The origin of the lower fourth molar in canids, inferred by individual variation

Introduction

During the evolution of mammalian dentition, the number of teeth usually declines (Davit-Béal, Tucker & Sire, 2009; Ungar, 2010; Jernvall & Thesleff, 2012). While there are possible examples of secondary acquisition of recently lost teeth (M₃ in callitrichine monkey: Scott, 2015), increased tooth number is a rare evolutionary event. Only whales (Cetacea), armadillos (Cingulata) and the bat-eared fox (Otocyon megalotis, Canidae, Carnivora) have evolved to increase the number of permanent teeth beyond the ancestral eutherian basic dental formula (I 3/3, C1/1, P 4/4 M 3/3) (Ungar, 2010). Some species, such as the manatee (Trichechus, Trichechidae, Sirenia), pigmy rock wallaby (Petrogale concinna, Macropodidae, Marsupialia), and silvery mole-rat (Heliophobius argenteocinereus, Bathyergidae, Rodentia), exhibit continuous horizontal replacement of teeth (Ungar, 2010; Gomes Rodrigues et al., 2011). However, only the bat-eared fox (O. megalotis) has shown an increase in tooth number, accompanied by neither morphological simplification nor continuous horizontal replacement. The dental formula in the bat-eared fox is I 3/3, C1/1, P 4/4 M 3–4/4–5, with these animal usually possessing an upper M³ and lower M₄ that ancestral Canidae lacked (Sillero-Zubiri, 2009; Ungar, 2010), and this is considered a rare case of increased functional teeth number beyond that of the extant eutherian basic dental formula (Sillero-Zubiri, 2009; Ungar, 2010).

Bat-eared foxes are primarily insectivorous (Sillero-Zubiri, 2009), with a molar morphology comprising an undeveloped carnassial blade, equally sized molars (in relation to the other canids) and increased number of molars, which are attributable to an adaptation to an insectivorous diet (Wang & Tedford, 2008; Asahara, 2013; Asahara et al., 2016). It has been proposed that this dentition is suitable to a diet of insects that are small relative to the body size of the bat-eared fox, with a larger molar row grinding surface that enables greater chewing efficiency (Asahara, 2013; Asahara et al., 2016). However, this does not explain the presence of the fourth molar, the developmental origin of which remains unclear.

Individual variations in tooth number (supernumerary and missing teeth) have been reported in many mammals (e.g. Miles & Grigson, 1990). In some cases, as discussed by Asahara, Kryukov & Motokawa (2012), individual variations may underpin evolution, that is, fixation of the variation could initiate a new dental formula. In the present study, the source of the M₄ in the bat-eared fox was investigated based on examination of supernumerary teeth in related species.

There are several reports of supernumerary molars in Canidae, including M³ and M₄ along the normally aligned tooth row in coyote (Canis latrans) (Hall, 1940; Paradiso, 1966) and gray fox (Urocyon cinereoargenteus) (Hall, 1940). Wolsan (1984) categorized two types of supernumerary tooth generation: (1) creation of additional tooth germ and (2) splitting of a tooth germ. Type 1 can explain most of the supernumerary teeth that appear in positions where the ancestor possessed teeth, whereas type 2 can explain the eruption of supernumerary teeth that possess abnormal morphology relative to adjacent teeth. While the M³ in the coyote can be regarded as type 1 and “atavistic,” M₄ is not readily explained in this manner. Although ancestral caniforms possessed M³ (Wang, 1994; Tomiya, 2011), the last ancestor possessing a functional M₄ may be as early as the Mesozoic period, since the common ancestor of Placentalia possessed only three lower molars (O’Leary et al., 2013).

A previous experimental study established a developmental model termed the inhibitory cascade model (IC model) to explain relative molar size among three lower molars in mammals as resulting from the balance of inhibition and activation molecules during dental development (Kavanagh, Evans & Jernvall, 2007). In this model, activation molecules from the mesenchyme stimulate the formation of distal molars (M₂ and M₃), which makes them larger in relation to M₁, whereas inhibition molecules secreted from M₁ suppress distal molars growth (M₂ and M₃) (Kavanagh, Evans & Jernvall, 2007). According to the balance of these two factors, the relative size of M₁, M₂, and M₃ (typically shown numerically as M₂/M₁ and M₃/M₁ size ratios; the size is defined as projected tooth area from occlusal view) result in a pattern of M₁ > M₂ > M₃, M₁ = M₂ = M₃, or M₁ < M₂ < M₃ (Kavanagh, Evans & Jernvall, 2007). This model has previously been applied to explain dental variation in Carnivora (Polly, 2007; Halliday & Goswami, 2013; Asahara, 2013; Asahara et al., 2016).

The IC model can also explain the loss of M₃ during murine evolution, such that greater inhibition and lower activation in experimental mice organs resulted in the disappearance of M₃, coinciding with changes in the M₂/M₁ and M₃/M₁ ratios. This corresponds to the dentition of the murine species Hydromys chrysogaster, which lacks the M₃ (Kavanagh, Evans & Jernvall, 2007). Asahara (2013) reported the relationship between dental anomalies (M₃ loss) and the IC model in canids, in which M₂/M₁ scores of individuals that lost M₃ were lower (indicating higher inhibition and lower activation) than the scores in normal individuals in local populations of raccoon dog (Nyctereutes procyonoides) and arctic fox (Vulpes lagopus) (Asahara, 2013). Evolutionary loss of M₃ in murines and canids is thus considered attributable to inhibition/activation dynamics of dental development. Conversely, Kavanagh, Evans & Jernvall (2007) reported one case in which a supernumerary molar (M₄) appeared in mice following suppression of inhibition molecules, indicative of an increase in molar number and altered inhibition/activation dynamics.

The working hypothesis for this study is that the fourth lower molars in the bat-eared fox and the supernumerary molar in some other canid species are generated by reduced inhibition and greater activation during dental development. This was tested by comparing the relative size of molars in locations of three canid species with some individuals possessing a supernumerary molar (M₄).

Materials and Methods

Mandible specimens (dentary bones and molars) of 451 Canis latrans (from the United States National Museum of Natural History), 153 Nyctereutes procyonoides (from the Primate Research Institute, Kyoto University, Japan), and 136 Urocyon cinereoargenteus (from the United States National Museum of Natural History) were examined. The collection of C. latrans was chosen based on an earlier study using the U.S. collection, which reported the presence of a supernumerary molar M₄ in some specimens (Paradiso, 1966). Presence of the supernumerary molar was determined by macroscopic observation. Photographs were taken of the occlusal view of the molar row with scales. Projected areas of M₁, M₂, and M₃ were manually measured using Image J software (NIH, Bethesda, MD, USA). The projected area is defined as molar size, as described by (Kavanagh, Evans & Jernvall, 2007; Asahara, 2013; Asahara, 2014). For specimens of N. procyonoides, most data were derived from Asahara (2014). The ratio of M₂ to M₁ size (M₂/M₁) and that of M₃ to M₁ size (M₃/M₁) were calculated to compare relative sizes among molars. According to the IC model (Kavanagh, Evans & Jernvall, 2007), lower inhibition and higher activation results in higher M₂/M₁ and M₃/M₁ scores, whereas increased inhibition and reduced activation lowers the M₂/M₁ and M₃/M₁ scores. M₁ size and M₂/M₁ and M₃/M₁ scores of individuals with normal dentition, without M₃, and with M₄ were compared by U-test. Comparisons were performed separately for each species and sample location (as defined by collection source), with analyses performed using Minitab 14 statistical software (Minitab Inc., PA, USA). For data of the bat-eared fox Otocyon megalotis in Fig. 2 is cited from Asahara (2013).

Results

Twelve individuals of C. latrans (2.66% of total 451 individuals), four individuals of N. procyonoides (2.61% of total 153 individuals), and five individuals of U. cinereoargenteus (3.68% of total 136 individuals) possessed an M₄ in the normally aligned molar row (Fig. 1), with occurrences differing according to species and specimen location (Table 1). Data relating to M₁ size and M₂/M₁ and M₃/M₁ scores are presented in Table 1 and Fig. 2. The distribution of specimens with normal dentition represented a typical pattern of interspecific variation among canids (Asahara, 2013) (Fig. 2). M₂/M₁ and M₃/M₁ scores per location and per species are also presented (Figs. 3–5).

No relationship was found between mean M₁ size and presence or absence of M₄ (Table 1). When specimens of all locations were combined, mean M₂/M₁ and M₃/M₁ scores were typically larger in individuals with M₄ than those with normal dentition (without M₄), with the exception of M₂/M₁ scores in the Gifu samples of N. procyonoides (Table 1). When each location was analyzed separately, no significant difference in M₂/M₁ scores existed between individuals with M₄ and with normal dentition; however, when all locations were pooled, individuals with M₄ showed significantly higher M₂/M₁ scores than individuals with normal dentition in C. latrans and U. cinereoargenteus (Table 1). M₃/M₁ scores of individuals with M₄ were significantly higher than those of individuals with normal dentition in the Gifu samples of N. procyonoides and the New Mexico samples of U. cinereoargenteus (Table 1). For the pooled samples of U. cinereoargenteus, M₃/M₁ scores of individuals with M₄ were significantly higher than those of individuals with normal dentition (Table 1).

Five individuals of C. latrans and thirteen individuals of N. procyonoides lacked M₃ (Table 1). M₁ size of individuals without M₃ is significantly larger than that of individuals with normal dentition in C. latrans. M₂/M₁ scores of individuals without M₃ were significantly smaller than those of individuals with normal dentition in N. procyonoides.

Discussions

The lack of a relationship between M₁ size and individuals with M₄ versus those with normal dentition indicates that absolute M₁ size does not affect the generation of the M₄. However, individuals with high M₂/M₁ and M₃/M₁ scores (and thus with relatively large M₂s and M₃s) did tend to possess a supernumerary molar M₄ (Table 1; Figs. 2–5). While these scores did not differ significantly in most within-location analysis except for M₃/M₁ scores in the Gifu samples of N. procyonoides and the New Mexico samples of U. cinereoargenteus (Table 1), these results could be affected by the low number of samples that possessed M₄. To overcome this limitation, analysis of combined sample-locations did establish that M₂/M₁ scores were larger in individuals with M₄ than those with normal dentition in C. latrans and U. cinereoargenteus (Table 1). These results support the hypothesis that altered inhibition/activation dynamics (i.e. lower inhibition and higher activation) is likely to direct the formation of the M₄ in these species.

Geographic variation can affect M₂/M₁ scores of pooled locations, such that locations with higher M₄ incidence simply have a larger M₂/M₁ score. This appears not to be an issue in this study, for two reasons. Firstly, a larger M₂/M₁ score in a location is indicative of a genetic background supporting lower inhibition and higher activation during molar development. Secondly, the ranges of mean M₂/M₁ and M₃/M₁ scores among locations were not larger than the difference between mean scores of individuals possessing or lacking the M₄ (Table 1). The data considered herein supports a causal relationship between inhibition and activation dynamics and the occurrence of M₄.

While the existence of M₄ correlates with high M₂/M₁ and M₃/M₁ scores, there were many individuals that exhibited high M₂/M₁ or M₃/M₁ scores but whose mandible did not possess M₄ (Figs. 1–5), indicating that inhibition/activation dynamics are likely to be only one of the causes of generating an extra molar. This interpretation is supported by the experimental suppression of inhibition in a mouse model resulting in only one case of the appearance of an extra molar (Kavanagh, Evans & Jernvall, 2007). Acquisition of M₄ is rare during evolution of placental mammals (Sillero-Zubiri, 2009; Ungar, 2010), and it is logical that several barriers to M₄ generation exist. Furthermore, occasional occurrence of M₄, albeit at a low rate, suggests that canids may possibly possess a genetic background that favors the generation of the M₄, and this may explain the evolution of an additional molar in the bat-eared fox O. megalotis.

M₂/M₁ scores were larger in individuals without M₃ than those with normal dentition in N. procyonoides (Table 1). The result accords with the proposed relationship between dental anomalies (M₃ loss) and the IC model reported by Asahara (2013). Significant differences in M₁ size found in C. latrans (Table 1) could be affected by the low number of samples including a large individual that lacked M₃.

Consistent with Paradiso (1966), this study found that the presence or absence of M₄ differed among locations of C. latrans. Furthermore, Gisburne & Feldhamer (2005) reported that M₄ was not observed among 510 specimens of Illinois U. cinereoargenteus, contrasting with the finding from this study of the presence of M₄ in a low proportion of Arizona and New Mexico gray foxes. Most studies of raccoon dog (N. procyonoides) have reported no M₄ molar in Japanese samples (Hata, 1972; Harada et al., 1989; Asahi & Mori, 1980; Nozaki, 1984; collectively 664 individuals), with a notable exception published by Machida & Saito (1986), who reported an M₄ in just one of 137 raccoon dogs from Saitama prefecture in Japan. Therefore, the presence of M₄ in the three canid species examined in this study is considered a rare anomaly.

Asahara (2013) and Asahara et al. (2016) reported that the relative molar size was reflective of the diet among species of canids. The more carnivorous canid species (such as Canis lupus) possess low M₂/M₁ scores and the more omnivorous species (such as N. procyonoides or U. cinereoargenteus) have higher scores (Asahara, 2013; Asahara et al., 2016). In addition, the insectivorous hoary fox Lycalopex vetulus has higher M₂/M₁ and M₃/M₁ scores than almost all other omnivorous species, and the insectivorous bat-eared fox O. megalotis has higher M₂/M₁ and M₃/M₁ scores than all other canids, indicating very low inhibition and high activation during molar development (Asahara, 2013). According to this pattern, the evolution of the insectivorous diet is associated with molars that are of more equal size (between M₁, M₂, and M₃) than the carnivorous or omnivorous species, with higher M₂/M₁ and M₃/M₁ scores reflecting low inhibition and high activation during molar development. The present study is consistent with the hypothesis that low inhibition and high activation is a cause of M₄ generation. Therefore, M₄ is considered to be an adaptive trait providing a larger total occlusal surface area to enable the canid to digest a large amount of insects, as discussed by Asahara (2013) and Asahara et al. (2016). It is hypothesized that the ancestral species began to consume an insectivorous diet, and it followed that a pattern of low inhibition and high activation during molar development evolved to generate more equally sized molars (between M₁, M₂, and M₃) than its ancestor, providing an evolutionary force supporting this adaptation to the insectivorous diet. Then, the inhibition/activation pattern generates M₄ by chance. M₄ is also an adaptive trait for insectivorous diet providing larger occlusal surface area and greater grinding function, therefore, presence of M₄ is naturally selected and fixed in the ancestral lineage of the bat-eared fox. While no fossil canid species has been to found to possess an M₄, several cases of a small supernumerary M³ has been reported in an individual of the Prototocyon genus (Petter, 1964; Petter, 1973), which may be closely related to (or belong to) Otocyon (Van Valen, 1964; Petter, 1964; Petter, 1973; Wang & Tedford, 2008; Tedford, Wang & Taylor, 2009; Werdelin & Peigne, 2010). This supports the argument that this genus was a transitional stage in which molar number increased. Further discoveries from the fossil record will be important in revealing the evolutionary history of the fourth molar.

Conclusions

The presence of a supernumerary molar M₄ observed in several canid species is influenced by low inhibition and high activation during molar development, with the presence of M₄ in the bat-eared fox O. megalotis originating from low inhibition and high activation due to evolutionary pressure relating to an insectivorous diet.

Supplemental Information