Puncture performance tests reveal distinct feeding modes in pinniped teeth

Marine mammals are secondarily aquatic vertebrates that are ideal for studying drivers of morphological transformations and ecological transitions across a group's evolutionary history (Pyenson, 2017; Uhen, 2007). Pinnipeds are a monophyletic clade of marine carnivorans that evolved from terrestrial ancestors approximately 27 million years ago (Berta et al., 2018). Their evolutionary history is documented by a series of stem pinniped fossils that capture many changes in morphology that are associated with a marine environment (Berta et al., 2018).

The rise and success of mammals is often linked to the origin of two dentition-related key innovations: a differentiated dentition and a tribosphenic molar. These innovations facilitate precise occlusion of the upper and lower dentition, which co-evolve with the rise of mastication, which is a hallmark of mammalian feeding (Gregory, 1921; Herring, 1993; Herring et al., 2001; Hiiemae, 2000; Ungar, 2010; Weijs, 1994). Mastication is a highly specialized behavior with a precise definition that requires a power stroke of the lower jaw as well as precise occlusion between the upper and lower dentition (Ahlgren, 1966; Herring, 1993; Herring et al., 2001; Hiiemae, 1978; Hiiemae, 2000; Weijs, 1994). Despite having descended from terrestrial mammals, extant pinnipeds lack a tribosphenic molar and do not masticate (Jones et al., 2013; Marshall and Goldbogen, 2016; Marshall and Pyenson, 2019). Although some pinnipeds do process prey by chopping, ripping or tearing with their teeth (Hocking et al., 2017a), they lack the precise occlusion and the talonid basin emblematic of mastication. Instead, many extant pinnipeds have a dentition that is secondarily reduced or simplified, resulting in pinnipeds being described as homodont or functionally homodont by numerous authors (Berta et al., 2018; Jones et al., 2013; Marshall and Pyenson, 2019; Uhen, 2018). These changes are associated with a return to the aquatic environment and an emphasis on the use of teeth for prey capture over prey processing (Marshall and Pyenson, 2019). These changes and the resulting performance of cheek-teeth for feeding by pinnipeds are of interest from an evolutionary and biomechanics perspective.

Although the dentition of extant pinnipeds has become simplified and lacks the precise occlusion characteristic of terrestrial mammals, and although pinnipeds do not masticate, there nevertheless remains a notable degree of variability in the morphology of their cheek-teeth (Fig. 1). The breadth of cheek-teeth morphology in pinnipeds ranges from simple conical pegs to complex, multi-cusped, trident-shaped teeth, which include varying degrees of wear. These morphologies potentially relate to the diversity of pinniped feeding modes, which include raptorial biters, suction specialists, filter feeders and multimodal generalists. These feeding modes can span multiple prey types including fish, squid, krill and tetrapods (Marshall et al., 2008; Marshall and Pyenson, 2019; Marshall et al., 2014; Pauly et al., 1998). In terrestrial mammals, dental morphology is often used as a proxy for studying feeding ecology, especially for fossil specimens (e.g. Damuth and Janis, 2011; Gill et al., 2014; Luo et al., 2011).

To understand the relationship of varying dental morphologies with feeding in pinnipeds, we conducted puncture performance tests, using models of the cheek-teeth of seven extant pinniped genera, as measured by the maximum force to puncture (F_max) and the maximum energy to puncture (E_max). We tested the hypothesis that different pinniped dental morphologies exhibit significant differences in their ability to puncture prey. We predicted that pinnipeds that use their teeth to capture prey raptorially would require the least force and energy to puncture, indicating that they are effective puncturers. Conversely, we predicted that pinnipeds that capture prey via suction would require the most force and energy to puncture and speculate that they may even be incapable of puncturing the prey item (ineffective puncturers). Finally, we predicted that more complex, multicusped teeth puncture prey intermediate to the simple, conical teeth of raptorial biters and that of suction feeders (i.e. moderate puncturers). In addition, we quantified the shape of cheek-teeth within each tooth row using orientation patch count rotated (OPCR) and Relief Index (RFI) and tested the hypothesis that these metrics correlate with puncture performance. We predicted that increased tooth complexity would be correlated with increased puncture performance (i.e. lower F_max and E_max).

To build a sample dataset with a phylogenetic context, we chose seven extant pinniped taxa (Families Phocidae and Otariidae) that spanned the breath of dental morphologies and known feeding ecologies. Our sample includes one specimen of each taxon, including four phocids (Lobodon carcinophagus, Hydrurga leptonyx, Ommatophoca rossii and Mirounga leonina) and three otariids (Zalophus californianus, Callorhinus ursinus and Arctocephalus pusillus). This taxonomic sampling includes specialized and generalized raptorial biters (Z. californianus, H. leptonyx, C. ursinus and A. pusillus) (Marshall et al., 2015), specialized suction feeders (Mirounga leonina and O. rossii) (Bryden and Felts, 1974; Hocking et al., 2013; Kienle and Berta, 2019; King, 1964) and filter feeders (L. carcinophagus) (Hocking et al., 2013; Klages and Cockcroft, 1989; Marshall and Pyenson, 2019; Ross et al., 1976). We chose only adult specimens with a complete dentition that lacked any obvious pathologies or obvious abnormal phenotype. Each of the genera selected represent a minimal number of extant species, with similar feeding modes, with the exception of Arctocephalus, which is a diverse genus that includes as many as eight species (Brunner, 2004) and encompasses a wide diversity of feeding modes (Hoskins et al., 2015). Our results are therefore only informative for A. pusillus, the only Arctocephalus species included in this study.

We scanned the skulls and mandibles in occlusion of each specimen using Nikon Metrology's combined 225/450 kV microfocus X-ray computed tomography (CT) walk-in vault system at National Technical Systems (Belcamp, MD, USA). CT slice thickness was 0.03 mm. DICOM files were processed in Mimics (Materialise NV, Leuven, Belgium) and 3D models of the cranium and mandibles were created, and are archived on Zenodo (https://zenodo.org/record/5236563). The skull of A. pusillus (NMV C5717) was scanned by the Evans EvoMorph lab using a Siemens Somatom GoUp medical CT scanner at Monash Biomedical Imaging, a technology research platform at Monash University, Australia, and is archived on Sketchfab (https://sketchfab.com/3d-models/skull-of-the-australian-fur-seal-24b0f9ae93d94fb79fb203cd6b4ec5f9). The skull of Z. californianus was scanned at the University of Texas High-Resolution X-Ray Computed Tomography Facility using a Varian Medical Systems ACTIS scanner. Access to these data was provided by Blaire Van Valkenburgh and Tim Rowe via MorphoSource (Duke University) under NSF IOB-0517748 and DBI-1902242.

Each 3D model was manually trimmed to include only the dental arcade of the post-canine cheek-teeth and then 3D printed at full scale by Shapeways (New York, NY, USA) using the ‘Versatile Plastic’ option and a natural finish. We manually inspected each print for quality and signs of anomalies during the printing process. We chose Versatile Plastic to ensure a consistent and standard material across each print, not to mimic the material properties of enamel. This ensured that our methodology is easy to replicate for both extant and fossil specimens for future work.

3D printed tooth rows were attached to 25.0 mm diameter rectangular steel tubing cut into sections ranging from 57.2 to 123.3 mm in length. Tooth rows were attached to these steel sections using Sonic-Weld epoxy putty (Tallahassee, FL, USA) and cured at room temperature (20–22°C) for 24 h. The 3D print-epoxy was reinforced with industrial-strength cyanoacrylate. The distance from the base of the steel tubing and the longest cusp apex ranged from 50.8 to 71.1 mm, and tooth cusps were set orthogonally to the steel section base.

Restaurant-grade market squid (Loligo sp.) was chosen as squid is a prey item that is consistently consumed by most pinnipeds (Pauly et al., 1998). Also, squid mantles represent a relatively homogeneous biological structure that allowed us to control for variation in tissue properties compared with the more heterogeneous structures of fishes. For example, Whitenack and Motta (2010) showed that varying fish scale thickness significantly changed tooth puncture performance of shark teeth. Squid were received and stored frozen, then thawed at room temperature for 4–6 h prior to puncture performance testing. We measured squid mantle lengths in ImageJ (Abràmoff et al., 2004) as the distance from the tip of the mantle collar to the posterior end of the mantle; these values ranged from 111.22 to 201.5 mm. Prior to testing, squid were placed on a wooden table lined with sandpaper (120 grit) to prevent movement during puncture tests following Whitenack and Motta (2010). To prevent inconsistent puncture through shifting of the internal pen, squid mantles were oriented 30 deg clockwise to their longitudinal axis and positioned so that the apex of the centermost tooth (rows with five teeth) or the location between the two centermost-teeth (rows with four teeth) was aligned with the center, or longitudinal axis of the squid mantle (Fig. 2). Each trial began with the longest apex of every tooth row positioned 50 mm above the squid mantle. A C-clamp was used to attach the steel tubing and tooth rows to the MTS grip and 2.5 kN load cell of the MTS Insight 5 system (Eden Prairie, MN, USA; Fig. 2). For each puncture trial, tooth rows were driven into the mantle of the squid at a displacement rate of 15.00 mm s⁻¹. The sampling rate of the MTS machine was 10 Hz.

For each tooth row, five squid were punctured for a total of 10 puncture performance trials per treatment (2 trials per squid). For the first trial for each individual squid, the pen was left intact and the tooth row driven in at the location where the center tooth (or the location between the two centermost teeth) aligned with 25% of the mantle length. In preparation for the second trial, the pen was manually removed from the mantle; no cutting was required. Here, the tooth row punctured the mantle at 75% of the mantle length, parallel to the first puncture site (Fig. 2). Load–displacement data were collected in TestWorks4 software (version 4.11 MTS). Maximum puncture force (F_max) and energy to maximum puncture force (E_max) were calculated using TestWorks4 software. Load–displacement curves from preliminary tests showed no drop in load at the end of the test when teeth punctured the squid mantle. Therefore, tooth row displacement, or the vertical distance moved, was terminated when the longest tooth apex of each row reached a distance of 3 mm above the wooden specimen support table (Fig. 2). This protocol ensured standardization of puncture distance and also prevented damage to the load cell or 3D printed models. Each 3D print was visually inspected for signs of deformation or degradation after each trial and no such problems were observed.

To further understand the extent to which morphology explained our performance results, we quantified the shape of the tooth crowns using two established morphometric measurements: OPCR and RFI (Table S1). OPCR is a geographic information system analysis developed by Evans et al. (2007) and refined by Evans and Janis (2014). This method measures the dental complexity of a tooth or tooth row by binning continuously adjacent faces of identical orientation into patches and then measuring the number of distinct patches on the model. Using this metric, higher OPCR scores indicate more variation in the topography of the tooth or tooth row, with many distinct patches each with distinct orientations. Conversely, lower OPCR scores represent a simpler model overall, with large patches of homogeneous regions all facing the same orientation. The lowest possible OPCR score is 8, which represents a perfect cone with faces all binned into 8 arbitrary directions.

Unlike OPCR, which measures dental complexity, RFI measures the overall height of the crown. The original RFI formula, first used by Ungar and M'Kirera (2003), was a simple ratio of the tooth crown's surface area to the 2D footprint area. More recently, authors have altered this formula so that RFI instead applies transformations to these two values (Boyer, 2008; Pampush et al., 2016). In each case, the method compares the 3D surface area of the tooth crown with the 2D footprint at the crown's base. Using this metric, a ratio of 1 indicates a very tall tooth, with a high 3D surface area relative to its 2D footprint, while a ratio that approaches 0 indicates a shorter tooth, with a large 2D footprint relative to its 3D surface area. Here, we follow the best practices suggested by Pampush et al. (2016) and use the weighted ratio version of the metric (Boyer, 2008).

We measured OPCR and RFI using the R package molaR (Pampush et al., 2016), which was created specifically for quantitative topographic analyses of teeth. The 3D surface models of each tooth row were trimmed to include only the crowns. Then, the models were simplified to a standard number of faces (1000) and oriented such that the occlusal surface was aligned with the z-axis. The molaR R package was used to measure OPCR and RFI following Pampush et al. (2016). Our R code is provided in the electronic supplemental materials (ESM).

All data were tested for normality using a Shapiro–Wilk test (Shapiro and Wilk, 1965) and subsequently log transformed. A Pearson correlation was conducted between squid mantle length and the maximum force to puncture (F_max) and maximum energy to puncture (E_max) to test whether mantle length significantly impacted puncture performance metrics (Figs S1, S2). To determine significant differences (α≤0.05) in F_max and E_max, we considered each tooth row separately, as they were used in the experimental trials. Therefore, because our study includes seven species, we considered 14 total tooth rows (seven upper tooth rows and seven lower tooth rows). A one-way MANOVA was performed to test the independent variable (tooth row) against the two dependent variables (F_max and E_max). A second one-way MANOVA was conducted to test the independent variable (squid pen in situ versus pen excised) against the two dependent variables (F_max and E_max). Both one-way MANOVAs used the entire dataset including all trials with the pen in situ and with the pen excised. A Pearson correlation plotted both metrics of dental complexity (OPCR and RFI) against F_max and E_max to test for correlation between dental complexity and puncture performance metrics. All statistical tests were conducted using the R package DPLYR (https://CRAN.R-project.org/package=dplyr).

The smallest squid mantle in these experiments measured 111.2 mm and the largest measured 201.5 mm. The mean (±s.d.) mantle length was 146.2±22.3 mm. To test whether the squid's mantle length potentially biased results, we conducted a Pearson correlation of mantle length against the maximum force to puncture and the maximum energy to puncture for each trial. We found no correlation between squid mantle length and maximum force to puncture (Pearson's r=−0.117, t=−0.974, d.f.=68, P=0.334). Similarly, we found no correlation between squid mantle length and maximum energy to puncture (Pearson's r=−0.080, t=−0.663, d.f.=68, P=0.510). Thus, squid mantle length had no correlation with the puncture performance metrics measured.

We conducted a one-way MANOVA to test for significant differences (α≤0.05) of the independent variable (tooth row) against the two dependent variables (F_max and E_max) using Pillai's trace test. Tooth row had a statistically significant association with both F_max and E_max (Pillai's trace=0.954, F_26,252, P<0.0001). A subsequent one-way MANOVA was conducted to test for significant differences (α≤0.05) of the independent variable (pen in situ versus pen excised) against the two dependent variables (F_max and E_max), also using Pillai's trace test. The presence of the pen had a statistically significant association with both F_max and E_max (Pillai's trace=0.113, F_2,137, P=2.59×10⁻⁴). Therefore, both independent variables had a significant association with both the maximum force to puncture (Fig. 3, Tables 1 and 2) and maximum energy to puncture (Fig. 4, Tables 3 and 4). Two post hoc Tukey's HSD tests were conducted to compare differences in the maximum force (Table S3) and maximum energy (Table S4) to puncture among the different tooth rows. These post hoc tests demonstrated significant differences (Table S2) between the best performing tooth rows (e.g. H. leptonyx lowers, Z. californianus lowers) and the worst performing tooth rows (L. carcinophagus lowers, O. rossii lowers).

Puncture performance data showed key patterns that broadly split the seven genera into three distinct performance groups. The most effective puncturers (lowest force and energy to penetrate) included C. ursinus, Z. californianus and H. leptonyx; each of these taxa had at least one set of teeth which punctured with overall low force values (for example, a mean from the five performance trials of less than 20 N with the pen in situ). The next group, which included A. pusillus, M. leonina and L. carcinophagus, are moderate puncturers. These taxa had some trials where the maximum force to puncture approached those of the first group, but altogether the mean and the standard error of the mean of the five trials was greater, suggesting that this second group were capable of strong puncture performance, but were more variable and less consistent than the first group. Finally, O. rossii was an ineffective puncturer that required the greatest maximum force to puncture compared with the other genera in this study; the mean maximum force to puncture of its five performance trials was >40 N with the pen in situ.

The lower dentition of five of the seven genera (Callorhinus, Zalophus, Arctocephalus, Mirounga and Hydrurga) in this experiment performed better than the upper dentition. This difference was particularly pronounced in C. ursinus and Z. californianus; each of these taxa had no overlap among any of the trials for their upper and lower dentition. Conversely, O. rossii and L. carcinophagus each performed better with the upper dentition, with L. carcinophagus in particular exhibiting a greater discrepancy between the force and energy required to puncture for the upper and lower tooth rows.

Values for OPCR – one of the two metrics used to quantify the overall shape of the tooth row – ranged from 47.0 (M. leonina) to 112.6 (H. leptonyx) (Figs 1 and 5). The multicusped pinnipeds (H. leptonyx and L. carcinophagus) exhibited the greatest OPCR values, whereas the more conical teeth had the lowest OPCR values (M. leonina and A. pusillus). We found no correlation between dental complexity (OPCR) and the maximum force to puncture (Pearson's r=−0.024, t=−0.083, d.f.=12, P=0.936), or between dental complexity (OPCR) and maximum energy to puncture (Pearson's r=0.198, t=0.702, d.f.=12, P=0.496).

RFI values ranged from 0.144 (O. rossii) to 0.393 (L. carcinophagus) (Figs 1 and 6). The lowest RFI values were obtained for O. rossii and M. leonina. Both taxa exhibited heavy wear of the dental crowns and lacked accessory cusps, which is typical of adult specimens of these species (King, 1983; C.D.M., personal observation). The multicusped teeth of H. leptonyx and L. carcinophagus resulted in similar RFI values to those of A. pusillus and Z. californianus, suggesting that multiple cusps alone do not necessarily increase RFI values. The teeth of C. ursinus yielded intermediate RFI values, suggesting that although the crown is quite tall, a relatively broad base lowers the RFI value. We found no correlation between crown height (RFI) and the maximum force to puncture (Pearson's r=−0.429, t=−1.644, d.f.=12, P=0.126), or between crown height (RFI) and the maximum energy to puncture (Pearson's r=−0.248, t=−0.887, d.f.=12, P=0.393).

Extant pinnipeds do not masticate and they lack a dentition specialized for precise occlusion (Marshall and Pyenson, 2019), such as that seen in ungulates, primates and rodents (Boyer, 2008; Evans and Janis, 2014; Evans et al., 2007; Ungar, 2010; Winchester et al., 2014). Despite this, pinnipeds, particularly phocids, display a high degree of variability in their cheek-teeth morphology (Fig. 1). Previous studies have sought to describe trends in the evolution of pinniped feeding by correlating dental morphotypes with specific feeding ecologies (Adam and Berta, 2002; Boessenecker, 2011; Churchill and Clementz, 2015a; Churchill and Clementz, 2015b; Hocking et al., 2017b; Hocking et al., 2017c; Kienle and Berta, 2016; Kienle and Berta, 2019; Kienle et al., 2017; King, 1961; King, 1969; King, 1983; Ross et al., 1976). However, to date there is a notable lack of experimental work that tests functional hypotheses. Here, we tested whether specific dental morphologies are better at prey puncture. Our results demonstrate significant differences in puncture performance (both F_max and E_max) among the seven genera in this study. Specifically, we identify three distinct functional groups based on their cheek-teeth morphology and puncture performance: effective puncturers, ineffective puncturers and moderate puncturers.

Callorhinus ursinus, Z. californianus and H. leptonyx were all effective puncturers. This group had the best performing set of cheek-teeth (defined as the minimal force or energy to puncture the prey item); they consistently required <20 N of force to puncture and sometimes <10 N. This group also encompasses the most disparate dental morphologies, from the conical teeth of C. ursinus to the multicusped teeth of H. leptonyx. Based on known feeding performance, C. ursinus is a specialist raptorial biter that does not employ suction to capture food (Marshall et al., 2015). Conversely, H. leptonyx feeds on many different types of prey, ranging from penguins to krill, and is capable of multimodal feeding that includes raptorial biting (including grip and tear), suction and filter feeding (Hocking et al., 2013; Kienle and Berta, 2016; Krause et al., 2015). Our results demonstrate that the dental morphologies of these taxa require less force and less energy to puncture prey.

The group of ineffective puncturers includes O. rossii and M. leonina. Both taxa required greater force to puncture the squid (>20 N) and exhibited more variation in performance, sometimes requiring forces in excess of 40 N. Although only a few feeding performance tests have been conducted (Naito et al., 2013), O. rossii and M. leonina are known suction feeders and their teeth are known to exhibit heavy wear from an early ontogenetic age (Bryden and Felts, 1974; Churchill and Clementz, 2015b; Kienle and Berta, 2016; King, 1964; Marshall and Pyenson, 2019; McGovern et al., 2019; Nordøy and Blix, 2000; van den Hoff and Thalmann, 2020). However, they typically swallow their prey whole, and the dental wear is the result of abrasion from suction, similar to that seen in walruses (Kastelein et al., 1994; Marshall and Pyenson, 2019). Thus, there is a potentially minimal loading environment on their cheek-teeth. Our results suggest that there may be little to no selective pressure for piercing on the morphology and function of their cheek-teeth.

The third group, comprising A. pusillus and L. carcinophagus, were moderate puncturers. Arctocephalus pusillus did not perform as well as C. ursinus and Z. californianus, despite having teeth that superficially resemble those of the more effective puncturers (tall, triangular teeth). As with the suction feeders in this study, A. pusillus puncture performance was more variable than that of the raptorial biters. Arctocephalus is a diverse genus that includes as many as eight species (Brunner, 2004). The taxon in our study, A. pusillus, is sometimes characterized as a benthic forager based on diving records (Arnould and Costa, 2006; Arnould and Hindell, 2001; Deagle et al., 2009), but is known to target a wide range of both benthic and pelagic prey types (Deagle et al., 2009; Gales and Pemberton, 1994; Hoskins et al., 2015; Hume et al., 2004; Kirkwood et al., 2008; Littnan et al., 2007), suggesting that it is best characterized as a generalist feeder. Some evidence suggests that prey type, rather than where in the water column the prey is obtained, is a greater factor in how the prey is processed (Hocking et al., 2016). Our results suggest that the teeth of C. ursinus and Z. californianus are better puncture performers relative to the more generalist teeth of A. pusillus. These results substantiate performance data that claim C. ursinus is a biting specialist (Marshall et al., 2015). The performance values for L. carcinophagus were the most variable of the taxa in our study: maximum force to puncture ranged from 21 to >80 N. Lobodon carcinophagus is known to ingest and subsequently filter krill, likely using suction and hydraulic jetting feeding modes (Klages and Cockcroft, 1989; Marshall and Pyenson, 2019; Ross et al., 1976). Our results suggest that, although the teeth of L. carcinophagus do sometimes perform well in puncture experiments, their teeth exhibit a high variability in their capacity to puncture.

Although our results indicate distinct patterns based on their capacity to puncture, none of the tooth rows in our study failed to pierce the squid. We chose squid because they are a prey item consistently consumed by pinnipeds (Pauly et al., 1998) and because we expected it to be a soft, homogeneous prey item. However, our results showed significant differences in the maximum force and maximum energy required to puncture squid between trials with and without the pen, indicating that the squid pen is harder than we anticipated. Franco-Moreno et al. (2021) have recently reported resistance values for squid that are comparable to those of some teleost fishes, suggesting that it may not be as homogeneous as previously thought. Now that a baseline of puncture performance using squid has been completed, follow up studies should use fish, especially considering that squid with the pen removed is not a biologically relevant sample prey item. Fish can have an overall greater tissue hardness relative to squid, and can provide a greater challenge to puncture performance (Franco-Moreno et al., 2021). Fish represent a more heterogeneous structure with a mix of hard and soft tissues. Such performance tests will likely discriminate feeding modes further, while establishing potential dietary constraints imposed by tooth morphology.

The presence of distinct cheek-teeth morphologies with distinct functional capabilities is interesting given that pinnipeds lack the talonid basin and precise occlusion characteristic of their terrestrial ancestors. Understanding our work in the broader context of the evolutionary biomechanics of mammalian feeding will require substantial comparative work. The physical constraints of feeding in an aquatic versus terrestrial system pose many challenges. However, there are interesting questions regarding the changes in feeding performance in the transition to aquatic environments that can be elucidated. Our experiments demonstrate that these distinct morphologies differ in their capacity to puncture, but not all pinniped species necessarily use the cheek teeth during feeding. Further work might compare puncture performance metrics with experimentally measured bite forces in pinnipeds, to test whether the forces required to puncture are within the range of what living pinnipeds can produce, and to link behavioral data with our experimental results. Finally, future work comparing the puncture performance and dental complexity of extant pinnipeds with that of stem pinniped ancestors and terrestrial carnivorans may elucidate the timing and mechanisms for the loss of mastication in pinnipeds.

We thank N. D. Pyenson and D. J. Bohaska for facilitating access to specimens at the NMNH. We also thank the Evans Morphology Lab, D. P. Hocking, and A. R. Evans for access to the A. pusillus dataset and B. Van Valkenburgh and T. Rowe for access to the Z. californianus dataset. Finally, we thank National Technical Systems (Belcamp, MD, USA) and C. Peitsch, R. Peitsch and C. Schueler for providing access and resources for microCT scanning. We thank R. Perkins III (TAMUG) for help with specimen preparation for puncture tests.