INTRODUCTION

Colobines from the equatorial forests of Africa form a monophyletic group (Fig. 1; Colobini Jerdon, 1867) of three genera (Colobus ‌Illiger, 1811, Piliocolobus ‌Rochebrune, 1877, and Procolobus ‌Rochebrune, 1877) that likely play a role of seed dispersers (Matsuda et al. 2013) and are documented as a source of subsistence for chimpanzees (Bugir et al. 2021). Molecular data characterized Colobus as the sister taxon to the Piliocolobus-Procolobus clade (Ting 2008; Perelman et al. 2011; Kuderna et al. 2023), with both clades documented sympatrically in western, central and eastern Africa (Fig. 1; Kingdon & Groves 2013). When documented in sympatry, eco-ethological data from the Taï Forest (Ivory Coast) highlighted that climbing and leaping is more frequent in Piliocolobusbadius ‌Kerr, 1792 than in Colobuspolykomos ‌Zimmermann, 1780, and leaf consumption higher in Pi. ‌badius compared to the specialized granivory and the processing of tough and fibrous food by Co. ‌polykomos (McGraw 1998; Daegling & McGraw 2001). Skeletal traits diagnostic of the Colobus and Piliocolobus-Procolobus clades were identified on the dental, mandibular, cranial, and postcranial anatomy (Schultz 1958; Verheyen 1962; Strasser 1992; Groves 2007; Cardini & Elton 2009a, b; Koyabu & Endo 2009; Dunham et al. 2015). Recent advances in the last decades were notably done in the identification of diagnostic dental and mandibular traits (Groves 2007; Hlusko et al. 2016; Pallas et al. 2019; Brasil et al. 2023; Pallas et al. 2024a, b). Among others, these studies highlighted the large lateral prominences of the corpus and the large inferior transverse torus of the symphysis, the angulated symphysis, the reduced area of the P4 relative to the M1 area, and the large tuberculum sextum (C6) of Piliocolobus and Procolobus compared to Colobus. Despite those diagnostic morphological and eco-ethological traits, the evolutionary history of extant Colobini is poorly known. This is especially true for the absence of fossil record for Piliocolobus and Procolobus, contrasting with the presence of Colobus remains in eastern African fossil deposits (Fig. 1; Frost 2001; Jablonski & Frost 2010).

Leaving aside scarce remains where taxonomy was established on isolated teeth (Eck 1977; Leakey 1987) or sites with dubious dating (Harrison & Harris 1996), a fossil species of Colobus, Colobusfreedmani ‌Jablonski & Leakey, 2008, which includes a female partial skeleton in its hypodigm along with a fairly complete putative male mandible (KNM-ER 6025), is known from the c. 1.9-1.4 Ma Okote and KBS members of the Koobi Fora Formation (Jablonski & Leakey 2008). An indeterminate Colobus species, hypothesized as related to extant Colobusangolensis ‌Sclater, 1860, is documented in the c. 600 Ka site of Asbole from a substantial collection of mandibular remains (Frost & Alemseged 2007). A large collection of Colobus specimens, likened to Co.guereza ‌Rüppell, 1835, from the Middle Awash research area is also known from the c. 100 Ka Faro Daba beds of the Daiwaitoli Formation in Ethiopia (Brasil et al. 2023). Another fossil population, partially described in Frost (2001) and possibly related to Colobusangolensisgiven its poorly dimorphic canines and heteromorphic incisors, is known from “Andalee” (see Brasil et al. [2023] for further information on site denomination and chronostratigraphic context). Finally, an isolated skull of Co. ‌guereza from Wadi Medani in central Sudan (Simons 1967; Jablonski & Frost 2010) and an undescribed partial skull from the Shungura Formation that may represent Colobus ‌(Assefa et al. 2008) complete our knowledge of the Colobus fossil record.

Relationships between fossil Colobus populations have been seldomly investigated and controversy remains regarding the taxonomic homogeneity of the Co. ‌freedmani sample from Koobi Fora and the taxonomic status of the Asbole specimens. Indeed, the Koobi Fora sample was described as Colobus by Jablonski & Leakey (2008) but previously regarded as Procolobus by Frost & Alemseged (2007), mainly based on the corpus robustness. In contrast, Jablonski & Leakey (2008) considered the robust corpus of the putative male KNM-ER 6025 as part of the range of variation of Colobus. Similarly, the colobine sample from Asbole was initially described as belonging to the Procolobus-Piliocolobus clade by Alemseged & Geraads (2000) and later revised to Colobus by Frost & Alemseged (2007). The species status of Asbole is indeterminate and its affinity with Co. ‌angolensis established primarily on biogeographical data and poor canine dimorphism. It has been recently suggested that the Colobus cf. guereza samples from Faro Daba and Asbole may represent a chronolineage (Brasil et al. 2023), giving support for population continuity in the northern part of Ethiopia during the last 600 Ka in addition to recognizing that the Asbole sample may include the earliest members of the Colobus ‌guereza lineage. Resolving the taxonomic affinity of the Asbole and Koobi Fora sample has hence implications on faunal dispersal along the East African Rift System during the Pleistocene, a time and place that witnesses the emergence of Homosapiens ‌Linnaeus, 1758.

Here, the mandibular colobine fossil record from Asbole and Koobi Fora is used to evaluate the possible presence of traits diagnostic of extant Colobini taxa that were previously disregarded. Univariate and multivariate analyses (exploratory and discriminant) are used along with hierarchical clustering established on lower teeth dimensions, symphyseal dimensions, and corpus shape to test the hypothesis that the specimens are phenetically closer to Colobus than to Procolobus and Piliocolobus. An alternative hypothesis would posit that the Asbole and Koobi Fora samples each represent a distinct taxon or, for the large collection from Asbole, a mix of different taxa. The range of variation present within the corpus morphology of the Asbole specimens is assessed using bootstrap analyses to test the hypothesis that only one taxon is present and whether sexual dimorphism can explain differences in corpus shape between fossil specimens, as proposed by Jablonski & Leakey (2008).

MATERIAL AND METHODS

Extant and fossil samples

The dimensions of the lower canine, premolars, and molars of n = 113 specimens and the symphyseal dimensions of n = 119 specimens extant Colobini were measured. The depth of the corpus, at mid-M1, of n = 233 extant Colobini specimens were also measured along with landmark digitization on the transverse cross-section of n = 178 corpus of extant Colobini. A list of the extant and fossil specimens used in this study is provided in Appendix 1. Only adult specimens were included in the analysis, and their adult status was asserted using the M3 eruption. Extant and fossil Colobini specimens were collected from the National Museums of Ethiopia (Addis-Abeba), the National Museums of Kenya (KNM, Nairobi), the Muséum national d’Histoire naturelle (MNHN, Paris), the Royal Museum for Central Africa (RMCA, Tervuren), the Bavarian State Collection of Zoology (ZSM, Munich), the Center for the Evolutionary Origins of Human Behavior (EHUB, Kyoto University, Inuyama), the Zoological Collection of the Kyoto University, the Smithsonian National Museum of Natural History (USNM, Washington), the American Museum of Natural History (New York), the Stony Brook University (New York), and the PALEVOPRIM Laboratory (University of Poitiers).

Data acquisition and morphometry

Fossil and extant specimens integrated in this study were scanned using the Artec Space Spider and Einscan Pro 2X surface scanners as well as using photogrammetry given photographs obtained with a Panasonic Lumix DC-GX9 camera. For the dental fossil data, I measured dental dimensions from 3D-generated surfaces using Avizo v7.0. The mesiodistal (MD) and buccolingual (BD) dimensions of the lower canine, premolars, and molars were measured to the nearest tenth of millimeters following the protocol illustrated in Figure 2A. Three dental ratios were used to evaluate the phenetic similarity between extant and fossil taxa. Precisely, these ratios quantify the MD elongation of the lower canine relative to the M2 area, the area of the P4 relative to that of the M1 and the length of the M3 relative to the M2 area.

The length of the symphysis, the breadth of the symphyseal transverse tori and the angulation of the symphysis relative to occlusal plane (Fig. 2B) were measured from pictures of sagittal cross-section of symphysis using ImageJ (Schneider et al. 2012) and following the protocol presented in Pallas et al. 2024a. In addition to differences in absolute angulation of the symphysis, the ratio of the symphyseal transverse torus (STT) breadth over the breadth of the symphyseal inferior torus (ITT) was used to investigate phenetic similarity between extant and fossil sample.

Three landmarks and seventy-five semilandmarks (Fig. 2C) were digitized on transverse cross-sections of the corpus of extant and fossil colobines following the protocol presented in Pallas et al. (2024b). To summarize, cross-section images at the M1/M2 junction were obtained using Avizo with landmarks and semilandmarks digitization done using tpsDig (Rohlf 2006). Three landmarks were placed at the 1) most proximal and buccal point of the corpus; 2) most inferior point of the corpus; and 3) most proximal and lingual point of the corpus. Seventy-five semilandmarks were then digitized equidistally along the corpus outline.

Repeatability of morphometric analyses

To verify the repeatability of the new protocol proposed here for the symphyseal morphometric analyses, four colobine specimens belonging to four different taxa (Nasalis ‌larvatus ‌Geoffroy, 1812; Colobus ‌satanas ‌Waterhouse, 1838; Pilioolobus ‌badius ‌Kerr, 1972; Procolobus ‌verus ‌Van Beneden, 1838) were measured five times each for each measurement by the same observer (LP). The measurement error estimate follows the computation presented in White & Folkens (2005) and a maximum error threshold of 5% was adopted. Briefly, the average value of the five replicates was calculated and then subtracted from each replicate value. The resulting cumulative differences were then divided by the number of measurements, which was five in this case, to obtain a percentage of error.

All results for replicability of the symphyseal measurements are presented in Appendix 2. The range of measurement error for symphyseal length (0.4-2.0%), breadth of the superior transverse torus (0.2-3.9%), breadth of the inferior transverse torus (0.4-0.9%), planum alveolare length (1.4-3.8%), and symphyseal angulation (2.6-5%) never exceeded 5% of error and are considered as repeatable.

A different approach was used to test for the repeatability of the 2D geometric morphometric analysis of the corpus shape. Repeatability is here considered as the proportion of the between taxa variance compared to the total observed variance (between and within taxa). Landmarks and semilandmarks were digitized by the same observer (LP) on four specimens of four different colobine taxa (same as previously). Landmarks and semilandmarks coordinates were subjected to a Generalized Procrustes Analysis and a principal component analysis (PCA) was run on the Procrustes coordinates (see below subsection below for more details). Repeatability was assessed on PCA scores using the rptGaussian function of rptR (Stoffel et al. 2017). For each PC scores, the within taxa variation is extremely low and nearly all the variation is explained by the between taxa variation for PC1 (95% confidence interval [CI] = 0.99-1), PC2 (95% CI = 0.98-1), and PC3 scores (95% CI = 0.97-0.99). This result is confirmed graphically in Appendix 3.

Statistical analysis

All statistical analyses were done using R v.4.4.1 (R Core Team 2024) and the interface RStudio v.2023.12.0 (RStudio Team 2020). All the packages and functions used can be found in Table 1.

Univariate analyses: First, phenetic similarity was evaluated by comparing dental ratio values between fossil specimens and extant Colobini. The formulas for the dental ratios can be found in Table 2. Such ratios were previously recognized as significantly distinct between extant Colobini taxa but given that additional specimens were added compared to Pallas et al. (2019), new statistical tests were computed to evaluate whether differences are still significant. Significance was evaluated using a one-way parametric analysis of variance (ANOVA) and non-parametric Kruskall-Wallis. Normality and homoscedasticity of the model residuals were evaluated prior to the ANOVAs using the Shapiro-Wilk and Bartlett tests, respectively. To identify pairs of taxa that differ significantly, the Tukey Honest Significant Differences (HSD) post-hoc test following ANOVA and the Dunn test following Kruskall-Wallis were used.

Principal Component Analyses and Generalized Procrustes Analysis: Following comparison of univariate data, a principal component analysis (PCA) was used to explore and order variation within the dental, symphyseal and corpus datasets. Data was transformed using the log-shape ratio method prior to PCA for the dental and symphyseal data while a Generalized Procrustes Analysis (GPA) was used for the corpus landmark dataset. The log-shape ratio method standardizes data and mitigates the confounding effect of size (Mosimann 1970). The GPA was applied on the 2D landmarks and semi-landmarks coordinates, permitting retaining shape data only by translating, scaling, and rotating the landmarks and semi-landmarks coordinates of each specimen.

Significant differences in PC scores between extant genera were examined using the parametric multivariate analysis of variance (MANOVA) and the non-parametric permutational multivariate analysis of variance (PERMANOVA). Normality and homoscedasticity were evaluated prior to MANOVA and PERMANOVA using the multivariate Shapiro-Wilk and Box M tests, respectively. Differences in PC scores between pairs of taxa were evaluated using Tukey HSD and Dunn post-hoc tests.

Sexual dimorphism in corpus shape: To evaluate sexual dimorphism in corpus shape, a MANOVA on the scores of the three PCs of the aligned Procrustes coordinates was run. Multivariate normality and homogeneity of covariances were evaluated prior to MANOVA using the multivariate Shapiro-Wilk and Box M tests, respectively. The MANOVA was followed by a Tukey HSD test to identify pairs of significantly distinct taxa.

Linear Discriminant Analyses (LDA): Following exploration and ordination of the variation, a linear discriminant analysis was used on the first four PCs of the dental, symphyseal and corpus PCAs to taxonomically classify fossil specimens and to highlight their phenetic affinities towards Colobus, Piliocolobus or Procolobus. Prior to inclusion of fossil specimens, a jackknife analysis was used to evaluate the accuracy of the model to correctly classify extant specimens.

Hierarchical clustering: Hierarchical clustering was used to assess the phenetic affinity of the Asbole specimens. A distance matrix was first computed using Euclidean distance on the scaled and centered values of the first two mean PC scores of dental shapes of the three extant Colobini genera in addition to the first two PC scores of corpus shape and the first three PC scores of symphyseal shape. The hierarchical clustering was computed using the Ward clustering method.

Two different hierarchical clustering were done given different operational taxonomic unit (OTU) for the Asbole specimens. The first clustering incorporates the mean value of all Asbole specimens for dental, corpus, and symphyseal shape. This view hence recognizes the Asbole colobine specimens as part of a single taxon. The second clustering includes only ASB-232-6 and ASB-100-1 from which a complete dataset for dental, corpus, and symphyseal shape is available.

Bootstrap analyses: To evaluate whether the range of variation of the corpus shape of the Asbole sample and the combined Asbole and Koobi Fora samples can be accommodated in the same genus, a bootstrap analysis was carried out based on comparison of the coefficient of variation. To this end, 10.000 random distributions were generated for Colobus, Procolobus, and Piliocolobus based on PC1 and PC2 scores obtained from the original dataset. Then, 10.000 random distributions were generated based on n = 6 and n = 7 specimens randomly selected for each extant genus. The number of specimens randomly selected corresponds to the number of specimens included in the Asbole sample (n = 6) and the combined Asbole and Koobi Fora samples (n = 7). Finally, coefficient of variation (i.e., standard deviation/mean) for each of the generated distribution of extant Colobini genera were computed and compared with that of the fossil sample.

A bootstrap analysis was also generated using corpus height below mid-M1 as many specimens from Asbole (n = 10) preserves this dimension. Compared to the corpus shape bootstrap, that of the mid-M1 depth of the corpus was evaluated at species level using n = 7 species of Colobini: Co. ‌angolensis, Co. ‌guereza, Co. ‌polykomos, Pi. ‌langi ‌Allen, 1925, Pi. ‌parmentieri ‌Colyn & Verheyen, 1987, Pi. ‌tholloni ‌Milne-Edwards, 1886, and Pro. verus. Corpus depth at M1 is also reported for the Faro Daba Co. ‌guereza specimens (Brasil et al. 2023), and its coefficient of variation compared to a combined Faro Daba, Koobi Fora and Asbole sample and also to the coefficient of variation of extant Colobini species.

Abbreviations

RESULTS

Univariate and multivariate dental analyses

Relative canine elongation: The index of the MD length of the canine relative to the M2 area statistically discriminates Procolobus and Piliocolobus from Colobus (Fig. 3A; Table 3). The canines of the former are mesiodistally elongated compared to the M2 area, having significantly higher index values (Table 3), and mesiodistally longer canines than Colobus. This condition reaches an extreme level in Procolobus, which has significantly longer canines relative to the M2 area than those of Colobus and Piliocolobus. The position of the Asbole colobines is ambiguous and apart from ASB-254, they occupy an intermediate position between the interquartile range of Piliocolobus and Colobus (Fig. 3A). Precisely, ASB-100-1 and ASB-232-6 have index values consistent with the interquartile of Piliocolobus and Colobus while ASB-80 and ASB-215-2 fall in the interquartile range of Colobus only and in the upper range of variation of Piliocolobus.

Area of the P4 relative to the area of the M1: The index of the area of the P4 relative to the area of the M1 statistically discriminate Procolobus and Piliocolobus from Colobus (Fig. 3B and Table 3), with the former two genera having a reduced P4 relative to M1 dimensions, and hence statistically lower index values compared to Colobus (Table 3). The Asbole colobine sample has a mean index value consistent with the interquartile range of Procolobus and Piliocolobus. Precisely, ASB-215-2 and ASB-232-6 are falling in the interquartile range of Piliocolobus and Procolobus while ASB-100-1 and ASB-25-6 are fitting with the interquartile range of Colobus (Fig. 3B). Similarly, Colobusfreedmani ‌KNM-ER 6025 falls in the interquartile range of Piliocolobus and Procolobus and in the first quartile of Colobus.

Relative M3 elongation: MD length of the M3 relative to the M2 area statistically discriminates Piliocolobus and Procolobus from Colobus (Fig. 3C and Table 3). Piliocolobus and Procolobus have significantly higher index values compared to Colobus, illustrating their longer M3 compared to the M2 area. The M3 of Procolobus is even longer, relative to the M2 area, than that of Piliocolobus. While ASB-76 conforms to the interquartile range of Colobus, the specimens ASB-232-6, ASB-80, and ASB-25-6 are within or close to the interquartile range of Piliocolobus. ASB-259 and ASB-100-1 are included in the range of variation of both Piliocolobus and Colobus.

Univariate and multivariate analyses of the mandibular symphysis

Breadth differential of the transverse tori: The differences in breadth between the STT and ITT differentiate extant Colobini, with Piliocolobus and Procolobus having tori of nearly equal breadth, and hence a significantly lower index compared to the marked breadth differential of the transverse tori of Colobus (Fig. 6A, Table 6 and Table 3). The Asbole colobines occupy an intermediate position, falling in the third quartile of Piliocolobus and first quartile of Colobus. However, they fall in the interquartile range of variation of Procolobus.

Symphyseal angulation: The symphysis of Piliocolobus is significantly more inclined than that of Colobus and Procolobus (Fig. 6B, Table 3 and Table 6). Apart from the obtuse symphysis of ASB-100-1, all other Asbole colobines have an inclination of the symphysis fitting with the interquartile range of variation of Procolobus and Colobus.

Univariate and multivariate analyses of the mandibular corpus

Taxonomic discrimination based on corpus shape: PC1 and PC2 account for 72% of the variance (Fig. 9). Colobus and Procolobus shows, on average, slightly negative scores on PC1 compared to the slightly positive scores of Piliocolobus. On PC2, Colobus has, on average, slightly negative scores, and contrast with the slightly positive scores of Piliocolobus and Procolobus (Table 8). Shape changes along PC1 (Fig. 9) are explained by an overall gracility of the corpus (high and thin in contrast to short and broad) from low scores (gracile corpus) to positive scores (robust corpus). Shape changes along PC2 (Fig. 9) rather reflects the development of the lateral prominences (positive scores) and the development of the submandibular fossa (negative scores). On PC1, the Asbole colobines have greatly (e.g., ASB-232-6) or slightly positive scores (e.g., ASB-25-6), hence better conforming with the Piliocolobus-Procolobus mean scores and its robust corpus (Fig. 9). A similar observation can be made for KNM-ER 6025. On PC2, the Asbole colobines also have greatly (e.g., ASB-80) or slightly (e.g., ASB-25-6) positive scores. KNM-ER 6025 has the highest positive scores of all the fossil colobines on PC2. Overall, the positive scores of the Asbole specimens and KNM-ER 6025 on PC2 demonstrate their enlarged lateral prominences, like Piliocolobus-Procolobus but distinct from Colobus (Fig. 9).

DISCUSSION

A reassessment of the taxonomy of Colobus ‌freedmani

This study demonstrated that the relatively small P4 and the corpus shape of Co. ‌freedmani ‌KNM-ER 6025 resemble Piliocolobus and Procolobus more than Colobus, although the fossil specimen still falls within the range of variation of extant Colobus. The initial justification for including KNM-ER 6025 to Colobus is dubious, as no traits clearly supporting its inclusion are provided in the original description of Jablonski & Leakey (2008). Sexual dimorphism in the development of the lateral prominences was advocated to explain the differences observed between the holotype of Co.freedmani (i.e., the damaged female mandible KNM-ER 44224) with the partial male mandible KNM-ER 6025 (Jablonski & Leakey 2008). The development of the lateral prominences of the corpus is indeed sexually dimorphic in extant Colobus as shown on the PC2 scores of the corpus outline analysis, but a trend (non-significant) for sexually dimorphic lateral prominences is also observed for Procolobusverus, demonstrating that this dimorphic pattern in corpus shape is not unique to Colobus. It was further hypothesized that Co. ‌freedmani was more sexually dimorphic than extant Colobus in corpus shape to accommodate KNM-ER 6025 in the hypodigm of Co. ‌freedmani. Unfortunately, no adequately preserved Colobus female specimens from Koobi Fora could have been integrated in our morphometric analysis to fully test this hypothesis. Moreover, assessing a level of dimorphism in a fossil population requires an extensive fossil sample and the Koobi Fora sample is currently too limited to test this hypothesis.

KNM-ER 6025 is an outlying specimen when compared to the range of variation of Colobus in corpus shape and its P4 area/M1 area ratio is also more consistent with that of Piliocolobus-Procolobus than with that of Colobus. Our data hence favor, on quantitative grounds, an attribution to the Piliocolobus-Procolobus clade for KNM-ER 6025 rather than to a male Co. ‌freedmani, supporting the initial hypothesis of Frost & Alemseged (2007). However, the hypothesis that two distinct taxa occur at Koobi Fora cannot be rejected given the low number of mandibular specimens available, with one taxon attributable to Colobus (the female mandible KNM-ER 44224) and another one to Piliocolobus – Procolobus (KNM-ER 6025). The humerofemoral proportion (i.e., humeral length over femoral length) of the partial female Co. ‌freedmani skeleton KNM-ER 5896 is more akin to that of Colobus than to that of Piliocolobus-Procolobus, highlighting the ambiguity of the taxonomic homogeneity of the small colobines from the Okote- KBS members of Koobi Fora (Pallas, in prep.). Future discoveries of small-sized colobines in the Okote, KBS, and Chari members of Koobi Fora or in penecontemporaneous fossil deposits of the Turkana Basin will be of prime value on this aspect. A closer look at postcranial traits distinguishing Piliocolobus from Colobus will also add more evidence to answer this question.

A reassessment of the taxonomy of Colobus from Asbole

Morphological similarities between the Asbole colobine specimens ASB-232-6 and ASB-256 with Piliocolobus-Procolobus and between ASB-100-1 and ASB-259 with Colobus, support their taxonomic allocation to the Piliocolobus-Procolobus clade and the Colobus clade, respectively.

Frost & Alemseged (2007) dismissed the attribution of Piliocolobus for the Asbole colobines based on the following traits: presence of infraorbital foramina of small diameters (foramina are larger in Piliocolobus), absence of maxillary fossae and ridges (present in Piliocolobus), gracile supraorbital tori (developed tori in male Piliocolobus), low and broad choanae (narrower in Piliocolobus), lack of a sagittal crest (present in male Piliocolobus), lack of a perforated pterygoid fossa (perforated fossae in Piliocolobus), deep mandibular corpus with expanded gonial area (shallower corpus in Piliocolobus), and absence of a median mental foramen (present in Procolobus but absent in Piliocolobus). In contrast, mandibular specimens of Asbole align with Piliocolobus-Procolobus and Colobus based on dental, corpus and symphyseal shape. Specifically, ASB-232-6 and ASB-25-6 support the presence of a colobine more similar to Piliocolobus-Procolobus at Asbole while ASB-100-1 and ASB-259 support the initial taxonomic hypothesis of Frost & Alemseged (2007). The bootstrap analysis computed based on the PC scores of the corpus shape and superoinferior dimension of the corpus at mid-M1 failed to demonstrate that the variation of the Asbole sample exceeds that of any extant Colobini genera, supporting the presence of a single taxon at Asbole. However, the combined dental, corpus, and symphysis evidence still provide support for the presence of a small colobine more like Piliocolobus-Procolobus than Colobus. Our data are also consistent with and complement those of Brasil et al. (2023) which demonstrated that the breadth-length proportions of the M2 of the Asbole sample is more similar to Piliocolobus and that the length of the P4 is greatly reduced compared to the length of the M1, a result consistent with our P4 area/M1 area ratio. The hierarchical clustering based on PC scores of dental, symphysis, and corpus analyses also showed that ASB-232-6 has phenetic similarities with Piliocolobus.

To accommodate the observations of Frost & Alemseged (2007) on the cranial and maxillary specimens from Asbole and to account for the phenetic affinity of several mandibular specimens with Colobus (e.g., ASB 100-1), this study partly supports the hypothesis that two taxa occur at Asbole with affinities to Piliocolobus-Procolobus on one side and Colobus on the other side. Given that the present study is limited to mandibular remains, these taxa are not named, but it supports an inclusion of ASB-232-6 and ASB-25-6 to cf. Piliocolobus sp. indet., as previously suggested (Alemseged & Geraads 2000), but their precise taxonomic designation awaits additional postcranial and cranial quantitative evidence. Unfortunately, the association of the female maxilla ASB-204A and mandible ASB-204B as well as the male maxilla and mandible ASB-207-1A and ASB-207-1B is not clearly asserted in Frost & Alemseged (2007) and most of the specimens presented in the initial study are unprepared, leaving several anatomical characteristics unknown. Future discoveries of securely associated cranial and mandibular elements will provide decisive insights into the taxonomic conundrum of the Asbole colobines.

Conclusively, the recognition of a fossil taxon similar to Piliocolobus-Procolobus in mandibular anatomy in Asbole is consistent with our conclusion regarding the KNM-ER 6025 mandible from Koobi Fora. The Colobini taxonomic diversity highlighted here at Asbole and Koobi Fora has implications on our understanding of the emergence of extant African colobine genera.

The emergence of extant African colobine genera in the fossil record

It has been suggested that the Colobus sample from Asbole was ancestral to that from Faro Daba, implying population continuity between the c. 600 Ka Colobus population from Asbole and the c. 100 Ka Colobus population from Faro Daba (Brasil et al. 2023). Brasil et al. (2023) showed that the marked sexual dimorphism of the Asbole colobines in canine dimensions was inconsistent with that of the Colobus cf. guereza from Faro Daba and from that of extant Co. ‌guereza species, hence suggesting that sexual dimorphism in canine dimensions was marked in the stem Co. ‌guereza lineage represented at Asbole. However the identification of specimens more similar to Procolobus-Piliocolobus than Colobus at Asbole does not support population continuity at the colobine community level as no Piliocolobus-Procolobus specimens are recorded at Faro Daba. However, Colobus specimens sampled at Asbole (ASB-100-1 and ASB-259) could be part of the Co. ‌guereza lineage and should be part of further analysis and comparison with Co. ‌guereza.

Extant eastern African Piliocolobus species are considered as relictual populations and currently confined to limited geographical ranges in Kenya, Zanzibar and Tanzania (Cardini & Elton 2009b). The identification of c. 1.5 Ma and c. 600 Ka Ma mandibles similar to Piliocolobus at Asbole and Koobi Fora demonstrate that the geographical range of Piliocolobus could once have been more expansive and extended into northern Ethiopia. Moreover, the bootstrap analysis established on corpus shape did not provide evidence for a distinct taxonomic designation between the Koobi Fora mandible KNM-ER 6065 and the Asbole sample, at least at the generic level. Future studies should investigate whether the specimens similar to Piliocolobus from Asbole and Koobi Fora represent extant Piliocolobus species, notably Pi. ‌gordonorum ‌Matschie, 1900, Pi. ‌rufomitratus ‌Peters, 1879 or Pi. ‌kirkii ‌Gray, 1868, and alternatively a new species of Piliocolobus.

A large hiatus is present in the fossil record of extant Colobini. According to molecular data, Colobus should be present in c. 8 Ma fossil deposits and members of the Piliocolobus-Procolobus clade should be sampled after c. 6 Ma. Prior to Co. ‌freedmani (c. 1.5 Ma), there is no consensus regarding the identification of fossil Colobini. At species level, the western African Colobus ‌satanas diverged molecularly from other Colobus species at c. 3.5 Ma. Molecular divergence dates between extant Piliocolobus species are more problematic but eastern African Piliocolobus (i.e., Pi. ‌rufomitratus, Pi. ‌gordonorum, and Pi. ‌kirkii) are part of the same clade which diverged from a central and western African clade at c. 3.5 Ma. Considering putative temporal lag between morphological speciation and molecular speciation, we should expect to identify Colobus and Piliocolobus specimens in eastern Africa after 3.5 Ma. A reevaluation of fossil specimens similar in size to extant Colobini after 3.5 Ma is needed to test whether Colobus and Piliocolobus are novel eastern African immigrants that appeared during the Pleistocene, as suggested by molecular data, or if older lineages could be identified in the eastern African fossil record. Considering that Colobini are forest-dependent taxa, extension of the geographical ranges of Colobini in the fossil record should also be correlated to paleoenvironmental changes and should hence track extension and contraction of forest covers along the East African Rift system. The identification of extant Colobini in the fossil record and the reevaluation of the small colobine fossil record has hence strong paleoenvironmental and evolutionary implications.

Acknowledgements

I greatly acknowledge the Ethiopian Heritage Authority at the National Museum of Ethiopia (NME) and Prof. Zeresenay Alemseged for granting access to the Asbole specimens. I am grateful to Sahle Melaku and Getahun Tekle for help and assistance at the NME and Dawit Seifu Hailemariam for preparation of the mandibles from Asbole. I am indebted to the National Museum of Kenya and specifically Tom Mukhuyu, Erus Justus and Jimmy Yatitch for help and assistance. I deeply especially acknowledge the support of curators and staff of the National Museums of Kenya (Osteology Division, Nairobi), the Royal Museum for Central Africa (Dr Emmanuel Gilissen and Mathys Rotonda from the Division of Mammals, Tervuren), the Center for the Evolutionary Origins of Human Behavior (Assoc. Pro. Takeshi Nishimura the Division of Mammals), the US National Museum of Natural History (D. Schlitter, Division of Mammals), and the Bavarian State Collection for Zoology (Dr Anneke van Heteren and Michael Hiermeier) for the study of skeletal collections under their care. I also thank the NSF BCS 1552848 attributed to Dr D. Boyer, and Assoc. Prof. Claire Terhune (University of Arkansas) for data sharing via the MorphoSource platform. I thank the Groupement de Recherche (GDR) Rift for financial assistance and the attribution of a scholarship that permitted to undertake this study. At last, Dr Brenda Benefit (New Mexico State University, Las Cruces) and an anonymous referee are also thanked for their remarks on a previous version of the article.

APPENDICES

Appendix 1 Appendix 1. — List of the extant and fossil specimens used in this study. 10.5281/zenodo.21537367 Genus/Acc. n° Dental analysis Colobus SNSB-ZSM 1911 2465 SNSB-ZSM 1911 2469 SNSB-ZSM 1911 2475 SNSB-ZSM 1911 2477 SNSB-ZSM 1912 1549 SNSB-ZSM 1951 263 SNSB-ZSM AM1269 SNSB-ZSM AM1270 SNSB-ZSM 1904 1542 SNSB-ZSM 1904 1543 SNSB-ZSM 1911 2470 SNSB-ZSM 1911 2471 SNSB-ZSM 1911 2476 SNSB-ZSM 1913 1168 SNSB-ZSM 1914 839 SNSB-ZSM 1914 847 SNSB-ZSM 1904 1541 SNSB-ZSM 1900 573 SNSB-ZSM 1909 P1 SNSB-ZSM 1904 1544 MNHN-ZM-AC-1983-97 MNHN-ZM-MO-1885-891 MNHN-ZM-AC-1983-98 MNHN-ZM-2009-359 MNHN-ZM-AC-1954-54 MNHN-ZM-MO-1939-702 MNHN-ZM-MO-1962-1390 MNHN-ZM-2009-358 MNHN-ZM-MO-1970-450 MNHN-ZM-MO-1962-2102 MNHN-ZM-2009-346 MNHN-ZM-MO-1939-701 MNHN-ZM-MO-1908-43 MNHN-ZM-MO-1970-60 MNHN-ZM-MO-1970-59 MNHN-ZM-MO-1890-194 MNHN-ZM-MO-1893-5 MNHN-ZM-MO-1962-123 MNHN-ZM-MO-1897-1505 MNHN-ZM-MO-1969-389 MNHN-ZM-MO-1969-379 MNHN-ZM-MO-1969-382 MNHN-ZM-MO-1969-384 MNHN-ZM-AC-1897-132 MNHN-ZM-MO-1969-380 MNHN-ZM-MO-1962-1398 MNHN-ZM-MO-1969-387 MNHN-ZM-MO-1969-386 MNHN-ZM-AC-1983-37 MNHN-ZM-MO-1908-191 MNHN-ZM-MO-1964-1517 MNHN-ZM-MO-1892-1010 MNHN-ZM-MO-1904-1960 MNHN-ZM-MO-1963-294 MNHN-ZM-MO-1892-1009 MNHN-ZM-MO-1892-1011 MNHN-ZM-MO-1962-986 MNHN-ZM-2009-345 MNHN-ZM-MO-1970-396 MNHN-ZM-MO-1886-108 MNHN-ZM-MO-1964-1514 MNHN-ZM-MO-1974-133 MNHN-ZM-MO-1892-1007 Piliocolobus KUPRI 10129#524 KUPRI 10127#524 KUPRI 10125#520 KUPRI 10124#519 KUPRI 10123#518 KUPRI 10122#517 KUPRI 10109#504 KUPRI 10115#510 KUPRI 10110#505 KUPRI 10112#505 KUPRI 10067#462 KUPRI 10183#4786 KUPRI 10070#465 KUPRI 10184#4787 KUPRI 10082#477 KUPRI 10052#447 KUPRI 10001#396 KUPRI 10095#490 KUPRI 10046#441 KUPRI 10188#4791 MNHN-ZM-MO-2002-103 MNHN-ZM-MO-1895-9 MNHN-ZM-MO-1939-705 MNHN-ZM-MO-1908-45BIS MNHN-ZM-MO-1939-706 MNHN-ZM-2009-288 MNHN-ZM-MO-1890-193 MNHN-ZM-MO-1961-114 MNHN-ZM-MO-1884-1428 Procolobus KUPRI 10053#388 KUPRI 10057#392 KUPRI 10055#390 KUPRI 10056#391 KUPRI 10052#387 KUPRI 10043#378 KUPRI 10045#380 KUPRI 10028#363 KUPRI 10029#364 KUPRI 10030#365 KUPRI 10051#386 KUPRI 10058#393 KUPRI 10060#395 KUPRI 10039#374 KUPRI 10038#373 KUPRI 10042#377 KUPRI 10041#376 KUPRI 10035#370 KUPRI 10011#346 KUPRI 10013#348 Corpus depth analysis Colobus MNHN-ZM-MO-1970-60 MNHN-ZM-MO-1970-59 MNHN-ZM-MO-1890-194 MNHN-ZM-MO-1893-5 MNHN-ZM-MO-1962-123 MNHN-ZM-MO-1897-1505 MNHN-ZM-MO-1970-62 RMCA 8107 RMCA 9392 RMCA 9409 RMCA 9673 RMCA 9675 RMCA 10547 RMCA 11475 RMCA 13600 RMCA 22408 RMCA 22410 RMCA 28554 RMCA 28561 RMCA 28107 RMCA 28110 RMCA 28106 RMCA 28789 RMCA 28790 RMCA 33621 RMCA 987 RMCA 1041 RMCA 1241 RMCA 1231 RMCA 1838 RMCA 1839 RMCA 2221 RMCA 6000 RMCA 8317 RMCA 8319 RMCA 8352 RMCA 8528 RMCA 8609 RMCA 10226 RMCA 37593 RMCA 8310 RMCA 14671 RMCA 19062 RMCA 20437 RMCA 28398 RMCA 37590 RMCA M6 RMCA 9912 RMCA 2147 RMCA 2202 MNHN-ZM-MO-1969-389 MNHN-ZM-MO-1969-379 MNHN-ZM-MO-1969-382 MNHN-ZM-MO-1969-384 MNHN-ZM-MO-1969-380 MNHN-ZM-MO-1962-1398 MNHN-ZM-MO-1969-387 MNHN-ZM-MO-1969-386 MNHN-ZM-MO-1969-383 MNHN-ZM-MO-1969-378 MNHN-ZM-MO-1902-1318 MNHN-ZM-AC-1983-37 MNHN-ZM-MO-1908-191 MNHN-ZM-MO-1972-350 MNHN-ZM-MO-1972-348 MNHN-ZM-AC-1897-132 RMCA 2800 MNHN-ZM-MO-1964-1517 MNHN-ZM-MO-1892-1010 MNHN-ZM-MO-1904-1960 MNHN-ZM-MO-1963-294 MNHN-ZM-MO-1892-1009 MNHN-ZM-MO-1892-1011 MNHN-ZM-MO-1962-986 MNHN-ZM-2009-345 MNHN-ZM-MO-1970-396 MNHN-ZM-MO-1886-108 MNHN-ZM-MO-1964-1514 MNHN-ZM-MO-1974-133 MNHN-ZM-MO-1892-1007 RMCA 2884 RMCA 3320 RMCA 3401 RMCA 3412 RMCA 5480 RMCA 5896 RMCA 8404 RMCA 8411 RMCA 8415 RMCA 8417 RMCA 8418 RMCA 8443 RMCA 11164 RMCA M2 MNHN-ZM-AC-1954-54 MNHN-ZM-MO-1939-702 MNHN-ZM-2009-358 MNHN-ZM-MO-1970-450 MNHN-ZM-MO-1962-2102 MNHN-ZM-2009-346 MNHN-ZM-MO-1939-701 MNHN-ZM-2009-308 MNHN-ZM-MO-1961-411 MNHN-ZM-MO-1958-710 MNHN-ZM-MO-1992-320 MNHN-ZM-MO-1961-1017 MNHN-ZM-2009-350 MNHN-ZM-2009-312 SNSB-ZSM 1914 MNHN 19641530 MNHN-ZM-AC-1983-97 MNHN-ZM-MO-1885-891 MNHN-ZM-AC-1983-98 MNHN-ZM-2009-359 MNHN-ZM-MO-1962-1390 MNHN-ZM-MO-1908-43 Piliocolobus MNHN-ZM-MO-1895-9 MNHN-ZM-MO-1939-705 MNHN-ZM-MO-1908-45BIS MNHN-ZM-MO-1962-1195 MNHN-ZM-MO-1939-706 MNHN-ZM-2009-288 MNHN-ZM-MO-1908-421 MNHN-ZM-MO-1905-420 MNHN-ZM-MO-1890-193 RMCA 346 RMCA 348 RMCA 6011 MNHN-ZM-MO-1884-1428 RMCA 0001 RMCA 0003 RMCA 0004 RMCA 0005 RMCA 0011 RMCA 0012 RMCA 0013 RMCA 0015 RMCA 0017 RMCA 0018 RMCA 0019 RMCA 0022 RMCA 0024 RMCA 18042 RMCA 10408 RMCA 11172 RMCA 11822 RMCA 12340 RMCA 20399 RMCA 20400 RMCA 0004 RMCA 0007 RMCA 0008 RMCA 0009 RMCA 0011 RMCA 0015 RMCA 0019 RMCA 0020 RMCA 0024 RMCA 0032 RMCA 0033 RMCA 0037 RMCA 0045 MNHN-ZM-MO-1961-114 RMCA 22108 RMCA 26445 RMCA 9516 RMCA 9517 RMCA 9520 RMCA 10288 RMCA 10291 RMCA 10549 RMCA 10553 RMCA 10604 RMCA 10605 RMCA 12789 RMCA 12790 RMCA 13339 Procolobus KUPRI 354 KUPRI 372 KUPRI 373 KUPRI 374 KUPRI 375 KUPRI 377 KUPRI 378 KUPRI 381 KUPRI 383 KUPRI 388 KUPRI 390 KUPRI 391 KUPRI 392 KUPRI 393 KUPRI 394 KUPRI 396 KUPRI 454 USNM 477327 Corpus shape analysis Colobus KAS 177 RMCA 5643 RMCA 8064 RMCA 8130 RMCA 9912 SBU RS 8110 SBU RS 8114 SBU RS 814 SBU RS 833 SBU RS 835 AMNH 52209 AMNH 52210 AMNH 52213 AMNH 52215 AMNH 52223 AMNH 52237 AMNH 52238 AMNH 52241 AMNH 52248 KUPRI 11372 KUPRI 8673 OM 3042 OM 3044 OM 3045 OM 3046 OM 3053 OM 3059 OM 3079 OM 3085 OM 3097 OM 3125 KUPRI 28851 KUPRI 28852 KUPRI 28853 KUPRI 28856 KUPRI 28857 KUPRI 28859 KUPRI 31491 KUPRI 35126 KUPRI 1046 KUPRI 1053 KUPRI 1067 KUPRI 1071 KUPRI 1078 KUPRI 1085 KUPRI 1086 KUPRI 1087 KUPRI 1103 KUPRI 1105 KUPRI 1115 KUPRI 1118 KUPRI 1121 KUPRI 1127 KUPRI 1135 KUPRI 1136 KUPRI 1137 MCZ 22624 MCZ 22629 RMCA 28850 RMCA 35124 RMCA 7313M234 RMCA 8107M100 RMCA 8107M101 RMCA 8107M102 RMCA 8107M103 RMCA 8107M105 RMCA 8107M106 RMCA 8107M107 RMCA 8107M108 RMCA 8107M110 RMCA 8107M114 RMCA 8107M118 RMCA 8107M120 RMCA 8107M126 RMCA 8107M127 RMCA 8107M128 RMCA 8107M129 RMCA 8107M134 RMCA 8107M144 RMCA 8107M155 RMCA 8107M156 RMCA 8107M159 RMCA 8107M164 RMCA 8107M166 RMCA 8107M171 RMCA 8107M172 RMCA 8107M87 RMCA 8107M88 RMCA 8107M89 RMCA 8107M90 RMCA 8107M91 RMCA 8107M92 RMCA 8107M93 RMCA 8107M94 RMCA 8107M95 RMCA 8107M96 RMCA 8107M99 RMCA 84037M38 USNM 477321 USNM 481784 USNM 481785 USNM 481786 SNSB-ZSM 1914263 Piliocolobus AMNH ED5 KAS 175 KUPRI 396 KUPRI 441 KUPRI 447 KUPRI 454 KUPRI 456 KUPRI 458 KUPRI 462 KUPRI 464 KUPRI 465 KUPRI 477 KUPRI 4786 KUPRI 4787 KUPRI 4791 KUPRI 490 KUPRI 500 KUPRI 507 KUPRI 518 KUPRI 520 KUPRI 522 KUPRI 524 MCZ 24080 MCZ 24775 MCZ 25627 MCZ 25810 MCZ 26552 MCZ 26553 MCZ 27108 MCZ 31939 RMCA 8107M97 RMCA 8107M98 RMCA 91069 USNM 378673 USNM 378674 USNM 481792 USNM 481795 USNM 452546 OM 4356 KAS 176 KAS 178 USNM 452644 Procolobus AMNH 89439 KUPRI 354 KUPRI 357 KUPRI 372 KUPRI 373 KUPRI 374 KUPRI 375 KUPRI 377 KUPRI 378 KUPRI 379 KUPRI 379 2 KUPRI 381 KUPRI 383 KUPRI 388 KUPRI 390 KUPRI 391 KUPRI 392 KUPRI 393 KUPRI 394 KUPRI 395 RMCA 02M66 RMCA 02M69 RMCA 84073 M176 USNM 477327 USNM 477331 Symphysis analysis Colobus MCZ 21151 MCZ 22624 MCZ 22626 MCZ 22629 MCZ 463680 RMCA 7313M234 RMCA 28851 RMCA 28852 RMCA 28853 RMCA 28856 RMCA 28857 RMCA 28859 RMCA 31491 RMCA 35126 KUPRI 1046 KUPRI 1067 KUPRI 1071 KUPRI 1085 KUPRI 1087 KUPRI 1103 KUPRI 1105 KUPRI 1121 KUPRI 1127 KUPRI 1053 KUPRI 1078 KUPRI 1086 KUPRI 1115 KUPRI 1118 KUPRI 1135 KUPRI 1136 KUPRI 1137 RMCA 5643 RMCA 8064 RMCA 8130 RMCA 9912 SBU-RS81-4 SBU-RS83-3 SBU-RS83-5 SBU-RS81-10 SBU-RS81-14 KAS 177 AMNH 52209 AMNH 52210 AMNH 52223 AMNH 52237 AMNH 52238 AMNH 52241 AMNH 52245 OM 3042 OM 3044 OM 3045 OM 3085 OM 3097 OM 3053 OM 3125 KUPRI 11372 KUPRI 8673 SNSB-ZSM 1914263 Piliocolobus AMNH ED5 MCZ 24080 MCZ 24775 MCZ 25627 MCZ 25810 MCZ 26552 MCZ 26553 MCZ 27108 RMCA 91-060-M-0069 USNM 378673 USNM 378674 USNM 481792 USNM 481795 KAS175 KUPRI 454 KUPRI 456 KUPRI 458 KUPRI 464 KUPRI 500 KUPRI 507 KUPRI 518 KUPRI 520 KUPRI 522 KUPRI 524 KUPRI 396 KUPRI 441 KUPRI 447 KUPRI 462 KUPRI 465 KUPRI 477 KUPRI 490 KUPRI 4786 KUPRI 4787 KUPRI 4791 USNM 452646 USNM 452644 OM 3046 KAS 178 Procolobus AMNH 89438 AMNH 89439 RMCA86-002-M-0066 RMCA86-002-M-0069 RMCA84-037-M-0176 USNM 477327 KUPRI 354 KUPRI 378 KUPRI 381 KUPRI 383 KUPRI 388 KUPRI 390 KUPRI 391 KUPRI 392 KUPRI 372 KUPRI 373 KUPRI 374 KUPRI 375 KUPRI 379 KUPRI 377 KUPRI 393 KUPRI 394 KUPRI 395

Appendix 2 Appendix 2 — Raw data and result of the replicability study of the morphometric symphyseal protocol. 10.5281/zenodo.21463104 SL Diff (Abs-Mean val) (Mean error- Mean)% STT Diff (Abs-Mean val) (Mean error-Mean)% ITT Diff (Abs–Mean val) (Mean error–Mean)% PL Diff (Abs–Mean val) (Mean error–Mean)% SA Diff (Abs–Mean val) (Mean error–Mean)% Nasalis ‌larvatus ‌AMNH 103365 1 20.3 0.047 7.3 0.01 7.64 0.022 11.43 0.083 55.81 0.595 2 19.87 0.383 7.29 0 7.67 0.052 10.98 0.533 56.73 1.515 3 19.85 0.403 7.29 0 7.62 0.002 11.19 0.323 56.12 0.905 4 20.43 0.177 7.28 0.01 7.59 0.028 11.77 0.257 55.23 0.015 5 19.83 0.423 7.26 0.03 7.75 0.132 11.14 0.373 56.46 1.245 6 21.24 0.987 7.32 0.03 7.44 0.178 12.57 1.057 50.94 4.275 Mean 20.25 0.403 1.991441738 7.29 0.013 0.182899 7.618 0.069 0.904251 11.513 0.438 3.802355 55.215 1.425 2.580820429 Colobus ‌satanas “ ‌MNHN 1914 263” 1 22.25 0.223 9.86 0.327 5.14 0.027 12.9 0.24 47.09 0.403 2 21.82 0.207 9.94 0.407 5.18 0.013 12.39 0.27 48.54 1.853 3 21.71 0.317 9.97 0.437 5.24 0.073 12.13 0.53 48.75 2.063 4 22.03 0.003 9.89 0.357 5.13 0.037 12.77 0.11 46.86 0.173 5 21.69 0.337 9.93 0.397 5.19 0.023 12.19 0.47 48.3 1.613 6 22.66 0.633 9.85 0.317 5.12 0.047 13.58 0.92 40.58 6.107 Mean 22.03 0.287 1.301452785 9.533 0.374 3.921275 5.167 0.037 0.709677 12.66 0.423 3.343865 46.687 2.362 5.059260317 Piliocolobus ‌badius ‌KUPRI 462 1 28.57 0.022 9.68 0.047 9.01 0.027 14.12 0.172 45.69 0.692 2 28.71 0.118 9.77 0.043 9.04 0.057 14.18 0.112 46.13 1.132 3 29.01 0.418 9.73 0.003 9 0.017 14.7 0.408 45.33 0.332 4 28.48 0.112 9.77 0.043 8.98 0.003 14.01 0.282 47.54 2.542 5 28.68 0.088 9.72 0.007 8.94 0.043 14.26 0.032 45.82 0.822 6 28.1 0.492 9.69 0.037 8.93 0.053 14.48 0.188 39.48 5.518 Mean 28.59 0.208 0.728650539 9.727 0.03 0.30843 8.983 0.033 0.371058 14.292 0.199 1.391642 44.998 1.839 4.087805721 Procolobus ‌verus ‌KUPRI 388 1 19.93 0.202 6.79 0.09 4.67 0.032 10.28 0.322 50.4 1.667 2 20.27 0.138 6.8 0.1 4.7 0.002 10.7 0.0982 48.02 0.713 3 20.07 0.062 6.75 0.05 4.69 0.012 10.46 0.147 50.56 1.827 4 20.12 0.012 6.78 0.08 4.65 0.052 10.57 0.037 49.81 1.077 5 20.18 0.048 6.78 0.08 4.82 0.118 10.58 0.022 48.27 0.463 6 20.22 0.088 6.7 0 4.68 0.022 11.02 0.418 45.34 3.393 Mean 20.13 0.092 0.455335707 6.767 0.067 0.985222 4.702 0.039 0.838946 10.602 0.172 1.624483 48.733 1.523 3.125854993

Appendix 3

Biplots of PC scores of the replicability study of the corpus shape 2D GM analysis.

Appendix 4

Distribution of coefficient of variation of extant Colobini species compared with that of the Asbole sample for PC scores of corpus shape.

Appendix 5

Distribution of coefficient of variation of extant Colobini species compared with that of the Asbole sample for PC scores of corpus depth at mid-M1.

Fig. 1. — A, Geographic distribution of extant Colobus ‌Illiger, 1811, Piliocolobus ‌Rochebrune, 1877 and Procolobus ‌Rochebrune, 1877; B, chronocladogram of extant Colobini with uncertainty in divergence dates modeled as a gray rectangle; C, chronostratigraphic frame of fossil Colobini (upper panel) along with 3D generated surfaces of Asbole and Koobi Fora mandibles discussed in the text (central panel) and geographic locations of fossiliferous deposits discussed in the text along the geographic range of extant Colobini (right panel). Scale bars: B, 20 mm; C, 10 mm. Illustration and photos: Laurent Pallas.10.5281/zenodo.21452006

Fig. 2. — Morphometric protocols for quantifying: A, dental dimensions; B, symphyseal dimensions; C, shape of the corpus. Illustration and photos: Laurent Pallas.10.5281/zenodo.21452010

Fig. 3. — A, Boxplots of the relative canine elongation of extant and fossil colobines; B, boxplots of the P4 area/M1 area ratio of extant and fossil colobines; C, boxplots of the relative length of the M3 of extant and fossil colobines. All boxplots with median (black bar), outliers (black dot), first quartile and third quartile. Illustration: Laurent Pallas.10.5281/zenodo.21463013

Fig. 4. — Boxplots of PC1 scores on the left panel and PC2 scores on the right panel of the PCA established on dental dimensions. A barplot illustrating the relative contribution of the first four PCs, in %, to the total variance is given on the central panel, along with a correlation circle for the PC1-PC2 biplot. Illustrations: Laurent Pallas.10.5281/zenodo.21452012

Fig. 5. — Biplot of the LD1-LD2 scores of the LDA established on PC scores of the dental dimensions. Illustration: Laurent Pallas.10.5281/zenodo.21479297

Fig. 6. — A, Boxplots of the breadth differential of the symphyseal tori of extant and fossil colobines; B, boxplots of the symphyseal angulation of extant and fossil colobines. All boxplots with median (black bar), outliers (black dot), first quartile and third quartile. Outlines of the symphysis of the Asbole colobines are given in the central panel. Illustrations: Laurent Pallas.10.5281/zenodo.21463015

Fig. 7. — Boxplots of PC1 scores on the upper panel and PC2 scores on the lower panel of the PCA established on symphyseal dimensions. A barplot illustrating the relative contribution of the first four PCs, in %, to the total variance is given on the central panel, along with a correlation circle for the PC1-PC2 biplot. Illustrations: Laurent Pallas.10.5281/zenodo.21463017

Fig. 8. — Biplot of the LD1-LD2 scores of the LDA established on PC scores of the symphyseal dimensions. Illustrations: Laurent Pallas.10.5281/zenodo.21452014

Fig. 9. — Biplot of the PC1-PC2 scores of the PCA established on aligned Procrustes coordinates of the corpus on the left panel. Shape changes associated with PC scores are given at the extrema of each PC axis. On the left panel, a biplot of the LD1-LD2 scores of the LDA established on PC scores of the corpus shape is given. On the lower panel, the mean shape of extant Colobini is given next to the corpus outline of fossil colobines discussed in the text. Illustrations: Laurent Pallas.10.5281/zenodo.21452016

Fig. 10. — Boxplots of PC1 scores on the upper panel and PC2 scores on the lower panel of corpus shape per sex and genus. All boxplots with median (black bar), outliers (black dot), first quartile and third quartile. Illustrations: Laurent Pallas.10.5281/zenodo.21452018

Fig. 11. — A, Dendrogram showing the results of the hierarchical clustering for the analysis taking into account Asbole specimens as a single OTU (upper dendrogram), taking into account dental, symphyseal, and corpus PCA scores of extant Colobini and the fossil ASB-232-6 and ASB-100-1 (middle dendrogram), and taking into account dental and corpus PCA scores of extant Colobini and the fossil ASB-232-6, ASB-100-1, and ASB-80; B, photographs and 3D generated surfaces of ASB-232-6 and ASB-100-1. ASB-232-6 is compared with Piliocolobus ‌Rochebrune, 1877 and ASB-100-1 is compared with Colobus ‌Illiger, 1811 on four traits: 1, symphyseal outlines of fossil specimens compared with Piliocolobus and Colobus; 2, outlines of P4 and M1 of fossil specimens compared with Piliocolobus and Colobus; 3, corpus outlines of fossil specimens compared with Piliocolobus and Colobus; 4, outlines of M2 and M3 of fossil specimens compared with Piliocolobus and Colobus. Scale bars: 10 mm. Illustration and photos: Laurent Pallas.10.5281/zenodo.21452020

Appendix 3. — Biplots of PC scores of the replicability study of the corpus shape 2D GM analysis. 10.5281/zenodo.21479303

Appendix 4. — Distribution of coefficient of variation of extant Colobini species compared with that of the Asbole sample for PC scores of corpus shape. 10.5281/zenodo.21479305

Appendix 5. — Distribution of coefficient of variation of extant Colobini species compared with that of the Asbole sample for PC scores of corpus depth at mid-M1. 10.5281/zenodo.21479307

Table 1. — List of the R package and associated functions in this article. 10.5281/zenodo.21463073 Analyses Packages Functions References GPA geomorph gpagen Adams et al. 2024 PCA on Procrustes coordinates (corpus) geomorph gm.prcomp – PCA on linear dimensions (dental and symphysis) stats prcomp R Core Team 2024 ANOVA stats aov – Kruskal-Wallis stats kruskal.test – MANOVA stats manova – Shapiro-Wilk stats shapiro.test – Bartlett stats bartlett.test – Tukey HSD stats TukeyHSD – Hierarchical Clustering stats hclust – PERMANOVA vegan adonis Oksanen et al. 2001 Multivariate Shapiro-Wilk rstatix mshapiro_test Kassambara 2023 Box M rstatix box_m – Dunn FSA dunnTest Ogle et al. 2023 rptR rptR rptGaussian Stoffel et al. 2017 LDA MASS lda Venables & Ripley 1997 Table 2. — Formulas of the dental ratios. Used formula: C1 MD/(M2MBL × M2DBL). 10.5281/zenodo.21463077 Descriptive statistics Relative canine elongation Colobus 0.18 (0.02) Piliocolobus 0.21 (0.04) Procolobus 0.27 (0.05) Asbole 0.20 (0.04) P4 area/M1 area Colobus 0.70 (0.06) Piliocolobus 0.64 (0.05) Procolobus 0.61 (0.06) Asbole 0.65 (0.05) Relative length of M3 Colobus 0.22 (0.02) Piliocolobus 0.27 (0.02) Procolobus 0.32 (0.03) Asbole 0.26 (0.03) Table 3. — Results of the statistical tests of the dental and symphyseal ratios. 10.5281/zenodo.21463081 Normality (Shapiro-Wilk) Homoscedasticity (Bartlett) Model (ANOVA or Kruskal-Wallis) Post-Hoc (Tukey HSD or Dunn) p. val. Relative elongation of C1 W = 1, p. val. < 0.05 K2 = 33, p. val. < 0.01 Kruskal-Wallis Colobus-Piliocolobus < 0.001 X2 = 50, Colobus-Procolobus < 0.001 p. val. < 0.01 Piliocolobus-Procolobus < 0.01 P4 area/M1 area W = 1, p. val. = 0.4 K2 = 0.8, p. val. = 0.8 ANOVA Colobus-Piliocolobus < 0.001 F = 25.9, Colobus-Procolobus < 0.001 p. val. < 0.001 Piliocolobus-Procolobus = 0.24 Relative length of M3 W = 1, p. val. = 0.7 K2 = 6, p. val. = 0.06 ANOVA Colobus-Piliocolobus = 0.06 F = 171, Colobus-Procolobus = 0.11 p. val. < 0.001 Piliocolobus-Procolobus = 0.06 Symphyseal tori breadth differential W = 10.9, p. val. < 0.001 K2 = 16, p. val. < 0.001 Kruskal-Wallis Colobus-Piliocolobus < 0.001 X2 = 63, Colobus-Procolobus < 0.001 p. val. < 0.01 Piliocolobus-Procolobus = 0.24 Symphyseal angulation W = 1, p. val. = 0.3 K2 = 5, p. val. = 0.07 ANOVA Colobus-Piliocolobus = 0.001 F = 10, Colobus-Procolobus = 0.392 p. val. < 0.001 Piliocolobus-Procolobus < 0.001 Table 4. — Results of the statistical tests of the dental and symphyseal PCA. 10.5281/zenodo.21463085 Multivariate Normality (Shapiro-Wilk multivariate extension) Homogeneity of covariance Model (MANOVA or Kruskal-Wallis) Post-Hoc (Tukey HSD or Dunn) p. val. PCA Scores Dental p. val. = 0.0013 p. val. = 0.023 MANOVA PC1 NA – F val. = 0.47, p. val. = 0.63 – – – – MANOVA PC2 Piliocolobus-Colobus < 0.001 F val. = 38.6, p. val. < 0.001 Piliocolobus-Procolobus < 0.05 Procolobus-Colobus < 0.001 – – MANOVA PC3 Piliocolobus-Colobus – F val. = 14.8, p. val. < 0.001 Piliocolobus-Procolobus < 0.05 Procolobus-Colobus < 0.001 PCA Scores Symphysis p. val. < 0.001 p. val. = 0.07 PERMANOVA PC1 Piliocolobus-Colobus < 0.001 X2 = 68, p. val. < 0.001 Piliocolobus-Procolobus < 0.05 Procolobus-Colobus < 0.001 – – PERMANOVA PC2 Piliocolobus-Colobus < 0.001 X2 = 61, p. val. < 0.001 Piliocolobus-Procolobus < 0.001 Procolobus-Colobus < 0.01 – – PERMANOVA PC3 NA – X2 = 0.1, p. val. = 0.7 – – Table 5. — Posterior probabilities of the Asbole specimens belonging to each extant Colobini genera, based on dental PCA scores. 10.5281/zenodo.21463087 Specimen Taxon Colobus Piliocolobus Procolobus ASB-80 Colobus sp. indet. 38% 58% 4% ASB-100-1 67% 14% 33% ASB-232-6 35% 60% 5% Table 6. — Descriptive statistics, per genus, for the symphyseal ratio and raw symphyseal measurement with standard deviation given in brackets. 10.5281/zenodo.21463089 Formula Descriptive statistics Torus transverse breadth differential STT breadth/ ITT breadth Colobus 1.61 (0.25) Piliocolobus 1.14 (0.14) Procolobus 1.25 (0.27) Asbole 1.30 (0.05) Symphyseal angulation Raw data Colobus 46.88 (3.66) Piliocolobus 44.06 (3.15) Procolobus 48.10 (4.82) Asbole 49.75 (3.13) Table 7. — Posterior probabilities of the Asbole specimens belonging to each extant Colobini genera, based on symphyseal PCA scores. 10.5281/zenodo.21463091 Specimen Taxon Colobus Piliocolobus Procolobus ASB-232-6 Colobus sp. indet. 31% 51% 18% ASB-215-2 51% 41% 8% ASB-94-1 53% 46% 1% ASB-100-1 58% 35% 7% Table 8. — Results of the statistical tests of the corpus shape PCA Scores Corpus (PC1+PC2+PC3 ~ Sex*Genera). 10.5281/zenodo.21463095 Multivariate Normality (Shapiro-Wilk multivariate extension) Homogeneity of covariance Model (MANOVA or Kruskal-Wallis) Post-Hoc (Tukey HSD or Dunn) p. val. p. val. = 0.06 p. val. = 0.7 MANOVA PC1 – – Sex F val. = 3.38, p. val. = 0.07 – – Genera F val. = 8.62, p. val. < 0.001 Piliocolobus-Colobus < 0.001 Procolobus-Colobus = 1 Procolobus-Piliocolobus < 0.01 Sex*Genera F val. = 2.67, p. val. = 0.07 F Piliocolobus-F Colobus = 0.03 F Piliocolobus-M Colobus = 0.01 M Procolobus-F Piliocolobus < 0.01 M Procolobus-M Piliocolobus = 0.01 MANOVA PC2 – – Sex F val. = 28.19, p. val. < 0.001 – – Genera F val. = 19.60, p. val. < 0.001 Piliocolobus-Colobus < 0.001 Procolobus-Colobus < 0.001 Procolobus-Piliocolobus = 0.96 Sex*Genera F val. = 0.67, p. val. = 0.5 M Colobus-F Colobus < 0.001 M Piliocolobus-F Piliocolobus = 0.7 M Procolobus-F Procolobus = 0.09 F Piliocolobus-F Colobus < 0.001 M Piliocolobus-F Colobus < 0.001 M Procolobus-F Colobus < 0.001 M Piliocolobus-M Colobus < 0.05 M Procolobus-M Colobus < 0.05 MANOVA PC3 – – Sex F val. = 0, p. val. = 1 – – Genera F val. = 1.62, p. val. = 0.20 – – Sex*Genera F val. = 2.67, p. val. = 0.13 – – Table 9. — Posterior probabilities of the Asbole specimens belonging to each extant Colobini genera based on corpus shape PCA scores. 10.5281/zenodo.21463097 Colobus Piliocolobus Procolobus KNM-ER 6025 Colobus ‌freedmani 25% 50% 24% ASB-100-1 Colobus sp. indet. 80% 7% 13% ASB-232-6 53% 23% 24% ASB-25-6 36% 42% 21% ASB-254 77% 9% 14% ASB-259 78% 11% 10% ASB-80 59% 13% 28% Table 10. — Summary table of the phenetic affinities of the considered Koobi Fora and Asbole colobines given each analysis. The rationale for assigning a specimen to a particular taxon is based on the absolute difference between the quantitative value of the specimen to the mean value of Colobus or Piliocolobus/Procolobus. Abbreviations: C, Colobus-like; P, Piliocolobus-like. 10.5281/zenodo.21463100 KNM-ER 3065 ASB-100-1 ASB-232-6 ASB-25-6 ASB-254 ASB-259 ASB-80 ASB-215-2 ASB-94-1 Dental Relative lower canine elongation NA C P NA P NA C C NA P4 area/M1 area P C P P P C P P NA Relative length of the M3 NA C P P NA P P NA NA LDA on dental dimensions NA C P NA NA NA P NA NA Symphyseal Breadth differential of the symphyseal tori NA P P NA NA NA NA P P LDA on symphyseal dimensions NA C P NA NA NA NA C C Corpus LDA on corpus shape P C C P C C C NA NA