Variations in dietary patterns of living sloths revealed by finite element analysis of jaws

Abstract

Although extinct sloths exhibited a wide range of dietary habits, modes of locomotion, and occupied various niches across the Americas, modern sloths are considered quite similar in their habits. The dietary habits of living sloths can be directly observed in the wild, and understanding the mechanical behavior of their jaws during chewing through finite element analysis (FEA) provides a valuable validation tool for comparative analysis with their extinct counterparts. In this study, we used FEA to simulate the mechanical behavior of sloth mandibles under lateral mastication loads, using it as a proxy for oral processing. Our research focused on the six extant sloth species to better understand their diets and validate the use of FEA for studying their extinct relatives. We found that all living sloths have the predominancy of low‐stress areas in their mandibles but with significant differences. Choloepus didactylus had larger high‐stress areas, which could be linked to a reduced need for processing tougher foods as an opportunistic generalist. Bradypus variegatus and Choloepus hoffmanni are shown to be similar, displaying large low‐stress areas, indicating greater oral processing capacity in a seasonal and more competitive environment. Bradypus torquatus, Bradypus pygmaeus, and Bradypus tridactylus exhibited intermediary processing patterns, which can be linked to a stable food supply in more stable environments and a reduced requirement for extensive oral processing capacity. This study sheds light on extant sloths' dietary adaptations and has implications for understanding the ecological roles and evolutionary history of their extinct counterparts.

The cluster analysis resulted in three groups of sloths divided by oral processing capacity. Choloepus didactylus experiences greater jaw stress than other species analyzed. Bradypus variegatus and Choloepus hoffmanni show similar oral processing patterns, suggesting competition for food, while Bradypus pygmaeus, Bradypus tridactylus, and Bradypus torquatus exhibit moderate oral processing capacity, with feeding strategies based on behavioral preferences instead of mandible specialization.

1. INTRODUCTION

In the past, the suborder Folivora comprised numerous members that showed significant diversity, both morphologically and in habits. These included various species such as giant ground sloths, semi‐aquatic sloths, and semi‐arboreal sloths, which inhabited different habitats across South and North America (McDonald, 2005). There are only two extant genera of sloths with similar strict arboreal and herbivore habits, Bradypus (Linnaeus, 1758) and Choloepus (Linnaeus, 1758) (Chiarello, 2008), which are restricted to the Neotropical forests of South and Central America (Chiarello, 2008). These animals are challenging to observe in their natural habitats due to their camouflage and habitat preference for very tall trees and dense rainforests.

Given the difficulty of directly observing these elusive animals, utilizing indirect paleobiological investigation techniques frequently employed in paleontology, such as dental mesowear (e.g., Saarinen & Karme, 2017) and biomechanical simulations (e.g., Bargo & Vizcaíno, 2008 ), can be beneficial for studying these species. These methods can provide, for example, valuable insights into the behavior, feeding habits, and locomotory adaptations of sloths.

The use of finite element analysis (FEA) has emerged as a powerful tool for reconstructing and assessing the diets of extinct species (e.g., Gill et al., 2014; Serrano‐Fochs et al., 2015; Wroe, 2018; Young et al., 2012). FEA is a computational technique used by researchers to model stresses and strains on an object, such as a skull or jaw, when exposed to different loads during biting or chewing (Rayfield, 2007). By comparing the predicted responses of the object with the observed morphology of fossil specimens, it becomes possible to deduce the types of food a species could consume and the mechanisms by which they were processed.

Initially designed for engineering, this method has been successfully applied to the field of paleobiology, becoming a primary tool for exploring the biomechanics of bites and discerning potential hunting habits in carnivores. (McHenry et al., 2007; Rayfield, 2007). However, FEA has recently been applied to the study of herbivorous species such as ungulates (e.g., Fletcher et al., 2010), rodents (DeSantis et al., 2020; Morris et al., 2022), marsupials (Sharp, 2015; Sharp & Rich, 2016), and xenarthrans (Melki et al., 2022; Serrano‐Fochs et al., 2015) to better understand their feeding habits and the evolution of their jaw and tooth adaptations.

This study aims to evaluate the feeding habits of the six extant sloth species from an ecomorphological standpoint using FEA to model the lateral chewing forces on their jaws. We aim to determine if the mechanical behavior observed is directly related to the reported feeding patterns of these species. Based on previous studies using FEA, we expect that more frugivorous taxa will exhibit higher overall stresses throughout the mandible, whereas more specialized folivorous taxa will display lower stress values. We also aim to analyze if the specific areas where higher and lower stresses are localized can be associated with differences in feeding habits. Although the dietary behaviors of living sloths can be observed in their natural habitats (even if with some difficulty), employing FEA to examine the mechanical aspects of their jaw during chewing could serve as a valuable validation tool to compare them with their extinct relatives.

2. MATERIALS

To construct the plane models, we gathered images from sloth mandibles in lateral and occlusal views. Pictures of specimens were taken by the senior author from the following specimens of the Mastozoological Collection of the Museu Nacional, in Rio de Janeiro, Brazil: Bradypus tridactylus (MN 2383), Bradypus variegatus (MN 26929), Bradypus torquatus (MN 5632), and Choloepus didactylus (MN 5018). The photographs were captured with a Samsung cellphone camera mounted on a portable tripod to ensure stability. Specimens were fixed using modeling clay and positioned at a 90° angle to the camera, effectively minimizing the parallax effect that could distort measurements. Images of Choloepus hoffmani (USNM 310364) and Bradypus pygmaeus (USNM 579177) were taken from the literature (Hayssen, 2008, 2011).

3. METHODS

The specimens' images were used to construct plane models using the open‐source software Fiji (Schindelin et al., 2012) and Rhinoceros 3D (version 7.0. Robert McNeel and Associates Seattle, WA). The resulting file was then exported to COMSOL Multiphysics® (v.5.3. for Windows; COMSOL AB, Stockholm, Sweden) for mesh construction and FEA. The decision to use planar models was influenced by two key factors: accessibility, particularly for specimens that were challenging to get access (especially B. pygmaeus) and simplicity of comparison. Utilizing planar models allowed for straightforward comparisons with minimal hardware requirements. Although tridimensional models are considered more accurate for testing scenarios and analyzing absolute loads (Smith et al., 2023), planar models are effective not only for qualitative comparisons of ecomorphological proxies (Melki et al., 2022) and quantitative assessments of stress (Marcé‐Nogué et al., 2017).

We selected the temporalis and masseter muscles (Turnbull, 1970), as representatives of the mastication loads (Melki et al., 2022). Similar to Melki et al. (2022), we will not include the analysis of the pterygoid in this study, as it would require considering loads that are not within the plane under examination. Insertion areas were outlined by examining muscle scars on specimens (Figure 1). Relative loads were calculated by the quasi‐homothetic transformation method (Marcé‐Nogué et al., 2013), with B. variegatus as the reference model (see Data S1 for measurements and load calculation). The selection of B. variegatus was based on the abundance of information for this species in the literature, its wider geographical distribution, and the accessibility of specimens.

FIGURE 1.

Muscle insertion and origin areas used for this study. (a) Bradypus variegatus; (b) Bradypus tridactylus; (c) Bradypus torquatus; (d) Bradypus pygmaeus; (e) Choloepus didactylus; and (f) Choloepus hoffmanni. The hatched areas indicate muscle that is behind the bone in lateral view.

All measurements were taken from pictures using Fiji software, and calculations were made using Microsoft Excel (Microsoft Corporation, v16.0). The mean thickness was calculated by measuring three homologous points on the mandible of each specimen in occlusal view. The physical parameters applied—the Young's modulus, or elasticity (E = 21 GPa) and Poisson's ratio (v = 0.45)—were taken from Macaca mulatta (Dechow & Hylander, 2000), while the bone density input (ρ = 1908 kg/m³) was based on the standard bone material from COMSOL Multiphysics. All structures were deemed isotropic, homogenous, and composed of linearly elastic material. The finite element mesh was defined as “very fine” (3000–9000 elements, as per software parameters—see Data S1 for mesh details and convergency test), composed only of triangular elements. Constraint points were set at the tip of the mandible and the top of the condyle in all directions to simulate the temporomandibular joint. The movement vector representing mandible closure was determined by connecting a straight line between the centroids of the muscle origin and insertion sites (Figure 1, see arrows in each model).

In paleontology, the FEA typically adopts a biomechanical focus, comparing various biomechanical scenarios and hypotheses (e.g., Janis et al., 2020; McHenry et al., 2007). Traditionally, warmer colors are linked to higher stress levels, while cooler colors are associated with lower stress levels (Marcé‐Nogué et al., 2017). Consequently, the interpretation of colors has often been associated with the robustness of the bone structure and the potential behaviors that the specimen could exhibit (Marcé‐Nogué, 2020). However, beyond identifying areas of higher or lower stress, we suggest using the von Mises stress diagram to demonstrate how loads are distributed, rather than just pinpointing the highest or lowest stress levels. The ecomorphological emphasis given by Melki et al. (2022; and further refined in this study) lies in discerning which patterns correlate with those that are associated with specific behaviors rather than analyzing absolute numerical values. We partially modified the interval method developed by Marcé‐Nogué et al. (2017) to suit the constraints and goals of our investigation, emphasizing a qualitative approach.

The intervals were established by utilizing the color scale of the von Mises stress diagram, following the standardization of stress values. The stress values of each specimen were normalized, with the highest stress value set to 1 and the lowest to 0, resulting in a scale from 0 to 1 (Melki et al., 2022). To visualize and compare homologous stress areas more effectively, we divided each mandible into three sectors labeled as S1, S2, and S3 (Figure 2). The areas were partitioned following a visual evaluation considering the distribution of stress areas in each model (Melki et al., 2022). The goal was to ensure consistency in homologous regions across specimens while analyzing significant stress areas. Two lines were delineated in each model: one from the ascendant ramus joining the mandible corpus to the inflexion point of the base of the mandible, and another from the midpoint of that line to the midpoint of the condyle and coronoid process (Figure 2). Three sectors are formed: S1, comprising the condyle and mandibular angle; S2, covering the ascending ramus area; and S3, including the most anterior portion of the mandibular body.

FIGURE 2.

Mandible partitioning. Example of the separation of the mandible model in sectors and stress magnitude areas using the Bradypus variegatus model. (S1) condyle and mandibular angle; (S2) ascending ramus area; and (S3) anterior portion of the mandibular body.

We measured each stress band separately within the sectors and identified three stress areas: (i) high‐stress area (HS), orange to red stress in the scale, with stress values ranging from 0.75 to 1.00; (ii) intermediary stress area (IS), cyan to yellow colors, with stress values from 0.26 to 0.74; and (iii) low‐stress area (LS), dark blue to blue colors, with stress values from 0.00 to 0.25 (see Figure 2 for a visual example). There are a total of nine areas, with HS1, IS1, and LS1 representing Sector 1, and so on. Area separation was achieved by adjusting stress range visualization in COMSOL Multiphysics, and areas were quantified using Fiji. The measurements were structured in a Microsoft Excel table and converted into proportions relative to the overall area of each model. Homologous models, such as B. variegatus (HS1) and B. torquatus (HS1), were compared in those areas.

The cluster analysis was applied to the proportion of stress areas using the Q‐mode and Euclidean coefficient with a bootstrap of N =10,000 to group similar models together. Finally, the cophenetic correlation coefficient was examined to determine the reliability and significance of the resulting groups. The cophenetic correlation coefficient is a measure used in hierarchical clustering analysis to assess the reliability of the dendrogram produced by the clustering algorithm, quantifying how well the dendrogram preserves the pairwise distances between the original data points (Kaesler, 1970). This metric reflects the degree of confidence in the accuracy of the clustering analysis and the formation of distinct groups: a cophenetic correlation coefficient close to 1 indicates that the dendrogram accurately reflects the similarities or dissimilarities between data points, while values close to zero or even negative values reflect the need to reconsider the analytical approach (Kaesler, 1970; Legendre & Legendre, 2012). All statistical analyses were conducted using the Paleontological Statistics Software (PAST) version 4.03 (Hammer et al., 2001).

4. RESULTS

All models, except C. didactylus, exhibit a similar distribution of stress areas throughout the mandible (Figure 3). As expected for herbivorous mammals, there is a noticeable prevalence of low‐stress areas (Fletcher et al., 2010). This pattern is especially evident in B. variegatus (Figure 3). All models showed minor stress peaks at the same anatomical regions: mandibular angle and coronoid (S1), ascendant ramus (S2), and anterior portion of the mandibular body (S3).

FIGURE 3.

Ecomorphological patterns of mandibles. Von Mises stress for the analyzed species (on the left) and divided by stress magnitude (on the right). After running the analysis, we standardized the stress scale so the highest stress value of each model would always be 1.00 (appearing as dark red) to exclude artificially inflated values and to permit better comparison between models, as we chose to measure color areas instead of punctual values.

Table 1 shows the results of the analysis on the relative percentage of stress in previously identified anatomical areas. Choloepus didactylus exhibits the largest stress area among the analyzed models (HA1 = 11.41%, HA2 = 4.54%, HA3 = 1.52%), whereas B. variegatus has the smallest stress area observed in this analysis (all low areas >93%), suggesting probable differences in oral processing capacities. On the other hand, B. torquatus and B. pygmaeus have similar area sizes, suggesting equivalent oral processing capacities.

TABLE 1.

Proportion of stress areas by sector for each model.

MODEL	HA1	IA1	LA1	HA2	IA2	LA2	HA3	IA3	LA3
B. variegatus	0.00	0.10	99.90	0.00	6.31	93.69	0.13	2.76	97.11
B. tridactylus	0.00	3.20	96.80	4.09	38.04	57.87	0.50	27.07	72.43
B. torquatus	0.05	4.57	95.39	1.71	28.49	69.80	0.42	19.46	80.12
B. pygmaeus	0.15	6.55	93.30	1.16	31.57	67.27	0.20	12.73	87.08
C. didactylus	11.41	58.72	29.87	4.56	40.95	54.49	1.52	58.52	39.96
C. hoffmanni	0.47	23.02	76.51	0.00	7.29	92.71	0.08	18.14	81.78

Note: All values are expressed as percentages. HA1 means the high‐stress area for sector 1, IA1 the intermediary stress area, LA1 the low‐stress area, and so forth. All high stress areas are colored red, intermediate stress areas in yellow and low stress areas in blue.

The cluster analysis, in turn, divided the species into two groups based on the relative percentage values of stresses observed in Table 1 (Figure 4), with C. didactylus being the external branch. Group 1 consists of C. hoffmanni and B. variegatus, while Group 2 consists of B. torquatus, B. pygmeus, and B. trydactylus. The cophenetic correlation coefficient of 0.969 suggests a significant difference between the groups. Most groups obtained bootstrap values exceeding 90, demonstrating reliable results (see Figure 4 for bootstrap values). The lowest bootstrap value (60, Group 1) indicates a significant difference in the values for the mandibular angle and coronoid area (S1), but the similarity between the S2 and S3 areas was stronger, bringing them closer together.

FIGURE 4.

Cluster analysis. C. didactylus is designated “Group 3,” colored pink for consistency. Group 1 is shown in green, while Group 2 is indicated in orange. Bootstrap values included. On the left is a small, simplified phylogeny for comparison. Phylogeny based on Presslee et al. (2019) with the addition of Bradypus crinitus (Miranda et al., 2022).

5. DISCUSSION

The results presented here align with the hypotheses proposed for herbivorous animals, whose jaws are typically described as “robust” (Fletcher et al., 2010), as indicated by the prevalence of low‐stress areas. Choloepus didactylus—an opportunistic generalist (Adam, 1999) and relatively more frugivorous (Saarinen & Karme, 2017) animal—is the only exception. Our data revealed that C. didactylus experiences greater stress on its jaws when faced with lateral chewing loads in comparison to other sloth species. This mechanical behavior may be related to a decreased requirement to process tougher foods, in line with a more frugivorous and opportunistic diet.

The association between C. hoffmanni and B. variegatus (Group 1; Figure 4) is particularly interesting as they belong to different genera, have a large geographic distribution, and have significant overlap in Central America (Kaup et al., 2021). Our analysis reveals that these two species possess the most robust mandibles among the living sloths, indicating that they employ a similar mode of oral processing during mastication and may engage in competition for food resources. Despite this similarity, B. variegatus demonstrates relatively more effectiveness than C. hoffmanni in food processing. This characteristic can be attributed to the significant differences in the posterior portion of their jaws, with C. hoffmanni displaying larger, higher and moderate stress areas (HA1 and IA1, see Table 1). The case of these two species highlights that some species in this study coexist in the same areas (see Figure 5) and potentially share resources, emphasizing the importance of considering the degree of interspecific competition when interpreting oral processing patterns.

FIGURE 5.

Distribution map of sloth species and their ecoregions of occurrence. Modified from Griffith et al. (1998) and Kaup et al. (2021).

According to Montgomery and Sunquist (1978), B. variegatus is likely the most competitive folivore in the Panamanian forests, forcing other local folivore species (mammalian or otherwise) to supplement their diet. Our findings partially support this observation, demonstrating that B. variegatus has the highest oral processing capacity among all the analyzed species. Additionally, our findings demonstrate that C. hoffmanni, which shares resources with B. variegatus at the location studied by Montgomery and Sunquist, displays a similar oral processing pattern, indicating a similar feeding strategy.

The oral processing pattern of B. variegatus may improve the animal's feeding efficiency, permitting less digestible leaves to be more easily digestible. This would be in line with their selective feeding behavior based not on specific plant species but on specific physical characteristics of any available trees in the environment (Montgomery & Sunquist, 1978). These characteristics include not only the overall digestibility of leaves (higher protein content, less fiber, and the presence of chemical defenses) but also the availability of young leaves throughout the year (leaf flush). More efficient oral processing would enable the animal to cope with different levels of seasonality by adjusting its diet to the most available resource, potentially playing a crucial role in facilitating the species' wide geographical distribution. Our results suggest that C. hoffmanni could also engage in this strategy, as it is also a species that spans across biomes (Sánchez‐Chavez, 2021), albeit more restricted than B. variegatus. Additionally, B. variegatus seems to dominate most of South and Central America, with other sloth species presenting a more restricted distribution pattern (Figure 5).

The grouping of B. pygmaeus, B. tridactylus, and B. torquatus (Group 2 of the cluster analysis) is noteworthy because these species have a large phylogenetic distance, with B. torquatus being the most basal species within the genus (Schetino et al., 2017). Furthermore, it is important to highlight that these three species exhibit distinct geographical distributions, which are more limited and isolated compared to other species (Kaup et al., 2021; Figure 5).

The oral processing pattern seen in the species of Group 2 appears to be an intermediary between Group 1 and C. didactylus. Unfortunately, B. torquatus is the only member of this group whose diet has been extensively studied (Giné et al., 2022; Mureb et al., 2023). However, this species can serve as a basis for understanding the processing patterns of the other two species. Bradypus torquatus has individual‐level preferences for consuming specific plant species, with a preference for young and easily digested foliage (Giné et al., 2022). This species inhabits regions with low seasonality (e.g., montane Atlantic Forest), where leaf flush is more constant throughout the year (Jones et al., 2014). Our result implies that B. torquatus employs a distinct feeding strategy from the one observed in the first group (C. hoffmanni and B. variegatus). Instead of prioritizing feeding efficiency through oral processing, B. torquatus relies on a consistent supply of young and easily digestible leaves, as observed in the literature (Giné et al., 2022), consistent with the interpretation of our analysis.

In contrast, there is limited data regarding the B. pygmaeus diet in literature, only been observed consuming Rhizophora mangle leaves (Anderson & Handley Jr, 2002). These leaves become less palatable as they mature due to salt accumulation (Hogarth, 2015). This suggests that these leaves are likely mainly consumed when they are younger, in a case similar to B. torquatus. Finally, due to a taxonomic history muddled by synonymy with part of the species B. variegatus, B. tridactylus has a poorly studied diet (Moraes‐Barros et al., 2010). However, based on the stress pattern observed here, we infer that it is likely that this species consumes food items with physical properties similar to the young leaves consumed by B. torquatus and B. pygmaeus.

Despite all being categorized as suspensory folivorous browsers, extant sloths demonstrate notable differences in our analysis. Upon reviewing our data, we identify three strategies that can be shown by the ecomorphological patterns of each group: Group 1 focuses on improving feeding efficiency through enhanced oral processing, reflected in mandibles with lower stress levels (referred to as “more bluish”), while the strategy of Group 2 involves a preference for easily digestible leaves, restricting the species to smaller geographic distributions in less seasonal areas. This is evidenced in the mandible's reduced ability to deal with higher loads, resulting in intermediary stress areas. Finally, the third strategy would be to complement the folivorous diet with fruits and other items, with C. didactylus displaying higher stress areas under the same loads. These strategies would materialize as a gradient inside the browsing habit, with Group 1 exhibiting a high degree of specialization to process various types of leaves throughout each season (permitting a broader distribution of the species) and Group 2 primarily focusing on consuming young, easily digestible leaves (restricting the species to more stable, smaller areas) while remaining predominantly folivorous. Choloepus didactylus leans closer toward a frugivorous diet. Additionally, there is a geographical distribution factor, wherein more widespread species appear to prioritize enhancing their feeding efficiency to expand their habitat range across various biomes.

This study can be related to the previous research conducted by Melki et al. (2022), which focused on comparing some extinct sloth species with generalized models of two extant genera. Melki et al. (2022) encompassed a wider variety of behaviors, such as grazing, which is not observed in extant sloths. The study also questioned the generalized “browser” behavior (Bargo & Vizcaíno, 2008), which comprises biomechanically distinct habits such as folivory, frugivory, and root‐eating (Solounias & Moelleken, 1993). In addition, the prevalent assumption that robust mandibles indicate grazing animals and less robust mandibles indicate browsing animals does not consistently align with observed results (Sharp, 2015). This assumption has been recognized as a substantial oversimplification (Clauss et al., 2008; Saarinen, 2019). The insights from this study suggest that FEA can offer a broader perspective beyond simply categorizing animals as “browsers” or “grazers,” making it a valuable tool for understanding dietary niches. Our objective in applying this analysis to extinct sloths is to utilize the knowledge acquired from studying the feeding behaviors of contemporary sloths to infer the behaviors of extinct species. Future endeavors may involve conducting additional comparisons among sloths using this ecomorphological approach.

6. CONCLUSION

The results presented in this study corroborate hypotheses regarding the robustness of jaws in herbivorous animals, as evidenced by the prevalence of low‐stress areas. Particularly noteworthy is the observation that C. didactylus, known for its opportunistic generalist behavior and relatively higher frugivory, experiences greater jaw stress during lateral chewing compared to other sloth species. This aligns with its observed dietary preferences, suggesting a reduced emphasis on efficient oral processing. The similarity in oral processing patterns between C. hoffmanni and B. variegatus, despite their distinct genera, hints at the competition for food resources. Our findings lend support to the notion of B. variegatus as the most competitive folivore in Panamanian forests, while also indicating a similar feeding strategy in C. hoffmanni. This strategy would be crucial for the widespread distribution of B. variegatus across diverse biomes in South and Central America.

The grouping of B. pygmaeus, B. tridactylus, and B. torquatus displays an intermediary oral processing pattern, exhibiting a distinct feeding strategy based on behavioral preferences rather than oral processing efficiency that favors feeding on young and more digestible leaves. Notably, this group comprises species with more geographically restricted distributions, confined to areas with minimal seasonal variations and consistent food availability throughout the year. This study highlights the variability in feeding strategies among sloth species and emphasizes the importance of ecomorphological approaches in understanding dietary niches. Furthermore, our findings suggest that FEA offers valuable insights beyond simplistic categorizations, providing a framework for studying both extant and extinct sloth species. Future research should explore these insights further through additional comparative analyses among sloths.

AUTHOR CONTRIBUTIONS

LBM and FHSB conceived the ideas and designed the methodology. LBM also analyzed the data and led the writing of the manuscript. LAS contributed to the analysis and methodology. LPB is the academic advisor of LBM and contributed with the means to accomplish the project as well as with the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

FUNDING INFORMATION

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) awarded the senior author's grant [88887.481061/2020‐00] that made this study possible.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1.

Melki, L.B. , Barbosa, F.H.S. , Alves‐Silva, L. & Bergqvist, L.P. (2024) Variations in dietary patterns of living sloths revealed by finite element analysis of jaws. Journal of Anatomy, 245, 583–592. Available from: 10.1111/joa.14099

DATA AVAILABILITY STATEMENT

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.