Theropithecus, and Mandrillus, with Parapapio as a likely medium-faced ancestor (this scenario supports that Lophocebus is an outgroup to Papio/Theropithecus). Alternatively, long faces may have evolved separately in all three lineages (Papio, Theropithecus, and Mandrillus) with short faces retained, or secondarily evolved, in Lophocebus and Cercocebus. A third scenario, less parsimonious based on available evidence, is that papionins shared a prognathic ancestor, currently undiscovered in the fossil record, and that Lophocebus and Cercocebus secondarily evolved shorter faces convergently.

In all scenarios, prognathic faces evolved convergently at least twice in papionins, and potentially three times, supporting that this trait is highly labile in this clade. The most parsimonious explanation is that Papio and Theropithecus shared a prognathic ancestor after the divergence of Lophocebus, and that long faces evolved twice in papionins (the first scenario described in the paragraph above). There continues to be debate about the phylogenetic relationships among papionins (e.g., Guevara and Steiper, 2014), with many researchers settling on a polytomy for Lophocebus/Papio/Theropithecus. Our work weighs in on this debate by providing morphological evidence that Lophocebus more likely diverged earliest in this clade, and prognathic faces evolved in a more recent common ancestor shared by Papio/Theropithecus.

It is possible that long-faced primates converged on relatively longer third molars through selection for longer teeth and/or selection for longer faces, as both traits have been hypothesized to assist in chewing and digesting a diet dominated by tough C4 plants and other abrasive materials (e.g., Cerling et al., 2011; Jablonski, 1993; Singleton, 2005). Future comparative studies looking at facial elongation in other mammals (e.g., Carnivora; Machado et al., 2018; Sears et al., 2007; Tamagnini et al., 2017), particularly herbivorous species (e.g., horses; Janis et al., 2024), as well as the biomechanics of chewing and muscular anatomy, may provide additional insights into the likely convergent evolution of prognathism in papionins.

Facial retraction in humans remains one of the most enigmatic changes in hominid cranial evolution. Decades of hypotheses attempting to explain the reduction of supraorbital robusticity and facial length in humans have focused on diet and the biomechanics of chewing, cognition and brain size, and other functional adaptations (e.g., Demes, 1987; Gómez-Robles et al., 2017; Preuschoft and Witzel, 2004). More recently, research has demonstrated the strong roles of craniodental integration and evolutionary allometry in shaping primate facial morphology. Although there is a large and growing body of research on cranial variation and modularity in anthropoids (e.g., Cummings et al., 2012; Grunstra et al., 2021; Kay and Kirk, 2000; Ravosa and Profant, 2000), key unanswered questions about the evolution of cranial anatomy remain, particularly for traits related to facial retraction.

Relative third molar lengths also reduced dramatically in Homo throughout the Plio-Pleistocene, leading to small maxillary and mandibular third molars in modern humans (e.g., Gómez-Robles et al., 2017; Ungar, 2012). High rates of agenesis of these teeth are also reported in humans, with occurrence of agenesis significantly associated with shorter facial and jaw lengths (e.g., Fekonja and Čretnik, 2022; Gkantidis et al., 2021). The identification of a common set of genes influencing both craniofacial development and tooth size has led researchers to hypothesize that reductions in prognathism and changes in tooth size proportions in human evolution are pleiotropically linked (e.g., Polychronis and Halazonetis, 2014). The impetus for reduced third molar length remains unexplained in human evolution, although a growing body of evidence suggests that these evolutionary changes may also be linked to changes in fetal growth and gestation (Monson et al., 2019b; 2022b).

Previous investigation of evolutionary changes in craniofacial and dental anatomy in fossil hominids has suggested that these traits may not be tightly linked (Gómez-Robles et al., 2017). However, this previous work focused on fossil samples and brain size rather than facial projection, and data for craniofacial and dental anatomy are often collected from unassociated specimens. Given the anatomical relationship between cranial and dental morphology seen in cercopithecids in this study, it is reasonable to hypothesize that the relatively reduced third molar length in humans may be associated with our shorter faces rather than resulting exclusively from any adaptive or functional role. Further analysis of morphological covariation between facial/palatal length and postcanine dental proportions across a broader sample of primates, particularly apes, will help test this hypothesis.