Mammalian Molar Complexity Follows Simple, Predictable Patterns

Mammalian molar complexity follows simple, predictable patterns

Keegan R. Seliga,1, Waqqas Khalida, and Mary T. Silcoxa

aDepartment of Anthropology, University of Toronto Scarborough, Toronto, ON M1C 1A4, Canada

Edited by Nils Chr. Stenseth, University of Oslo, Oslo, Norway, and approved November 19, 2020 (received for review May 4, 2020) Identifying developmental explanations for the evolution of com- been demonstrated that there are strong genetic controls that plex structures like mammalian molars is fundamental to studying account for the developmental cascade, with research linking the phenotypic variation. Previous study showed that a “morphoge- ICM and the sonic hedgehog (Shh) gene, among others (10). It netic gradient” of molar proportions was explained by a balance should be noted that there are cases where the ICM does between inhibiting/activating activity from earlier developing mo- not seem to account for molar size covariation. For example, lars, termed the inhibitory cascade model (ICM). Although this Roseman and Delezene (11) found that gorillas do not meet any model provides an explanation for variation in molar proportions, of the predictions of the ICM, and Carter and Worthington (12) what remains poorly understood is if molar shape, or specifically demonstrate that some hominoids and cercopithecins do not complexity (i.e., the number of cusps, crests), can be explained by meet all of the predictions either. However, Evans et al. (2) also the same developmental model. Here, we show that molar com- analyzed the ICM in apes and hominids and considered both the plexity conforms to the ICM, following a linear, morphogenetic deciduous premolars and a reversal of the inhibitory cascade gradient along the molar row. Moreover, differing levels of inhib- patterns in their analysis. By considering these two additional iting/activating activity produce contrasting patterns of molar factors, Evans et al. (2) demonstrated that hominoids do actually complexity depending on diet. This study corroborates a model meet the expectations of the ICM. for the evolution of molar complexity that is developmentally sim- The ICM provides a model to explain molar size covariation in ple, where only small-scale developmental changes need to occur many taxa. What it does not consider, however, is the potential to produce change across the entire molar row, with this process for adaptation in patterns of molar complexity. Molar shape, and being mediated by an animal’s ecology. The ICM therefore pro-

particularly the number of features present on the occlusal sur- EVOLUTION vides a developmental framework for explaining variation in mo- face such as crests, cusps, and crenules (i.e., occlusal complexity) lar complexity and a means for testing developmental hypotheses covaries with feeding behavior (13–20). Kavanagh et al. (1) ob- in the broader context of mammalian evolution. served some apparent morphological differences in the dentition of murines: Taxa that had more equally sized molars were gen- Euarchonta | three-dimensional orientation patch count rotated | erally herbivorous and had relatively more complex teeth, occlusal morphology whereas faunivorous taxa had simple teeth that became relatively reduced in size moving distally. Following on this result, we ammalian molars develop from mesial to distal, with the provide further, empirical evidence that that differences in levels Mdental lamina of the first molar (M1) extending posteri- of inhibition/activation in the molar row in other mammals can orly, giving rise to the second molar (M2). This means that in- be explained by differences in diet (1). There is some indication hibition/activation of earlier structures has a cumulative effect on later developing teeth, as more developmental events must take Significance place for later structures to develop (1–3). A relationship has been identified between the size (surface area) of the earlier Recent studies have sought simple developmental models to developing molars and that of the later developing molars, be- explain variation in complex structures like mammalian teeth. lieved to be a product of the balance between inhibitors (sig- The inhibitory cascade model (ICM) provides an explanation for naling molecules that impede the development of subsequent the “morphogenetic gradient” of molar size, where a balance teeth) and activators (signaling molecules that promote the de- between signaling molecules in developing teeth produces velopment of subsequent teeth) in the earlier developing struc- predictable, linear proportions across the molars. However, it tures (1). The M1 acts as an inhibitor on the M2, reducing the is unknown if molar form (specifically complexity) can be size of the later developing tooth in vivo. When the developing explained by the ICM. We demonstrate that molar complexity M2 is cut from the posterior end of the M1 in vitro earlier than follows a linear, predictable gradient along the molar row. these structures would naturally separate, inhibition from M1 is Moreover, whether the pattern is of increasing or decreasing reduced, thus M2 and subsequently the third molar (M3) de- complexity moving from front to back is dependent on the velop to be relatively larger (1). Kavanagh et al. (1) recognized animal’s diet. This study provides a developmental framework this mechanism as the inhibitory cascade model (ICM), wherein for explaining patterns of morphological complexity in molars a balance between inhibitors/activators controls molar propor- and for testing developmental and ecological hypotheses tions. Kavanagh et al. (1) argued that if molars follow a mor- in mammals. phogenetic gradient, then a ratio of third molar size to first molar size regressed against a ratio of second molar size to first molar Author contributions: K.R.S. and M.T.S. designed research; K.R.S. performed research; size should produce a positive, linear regression with a slope of K.R.S. and W.K. analyzed data; K.R.S. wrote the paper; W.K and M.T.S. edited the man- 2.0 and an intercept of −1.0. Similarly, Evans et al. (2) suggested uscript; and M.T.S. oversaw the project. that if the size of the M2 is the average of the M1+M3, or the The authors declare no competing interest. relationship between the molars is linear as posited by the ICM, This article is a PNAS Direct Submission. then the relative size of the molars conforms to the ICM as well. Published under the PNAS license. As a whole, the ICM provides both a mechanism and a predictive 1To whom correspondence may be addressed. Email: [email protected]. tool for explaining the patterning of molar proportions. This This article contains supporting information online at https://www.pnas.org/lookup/suppl/ model has been found to be a strong predictor of molar pro- doi:10.1073/pnas.2008850118/-/DCSupplemental. portions across many mammalian clades (2, 4–9). It has also Published December 28, 2020.

PNAS 2021 Vol. 118 No. 1 e2008850118 https://doi.org/10.1073/pnas.2008850118 | 1of5 Downloaded by guest on September 26, 2021 that primates of differing diets show contrasts in the pattern of Results dental morphology across the molar row, with complexity either Analysis of 40 extant euarchontan lower molar toothrows increasing or decreasing moving distally (9), but this pattern has (M1–M3, n = 120) shows that molar complexity conforms to the not previously been examined using the ICM. In taxa where predictions of the ICM (see SI Appendix, Table S3 for raw data). higher occlusal complexity in the molar row helps in the break- For the full sample (Euarchonta), there is a linear increase in down of mechanically challenging foods, such as in folivores, we mean complexity moving distally along the molars (ordinary least predict that these taxa will experience lower levels of inhibition/ squares [OLS] regression R2 = 0.9994). Reduced major axis re- higher levels of activation, which will lead to a linear increase in gression of M3/M1 complexity regressed against M2/M1 com- complexity moving distally from M1 to M3. In taxa where greater plexity is positive and significant (P = 0.00433, R2 = 0.34094), molar complexity does not provide as great a functional advan- and the predicted values of the ICM fall within the 95% CI for tage for comminuting food, as in frugivores, we predict that in- our model (Fig. 1 and Table 1). Although molar size does follow hibition will be higher and, therefore, the distal molars will be the predictions of the ICM as well (reduced major axis [RMA] relatively less complex than the mesial molars. regression P ≥ 0.0001, R2 = 0.66097), with the predicted values During the early, patterning period of occlusal development, of the ICM falling within the 95% CI for our model (Fig. 1 and nonmitotic signaling centers present in the tooth germ, known as SI Appendix, Fig. S1 and Table 1), the relationship between enamel knots, influence the formation of the developing dental molar size is not as strongly linear as in the case of molar complexity epithelium by releasing proteins that inhibit or activate cellular (OLS regression R2 = 0.6661), particularly in the case of the fru- development (1, 9, 21–23). In multicuspate teeth, the position, givores (OLS regression R2 = 0.5459) and folivores (OLS regression development, and eventual apoptosis of the primary enamel R2 = 0.4912). Regressions also suggest that there is no relationship knots triggers the formation of the secondary enamel knots, between molar surface area and complexity (M1 RMA regression which also act as signaling centers. The secondary enamel knots P = 0.091228, R2 = 0.13601; M2 P = 0.11669, R2 = 0.11851; M3 P = appear at the sites of future-forming cusps, meaning their loca- 0.14222, R2 = 0.10541), meaning complexity is independent from tion provides the first blueprint of species-specific cuspal pat- surface area (SI Appendix,Fig.S2). The M3 is the smallest tooth in terning (24, 25). The development of further enamel knots at the each dietary guild, even when it is the most complex. Therefore, a margins of the signaling fields of the successional teeth inhibit smaller tooth does not necessarily make for a simpler tooth. This and activate development in the later forming teeth. There also suggests that although both complexity and size conform to the exists a relationship between the earlier and later developing ICM, the processes and inhibiting/activating factors that govern the molar cusps, with the positioning and proportions of earlier development of molar proportions and molar complexity are at developing cusps affecting later developing distal cusps within a least partially independent from one another. molar (3, 9). Harjunmaa et al. (26) demonstrated that dental Whereas each dietary guild follows the predictions of the ICM morphology, particularly occlusal complexity, can be influenced in being linear (Fig. 1), there is a contrast in the pattern followed dramatically by manipulating some of the same inhibitor/acti- by the frugivores and folivores, where the folivores show a linear vator signaling pathways in vitro that are known to control for increase in complexity moving distally and the frugivores show a the proportions of molar size under the ICM. Because of the linear decrease (SI Appendix,Fig.S2). Although the insectivores and apparent relationship between inhibitor/activator activity affect- omnivores also display an increase in complexity moving distally, the ing both the patterning of tooth size and molar complexity, we folivores are characterized by an M3 that is far more complex rel- propose that molar complexity also follows patterns predicted by ative to the other molars when compared to the other dietary guilds. the inhibitory cascade. Because the patterning phase of dental development is short and occurs earlier compared to the period Discussion of crown maturation when amelogenesis takes place and the final Our results suggest that the ICM not only explains the size and size of the tooth is determined (27), complexity may be a more proportion of mammalian molars, but also explains aspects of robust measure of the ICM compared to size because the later molar morphology, specifically complexity. Developmental events and longer stage of development is potentially more susceptible that affect the complexity of M1 seem to accumulate and affect to developmental and external factors. M2 and, subsequently, M3 based on varying levels of inhibition/ In several mammalian clades, the M3 is the most variable activation signaled from earlier developing structures. These tooth in the molar row in terms of morphology, size, and initi- results therefore support the notion that the evolution of molar ation (27–29) attributed to the greater number of developmental complexity is developmentally simple (3). Changes in the pat- events that must take place for the M3 to develop (1, 9) and the terning of enamel knots and their expression of inhibitor/acti- potential for environmental factors to affect the M3 during its vator signaling molecules accumulates and manifests in overall long period of development. Small amounts of variation across changes to the complexity of the molar row. It should be noted, these events combine to produce higher levels of variability in however, that in thick enameled taxa such as orangutans char- later developing structures on the M3. However, it has been acterized by crenulated enamel, later developmental events like hypothesized that the M3 should also be the least morphologi- enamel deposition may also influence occlusal complexity. cally complex tooth in the molar row in order to maintain precise Our results provide further evidence that teeth are highly occlusion, even if variability is high (26), although there are evolvable structures, where only small-scale changes in devel- known exceptions to this as in the case of many suids with very opmental processes need to occur to produce large-scale changes complex M3s (13). As such, the ICM provides a possible in molar complexity (3). This may explain why even closely re- mechanism for producing this pattern of morphology. lated taxa within different dietary guilds follow different patterns Euarchontans (primates, colugos, and treeshrews) are devel- of molar complexity in the tooth row (Fig. 2). The ICM provides opmentally and ecologically diverse, having diverged some >83 a developmental explanation for these differing patterns of million years ago (30), making them an ideal group for studying molar row complexity. For example, folivorous taxa are charac- patterns of dental morphogenesis and functional morphology terized by a prominent linear increase in complexity moving among mammals. Diet is also commonly invoked as a prime distally, whereas the frugivores show a linear decrease. Under mover for the evolution of Primates (31–34). Therefore, we ex- the ICM, a pattern of M1 < M2 < M3 indicates a weak level of amined molar complexity in a sample of euarchontans, which inhibition/high level of activation. Frugivores, characterized by a may provide a model for connecting patterns of morphology and M1 > M2 > M3 pattern (1), would exhibit a strong level of in- development in the broader context of mammalian evolution. hibition/low level of activation. As a whole, our sample shows a

2of5 | PNAS Selig et al. https://doi.org/10.1073/pnas.2008850118 Mammalian molar complexity follows simple, predictable patterns Downloaded by guest on September 26, 2021 EVOLUTION

Fig. 1. Reduced major axis regression and plot of mean values showing both molar complexity and size conforming to the expectations of the ICM. Results of the RMA regression of M2/M1 complexity against M3/M1 complexity (A) and M2/M1 surface area against M3/M1 surface area (B), with the mean values plotted for each dietary guild, conforming to the inhibitory cascade model in being linear. Bars represent one SD from the mean. (C) Plot showing the mean values for the measurement of molar complexity at each tooth position. R2 statistics from the OLS regressions following Evans et al. (2) for Euarchonta = 0.9994; frugivores = 0.3370; omnivores = 0.9890; insectivore = 0.8317; folivores = 0.9757. Note the distally increasing dental complexity among the folivores and decreasing complexity among frugivores. This implies that the folivores exhibit comparatively lower inhibition whereas the frugivores, which plot lower, experience higher inhibition from earlier developing molars. The insectivores and omnivores exhibit relatively even molar complexity across the tooth row. (D) Plot showing the mean values for the measurement of molar surface area at each tooth position. R2 statistics from the OLS regressions for Euarchonta = 0.6661; frugivores = 0.5459; omnivores = 0.9247; insectivore = 0.7005; folivores = 0.4912. “Euarchonta” includes the complete sample. Note that in the case of each dietary guild, the M3 is the smallest tooth, even in cases where it is the most morphologically complex (i.e., folivores).

linear increase in complexity moving distally. This is in violation effective breakdown of food seems to have outweighed the po- of expectations that there would be a constrained and lower level tential selective pressure for limiting complexity on the M3 to of complexity in the M3 because of its greater degree of vari- preserve precise occlusion. Our research, therefore, provides a ability (27–29), which highlights the important role that feeding potential developmental explanation for the discrepancy be- ecology plays in the development of dental complexity. If there is tween having high M3 complexity and the need for preserving selective pressure for constraining complexity of the M3 (14), precise occlusion. Future research should look further into the then the selective advantage of greater molar complexity for the question of high variability and low complexity of the M3.

Table 1. Results of the RMA regression for molar complexity and molar area following Kavanagh et al. (1) RMA regression: M2/M1-M3/M1 RMA regression: M2/M1-M3/M1 (molar complexity) (molar area)

Slope a 1.8235 1.8079 Intercept b −0.82439 −1.0594 95% bootstrapped CIs, n 1,999 1,999 Slope a (0.65941, 2.3876) (1.404, 2.2266) Intercept b (−1.4065, 0.36602) (−1.4721, -0.651) Correlation: R 0.5839 0.813 R2 0.34094 0.66097 t 3.2165 6.2443 P 0.00433 4.25e-06 Permutation P 0.0044 0.0001

Selig et al. PNAS | 3of5 Mammalian molar complexity follows simple, predictable patterns https://doi.org/10.1073/pnas.2008850118 Downloaded by guest on September 26, 2021 adaptations to frugivory may explain why some taxa deviate from the ICM. Although we found that our sample as a whole conforms to the ICM, the frugivores show low R2 values in the measurement of both surface area and complexity, consistent with the findings of Carter and Worthington (12). Again, however, it should be noted that when deciduous premolars and a reversal of the inhibitory cascade are considered, hominoids and cercopithecins do follow the expectations of the ICM (2). It is also noteworthy that both molar complexity and molar size conform to the ICM but seem to do so in different ways. For example, all dietary guilds have molars that decrease in size moving distally, however, the folivores show an increase in complexity. Only the frugivores have molars that decrease in both size and complexity moving distally. This suggests that whereas the ICM explains both molar proportions and complexity, different developmental events must take place and potentially different signaling molecules must be responsible for controlling aspects of size and aspects of morphology, pointing to a decoupling of the development of size and morphology in the molars. Overall, our results point to a developmental mechanism for explaining patterns of variation in molar morphology across taxa, particularly of differing diets. This has powerful implications for explaining patterns of phenotypic variation in that it allows you to make predictions and explain patterns in the fossil record with a genetic framework. It also provides additional means for un- derstanding how diet may be linked to the processes of development. Given that other structures such as limbs, digits, and somites also develop following predictions of the ICM (36), our results may provide a means of examining the development of complex segmented organs beyond the dentition. Methods Sample. Our sample includes the lower M1 through M3 for 40 extant euarchontan specimens representing all three orders, and a sample of 13 families and 22 species (SI Appendix, Tables S1 and S2). The mandibles were downloaded as microcomputed tomography (micro-CT) scans from Mor- Fig. 2. Representative meshes of Alouatta palliata (MCZ 5323) (A) and phoSource, an online repository for CT data (37) or they were scanned on a Ateles geoffroyi (MCZ 5354) (B) shown with their 3D-OPCR maps. Patch Nikon XT H 225 ST High Resolution X-ray CT Scanner at the Shared Materials count values are the mean values for each taxon. Note that even among two Instrumentation Facility at Duke University. Specimens with minimal to no closely related taxa, which are both within the family Atelidae, differences in wear (no exposed dentine in the occlusal basin) that had a complete tooth diet produce a contrasting pattern in the expression of complexity along the row on one side of the mouth were chosen. Each taxon within the sample molar row. The folivore shows an increase in complexity moving from M1 to was placed within one of four dietary guilds (folivory, insectivory, omnivory, M3, whereas the frugivore shows a decrease. (Scale bars = 1 mm.) and frugivory) based on the mechanical properties of the principal food- stuffs consumed (SI Appendix, Table S2 for references). Each dietary cate- gory is represented by 10 specimens, with four to seven species in each In order to efficiently process chitinous exoskeletons, insecti- dietary guild. vores require teeth with taller cusps and higher shearing crests, Topographic Analysis. We used a dental topographic method known as three- leading to greater complexity (16, 18). Therefore, it might be dimensional orientation patch count rotated (3D-OPCR) to measure molar predicted that the insectivores would show a similar pattern to surface complexity (17, 38, 39). This metric breaks up the tooth surface based the folivores, with molar complexity increasing moving distally. on the number of discrete patches present. As the mesh triangles within a Within Euarchonta, however, most “insectivores” consume at patch begin to face a new direction based on the eight cardinal directions, least some fruit (with the exception of tarsiids) (35). This likely they constitute a new patch; as such, a higher patch count reflects greater explains why the insectivorous taxa are characterized by a pattern complexity. We measured 3D-OPCR using MorphoTester (38) with the min- most like the omnivores, in which levels of complexity remain imum patch count set to five triangles. MorphoTester also produces a similar along the toothrow. It could be predicted that complexity measurement of two-dimensional surface area of the mesh, which we used as a measurement of tooth size. may increase moving distally in insectivorous taxa in mammalian clades that rely exclusively on insects. Mesh Preparation. Each micro-CT scan was prepared using Avizo 9.0 (33) Our sample also conforms to the ICM based on the mea- following conventional methods for topographic analysis (15, 16, 40–42). surement of surface area, with the inhibitory effect of M1 being First, each tooth was segmented to isolate it from adjacent teeth. Un- relatively strong among all dietary guilds. This is noteworthy as smoothed surfaces were generated for each isolated molar, which were then variation in molar proportions are either not consistent, or only cropped along the cervix to isolate the crown (16, 41). Surfaces were then partially consistent, with the ICM in some primate taxa such as simplified to 10,000 faces and smoothed 100 iterations with the lambda set hominoids and cercopithecins (11, 12). The discrepancy between at 0.6 (15). It should be noted that previous analysis of the relationship between simplification and OPCR (42) has shown that face count should be these findings and our own may be due to the lack of hominoids held constant, as opposed to the resolution of the mesh, as OPCR performs and inclusion of only a single cercopithecoid in our analysis. better if a constant face count is used. An argument could be made that However, Carter and Worthington (12) also argue that diet may smaller teeth (such as the m3) should be simplified to a lower face count play a role in which taxa adhere to the ICM, noting that because they are represented by the same number of triangles as larger

4of5 | PNAS Selig et al. https://doi.org/10.1073/pnas.2008850118 Mammalian molar complexity follows simple, predictable patterns Downloaded by guest on September 26, 2021 teeth. Not doing so effectively increases the resolution of small teeth at the Copes provided access to the Museum of Comparative Zoology data with cost of the larger teeth. However, Berthaume et al. (43) demonstrated that funding from National Science Foundation (NSF) Doctoral Dissertation there is a taxonomic and dietary effect in which the molars of different taxa Improvement Grant 0925793 and the Wenner Gren Foundation. Greg and of different dietary guilds react differently to varying levels of simpli- Watkins-Colwell provided access YPM MAM 014402, the collection of which fication. Moreover, the complexity of a tooth has a direct effect on the face was funded by NSF Grants DBI-1701769, DBI-1701714, and BCS-1552848. count for a given tooth mesh, meaning that accounting for size could have a Lauren Gonzales provided access to the scans of USNM 282339, USNM confounding effect on the morphology that this study is aiming to measure. 282761, USNM 269839, USNM 100550, USNM 200279, and USNM 100512, the Therefore, choosing a face count value based on the size of the tooth is collection of which was funded by the Louis Seymour Bazett Leakey associated with as many, if not more, assumptions as holding face count Foundation. NSF Grants BCS 1440742 and 1552848 to Doug Boyer supported constant. Because OPCR was shown to perform better under a constant face the processing and upload of these files to MorphoSource. The scanning and count, we chose to simplify all meshes to 10,000 faces. Once meshes were upload of AMNH 100843, AMNH 31265, AMNH 56530, and FMNH 141132 simplified, they were oriented occlusally and saved as .ply files. was funded by NSF Grants BCS 1317525 (to Doug Boyer and Erik Seiffert), 1552848 (to Doug Boyer) and Duke University Trinity College of Arts and Sciences. Chris Wall provided access to USNM 488052, USNM 488055, and Statistical Analyses. All statistical analyses were performed in PAST 3.26 (44). UNSM 481107 with funding from NSF Grant BCS 1304045 and a research We used RMA regression to examine if molar complexity conforms to the grant from Duke University Trinity College of Arts and Sciences. All micro-CT predictions of the ICM following Kavanagh et al. (1). We also calculated the 2 scans included in this analysis are available on MorphoSource.org, Duke Uni- OLS R statistic to test if molar complexity follows a linear pattern moving versity. We thank Tamara Kumpan, Sergi López-Torres, and the M.T.S. Lab- from M1 to M3 following Evans et al. (2). In cases where species were rep- oratory for their helpful discussions. We thank three anonymous reviewers resented by more than one specimen, we used the species mean values for for their helpful comments that substantively improved this manuscript. This those taxa in all analyses. research was supported by a grant from the University of Toronto Scarbor- ough International Research Collaboration Fund and a Natural Sciences and Data Availability. All data are available for download from MorphoSource (37). Engineering Research Council of Canada Discovery Grant (to M.T.S.), as well as a Pilot Research Grant from the Department of Anthropology and a ACKNOWLEDGMENTS. We thank the curators and collection staff who made School of Graduate Studies Research Travel Grant from the University of specimens available for upload to MorphoSource. Lynn Lucas and Lynn Toronto (to K.R.S.).

1. K. D. Kavanagh, A. R. Evans, J. Jernvall, Predicting evolutionary patterns of mam- 24. J. Jernvall, I. Thesleff, Reiterative signaling and patterning during mammalian tooth malian teeth from development. Nature 449, 427–432 (2007). morphogenesis. Mech. Dev. 92,19–29 (2000). 2. A. R. Evans et al., A simple rule governs the evolution and development of hominin 25. J. Jernvall, P. Kettunen, I. Karavanova, L. B. Martin, I. Thesleff, Evidence for the role of tooth size. Nature 530, 477–480 (2016). the enamel knot as a control center in mammalian tooth cusp formation: Non- 3. J. Jernvall, Linking development with generation of novelty in mammalian teeth. dividing cells express growth stimulating Fgf-4 gene. Int. J. Dev. Biol. 38, 463–469 EVOLUTION Proc. Natl. Acad. Sci. U.S.A. 97, 2641–2645 (2000). (1994). 4. K. Schroer, B. Wood, Modeling the dental development of fossil hominins through 26. E. Harjunmaa et al., On the difficulty of increasing dental complexity. Nature 483, the inhibitory cascade. J. Anat. 226, 150–162 (2015). 324–327 (2012). 5. M. M. Skinner, P. Gunz, The presence of accessory cusps in chimpanzee lower molars is 27. G. R. Scott, C. G. Turner II, G. C. Townsend, M. Martinón-Torres, The Anthropology of consistent with a patterning cascade model of development. J. Anat. 217, 245–253 Modern Human Teeth: Dental Morphology and its Variation in Recent and Fossil (2010). Homo sapiens (Cambridge University Press, 2018), vol. 79. 6. A. M. C. Couzens, A. R. Evans, M. M. Skinner, G. J. Prideaux, The role of inhibitory 28. P. D. Gingerich, B. H. Smith, K. Rosenberg, Allometric scaling in the dentition of dynamics in the loss and reemergence of macropodoid tooth traits. Evolution 70, primates and prediction of body weight from tooth size in fossils. Am. J. Phys. An- – 568 585 (2016). thropol. 58,81–100 (1982). 7. M. Asahara, Unique inhibitory cascade pattern of molars in canids contributing to 29. T. M. Smith et al., Dental evidence for ontogenetic differences between modern – their potential to evolutionary plasticity of diet. Ecol. Evol. 3, 278 285 (2013). humans and Neanderthals. Proc. Natl. Acad. Sci. U.S.A. 107, 20923–20928 (2010). 8. T. J. Halliday, A. Goswami, Testing the inhibitory cascade model in Mesozoic and 30. T. E. Roberts, H. C. Lanier, E. J. Sargis, L. E. Olson, Molecular phylogeny of treeshrews Cenozoic mammaliaforms. BMC Evol. Biol. 13, 79 (2013). (Mammalia: Scandentia) and the timescale of diversification in Southeast Asia. Mol. 9. J. M. Winchester, “Molar topographic shape as a system for inferring functional Phylogenet. Evol. 60, 358–372 (2011). morphology and developmental patterning in extant cercopithecoid primates,” PhD 31. R. W. Sussman, Primate origins and the evolution of angiosperms. Am. J. Primatol. 23, dissertation, State University of New York at Stony Brook, Stony Brook, NY (2016). 209–223 (1991). 10. N. Navarro, A. Murat Maga, Genetic mapping of molar size relations identifies in- 32. R. W. Sussman, P. H. Raven, Pollination by lemurs and marsupials: An archaic co- hibitory locus for third molars in mice. Heredity 121,1–11 (2018). evolutionary system. Science 200, 731–736 (1978). 11. C. C. Roseman, L. K. Delezene, The inhibitory cascade model is not a good predictor of 33. R. A. Barton, Visual specialization and brain evolution in primates. Proc. Biol. Sci. 265, molar size covariation. Evol. Biol. 46, 229–238 (2019). 1933–1937 (1998). 12. K. E. Carter, S. Worthington, The evolution of anthropoid molar proportions. BMC 34. R. W. Sussman, D. Tab Rasmussen, P. H. Raven, Rethinking primate origins again. Am. Evol. Biol. 16, 110 (2016). J. Primatol. 75,95–106 (2013). 13. A. Souron et al., Three-dimensional dental microwear texture analysis and diet in 35. J. G. Fleagle, Primate Adaptation and Evolution (Academic Press, 2013). extant Suidae (Mammalia: Cetartiodactyla). Mammalia 79, 279–291 (2015). 36. N. M. Young, B. Winslow, S. Takkellapati, K. Kavanagh, Shared rules of development 14. P. D. Gingerich, M. J. Schoeninger, Patterns of tooth size variability in the dentition of predict patterns of evolution in vertebrate segmentation. Nat. Commun. 6, 6690 primates. Am. J. Phys. Anthropol. 51, 457–465 (1979). (2015). 15. J. M. Winchester et al., Dental topography of platyrrhines and prosimians: Conver- 37. D. M. Boyer, G. F. Gunnell, S. Kaufman, T. McGeary, MorphoSource–Archiving and gence and contrasts. Am. J. Phys. Anthropol. 153,29–44 (2014). – 16. J. M. Bunn et al., Comparing Dirichlet normal surface energy of tooth crowns, a new sharing 3D digital specimen data. J. Paleontol. 22, 157 181 (2016). technique of molar shape quantification for dietary inference, with previous methods 38. J. M. Winchester, MorphoTester: An open source application for morphological to- in isolation and in combination. Am. J. Phys. Anthropol. 145, 247–261 (2011). pographic analysis. PLoS One 11, e0147649 (2016). 17. D. M. Boyer, Relief index of second mandibular molars is a correlate of diet among 39. Visualization Sciences Group, Avizo, Version 9.0 (Mercury Computer Systems, Bur- prosimian primates and other euarchontan mammals. J. Hum. Evol. 55, 1118–1137 lington, MA, 2009). (2008). 40. K. R. Selig, E. J. Sargis, M. T. Silcox, The frugivorous insectivores? Functional mor- 18. A. R. Evans, G. P. Wilson, M. Fortelius, J. Jernvall, High-level similarity of dentitions in phological analysis of molar topography for inferring diet in extant treeshrews – carnivorans and rodents. Nature 445,78–81 (2007). (Scandentia). J. Mammal. 100, 1901 1917 (2019). 19. R. F. Kay, The functional adaptations of primate molar teeth. Am. J. Phys. Anthropol. 41. K. A. Prufrock, D. M. Boyer, M. T. Silcox, The first major primate extinction: An 43, 195–216 (1975). evaluation of paleoecological dynamics of North American stem primates using a 20. W. K. Gregory, Origin and Evolution of the Human Dentition (Williams and Wilkins, homology free measure of tooth shape. Am. J. Phys. Anthropol. 159, 683–697 (2016). 1922). 42. J. D. Pampush et al., Introducing molaR: A new R package for quantitative topo- 21. Y. Kassai et al., Regulation of mammalian tooth cusp patterning by ectodin. Science graphic analysis of teeth (and other topographic surfaces). J. Mamm. Evol. 23, 309, 2067–2070 (2005). 397–412 (2016). 22. J. Jernvall, S. V. Keränen, I. Thesleff, Evolutionary modification of development in 43. M. A. Berthaume, J. Winchester, K. Kupczik, Effects of cropping, smoothing, triangle mammalian teeth: Quantifying gene expression patterns and topography. Proc. Natl. count, and mesh resolution on 6 dental topographic metrics. PLoS One 14, e0216229 Acad. Sci. U.S.A. 97, 14444–14448 (2000). (2019). 23. I. Salazar-Ciudad, J. Jernvall, A gene network model accounting for development and 44. Ø. Hammer, D. A. T. Harper, P. D. Ryan, PAST: Paleontological statistics software evolution of mammalian teeth. Proc. Natl. Acad. Sci. U.S.A. 99, 8116–8120 (2002). package for education and data analysis. Palaeontol. Electronica 4, 9 (2001).

Selig et al. PNAS | 5of5 Mammalian molar complexity follows simple, predictable patterns https://doi.org/10.1073/pnas.2008850118 Downloaded by guest on September 26, 2021