One of the major goals of evolutionary biology is to identify the mechanisms and forces responsible for phenotypic variation. The assembly of a phenotype into modules or developmental fields, which are semiautonomous units in development and function, seems to be one of the strategies to increase species' capacity to produce phenotypic variation. The concept of modular evolution has been largely used in evolutionary developmental biology (evo-devo) studies, and has helped to increase our understanding of how the evolution of morphological traits can be influenced by developmental process (Klingenberg et al., 2003; Klingenberg, 2009; Laffont et al., 2009). The concept of module has been defined in several ways, but generally it may be defined as a self-organizing morphologic trait (e.g., eye field, limb field, dental field), which differentiates in response to several inductive genetic factors. A developmental field is composed of a group of cells able to respond as a coordinated unit to discrete, localized biochemical signals leading to the development of specific morphological structures or organs. Although a module is highly integrated internally and relatively independent from other modules, as a part of a higher order hierarchical organization, they must connect and interact with other parts of the system. The fact that a module may sometimes split into submodules, as seen in the development of many serially homologous structures such as vertebrate limbs (Chiu and Hamrick, 2002), teeth (Stock, 2001), and vertebrae (Buchholtz et al., 2007), shows that these structures are not totally independent units (Klingenberg, 2010).
Teeth are an important model in the field of evolutionary developmental biology (Jernvall et al., 2000). Mammalian dentition fits within the concept of modularity. Based on discontinuous shape patterns within the dentition and independent shape changes in evolution mammalian dentition has been divided into incisor, canine, premolar, and molar fields (Butler, 1978; Dahlberg, 1945; Townsend et al., 2009). Evidences for the existence of dental fields have been supported by gene expression analysis of dental development in mice (McCollum and Sharp, 2001). The modularity of mammalian dentition has been demonstrated by analyzing families with mutations in genes that affect the pattern of dentition in humans (Line, 2001, 2003), and more recently by studies showing that the relative size and number of molar teeth can be predicted by interactions that occur exclusively among developing teeth within this field (Kavanagh et al., 2007; Renvoisé et al., 2008).
Most experimental studies that use mammalian dentition as a model to connect development and evolution have focused on variations of size and shape of molar teeth, considering this region as an isolated and independent module (Kavanagh et al., 2007; Renvoisé et al., 2008; Koh et al., 2010). It is worth mentioning that the upper dentition is formed on two distinct developmental processes (Lumsden and Buchanan, 1986). Upper incisors are formed on the primary palate (premaxilla), derived from the embryonic frontonasal process, whereas canine, premolar and molar classes develop on the maxillary process derived from the first pharyngeal arch (Nanci, 2007). Therefore, premolar and molar modules can be seen as part of a larger and hierarchically superior module (i.e., the alveolar process derived from the maxillary process, termed herein as PALATE2). Although the relative sizes of molar teeth are dependent on the effect of local factors, it is plausible that the absolute size and number of posterior teeth will also be related to more general factors such as the size and shape of the jaws and the space required by other tooth classes. Furthermore, in the present work we seek to understand to what extent the changes in the relative size of premolar and molar teeth connect with the variations in the dental formula that occurred during primate evolution.
MATERIALS AND METHODS
Scaled cranial photographs of 85 primates were obtained from the inferior view of the Mammalian Cranial Photographic Archive (Takahashi et al., 2006, http://macro.dokkyomed.ac.jp/mammal/en/mammal.html). In the inferior view, the palatine plane was set horizontally, corresponding to the occlusal plane formed by the maxillary molar and premolar tips. According to Takahashi et al. (2006), in each cranium, the orbitomeatal or palatine plane of which was placed horizontally, was photographed from six different angles (anterior, posterior, left, right, superior, and inferior) at a long distance from the camera through a telephoto or telemacro lens. The long-distance shot decreases perspective distortion that may lead to measurement errors when studying crania profiles. For the cranial images whose anatomical direction is deviated from the lens' optical axis due to technical inaccuracy, adjustments in the 3D rotations on the cranium's anatomical axes have been repeated until the setting error is minimized. No digital enhancement has been applied to captured images, which are JPEG compressed and of 6 mega pixels.” All the measurements were done by the same individual (MMR).
In this study eleven families of primates were included, one specimen of each species. Species were classified according to Groves (2001). Figure 1 in Supporting Information describes the families and species included in this article. To minimize sexual dimorphism in dental variability only male skulls were chosen. The specimens were classed into three groups that represent the three patterns of dentition found within the primates studied: (1) 3PM2M which corresponds to species with three premolars and two molars; (2) 3PM3M corresponding to species with three premolars and three molars; and (3) 2PM3M species which have two premolars and three molars (Supporting Information Fig. 1).
The Image J software (http://rsbweb.nih.gov/ij/) was used for the measurements of upper jaw and teeth. Each premolar (PMnL) and molar (MnL) length was measured by the maximum mesiodistal diameter taken on the occlusal surface between the mesial and distal contact points using the straight line tool. Premolar tooth-row length (PML) and molar tooth-row lengths (ML) were obtained by measuring the distance from the mesial contact point of the first tooth (second or third premolar for premolar teeth and first molar for molar teeth) to the distal contact point of the last tooth (fourth premolar for premolar teeth and second or third molar for molar teeth). The distance from the mesial surface of the canine to the last premolar, measured using the straight line tool in Image J, plus molar tooth-row length (ML) was referred here as PALATE2. This distance correlates in size to the alveolar process derived from the embryonic maxillary process. Dimensions were obtained by the average of the left and right sides in each specimen analyzed. More detailed information about the measurements can be seen in Fig. 1.
The analysis performed in this study used the following parameters:
PMnL vs. PML
MnL vs. ML
PML/PALATE2 vs. PMnL/PML
ML/PALATE2 vs. MnL/ML
(PML = length of premolar tooth-row, ML = length of molar tooth-row, PALATE2 = length of alveolar process derived from the maxillary process, n = position of tooth (i.e., = 1, 2, or 3 for molars, or 2, 3, 4 for premolars). Therefore, PML/PALATE2 is the size of premolar teeth relative to PALATE2. PMnL/PML is the size of each premolar tooth relative to the premolar tooth row. Accordingly, ML/PALATE2 is the size of molar teeth relative to PALATE2. MnL/ML is the size of each molar tooth relative to the molar tooth row.
Kruskal–Wallis test followed by Student–Newman–Keuls post hoc multi comparison analysis were used to compare values among the different groups. Linear regression analysis was used to obtain the slope (β coefficient) of the regression line (least-squares method). All statistical calculations were performed using the BioEstat statistical package (http://www.mamiraua.org.br). Differences were considered statistically significant when P < 0.05.
The Variation in the Relative Sizes for Each Post-Canine Tooth is Dependent on the Pattern of Dentition
In this analysis we seek to understand how the variations in the size of premolar and molar fields are related to the variation in each tooth size within its respective field, and if the pattern of variation is related to the dentition pattern.
When the MnL data were plotted against ML it was possible to observe that the three patterns of postcanine dentition (3PM2M, 3PM3M, and 2PM3M) had distinct patterns of molar variation (Fig. 2), as shown by the slope of regression line (β) (Table 1). As expected, 3PM2M animals had the highest β coefficients (i.e., angle of the regression slope relative to x axis), since these animals have only two molars to fit in the molar field. In the 2PM3M animals the smallest β coefficient was that of M1 (0.30), whereas M2 and M3 had approximately similar coefficients (0.34 and 0.35, respectively, Table 1). In the 3PM3M group M1 and M2 had higher β values (0.38) than M3 (0.29).
Table 1. Characteristics of linear regression lines of molar teeth
β is the slope of regression line, ICβ is the confidence interval the slope of regression line and R2 is the coefficient of determination of the regression model.
Similar to molars, the analysis of premolar field (Fig. 3) showed that 2PM3M animals had the highest β coefficients, since these animals have only two premolars to fit in the premolar field. In 3PM2M animals the highest β coefficient was of PM2 (0.38), while in 3PM3M animals this tooth presented the smallest β coefficient (0.31) (Table 2).
Table 2. Characteristics of linear regression lines of premolar teeth
β is the slope of regression line, ICβ is the confidence interval the slope of regression line and R2 is the coefficient of determination of the regression model.
Variations in the Relative Sizes of Molar and Premolar Fields in Relation to Secondary Palate Size
The aim of this analysis was to observe how the relative sizes of premolar and molar fields are associated with the pattern of dentition. When the MnL/ML vs. ML/PALATE2 data were plotted it was possible to observe that the three patterns of dentition (3PM2M, 3PM3M, and 2PM3M) could be distinguished from one another (Fig. 4). The distinction was most evident when observing the M1L/ML vs. ML/PALATE2. Table 3 presents all the ML/PALATE2 ratios.
Table 3. ML/PALATE2 and PML/PALATE2 ratios
The MnL/ML ranges also differed among the three patterns of posterior dentition. The means of the M1L/ML ratios in 3PM2M, 3PM3M, and 2PM3M groups were 0.59, 0.38, and 0.32, respectively (P < 0.05 A large M1L/ML ratio, as seen in Callithrichidae, would result in smaller M2L/ML and absent M3). Differences in the M2L/ML ratios were smaller than for M1L/ML and M3L/ML. The means for the M2L/ML ratios of 3PM2M, 3PM3M and 2PM3M groups were 0.41, 0.36, and 0.36, respectively.
Loss of M3 in the family Callithrichidae (3PM2M) was accompanied by a steep increase in the M1L/ML ratios compared with animals with M3s. A smaller increase in the M2L/ML ratio was also noted (Fig. 4). Despite some overlapping in the ML/PALATE2 ratio among the three groups of primates there was a clear association between the relative size of the molar region in relation to PALATE2 (ML/PALATE2) and the presence or absence of the M3 and PM2. Species lacking upper M3 have a ML/PALATE2 ratio smaller than 0.4, while species with ML/PALATE2 ratios larger than 0.49 tend to lack PM2 (P < 0.05, Kruskal–Wallis test followed by Student–Newman–Keuls post hoc multi comparison analysis). Within species having M3 the animals with three premolars (3PM3M) tend to have M1 considerably larger than M3, whereas animals with two premolars (2PM3M) tend to have M1 of the same size or slightly smaller than M3 (Fig. 4).
When the PMnL/PML vs. PML/PALATE2 data were plotted it was possible to observe that 3PM2M, 3PM3M, and 2PM3M animals presented clearly distinct PML/PALATE2 ranges (P < 0.05, Kruskal–Wallis test followed by Student–Newman–Keuls post hoc multi comparison analysis) (Fig. 3). Table 3 presents all the PML/PALATE2 ratios.
Different from the results for the molars the PMnL/PML ranges were similar for animals with 3PM2M and 3PM3M but were significantly smaller than those with 2PM3M (P < 0.05, Kruskal–Wallis test followed by Student–Newman–Keuls post hoc multi comparison analysis). The means of the PM2L/PML ratios in 3PM2M and 3PM3M groups were 0.34 and 0.34, respectively. The means for the PM3L/PML ratios of 3PM2M, 3PM3M, and 2PM3M groups were 0.33, 0.33 and 0.50, respectively; and the PM4L/PML ratio means for 3PM2M, 3PM3M, and 2PM3M groups were 0.33, 0.33 and 0.50, respectively.
Loss of PM2 in 2PM3M animals was accompanied by a steep increase in the PM3L/PML and PM4L/PML ratios (Fig. 5). There is a limit for the relative size of the premolar region in relation to the secondary palate (PML/PALATE2) that determines the presence of the PM2. Species lacking upper PM2 have a PML/PALATE2 ratio smaller than 0.32. Figure 6 shows representative drawings of the relative sizes of the teeth in the three types of postcanine dentition (3PM2M, 3PM3M, and 2PM3M) found in primates.
Molar teeth have a major role in the masticatory process, as they are the major elements responsible for crushing and grinding. It has been shown that the relative sizes of molar teeth depend on the interactions between activators and inhibitors during tooth development, where inhibitors are diffusible molecules secreted by the predecessor tooth germ. Inhibitors will delay the initiation of tooth development, resulting in smaller teeth (Kavanagh et al., 2007). Accordingly, since the first molar is the first permanent tooth to develop in the primate postcanine dentition, this tooth will have a major role on the development of its successor teeth. A large M1L/ML ratio, as seen in Callithrichidae, would result in smaller M2L/ML and absent M3, leading to small ML/PALATE2. Animals with small ML/PALATE2 ratio have relatively large PML/PALATE2 ratio and three premolars. On the other extreme are the animals with small M1L/ML ratio (ex: Pongidae). These animals have relatively larger M2 and M3 with an increased ML/PALATE2 ratio resulting in reduced PML/PALATE2 ratio, explaining the presence of only two premolars. In this sense the premolar and molar fields can be seen as submodular structures that function as an integrated part within a larger module (i.e., PALATE2). Interestingly, the number of premolar teeth can be predicted with a high degree of certainty by simply measuring the animal's M1L/ML ratio. Thirty four out of the 37 animals (91.9%) with 3 premolars had a M1L/ML ratio >0.36, and 45 out of the 47 species (95.7%) with 2 premolars had a M1L/ML ratio smaller than 0.36. It is likely that this association is related to the important functional and developmental roles played by this tooth.
Our results show that the loss of M3 (PM3M2 group, Table 1) resulted in a greater increase and higher β coefficients in M1L/ML compared to M2L/ML (Fig. 4 and Table 1, respectively). These observations fit in the general model of tooth proportions proposed by Kavanagh et al. (2007) that connects the timing of molar initiation with tooth size and number. In this model the timing of molar initiation is regulated by diffusible inhibitors produced by the predecessor teeth. Because M1 is the first molar to be formed this teeth will ultimately determine the size and number of successor molars, where increased inhibition would result in smaller posterior teeth and eventually the loss of M3. Similarly in 2PM3M animals the loss of PM2 was accompanied by a steep increase in the PM3L/PML and PM4L/PML ratios (Fig. 5), suggesting that a similar inhibitory model is also modulating the growth of premolars. In fact, the analysis of PM3M2 animals shows a progressive decrease in β coefficients from P2 to P4 (Table 1), which is consistent with a cascade of decreased inhibition, where P4 is in most cases the first teeth to be formed in Callitrichids (Swindler 2002). Interesting, M1L/ML ratio is nearly always larger in species with 3 premolars than in those with 2. This can be explained by the fact that larger M1 is associated with smaller M2 and M3 teeth and consequent smaller ML, allowing more space for premolar field.
Although we have used only one specimen for each species our results clearly show that the specimens with similar pattern of dentition are clustered together. Additionally, it is important to mention that the present work analyses the ratio between dental and maxillary sizes in the same specimen. It has been shown that the ratios between postcanine parameters show small interspecific variation in primates (Pirie, 1978), and it is plausible to assume that ratios in postcanine region will present even smaller interspecific variation. It is worth mentioning that our analysis was restricted to male specimens that eliminate variations due to sexual dimorphism.
The first primates are believed to have appeared during the beginning of the Eocene at ∼55 million years ago (Franzen et al., 2009). The diversification of primate species seems to have been strongly influenced by the adaptation to new diet-based adaptive zones (Sussman, 1991; Fleagle, 1999). Diet has been strongly associated with absolute size differences in primates, as it will impose metabolic and foraging constraints (Marroig and Cheverud, 2005; Marroig, 2007). The shape and relative size of postcanine dentition in primates seems to be influenced by dietary constraints, such as food size, shape, abrasiveness and protein levels (Kay, 1975; Pirie, 1978; Lucas et al., 1986). Our results show that the variation in the pattern of primate dentition is directly correlated with the relative size of molar field. The posterior dentition of primitive primates was likely to be composed of three premolars and three molars (Fleagle 1999, Franzen et al., 2009). The loss of M3 in Callitrichids has been associated with an emphasis on incisal biting and consequent decrease of selective pressure on postcanine dentition (Ford, 1980; Anapol and Lee, 1994). Within a certain range, the variations in the relative length of premolar or molar fields (PML/PALATE2 or ML/PALATE2) are accomplished by an increase or decrease in the lengths of the teeth within its field. The relative size of each tooth within its field (PMnL/PML or MnL/ML) showed a distinct rate of variation, and the variation rate for a tooth depends on the pattern of postcanine dentition. When the relative length of the field reaches a threshold limit, the accommodation of posterior teeth in the maxillary region is achieved by the loss of PM2 and an increase in the relative size of M3 (M3L/ML ratio, upper threshold limit), or by the loss of M3 and an increase in the relative size of M1 (M1L/ML ratio, lower threshold limit). It is worth mentioning that Tarsius is the major exception in this rule, where the increase in ML/PALATE2 ratio to 0.56 did not cause the loss of M3. In fact, Tarsius' dentition was shown to be unusual among primates. Tarsius molar teeth fall well above the general tooth size versus body mass scaling axis of primates, and are more similar to the insectivore's scaling axis, a position consistent with their insectivorous habits (Gingerich, 1984). This indicates that the developmental programming that regulates the relative size of postcanine teeth may differ among mammalian orders, and the predictions made here by the analysis of primate dentition may not apply to all mammals. It is possible that other mammals that have similar dietary behaviors to other, less insectivorous primates will fit the predictions better than tarsiers.
It is important to mention that the results presented here were obtained with the analysis of male specimens and that in some species, such as gorillas and baboons, the sizes of canine teeth are considerably larger in males than females (i.e., sexual dimorphism). In this sense, it is expected that PALATE2 would be larger in sexually dimorphic species compared to those whose males have smaller canines. Thus, the relative sizes of PML and ML will be reduced in those species that are more sexually dimorphic. It is possible, therefore, that the analysis of female specimens would have produced smaller variations in PML/PALATE2 and ML/PALATE2 ratios than male specimens.