The functional significance of morphological changes in the dentitions of early mammals

Abstract

The Mesozoic marked a time of experimentation in the tooth morphology of early mammals. One particular experiment involved the movement of three points, or cusps, on the surface of a molar tooth from a line into a triangle. This transition is exemplified by two extinct insectivorous mammals, Morganucodon (cusps in a line) and Kuehneotherium (cusps in a triangle). Here we test whether this difference in cusp arrangement, alongside cusp heights and angles between cusps, is associated with differences in the ability of the teeth to fracture proxy-insect prey. We gathered measurements from molar teeth of both species and used them to create physical models. We then measured the force, time and energy at fracture and peak force, and the amount of damage inflicted by the models on hard and soft gels encased in a tough film that mimicked the material properties of insects. The Morganucodon model required less force and energy to fracture hard gels and reach peak force compared with Kuehneotherium. Kuehneotherium required a similar time, force and energy to fracture soft gels but reduced the time, force and energy to reach peak force. More importantly, Kuehneotherium also inflicted more damage to both the hard and the soft gels. These results suggest that changes in dental morphology in some early mammals was driven primarily by selection for maximizing damage, and secondarily for maximizing biomechanical efficiency for a given food material property.

1. Introduction

The transition from a simple to complex dentition occurred multiple times during the evolution of mammals and has long been considered as one of the major keys to their success [1–3]. Following the evolution of homeothermy early in the Mesozoic, increased metabolic demands produced strong selection for those mammals with teeth that could process food efficiently [4]. As a result, the morphological diversity and complexity of teeth increased, allowing mammals to access a greater range of diets [5]. During the Late Triassic and Jurassic many new tooth morphologies arose and replaced the simpler unicuspid teeth found in therocephalids (e.g. Promoschorhynchus [6]). These new groups included triconodonts, symmetrodonts and docodonts, whose molars exhibited three main cusps arranged in either lines or triangles, and multituberculates, which had molars with multiple cusps arranged in parallel rows (figure 1).

Figure 1.

Phylogeny of select Mesozoic mammals with lower molar morphologies (based on [7–9]). P, Permian. Position of Kuehneotherium is poorly resolved but always falls out as more derived than Morganucodon (grey dashed branches).

By the Late Cretaceous, most mammalian clades had either multituberculate molars or a new tooth morphology known as the tribosphenic molar. What these derived and very different forms had in common was the ability to cut and grind food. These molars were more functionally and morphologically complex compared to their earlier single-cusped counterparts because of the increased number of shearing crests and surfaces which could process a greater variety of food with heterogeneous material properties. Multituberculate molars cut and crushed food as rows of cusps on the lower molars moved past those on the upper molars [3]. Lower tribosphenic molars had three primary cusps arranged in a triangle to serve as a cutting surface (the trigonid), and additional cusps that extended distally to produce a basin-like crushing surface (the talonid). The trigonid and talonid interdigitated with three-cusped upper tribosphenic molars to produce an efficient cutting and crushing complex [10]. The evolution of both types of molars with multiple, interdigitating cusps was accompanied by increased dietary and ecological diversity [5]. Understanding the biomechanical consequences of the transition to these complex cusp arrangements could provide insights into the performance parameters that may have been the object of selection in driving increased molar tooth complexity and the attendant dietary diversification of mammals.

The first step in the evolution of the tribosphenic molar involved the movement of three cusps on the molar crown from a linear (triconodont) arrangement to a triangular (symmetrodont) arrangement that formed the trigonid [11–13]. Triconodonts and symmetrodonts form paraphyletic groups among basal mammals, and their relationships are still poorly resolved [14,15]. Qualitative studies suggest that the transition from a linear to triangular arrangement of cusps increased the number of shearing surfaces between the upper and lower teeth, perhaps facilitating food processing [10,16]. Morganucodon (a triconodont) and Kuehneotherium (a symmetrodont) spanned the Late Triassic–Early Jurassic and exemplify molar teeth with linear and triangular arrangements of cusps (figure 2). Other differences between their molars include the sizes of small accessory cusps, the angles between the primary and accessory cusps (notch angle), the size of the cingulum (a shelf of enamel surrounding the molar crown), and the way in which the teeth wear with use [12,17]. Most studies of early mammal teeth focus on studying relationships based on tooth morphology [11,18], patterns of wear on the teeth [17] and simulations of the chewing cycle [19–21]. A single experimental study based on Morganucodon and Kuehneotherium demonstrated that the cingulum functions to dissipate stress away from the molar roots, reducing the likelihood of tooth breakage [22]. No other quantitative study has documented how linear and triangular cusp arrangements translate into differences in feeding performance. A detailed investigation of the mechanical performance of linear and triangular cusp arrangements in processing food can offer insights into why selection favoured the transition from a triconodont to symmetrodont arrangement of cusps.

Figure 2.

Schematic and computed tomography (CT) images of the teeth used in this study. Schematic drawings and CT scans of Morganucodon watsoni (a) and Kuehneotherium praecursoris (b) from Late Triassic–Early Jurassic fissure fills in Glamorgan, Wales, UK. Upper scans, lingual view; middle scans, buccal view; lower scans, occlusal view. See supplementary information for scanning and specimen details.

Here we test the hypothesis that the transition from a triconodont to a symmetrodont molar morphology increased the efficiency of food breakdown using physical models of the upper and lower molar tooth rows based on the molars of Morganucodon and Kuehneotherium. We compare the force, displacement (as a proxy for time) and energy at two distinct phases of biting a food item: the point of first fracture (when cusps first puncture though the food item) and at the point of maximum force. First fracture indicates the point at which a tooth initiates a crack in a food item, while the point of maximum force offers a window onto how the teeth penetrate and propagate cracks through a food item. By reducing the work required to process food, energy can be invested elsewhere (foraging, reproduction, etc.). We also record the displacement at which fracture occurred and the maximum force that was attained. Assuming that displacement occurs at a steady rate, it can be considered as a proxy for time. A tooth that can fracture a food item more quickly could potentially process relatively more food in a shorter amount of time. Mesozoic mammals were likely homoeothermic and had high metabolic rates that required a great deal of food to maintain. They were under strong selection for energy efficiency [4]: any reduction in force, time or energy expended during food processing could have been beneficial.

Previous studies indicate that Kuehneotherium likely preyed on softer insects compared with Morganucodon [3,12,23]. This is evidenced by different molar wear patterns between the mammals, and a more slender jaw morphology in Kuehneotherium that was not as capable of withstanding high bite forces [23]. To replicate this, we used hard and soft gels encased in a tough film that together exhibit material properties similar to those of the invertebrate prey that these mammals probably consumed (e.g. beetles and scorpion flies) [23,24]. Insects are structurally complex, requiring multiple fractures to process them completely. Both hard and soft invertebrates require a combination of cusps and crests (figure 2) to initiate and propagate cracks through the entire body [25]. Given these differences in diet, we predict that the molar models of Kuehneotherium will process soft food items more efficiently while the molar models of Morganucodon will process hard food items more efficiently. Efficiency is measured as the ability of a model to reduce energy, force and time during a bite. We also predict Kuehneotherium will inflict more damage to the gels given the more complex triangular arrangement of cusps on the crown. These aspects of performance make intuitive sense from the classical perspective that teeth are tools optimized for breaking apart food.

2. Material and methods

2.1. Molar measurements, computer-aided design and insect model construction

Eighteen morphological measurements of Morganucodon watsoni (n = 11) and Kuehneotherium praecursoris (n = 11) were taken from isolated lower molars (electronic supplementary material, figure S1; table S1). In order to identify the variables that most clearly differentiate between the two species, we assessed correlations among all variables within each species (electronic supplementary material, figures S2–S3; table S2). We then compared each variable between Morganucodon and Kuehneotherium using t-tests (electronic supplementary material, table S3). The variables that differed most between the two species, and were not correlated with other variables, were used as the foundation for computer-aided design (CAD) models of molar teeth for each species. These key variables were position of the accessory cusps (linear or triangular), notch angle and height of the b-cusp. We used species means of these three variables to build a model for each species. Values for the remaining 15 variables in both models were based on the pooled sample from both species.

We created CAD models for each species using Creo Parametric 2.0 (PTC, Inc.) and isometrically scaled them to one another by making the height of the primary cusp the same for both species (electronic supplementary material, figure S5; tables S4–S5). Upper and lower molars are essentially identical within species, so we used the same models for both upper and lower tooth rows. We used three upper and four lower molars for both models. Cusp tips came to relatively sharp points in both models, which may initiate crack propagation more easily than the somewhat more blunt cusps of real teeth. We scaled up the models by a factor of 10 and printed them in three-dimensions using a Dimension uPrint SE Plus printer, using an acrylonitrile butadiene styrene (ABS) plastic and a printed layer resolution of 254 µm (figure 3b).

Figure 3.

CAD models and experimental set-up. (a) CAD models of each molar row. (b) Experimental set-up using Kuehneotherium and PDMS-PET proxy food item.

We constructed two proxy food items of different stiffnesses to mimic a hard and a soft food item with material properties based on values gathered from the literature ([24]; electronic supplementary material, table S6). We use the term ‘soft’ to describe food items that exhibit low stiffness, and the term ‘hard’ to describe food items with high stiffness (electronic supplementary material, table S6). Our hard food item consisted of a poly(dimethyl siloxane) [PDMS] elastomer gel (Sylgard^® 184, Dow Corning, Inc.), encased in a polyethylene terephthalate (PET) film (GmbH, Hostaphan^® TT, Mitsubishi Polyester Film) (electronic supplementary material, figure S6a). Our soft food item consisted of a poly(methyl methyacrylate) [PMMA]-poly(n-butyl acrylate) [PnBA]-[PMMA] triblock copolymer gel (Kurarity™, Kuraray Co.), encased in a PET film (electronic supplementary material, figures S6b–S7). The ability of the gel complexes to resist compressive force during biting depends on the tensile strength of the film, the tensile stiffness of the film and the compressive stiffness of the inner gel. The gels model the ability of an entire insect to resist compressive force. We altered the fracture properties of the food item by changing the material properties of gels and keeping PET film material properties constant. We changed the material properties of the gels as this approach is technically easier and accomplishes the same final goal: to change the material properties of the entire proxy food item. We confirmed this by testing the material properties of the hard and soft food items (electronic supplementary material, table S6). For further explanation of stiffness calculations and gel construction, see the electronic supplementary material.

2.2. Physical testing

Lower tooth rows were affixed to the base of an Instron machine (model 5500R) and the upper tooth rows were attached to a 50 N load cell via a plate that could be moved in the x–y plane. This allowed us to consistently place the upper tooth row into precise alignment with the lower tooth row, so to replicate the orthal pattern of occlusion hypothesized for Morganucodon and Kuehneotherium [17]. Although the upper and lower molar models occlude during the production of the fractures they do not come into full contact, so some elements of the masticatory cycle such as shearing forces are not modelled in this study. Chewing is also not modelled in this study although, given the material properties of insect exoskeleton, much of the processing likely occurs in the first few chewing cycles [26]. Beginning with the upper and lower teeth just in contact with the upper and lower surfaces of the gel, we displaced the upper molar tooth row 8.5 mm for the soft gel and 10.6 mm for the hard gel at a constant rate of 0.15 mm s⁻¹ (electronic supplementary material, figure S9). These distances ensured that the teeth punctured the gel but did not come into contact with the base-plate, which could have damaged the models and/or the load cell. We used a constant rate of displacement for all trials as changing the rate can impact fracture properties of elastic materials, such as PDMS and triblock, over the timescale of our puncture experiments [27]. For each species, we conducted 12 trials using the soft food item and 10 trials using the hard food item. There was no evidence that the force or energy decreased with additional tests, indicating that the models were not experiencing wear that affected the results (electronic supplementary material, figure S8).

We extracted the force, energy (work done), time at initial fracture and point of maximum force from the force–displacement curve generated by each trial (electronic supplementary material, tables S7–S8), and compared the values of these mechanical performance variables between Morganucodon and Kuehneotherium using t-tests (electronic supplementary material, tables S9–S10).

To compare the damage inflicted on the gel by the Morganucodon and Kuehneotherium models, we photographed the hard and soft gels after trials were completed and measured the length of fractures in the film using ImageJ (electronic supplementary material, figure S10–S11) [28]. We compared the length of fractures in the cuticle film between Morganucodon and Kuehneotherium using t-tests (electronic supplementary material, table S11). All statistics were performed in R v. 3.1.0 [29].

3. Results

There were differences in the force–displacement curves for the two species (figures 4 and 5). In soft foods (figure 5a–c), initial fracture was achieved at a similar time (figure 5a, p = 0.09), with a similar force (figure 5b, p = 0.47) and with expenditure of energy (figure 5c, p = 0.33) by the Morganucodon and Kuehneotherium models. The mean maximum force for the Kuehneotherium model occurred after less time (figure 5a, p = 0.0001), with less force generated at the maximum force point (figure 5b, p = 0.002) and with less energy expenditure (figure 5c, p = 0.0009) compared with the Morganucodon model. In sum, Kuehneotherium was more efficient at reducing the time, force, and energy to reach maximum force in soft foods. There was no difference between the models in their ability to initiate fracture.

Figure 4.

Force landscape for gel puncture trials. (a) Soft food item, (b) hard food item. *Denotes initial fracture in Morganucodon, † in Kuehneotherium. § denotes the peak of maximum force in Morganucodon, ‡ in Kuehneotherium.

Figure 5.

Time (displacement), force and energy boxplots for both species. Kuehneotherium, grey boxplots; Morganucodon, white boxplots. (a–c) Soft food items; (a) boxplots of displacement at point of initial fracture and maximum force, (b) boxplots of force at fracture and maximum force, (c) boxplots of energy at point of initial fracture and maximum force. (d–f) Hard food items (d) boxplots of displacement at point of initial fracture and maximum force, (e) boxplots of force at fracture and maximum force, (f) boxplots of energy at point of initial fracture and maximum force. *p < 0.05, **p < 0.01, ***p < 0.001, n.s., not significant.

In experiments with the hard food items (figure 5d–f), the Morganucodon model achieved initial fracture more quickly (figure 5d, p = 0.0008), with lower forces (figure 5e, p = 0.0002) and with lower expenditures of energy (figure 5f, ) compared with the Kuehneotherium model. The Morganucodon and Kuehneotherium models achieved maximum force at similar times (figure 5d, p = 0.75), but the Morganucodon model exhibited higher forces (figure 5e, p = 0.005) and larger expenditures of energy (figure 5f, p = 0.01). In sum, Morganucodon was more efficient at reducing the force and energy to initiate fracture and reach maximum force in hard foods.

We evaluated the amount of damage the models inflicted on the proxy food items by comparing the lengths of the fractures caused by the Morganucodon and Kuehneotherium models. The fractures made by Kuehneotherium were significantly longer than those caused by Morganucodon (soft, ; hard, p = 0.01, figure 6a), suggesting that Kuehneotherium could cause substantially more damage to food items than Morganucodon.

Figure 6.

Cuticle damage. (a) Boxplot illustrating the total length of cuticle puncture for both species (M, Morganucodon, white; K, Kuehneotherium, grey), *p < 0.05, ***p < 0.001. (b) Schematic to illustrate where forces are being exerted on the gel at the point of maximum force (white = upper cusps; black = lower cusps). After the primary cusp a fractures the cuticle the accessory cusps come into contact to further propagate cracks through the food item. In Morganucodon, the accessory cusps contact the cuticle sequentially, the c cusp followed by the b cusp, such that force is initially concentrated over a smaller area. In Kuehneotherium, both accessory cusps contact the cuticle simultaneously, forming a triangle and distributing force over a wider area.

4. Discussion

Using models of molar tooth rows for two Triassic mammals, we found many biomechanical differences between the Morganucodon and Kuehneotherium models in their ability to process hard and soft food items. We found some evidence to support our hypothesis that the transition from a triconodont to a symmetrodont molar morphology increases the efficiency of food breakdown, but the results are not straightforward. The biomechanical parameters we measured distinguished between the hard and soft diets, while the damage parameter distinguished molar complexity. The triconodont, Morganucodon, was more efficient at consuming hard prey. Both Morganucodon and the symmetrodont, Kuehneotherium, were similarly efficient at initiating fracture in soft food items, although Kuehneotherium reached maximum forces more quickly and with less energy. However, Kuehneotherium could inflict more damage than Morganucodon on both hard and soft food items. The more complex molars of Kuehneotherium provide the ability to inflict more damage and expose more surface area for digestive enzymes, which may be a critical target of selection given the greater metabolic demands experienced by early mammals.

The Morganucodon model was better able to process hard food compared to Kuehneotherium (figure 5d–f). The linear alignment of cusps alongside the blade-like occlusion of Morganucodon served to concentrate the forces over a smaller area (figure 6b). By concentrating the bite forces onto a small area, the molar can maximize the pressure and stress on the food for a given force [16,25]. Once fracture occurs in the hard food item, we hypothesize that much of the energy is expended through propagating the crack in the film and fracturing through the thickness of the sample, rather than deforming the food item [30]. Owing to the linear alignment of the cusps in Morganucodon, much of the energy was expended through propagating the crack in the film along the length of the food item. The thin, blade-like cusps of Morganucodon penetrated through the hard food item and propagated cracks more easily than Kuehneotherium. The ability of the Morganucodon models to more efficiently process hard food items could be beneficial for penetrating the hard insects that they are thought to specialize on [23].

The Kuehneotherium molar was better able to process soft food compared to Morganucodon (figure 5a–c). Like hard food items, energy is expended in initiating cracks in soft food items, but a larger percentage of energy is expended in deforming a soft food item. Deformation of the soft food item can occur at a considerable distance away from the initial crack. There is less advantage to concentrating force in a single area when high deformation occurs, and critical stress concentrations can be achieved with a relatively larger area of contact [16]. This may explain why Morganucodon and Kuehneotherium were equally able to initially fracture soft foods given the larger contact area of Kuehneotherium (figure 6b). The Kuehneotherium model could distribute force over a larger area because the height of the b-cusp is equal to the c-cusp, therefore, they contact the food item at the same time adjoining the a cusp to form a triangle. Morganucodon has a smaller b-cusp, which caused the cusps to act like a series of blades in our modelling experiments. The propagation of cracks in the soft food item is more easily achieved by a cusp than a blade as the crests of a cusp act as small blades to divide material following crack initiation [16]. The linear alignment of crests in Morganucodon are at a disadvantage, here and this may go some way to explaining the increased efficiency of Kuehneotherium in propagating cracks through the soft food. Given previous studies investigating notch angles, Kuehneotherium's more acute notch angles could be more effective than obtuse angles at reducing the amount of energy required to process food [31,32]. Energy reduction is achieved by localizing strain to smaller areas of the food item and reducing deformation. Acute notch angles also increase the ‘trapping ability’ of teeth, keeping the food item stationary while the cusps are driven through it [31]. The effect of notch angle on food item fracture may be reduced in hard food items because lower deformation of the food will produce less contact between the notch angles and food item.

The triangular arrangement of cusps in Kuehneotherium increased the damage inflicted on both hard and soft food items given their longer shearing crests. In biological terms, this translates into breaking food into more, smaller pieces in a single bite. Damage to a food item increases the surface area that contacts digestive enzymes, which allows more rapid extraction of nutrients in the digestive tract. Rapid nutrient extraction is important in meeting the high metabolic demand of homeothermy. However, Kuehneotherium caused more damage at the expense of higher forces, energy and time required to process hard food items. The jaw of Kuehneotherium was more gracile and so could probably not produce the higher forces required to penetrate hard foods and would likely favour soft foods. This is supported by tooth microwear studies which demonstrated how the molars of Kuehneotherium typically exhibit smaller pits and scratches representative of a softer diet [23]. In soft food items, Kuehneotherium caused more damage for a similar amount of time, energy and force, and was also able to propagate cracks more efficiently compared to Morganucodon. Strait [33] predicted that the teeth of insectivores that specialize on soft foods exhibit longer shearing surfaces as they can contact and divide more food per chewing cycle. It is likely that there are also additional advantages associated with triangular cusp arrangements (e.g. increased capture areas and food trapping ability).

Damage may be the most important parameter in this assessment of early mammal tooth morphology and function. Morganucodon may exhibit low forces and energy following a feeding trail, but if the food item were still intact then it would become more difficult to extract useable energy from the food item during digestion. Kuehneotherium caused the greatest amount of damage to both hard and soft gels. This is the only biomechanical parameter we measured where one model was superior regardless of food item stiffness. The damage inflicted to a food item may be easier for selection to act upon because it has a direct relationship to the amount of useable energy that can be extracted from the food item [16,34,35]. The transition to complex teeth improved the ability to inflict damage on a food item that transcends the biomechanical variables we measured here. Complex teeth with different morphologies are often specialized for biomechanical tasks associated with the material properties of the foods that the animals eat [21,25]. For example, horse teeth perform well at cutting and grinding grass, whereas cat teeth are good at cutting and piercing meat, they are both complex dentitions with very different morphologies that are specialized for damaging food with a specific material property [36]. Kuehneotherium is more complex than Morganucodon hence performs better at inflicting damage on a food item, but the biomechanical variables show that Kuehneotherium was tuned to soft food and Morganucodon to hard food.

5. Conclusion

Here we show, by testing physical models of tooth rows, that increased tooth complexity can have a significant impact on the biomechanics of food processing. We find evidence for dietary specialization based on molar morphology; the triangular molars of Kuehneotherium are more suited to processing soft foods while the blade-like molars of Morganucodon are more suited to processing hard foods. We also highlight the importance of incorporating food item damage into studies of dental functional evolution, as this may be an important target of selection. This study provides more evidence to suggest stem mammals exhibit more morphological and functional diversity than previously thought, and specializations in the teeth of early mammals could lead to trophic diversification and niche partitioning. Changes to tooth morphology can originate in relatively simple developmental shifts that alter the position and number of cusps on the molar crown [37,38]. Subsequent selection for differences in molar structure that affect food processing led to the close association between molar morphology and diet we see in modern mammals [36]. Owing to the intimate relationship between molar form and function and the rising metabolic costs of homeothermy, the ability to break food into small pieces was likely a significant factor in driving the trophic diversification of Mesozoic mammals and their living descendants [2,23].

Data accessibility

All data presented in this study are available in the supplementary text and tables.

Authors' contributions

An.J.C. designed study, collected data and wrote paper, M.J.I. designed study, collected data and co-wrote paper, Al.J.C. designed study and co-wrote paper, E.R.D. designed study and co-wrote paper.

Competing interests

We declare we have no competing interests.

Funding

M.J.I. and Al.J.C. would like to thank the Human Frontiers Science Program (RPG0034/2012) for funding. An.J.C. was supported by a NSF DDIG no. 1501385 and a UMass Natural History Collections grant.