Diet-related differences in craniodental morphology between captive-reared and wild coyotes, Canis latrans (Carnivora: Canidae)

Diet-related differences in craniodental morphology between captive-reared and wild coyotes,... Abstract In mammals, skull loading incurred during feeding and prey capture shapes craniodental morphology as a result of bone plasticity, but the degree to which skull form is influenced by mechanical loading is poorly understood. Here, we tested how craniodental morphology is affected by differences in diet and age between wild and captive-reared coyotes (Canis latrans) raised on a relatively soft diet lacking bone. For both samples, we quantified tooth wear, external skull dimensions and frontal sinus morphology. Skull dimensions were similar between both populations, but captive coyotes had significantly broader zygomatic widths and palates, and lesser tooth wear and tooth fracture. However, differences observed between captive and wild coyotes are less than those observed in larger, more hypercarnivorous carnivorans. Frontal sinus volume did not differ significantly between populations, but was significantly correlated with external cranial dimensions in wild, but not captive coyotes. We hypothesize that greater mechanical demands of a wild diet, as evidenced by greater tooth wear, resulted in a stronger correlation between frontal sinus form and external skull form in wild coyotes. In captive coyotes, reduced skull loading because of a softer diet reduces constraints on cranial morphology and results in less covariation between internal and external cranial morphology. INTRODUCTION Bone is a highly plastic tissue that is remodelled throughout an organism’s life in response to how it is loaded. This results in skeletons seemingly optimized to meet the mechanical demands of diet, locomotion and support (Wolff, 1892). In mammals, skull morphology is influenced by the mechanical properties of food, as well as prey capture and feeding behaviour (Herring, 1993; Van Valkenburgh, 1999; Samuels, 2009; Dumont et al., 2012; Santana et al., 2012). However, the degree to which skull shape within a species is affected by differences in food texture is poorly understood. The impact of the mechanical properties of food on mammal skull morphology is perhaps best understood from a limited number of studies comparing wild and captive-reared conspecifics (see review by O’Regan & Kitchener, 2005). Captive individuals included in these studies were typically reared on commercially produced foods with softer textures that presumably load the skull to a lesser degree than the wild-type diet, and captive individuals often differ in skull morphology from their wild counterparts. For captive lions (Panthera leo), leopards (Panthera pardus) and tigers (Panthera tigris), the consumption of a softer diet appears to result in atypical development of the zygomatic arches and palate, lesser doming of the dorsal roof of the skull, and a reduced size of crests that serve as attachment sites for the muscles that close the jaws and control head movement during prey capture and feeding (Hollister, 1917; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014). Cheetahs (Acinonyx jubatus) fed softer foods in captivity are more likely to suffer from malocclusion of the teeth attributable to minimal use and consequent atrophy of the masticatory muscles (Fitch & Fagan, 1982). However, in these studies, precise age and diet information are often unknown and could bias or obscure patterns in the data. Although the aforementioned studies suggest that external skull dimensions differ within species owing to differences in the mechanical properties of food, there is little information on how differences in skull loading impact the internal architecture of the skull. In mammal skulls, pneumatic cavities called paranasal sinuses are commonly found in the bones surrounding the nasal chamber, including the frontal, maxilla, ethmoid and sphenoid bones. Paranasal sinuses appear to develop in areas where bone is mechanically unnecessary and promote even dissipation of stress across the skull during use (Witmer, 1997, 1999; Farke, 2007, 2008, 2010; Tanner et al., 2008). In agreement with this, the morphology of the frontal bone and frontal sinus in Carnivora is correlated with diet and ecology across species (Curtis and Van Valkenburgh, 2014; Curtis et al., 2015). The frontal sinuses of carnivorans best adapted to produce high bite forces are typically large and underlie dome-shaped frontal bones, which appear to aid in uniform distribution of stress across the skull during feeding (Tanner et al., 2008; Tseng & Wang, 2010, 2011). The most extreme examples of this are seen in bone-cracking hyenas (Crocuta crocuta), bamboo-eating panda bears (Ailuropoda melanoleuca) and extinct bone-cracking canids (Borophaginae), in which the frontal sinuses pneumatize the sagittal crest (Davis, 1964; Wang, Tedford & Taylor, 1999; Tanner et al., 2008; Tseng & Wang, 2010, 2011; Curtis and Van Valkenburgh, 2014; Curtis et al., 2015). However, it remains unclear whether the link between frontal sinus morphology and cranial shape differences related to dietary differences persists at the intraspecific level and what role age plays in this relationship. Witmer (1997) hypothesized that development and maintenance of pneumatic cavities, including frontal sinuses, represents a compromise between pneumatic epithelia that opportunistically resorb bone and bone deposition in response to biomechanical loading. If this is the case, then similarity in sinus form among individuals within a species should be consistent with similar loading regimes on the skull. Correspondingly, differences in the mechanical properties of the diet among individuals should result in intraspecific variation in sinus form. In a study examining frontal sinus morphology in Carnivora that included 31 wild-caught and two captive-reared animals (one bush dog, Speothos venaticus and one snow leopard, Uncia uncia), Curtis & Van Valkenburgh (2014) found evidence suggesting that sinus morphology is affected by intraspecific differences in skull loading. Huxley (1880) reported that bush dogs have large frontal sinuses, but the skull of a captive bush dog (Speothos venaticus) included in the study by Curtis & Van Valkenburgh (2014) had no frontal sinuses. Bush dog skulls are adapted for subduing relatively large prey (Van Valkenburgh & Koepfli, 1993), and a softer diet in captivity might have affected development of the feeding musculature and skull in ways that resulted in a lack of space for a frontal sinus to develop in that particular individual. In addition, a wild-caught male raccoon dog (Nyctereutes procyonoides) examined by Curtis & Van Valkenburgh (2014) had small frontal sinuses, whereas a wild-caught female lacked frontal sinuses. The skull of the male showed greater development of muscle attachment sites for the jaw-closing muscles than the female, which also suggests that frontal sinus development might be impeded or unnecessary when the skull is subjected to lesser loading. However, with such small intraspecific sample sizes, it is not clear whether these differences are genuinely a result of differences in food texture and consequent skull loading. Other authors have reported that captive adult spotted hyenas (Crocuta crocuta) retain a cub-like appearance and lack the domed frontal bones, large sagittal crests and well-developed zygomatic arches typically present in their bone-eating wild counterparts (Binder & Van Valkenburgh, 2000; West-Eberhard, 2003: see caption for fig. 3.3). The domed frontal bones of wild spotted hyenas cover impressively large frontal sinuses that completely fill the sagittal crest, and it is likely that captive individuals fed on softer diets retain a cub-like appearance owing to a lack of development of the frontal sinus and sagittal crest, although this has yet to be quantified. To gain a better understanding of how external and internal skull shape are affected by differences in the mechanical properties of food, we investigated intraspecific variation in skull and frontal sinus morphology in the coyote (Canis latrans). We used a sample of captive-reared individuals with associated age and diet information. These individuals were raised in controlled conditions and fed soft, commercially produced food for a study of cranial growth and development (La Croix et al., 2011a, b). We compared the skull and sinus morphology of our captive-reared sample with that of a sample of wild coyotes with associated age and diet data (Johnson, 1978; Blood, Matson & Patten, 1985; Bartel & Knowlton, 2005). This allowed us to test whether skull dimensions and frontal sinus morphology vary in response to dietary differences within species, and whether frontal sinus morphology is remodelled with age. The diet of the wild coyotes (largely lagomorphs and rodents) is expected to have demanded higher bite forces on average and subjected their skulls to greater stress and strain than the soft commercial food fed to the captive coyotes. Skull morphology in captive coyotes fed a soft diet might be less functionally constrained than when subjected to the mechanical challenges of a wild diet. As a result, we expect skull shape to be more variable in captive coyotes and to show lesser development of attachment sites for the feeding musculature, as seen in similar comparisons in other taxa (Hollister, 1917; Fitch & Fagan, 1982; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014). We also expect to observe more pronounced doming of the frontal bones and larger frontal sinuses in wild coyotes that would indicate that their skulls are better shaped to cope with a more mechanically challenging diet (Davis, 1964; Wang et al., 1999; Tseng & Wang, 2010, 2011; Curtis and Van Valkenburgh, 2014; Curtis et al., 2015). With reduced skull loading in captive coyotes, we expect frontal sinus size and shape to be more variable and less strongly correlated with external skull shape than in wild coyotes, reflecting relaxed constraints on skull performance. This is the first study to test for intraspecific differences in sinus morphology related to differences in diet and is the first to look at the impact of age on sinus morphology in a non-human species. The results should improve our understanding of the role of mechanical loading in shaping mammal skull morphology. MATERIAL AND METHODS Specimens We sampled specimens from two populations: wild-caught and captive-reared C. latrans (Supporting information, Table S1). The wild-caught population consisted of 12 skulls of male C. latrans collected in Kern County, CA, USA between the years 1969 and 1973 that were aged by counts of cementum annuli (Blood et al., 1985). We selected our wild sample to include only those skulls that were scored as having negligible error in age estimation by Blood et al. (1985). The diet of this population consisted predominantly of lagomorphs and rodents with some livestock supplementation, and bone consumption was common (Cypher, Spencer & Scrivner, 1994). The captive-reared sample consisted of 12 skulls from male C. latrans drawn from a captive population maintained at the Logan Field Station in Millville, UT, USA (La Croix et al., 2011a, b). This population was established using wild coyotes trapped as pups in Idaho and Utah, where their natural diet is predominantly lagomorphs and rodents (Johnson, 1978; Bartel & Knowlton, 2005). The diet of the captive-reared sample consisted of commercially produced wet food for fur-bearing mammals, and never included bones, other foods or chew toys (La Croix et al., 2011a, b). Specimens ranged in age from 6 months (by which time the skull size and shape have matured; La Croix et al., 2011a, b) to 13 years. We age matched our captive and wild samples as closely as possible, and populations did not differ significantly in age (Student’s paired t-test, t = 1.19, d.f. = 10, P > 0.05). We included only males because male and female coyotes mature at different rates and are sexually dimorphic (La Croix et al., 2011a, b;,Blood et al., 1985). Although a sample size of 12 individuals per population is not particularly large, previous studies including similar sample sizes have shown differences in captive and wild individuals (e.g. Hartstone-Rose et al., 2014), and the time required to collect and process computed tomography (CT) data in non-trivial. One specimen (MSU 36583) in the captive sample was an outlier in frontal sinus volume, with considerably larger frontal sinuses than all other skulls in our sample, and showed a prominently domed frontal bone that distinguished it qualitatively from other skulls in our sample. Consequently, this specimen was excluded from all quantitative analyses (Supporting information, Table S1). To make sample sizes even for paired analyses, we randomly selected LACM 43391 (one of two wild specimens aged 13 years old) using a coin-flip for exclusion. LACM 43391 was included in all other analyses. Dental indicators of food texture The degree of tooth wear and tooth fracture, as well as differences in wear and fracture among tooth types (e.g. canines vs. molars), are indicators of food texture and feeding behaviour in mammalian carnivores (Van Valkenburgh, 1988, 2009). We quantified the amount of tooth wear in our captive and wild coyote samples using five categories: slight, slight–moderate, moderate, moderate–heavy and heavy (Van Valkenburgh, 1988, 2009). We documented the tooth wear stage for each tooth type [incisors, canines, premolars (P1–P3), carnassials (P4) and molars] and an overall tooth wear stage for each individual. We assessed tooth fracture frequency by counting the number of teeth broken in life per individual. A tooth was counted as broken if it showed wear on the broken surface indicating that the fracture occurred antemortem. We used pairwise Wilcoxon signed rank tests and a Bonferroni correction to compare tooth wear stage and Pearson’s χ2 test with a continuity correction to test for differences in tooth fracture frequency between wild and captive coyotes in R v. 3.4.1 (R Core Team, 2017). We expected to observe significantly greater amounts of tooth wear and tooth fracture in wild coyotes than in captive coyotes, owing to a more mechanically challenging diet in wild coyotes and consequent differences in skull loading between groups. We also expected to see tooth wear and tooth fracture increase at a greater rate with age in wild than in captive coyotes. To test how tooth wear and tooth fracture vary with age, we used ordinary least squares regressions to estimate the relationship between these variables and age and to test the strength of the relationships. To test for differences in the rate of tooth wear and tooth fracture accumulation between populations, we performed a slope test using the ‘smatr’ package in R (Warton et al., 2012; R Core Team, 2017). Computed tomography scanning We CT scanned all skulls in the UCLA School of Dentistry using a dental Cone-Beam CT scanner, with slice thickness of 0.03 mm for all scans. We segmented skulls from CT scans by manually thresholding bone from air using Mimics v. 17 (http://biomedical.materialise.com, last accessed 15 January 2018. Materialise, 2014) specialized imaging software, and generated volumetric models of skulls for use in morphometric analyses. The CT scan parameters are available from the corresponding author. Skull size and shape We quantified skull dimensions in captive and wild coyotes in two ways. First, we took a set of 16 linear measurements (Table 1; raw data provided in Supporting information, Table S1, Fig. S1) from volumetric models of each skull to the nearest 0.01 mm in Mimics (Materialise, 2014). Measurements included those used by previous authors to compare differences in skull dimensions between captive and wild individuals (Hollister, 1917; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004), as well as additional measures that captured the dimensions of the frontal bone, rostrum and braincase. Thus, we were able to test how our results fit within the context of prior work on differences in skull morphology between captive and wild conspecifics. Table 1. Descriptions of linear measurements taken from coyote crania Measurement  Description  WPOP  Width across postorbital processes of the frontal bone  OOL  Occiput to orbit length  CBL  Condylobasal length  ZW  Zygomatic width  JWC  Palate width between medial borders of canine alveoli  JWCn  Palate width between anterolateral borders of carnassial alveoli  PL  Palate length from anteriormost midsagittal point on incisive to posterior margin of palatine  M1O  Depth from anterolateral alveolus of M1 to the orbit  HJ  Dorsoventral height of the jugal  WFL  Width between the fronto-lacrimal sutures  POC  Width of postorbital constriction  FL  Mid-sagittal length of frontal bone  MW  Mastoid width  LCM  Length from tip of canine to mandibular fossa  WMF  Width across mandibular fossae  BCL  Basicranial length  Measurement  Description  WPOP  Width across postorbital processes of the frontal bone  OOL  Occiput to orbit length  CBL  Condylobasal length  ZW  Zygomatic width  JWC  Palate width between medial borders of canine alveoli  JWCn  Palate width between anterolateral borders of carnassial alveoli  PL  Palate length from anteriormost midsagittal point on incisive to posterior margin of palatine  M1O  Depth from anterolateral alveolus of M1 to the orbit  HJ  Dorsoventral height of the jugal  WFL  Width between the fronto-lacrimal sutures  POC  Width of postorbital constriction  FL  Mid-sagittal length of frontal bone  MW  Mastoid width  LCM  Length from tip of canine to mandibular fossa  WMF  Width across mandibular fossae  BCL  Basicranial length  For tooth fracture: total = mean raw number of broken teeth; percentage broken = mean (number of broken teeth/total teeth present). View Large As linear morphometrics cannot reveal changes in the relative positions of anatomical structures, we also used geometric morphometric techniques as a second method to investigate skull shape. We placed 20 landmarks (Fig. 1; Supporting information, Table S2) on the left sides of volumetric models of crania using the MedCAD module in Mimics (Materialise, 2014). Unilateral landmarking is time efficient, produces accurate size estimates for bilateral structures, and results in fewer redundant data points than bilateral landmark configurations. However, recent work suggests that landmarking only one side may result in slightly lower accuracy in estimating shape (Cardini, 2017). We divided landmarks into two sets; one that delimited frontal bone dimensions and the second that described overall skull shape minus the frontal bone (Supporting information, Table S2). Dividing the landmark sets in this way allowed us to test whether frontal sinus morphology is more strongly correlated with the morphology of the frontal bone than the overall skull, as has been shown before in interspecific studies (Farke, 2007, 2010; Curtis & Van Vakenburgh, 2014). Figure 1. View large Download slide Landmarks used for geometric morphometric analyses of skull and frontal bone shape, respectively. Landmarks included in the ‘skull’ dataset are in black; landmarks included in the ‘frontal bone’ dataset are in white. Landmark descriptions are given in the Supporting information (Table S2). Figure 1. View large Download slide Landmarks used for geometric morphometric analyses of skull and frontal bone shape, respectively. Landmarks included in the ‘skull’ dataset are in black; landmarks included in the ‘frontal bone’ dataset are in white. Landmark descriptions are given in the Supporting information (Table S2). We performed a Procrustes fit to align, rotate and scale landmark configurations to a common centroid size in MorphoJ v. 1.06b (Klingenberg, 2011). We extracted centroid sizes for the frontal bone and skull, respectively, for each specimen as a proxy for frontal bone and skull size in subsequent analyses (raw data provided in Supporting information, Table S1). We used Student’s paired t-tests and Bonferroni post hoc tests to test for significant differences among our 16 linear measurements, as well as skull and frontal bone centroid sizes between captive and wild coyotes in R (R Core Team, 2017). We used Levene’s test to test for significant differences in variance among these same variables in R using the ‘levene.test’ function in the package ‘car’ (Levene, 1960; Fox & Weisberg, 2011; R Core Team, 2017). We examined skull shape variation in captive and wild coyotes using principal components analysis (PCA) on a covariance matrix generated from Procrustes-fitted landmark data from the entire skull and the frontal bone, respectively, in MorphoJ (Klingenberg, 2011). We used a broken stick test to determine significant principal components (PCs) using the ‘bstick’ function in the ‘vegan’ package in R (Frontier, 1976; Jackson, 1993; Oksanen et al., 2017). To test for differences between skull and frontal bone shape, respectively, between captive and wild coyotes, we used a MANOVA with a Pilllai approximation to estimate an F-value on all significant PCs in R (R Core Team, 2017). To test for differences in variance in skull and frontal bone shape between captive and wild coyotes, we used an analysis of multivariate homogeneity of group dispersions on a distance matrix computed from Procrustes coordinates for skull and frontal bone shape, respectively, using the ‘betadisper’ function in the package ‘vegan’ in R (Anderson, 2006; Anderson, Ellingsen & McArdle, 2006; Oksanen et al., 2017). This method is a multivariate analogue of a Levene’s test that uses dissimilarity measures (Anderson, 2006; Anderson et al., 2006). In both linear morphometric and three-dimensional geometric morphometric analyses, we expected wild coyotes to show less skull shape disparity than captive coyotes because of the greater functional demands of a harder diet. We also expected wild coyotes to show greater development of attachment sites for feeding musculature, such as the zygomatic arches and sagittal crest, as well as enhanced doming of the frontal bone. Sinus size and shape We segmented left and right frontal sinuses from CT scans and estimated total frontal sinus volume (in cubic millimetres) using Mimics (Materialise, 2014). We used Student’s paired t-tests to test for significant differences in sinus size between populations and Levene’s tests to test for differences in variance of frontal sinus volume in R (R Core Team, 2017). Given that wild coyote skulls are subjected to more diet-related loading than captive coyote skulls, we expect to observe larger frontal sinuses in wild coyotes, as well as less variability in sinus size in wild coyotes owing to greater constraints on skull function. We quantified frontal sinus shape using spherical harmonics software (SPHARM v. 2, http://www.enallagma.com/SPHARM.php, last accessed 15 January 2018; Shen & Makedon, 2006; Shen, Fared & McPeek, 2009), following the methods of Curtis and Van Valkenburgh (2014). We performed SPHARM analyses on right frontal sinuses only, and rotated, aligned and scaled all frontal sinuses to a common centroid size in SPHARM before shape analysis. SPHARM approximates the shapes of three-dimensional surfaces by building upon the equation for a unit sphere, and is particularly useful for quantifying the shape of structures, such as frontal sinuses, that lack definitive and homologous landmarks. SPHARM generates a series of coefficients, on which multivariate statistics can be performed to quantify shape disparity among a set of surfaces, as we do here for frontal sinuses (Shen & Makedon, 2006; Shen et al., 2009). SPHARM coefficients are provided in the Supporting information (Table S1). To explore shape variation in frontal sinus shape in captive and wild coyotes, we performed a PCA on sinus SPHARM coefficients using the functions available in SPHARM software. We used a broken stick test to determine significant PCs using the ‘bstick’ function in the ‘vegan’ package in R (Frontier, 1976; Jackson, 1993; Oksanen et al., 2017). To test for frontal sinus shape differences between populations, we conducted a MANOVA with a Pilllai approximation to estimate an F-value on all significant PCs in R. As an additional test of shape differences between captive and wild coyotes, we conducted a Mahalanobis distance test with a permutation test (10000 rounds) to estimate a P-value in MorphoJ (Klingenberg & Monteiro, 2005; Klingenberg, 2011). If shape showed significant differences between populations, we conducted a discriminant function analyses on Procrustes coordinates and used a ‘leave-one-out’ cross-validation to test the reliability of the discrimination using the functions available in MorphoJ (Klingenberg, 2011). To test for differences in skull and frontal bone shape variance between captive and wild coyotes, we performed an analysis of multivariate homogeneity of group dispersions with a distance matrix based on SPHARM coordinates using the ‘betadisper’ function in the package ‘vegan’ in R (Anderson, 2006; Anderson et al., 2006; Oksanen et al., 2017). Similar to our expectations for skull shape, we expected frontal sinus shape to show less variation in wild coyotes than in captive coyotes. This is because we presume that wild coyotes have greater functional demands on skull performance. Changes in frontal sinus size with age and tooth wear To test whether frontal sinus size is correlated with age or tooth wear, we plotted the cube root of total frontal sinus volume against age and tooth wear, respectively. Preliminary scatter plots of frontal sinus size against age and tooth wear, respectively, suggested that frontal sinus size initially increases then decreases with age and tooth wear, and suggested that these data might covary in a non-linear fashion. Therefore, we fitted these relationships with three separate regression models (linear, quadratic with a linear term, and quadratic without a linear term) in R (R Core Team, 2017). We used Akaike information criterion (AIC; Akaike, 1973) scores to rank which model best fitted the data for both captive and wild coyotes, with the lowest AIC score representing the model that best fitted the data. We examined the relationship with both age and tooth wear, despite the fact that they covary, because sinus morphology might show different relationships with each. Covariation between frontal sinus size and external skull dimensions We examined the relationship between log10-transformed frontal sinus volume and linear cranial measurements using reduced major axis (RMA) regression. We performed RMA regressions using the ‘smatr’ package in R (Warton et al., 2012; R Core Team, 2017). RMA regression analysis is preferred over ordinary least squares when both variables are measured with error, and the allometric relationship between two variables is the desired outcome rather than the ability to predict one variable from another (Warton et al., 2006; Smith, 2009). If frontal sinus morphology is affected by diet-related skull use, we expected sinus volume to be correlated more strongly with cranial measurements in wild coyotes than in captive coyotes. RESULTS Dental indicators of food texture As expected, tooth wear and tooth fracture patterns were significantly different between wild and captive coyotes, supporting our assumption that food texture and consequent levels of skull loading differed between captive and wild individuals (Table 2). Although mean overall tooth wear did not differ significantly between wild and captive coyotes, wild coyotes had significantly greater wear for all individual tooth types (pairwise Wilcoxon signed-rank tests, P < 0.05), except the incisors (pairwise Wilcoxon signed-rank test, P > 0.05; Table 2). Total tooth fracture (Pearson’s χ2 test, χ2 = 10.51, P = 0.001) was significantly higher in wild coyotes. Table 2. Summary statistics from paired Wilcoxon signed rank tests comparing tooth wear stage in captive vs. wild coyotes Tooth wear  Captive mean  Wild mean  P-value  Total  2.36  2.82  NS  Incisors  2.45  2.91  NS  Canines  2.36  3.00  0.03  Premolars  2.36  3.00  0.04  Carnassials  2.18  2.82  0.03  Molars  2.18  2.82  0.03  Tooth wear  Captive mean  Wild mean  P-value  Total  2.36  2.82  NS  Incisors  2.45  2.91  NS  Canines  2.36  3.00  0.03  Premolars  2.36  3.00  0.04  Carnassials  2.18  2.82  0.03  Molars  2.18  2.82  0.03  View Large Tooth wear was positively and significantly correlated with age in both wild and captive coyotes (Fig. 2A, Table 3). Tooth wear was more strongly correlated with age in wild than captive coyotes (R2 = 0.93, P < 0.0001 for wild; R2 = 0.72, P = 0.0008 for captive). Although the slope of this relationship was higher in wild than captive coyotes (wild slope = 0.27, captive slope = 0.19), this difference was not significant at the 0.05 level. Figure 2. View largeDownload slide Ordinary least squares regressions of (A) tooth wear stage against age and (B) raw number of broken teeth against age. Regression statistics are given in Table 3. Figure 2. View largeDownload slide Ordinary least squares regressions of (A) tooth wear stage against age and (B) raw number of broken teeth against age. Regression statistics are given in Table 3. Table 3. Regression statistics for total tooth wear against age and total tooth fracture against age Regression  Slope  y-intercept  R2  P-value  Tooth wear vs. age          Captive  0.19 (0.10–0.28)  1.28 (0.65–1.92)  0.72  0.0008  Wild  0.27 (0.22–0.33)  1.31 (0.91–1.70)  0.93  < 0.0001  Tooth fracture vs. age          Captive  0.22 (−0.05 to 0.50)  0.85 (−1.08 to 2.79)  0.27  NS  Wild  0.67 (0.32–1.03)  0.85 (−1.78 to 3.64)  0.64  0.003  Regression  Slope  y-intercept  R2  P-value  Tooth wear vs. age          Captive  0.19 (0.10–0.28)  1.28 (0.65–1.92)  0.72  0.0008  Wild  0.27 (0.22–0.33)  1.31 (0.91–1.70)  0.93  < 0.0001  Tooth fracture vs. age          Captive  0.22 (−0.05 to 0.50)  0.85 (−1.08 to 2.79)  0.27  NS  Wild  0.67 (0.32–1.03)  0.85 (−1.78 to 3.64)  0.64  0.003  Tooth fracture represents total number of broken teeth. The 95% confidence intervals are given for slope and y-intercept. P-values are from a test of the null expectation that slope = 0. View Large Tooth fracture was significantly correlated with age in wild but not in captive coyotes (Fig. 2B; R2 = 0.64, P = 0.003 for wild; R2 = 0.27, P > 0.05 for captive). All captive individuals > 7 years of age had fewer broken teeth (fewer than five broken) than age-matched wild individuals in the same age cohort (seven to 15 broken). We also observed that the pulp cavities of the teeth in the oldest individuals from our wild sample, most of which exhibited the greatest amount of tooth wear and fracture, were almost completely or completely filled with secondary dentin, whereas in the captive sample, pulp cavities were still clearly visible in CT scans (Fig. 3). Figure 3. View largeDownload slide Comparison of pulp cavity of the first upper molar (M1) taken at the paracone between captive and wild coyotes, showing that the pulp cavities of the oldest wild coyotes in our samples are filled almost completely with secondary dentin in comparison to those of captive coyotes, in which there are visible pulp cavities even at a similar age and tooth wear stage. Figure 3. View largeDownload slide Comparison of pulp cavity of the first upper molar (M1) taken at the paracone between captive and wild coyotes, showing that the pulp cavities of the oldest wild coyotes in our samples are filled almost completely with secondary dentin in comparison to those of captive coyotes, in which there are visible pulp cavities even at a similar age and tooth wear stage. Skull size and shape We found that skull and frontal bone centroid sizes did not differ significantly between captive and wild coyotes (Table 4; Student’s paired t-tests, P > 0.05), nor did they differ significantly in variance (Supporting information, Table S3; Levene’s tests, P > 0.05). Nevertheless, we did observe significant differences in the size of three of 16 linear measurements between captive and wild coyotes. Zygomatic width (Table 4; Student’s paired t-test, t = 2.25, d.f. = 10, P = 0.05), occiput to orbit length (Table 4; Student’s paired t-test, t = 2.63, d.f. = 10, P = 0.03) and palate width at the carnassial (Table 4; Student’s paired t-test, t = 5.69, d.f. = 10, P = 0.0002) were all significantly larger in captive than in wild coyotes. Contrary to expectation, the variance in linear cranial measurements did not differ significantly between captive and wild coyotes (Supporting information, Table S3; Levene’s test, all variables P > 0.05). Table 4. Summary statistics from Student’s paired t-tests comparing age, frontal sinus volume, centroid size and linear measurements in captive vs. wild coyotes Trait  Captive mean  Wild mean  d.f.  t  P-value  Age  5.74  5.54  10  1.19  NS  Frontal sinus volume  2712.76  2779.94  10  −0.32  NS  Skull centroid size  240.32  237.85  10  1.13  NS  WPOP  46.99  46.55  10  0.23  NS  OOL  108.64  106.22  10  2.63  0.03  CBL  184.03  183.08  10  0.52  NS  ZW  98.24  96.62  10  2.25  0.05  JWC  30.58  31.13  10  −1.02  NS  JWCn  53.84  51.54  10  5.69  0.0002  PL  95.44  96.19  10  −0.48  NS  M1O  26.39  27.21  10  −1.26  NS  HJ  11.11  11.86  10  −1.63  NS  WFL  39.01  38.53  10  0.52  NS  POC  35.43  34.52  10  1.05  NS  FL  52.65  52.45  10  0.26  NS  MW  55.52  54.73  10  1.00  NS  LCM  124.11  123.13  10  0.79  NS  WMF  85.30  85.04  10  0.39  NS  BCL  46.32  46.36  10  −0.06  NS  Trait  Captive mean  Wild mean  d.f.  t  P-value  Age  5.74  5.54  10  1.19  NS  Frontal sinus volume  2712.76  2779.94  10  −0.32  NS  Skull centroid size  240.32  237.85  10  1.13  NS  WPOP  46.99  46.55  10  0.23  NS  OOL  108.64  106.22  10  2.63  0.03  CBL  184.03  183.08  10  0.52  NS  ZW  98.24  96.62  10  2.25  0.05  JWC  30.58  31.13  10  −1.02  NS  JWCn  53.84  51.54  10  5.69  0.0002  PL  95.44  96.19  10  −0.48  NS  M1O  26.39  27.21  10  −1.26  NS  HJ  11.11  11.86  10  −1.63  NS  WFL  39.01  38.53  10  0.52  NS  POC  35.43  34.52  10  1.05  NS  FL  52.65  52.45  10  0.26  NS  MW  55.52  54.73  10  1.00  NS  LCM  124.11  123.13  10  0.79  NS  WMF  85.30  85.04  10  0.39  NS  BCL  46.32  46.36  10  −0.06  NS  For measurement abbreviations, see Table 1. View Large A PCA on the covariance matrix generated using Procrustes coordinates for skull shape excluding the frontal bone revealed seven significant PCs. A plot of PC2 against PC1 (together explaining 35.91% of cumulative shape variation) showed that captive and wild coyotes shared similar shape space (Fig. 4A). PC1 (19.60% of total skull shape variation) separated individuals with dorsoventrally shallow, mediolaterally slender and downturned rostra, slender zygomatic arches, poorly developed sagittal crests and small mastoid processes (positive scores) from individuals with deep, broad upturned rostra, broad zygomatic arches, well-developed sagittal crests and large mastoid processes (negative scores). PC2 (16.31% of total skull shape variation) separated individuals with proportionally short and dorsoventrally deep rostra, broad zygomatic arches and a well-developed sagittal crest (positive scores) from individuals with proportionally long, dorsoventrally shallow rostra and poorly developed sagittal crests (negative scores). A MANOVA on the first seven PCs showed that skull shape is significantly different between populations (Pilllai approximation, d.f. = 7,15, F = 3.39, P = 0.02). A Mahalanobis distance test also revealed that skull shape differs significantly between captive and wild coyotes in our sample (Mahalanobis distance = 5.42, T2 = 168.70, P = 0.01). The single discriminant function successfully separated captive and wild coyotes. However, in leave-one-out cross-validation tests, the discriminant function was able to classify skull shape correctly only 57% of the time, which suggests that separation of wild and captive coyotes by the discriminant function is likely to be an artefact of low sample sizes and a large number of variables in our dataset (Table 5). Figure 4. View largeDownload slide (A) Principal components analysis on Procrustes coordinates for skull shape. Principal component (PC) 1 accounted for 19.60% of total shape variation, and PC2 accounted for 16.31% of total shape variation. (B) Principal components analysis on Procrustes coordinates for frontal bone shape. PC1 accounted for 36.77% of total shape variation, and PC2 accounted for 23.96% of total shape variation. Wild and captive coyotes are denoted as in Figure 2. Figure 4. View largeDownload slide (A) Principal components analysis on Procrustes coordinates for skull shape. Principal component (PC) 1 accounted for 19.60% of total shape variation, and PC2 accounted for 16.31% of total shape variation. (B) Principal components analysis on Procrustes coordinates for frontal bone shape. PC1 accounted for 36.77% of total shape variation, and PC2 accounted for 23.96% of total shape variation. Wild and captive coyotes are denoted as in Figure 2. Table 5. Summary of results from a discriminant function analysis of skull shape with a leave-one-out cross-validation (classification analysis) Group  Correct  Misclassified  Total  Percentage correct  Discriminant function          Wild coyotes  12  0  12  100  Captive coyotes  11  0  11  100  Total  23  0  23  100  Classification analysis  Wild coyotes  9  3  12  75  Captive coyotes  4  7  11  36  Total  13  10  23  57  Group  Correct  Misclassified  Total  Percentage correct  Discriminant function          Wild coyotes  12  0  12  100  Captive coyotes  11  0  11  100  Total  23  0  23  100  Classification analysis  Wild coyotes  9  3  12  75  Captive coyotes  4  7  11  36  Total  13  10  23  57  View Large A PCA on a covariance matrix generated using Procrustes coordinates for frontal bone shape alone revealed three significant PCs. Captive coyotes occupied a greater amount of shape space in a plot of PC2 against PC1 (together explaining 60.73% of cumulative shape variation), but this difference in shape variance was not significant (Fig 4B; multivariate homogeneity of group dispersions test, P > 0.05). PC1 (36.77% of total frontal bone shape variation) distinguished frontal bones with slender postorbital processes that were dorsoventrally shallow with flat dorsal profiles (positive scores) from those with wide postorbital processes that were dorsoventrally deep with more domed profiles (negative values). PC2 (23.96% of total frontal bone shape variation) distinguished frontal bones with wider postorbital processes that were dorsally convex (positive scores) from those with slender postorbital processes that were dorsally flat (negative scores). Despite falling within the same shape space as captive coyotes, wild coyote frontal bone shape tended to plot slightly more positively along PC2, but a MANOVA of the first three PCs showed no significant differences in frontal bone shape between populations (Pilllai approximation, F = 1.85, d.f. = 3, 19, P > 0.05). A Mahalanobis distance test also showed no significant differences in frontal bone shape between captive and wild coyotes in our sample (Mahalanobis distance = 3.71, T2 = 78.86, P = 0.014. Sinus size and shape Contrary to expectations, frontal sinus volume did not differ significantly between captive and wild populations (Table 4; Student’s paired t-test, t = −0.42, d.f. = 10, P > 0.05), and variance in sinus size also did not differ between populations (Supporting information, Table S3; Levene’s test, d.f. = 1, 20, P > 0.05). Qualitatively, frontal sinus morphology was broadly similar among all specimens examined (Figs 5, 6). In general, the frontal sinuses were centred at the postorbital processes, and never reached the margin of the fronto-parietal suture as seen in larger hypercarnivorous canids, such as the gray wolf (Canis lupus; Curtis & Van Valkenburgh, 2014). However, in the wild-caught coyotes, the frontal sinuses tended to pneumatize further posteriorly relative to the postorbital processes than those of the captive animals. In both groups, the posterior margin of the frontal sinuses on each side of the midline typically showed what appeared to be three lobes of pneumatization separated by two small bony trabeculae. Anteriorly, the frontal sinuses on each side of the midline usually had two lobes that differed little within and between captive and wild coyotes; both displayed a medial lobe where the sinus was connected to the nasal chamber via an ostium located posterior to frontoturbinal 2 (sensuPaulli, 1900), and a lateral lobe that expanded anteriorly from the postorbital process. Figure 5. View largeDownload slide Dorsal views of sinuses within the skulls of wild coyotes. Age of animal and tooth wear stage are listed beneath each skull. LACM 43391 is shown with only a right frontal sinus because the left frontal bone had a gunshot wound. Figure 5. View largeDownload slide Dorsal views of sinuses within the skulls of wild coyotes. Age of animal and tooth wear stage are listed beneath each skull. LACM 43391 is shown with only a right frontal sinus because the left frontal bone had a gunshot wound. Figure 6. View largeDownload slide Dorsal view of frontal sinuses within the skulls of captive-reared coyotes. Age of animal and tooth wear stage are listed beneath each skull. Figure 6. View largeDownload slide Dorsal view of frontal sinuses within the skulls of captive-reared coyotes. Age of animal and tooth wear stage are listed beneath each skull. A PCA on SPHARM coefficients describing frontal sinus shape revealed four significant PCs. Both groups overlapped widely in morphospace (Fig. 7). PC1 (21.2% of total frontal sinus shape variation) contrasted individuals with dorsoventrally deep, dorsally convex frontal sinuses (positive values) from those with dorsoventrally shallow, less dorsally convex (negative values), and showed a weak but significant relationship with frontal sinus volume (R2 = 0.30, P = 0.007). PC2 accounted for 13.9% of total frontal sinus shape variation, and again separated individuals based on dorsoventral depth and complexity, and also showed a weak but significant correlation with frontal sinus volume (R2 = 0.20, P = 0.03). In this case, sinuses that plotted negatively were dorsoventrally deep and pneumatized further posteriorly across the braincase. Wild coyotes tended to plot more positively along PC1 and ore negatively along PC2 than captive coyotes, corroborating our qualitative observations that the frontal sinuses tend to pneumatize further posteriorly and to be deeper dorsoventrally and more convex dorsally in our wild sample. A MANOVA of the first four PCs detected no differences in frontal sinus shape between captive and wild coyotes (Pilllai approximation, d.f. = 4, 18, F = 1.79, P > 0.05), and frontal sinus shape variance also did not differ between both groups in our sample (multivariate homogeneity of group dispersions test, P > 0.05). Changes in frontal sinus size with age and tooth wear There was no significant relationship between frontal sinus volume and either age or tooth wear in our captive sample. However, within the wild sample, there was a significant non-linear relationship between frontal sinus volume and tooth wear (P = 0.04; Table 6, Fig. 8). The relationship between frontal sinus volume and age in our wild sample was similar in shape to that for tooth wear, but did not quite achieve significance (P = 0.09; Table 6, Fig 8). In the wild sample, the significant relationship between frontal sinus volume and tooth wear was best modelled as a quadratic function including a linear term (Fig. 8). Frontal sinus volume increased as tooth wear stage increased from slight to moderate, but then frontal sinus volume declined as wear stage progressed. Table 6. Comparison of regression models of frontal sinus volume against tooth wear and age for the wild sample (N = 12)   R2  P-value  AIC score  TSV⅓ vs. tooth wear        Quadratic with linear term  0.51  0.04  44.36  Quadratic without linear term  0  NS  50.79  Ordinary least squares  0.04  NS  50.31  TSV⅓ vs. age        Quadratic with linear term  0.42  NS  46.36  Quadratic without linear term  0  NS  50.82  Ordinary least squares  0.03  NS  50.48    R2  P-value  AIC score  TSV⅓ vs. tooth wear        Quadratic with linear term  0.51  0.04  44.36  Quadratic without linear term  0  NS  50.79  Ordinary least squares  0.04  NS  50.31  TSV⅓ vs. age        Quadratic with linear term  0.42  NS  46.36  Quadratic without linear term  0  NS  50.82  Ordinary least squares  0.03  NS  50.48  Lowest Akaike information criterion score represents the model that best fits the data. TSV, total frontal sinus volume. View Large Figure 7. View largeDownload slide Principal components analysis on SPHARM coefficients. Principal component (PC) 1 accounted for 21.22% of total shape variation, and PC2 accounted for 13.93% of total shape variation. Dorsal and medial views of skulls showing sinus shapes at the extremes for each PC along each axis. Wild and captive coyotes are coded as in Figure 2. Figure 7. View largeDownload slide Principal components analysis on SPHARM coefficients. Principal component (PC) 1 accounted for 21.22% of total shape variation, and PC2 accounted for 13.93% of total shape variation. Dorsal and medial views of skulls showing sinus shapes at the extremes for each PC along each axis. Wild and captive coyotes are coded as in Figure 2. Figure 8. View largeDownload slide Plot of frontal sinus volume against tooth wear stage. The curve represents a quadratic equation with a linear term that best fitted this relationship in wild coyotes. The relationship between sinus size and tooth wear was not significant in captive coyotes. Regression statistics are given in Table 6. Data for wild and captive coyotes are coded as in Figure 2. Figure 8. View largeDownload slide Plot of frontal sinus volume against tooth wear stage. The curve represents a quadratic equation with a linear term that best fitted this relationship in wild coyotes. The relationship between sinus size and tooth wear was not significant in captive coyotes. Regression statistics are given in Table 6. Data for wild and captive coyotes are coded as in Figure 2. Covariation between frontal sinus size and external skull dimensions In our wild sample, frontal sinus volume was positively correlated with width across the postorbital processes, zygomatic width, width between the fronto-lacrimal sutures on the orbit, width across the mandibular fossae, and basicranial length (Table 7). Among these five cranial measurements, frontal sinus volume scaled with positive allometry (slope > 3) in all but width across the postorbital processes, where the scaling coefficient was not significantly different from isometry (slope = 3; Table 7). We found no significant correlations between frontal sinus volume and cranial measurements in captive coyotes. Table 7. Regression statistics for log10/log10 reduced major axis regression of total frontal sinus volume against individual cranial measurements that showed a statistically significant relationship in wild coyotes TSV vs.  Slope  y-intercept  R2  WPOP  3.38 (2.05–5.58)  −1.88 (−4.82 to 1.07)  0.52  ZW  9.68 (6.72–13.93)  −15.46 (−22.60 to −8.30)  0.76  WFL  6.63 (3.91–11.24)  −6.76 (−12.57 to −0.94)  0.46  WMF  8.86 (6.35–12.37)  −13.33 (−19.15 to −7.53)  0.80  BCL  6.88 (4.04–11.70)  −7.70 (−14.08 to −1.33)  0.46  TSV vs.  Slope  y-intercept  R2  WPOP  3.38 (2.05–5.58)  −1.88 (−4.82 to 1.07)  0.52  ZW  9.68 (6.72–13.93)  −15.46 (−22.60 to −8.30)  0.76  WFL  6.63 (3.91–11.24)  −6.76 (−12.57 to −0.94)  0.46  WMF  8.86 (6.35–12.37)  −13.33 (−19.15 to −7.53)  0.80  BCL  6.88 (4.04–11.70)  −7.70 (−14.08 to −1.33)  0.46  The 95% confidence intervals are given for slope and y-intercept. TSV, total frontal sinus volume (in cubic millimetres). View Large DISCUSSION Previous explorations of the impact of differences in diet and associated feeding behaviour on skull shape within mammal species have been limited by a lack of age and diet information for specimens (however, see Forbes-Harper et al., 2017), as well as limited information on source populations for captive individuals (Hollister, 1917; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014). Without these data, it is difficult to tease apart the relative effects of age, material properties of food, and geographical variation on cranial shape. Here, we were able to use a unique sample of wild and captive-reared coyotes obtained from a geographically similar source population. Both populations also had associated age and diet information, allowing us to quantify the effects of food texture (relatively soft vs. relatively hard) on tooth condition and cranial dimensions, including the size and shape of the frontal sinuses. We found evidence that craniodental morphology was affected by differences in loading, but differences were minimal. This indicates that coyote skull morphology, and probably biomechanical performance, are not greatly affected by reduced mechanical loading. However, there were marked differences between the two samples in tooth wear and fracture that were consistent with more regular use of higher bite forces to eat more mechanically challenging food in the wild sample than the captive animals, all of which were fed processed commercial pet food without bones. As expected, we observed a greater degree of tooth wear and tooth fracture frequency in wild coyotes than in captive coyotes, supporting the assumption that our wild sample consumed a more mechanically challenging diet. Wild coyotes showed a higher frequency of tooth fracture, and all teeth, except the incisors, showed much greater wear than those of captive coyotes. Increased tooth wear and fracture in other wild carnivorans are thought to reflect more bone consumption (greater carcass utilization), as well as more rapid consumption of food as a consequence of competition for food access (Van Valkenburgh, 1988, 2009). Tooth fracture, especially of the canine teeth, could also occur during prey capture or intraspecific and interspecific combat. However, the presence of fractured cheek teeth (premolars, molars) as well as the co-occurrence of tooth fracture with increased wear on non-fractured teeth suggests that much of the breakage occurs during feeding. Although we have no data on levels of intraspecific competition in our wild coyote sample, it is clear that their diet of rodents and lagomorphs included more bone than the processed wet food consumed by the captive individuals. Wild coyotes also had a greater amount of secondary dentin filling the pulp cavities (sometimes completely), and this is likely to reflect repeated trauma to the teeth, given that trauma induces deposition of secondary dentin by odontoblasts along the periphery of the tooth pulp cavity (Klugh, 2010). This observation further supports a more mechanically challenging diet in wild coyotes. We observed a few subtle, yet significant, differences in skull dimensions between captive and wild coyotes. Similar to previous studies that used linear measurements to compare skull morphology between captive and wild conspecifics (e.g. O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004), as well as one three-dimensional geometric morphometric study (Hartstone-Rose et al., 2014), the two groups differed in zygomatic width and palate width, as well as occiput to orbit length. As in lions (O’Regan, 2001; Zuccareli, 2004; Hartstone-Rose et al., 2014), zygomatic width was broader in captive individuals, and like lions, the absence of normal muscle forces experienced by wild counterparts resulted in the zygomatic arches being somewhat circular in cross-section, rather than mediolaterally compressed, as in typical wild lions and coyotes. However, these differences were less pronounced in coyotes. We found no significant differences in variance for any traits between captive and wild coyotes in our sample, which does not support the hypothesis that the skulls of captive coyotes are less morphologically constrained than wild conspecifics. However, with a larger sample size, it is possible that results would show that frontal bone shape is more variable in captive individuals (see Fig. 4B). Although we expected to find that wild coyotes had larger frontal sinuses than captive coyotes, the two samples did not differ significantly in sinus size, suggesting that differences in food texture do not impact frontal sinus morphology. However, there were slight qualitative differences in frontal sinus shape between the coyote samples, such as the tendency for the frontal sinuses of wild individuals to be deeper dorsoventrally and to extend further posteriorly. We also observed the greatest frontal sinus volume among all coyotes sampled in a captive individual (MSU 36583), which was an outlier in our sample and excluded from all statistical analyses. This suggests that patterns of bone deposition and removal might alter spatially in response to differences in diet-related loading of the skull, and that aberrant morphologies might be more common in captive individuals. We suspect that a larger sample size could reveal significant differences in sinus shape or size not visible in our sample. Frontal sinus volume increased with age and tooth wear up to a point in our wild sample but then declined as they aged further. The decrease in frontal sinus volume associated with heavier tooth wear in the oldest wild coyotes in our sample might reflect recruitment of proportionally greater bite forces to apprehend prey and/or process food with heavily worn or missing teeth. This hypothesis is supported by qualitative observations that sagittal crest size was largest in the oldest wild individuals (Fig. 9). An enlarged sagittal crest reflects greater development of the temporalis muscle (the primary jaw-closing muscle in carnivores). Production of higher bite forces is likely to subject the skull to higher strain and could induce bone deposition in the frontal region, resulting in a decrease in sinus volume. Sagittal crest size and tooth wear are also reported to be greater in older adult wild pumas (Puma concolor), but whether their sinuses are affected is unknown (Gay & Best, 1996). This pattern contrasts with our hypothesis that we should observe larger frontal sinuses in individuals with more mechanically challenging diets. However, one species of canid (the African wild dog, Lycaon pictus) has relatively small sinuses inside relatively thick frontal bones, and their skulls are well adapted to bring down large, struggling prey (Slater, Dumont & Van Valkenburgh, 2009; Curtis & Van Valkenburgh, 2014). Thus, there might be other ways in which sinus morphology can be remodelled to improve skull function. Given the limits of our sample size, it is possible that the apparent increase and decrease of frontal sinus volume with age might not be real. The initial upward slopes in the relationship between sinus volume and age or tooth wear, respectively, are driven by the youngest coyotes in our sample (between 6 months and 2 years age) and could be linked to ontogenetic changes rather than changes because of skull loading. Likewise, the downward slope in older individuals might be an artefact of over-fitting models to our relatively small dataset. A better understanding of the relationships among skull shape, frontal sinus size and muscle development would require a larger sample that also includes more landmarks along the sagittal and lamboidal crests. Figure 9. View largeDownload slide Differences in the development of the sagittal crest in (A) captive vs. (B) wild coyotes. Dorsal views (top), with red bracket denoting the length of the sagittal crest, and curve extending to the postorbital process delimiting the temporal line. Lateral views (bottom), with red line delimiting the base of the sagittal crest. Note the greater development of the sagittal crest in the wild coyote in both views. Figure 9. View largeDownload slide Differences in the development of the sagittal crest in (A) captive vs. (B) wild coyotes. Dorsal views (top), with red bracket denoting the length of the sagittal crest, and curve extending to the postorbital process delimiting the temporal line. Lateral views (bottom), with red line delimiting the base of the sagittal crest. Note the greater development of the sagittal crest in the wild coyote in both views. The significant correlation we observed between frontal sinus size and external cranial morphology in wild coyotes, but not captive-reared coyotes, suggests that the development and maintenance of internal and external skull morphology might be more integrated in wild individuals. This suggests that the forces experienced by wild coyote skulls could induce remodelling of frontal sinus morphology to improve mechanical efficiency. This could be tested by comparing patterns of stress and strain across the frontal region in captive and wild individuals using finite element methods. Overall, our study suggests that coyote skull and frontal sinus morphology is not greatly impacted by reduced skull loading. Given two coyote skulls of similar age, one from our captive sample and one from our wild sample, it would be difficult to determine their origin from their external or internal appearance, especially if they lacked any teeth. However, the greater mechanical demands of a wild diet, as evidenced by heavier tooth wear and tooth fracture, may have resulted in a stronger relationship between sinus form and external skull morphology in wild coyotes. In the captive sample, a softer diet and the consequent reductions in skull loading could have relaxed the relationship between internal and external cranial morphology. However, it is unlikely that these differences translate to drastic differences in biomechanical performance of the skull. It would be interesting to explore how skull and sinus morphology is affected under greater, rather than lesser, loading conditions. For example, some coyote populations frequently take prey much larger than lagomorphs, including white-tailed deer, which undoubtedly subject their skulls to much greater stress and strain (Lingle, 2002). The impact of mechanical loading on skull form might also be more pronounced in larger carnivore species with more specialized diets, such as spotted hyenas that crack open bones with their teeth, or lions that can bring down buffalo (Syncerus caffer) and even elephants (Loxodonta africana). Prior comparisons of the skulls of captive and wild large carnivorans reported more pronounced differences in cranial dimensions (e.g. Fitch & Fagan, 1982; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014), which suggests that skull loading plays a larger role in shaping skull form in these taxa. As a smaller-bodied generalist, wild coyotes probably do not subject their skulls to loads that vary in magnitude to the degree experienced by species that regularly hunt large, struggling prey or consume very hard foods, which might explain the smaller differences in cranial form we observed between captive and wild coyotes. This suggests that the effects of a non-wild diet lacking in bones or other tough foods probably has much greater negative impacts on craniodental morphology and function in large than small carnivores in captivity. This could also have important implications in wild individuals if their typical prey becomes scarce, as is believed to have occurred for many large carnivorans towards the end of the Pleistocene as large prey species went extinct (see review by Ripple & Van Valkenburgh, 2016). Might atypical skull development in large carnivorans occur if their large herbivorous prey becomes scarce and, consequentially, skull loading is reduced? This would be likely to make it more difficult for large carnivorans to switch successfully to smaller prey and compete with smaller carnivorans that are better adapted to subdue and eat smaller prey. Similar studies in a broader sample of taxa, especially among wild populations, would greatly improve our understanding of how and to what degree phenotype is shaped by environmental conditions in mammals and how this might affect feeding performance and, consequently, fitness. ACKNOWLEDGEMENTS The authors are grateful to L. Parker in the UCLA School of Dentistry for her help with scanning all skulls included in this study. We also thank L. Abraczinskas at Michigan State University, Lansing and J. Dines at the Los Angeles County Museum of Natural History for their help with specimens and associated data. We thank D. Bird, J. Wolf and J. Pajoli for insightful discussions and many helpful suggestions, and we thank the three anonymous referees and N. Warburton for their constructive feedback that greatly improved the quality of this manuscript. The authors declare no conflicts of interest. Funding was provided by National Science Foundation (NSF) IOB-0517748 to B.V.V. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Figure S1. Linear measurements taken from skulls. Descriptions and codes for linear measurements are given in Table 1. Table S1. (Attached file: Coyote_Skull_Raw_Data.xls). Raw data collected from wild (LACM) and captive (MSU) specimens included in this study. Specimen, specimen number. Museum: MSU, Michigan State University; LACM, Los Angeles County Museum of Natural History. Total tooth wear: ib (broken incisors), cb (broken canines), pb (broken premolars), cnb (broken carnassials), mb (broken molars), tb (total broken teeth), tt (total teeth), tw (total tooth wear stage), iw (incisors tooth wear stage), cw (canine wear stage), pw (premolar wear stage), cnw (carnassial wear stage), mw (molar wear stage), TSV (total frontal sinus volume, in mm3), SKCS (skull centroid size), FCS (frontal bone centroid size). Remaining columns represent raw data for linear measurements, followed by landmark coordinates (x1, y1, z1,…), and SPHARM coordinates (coordx1, coordyy1, coordz1, …). Codes for linear measurements are given in Table 1. Table S2. Descriptions of landmarks used for geometric morphometric analysis of skull shape. Table S3. Variance for traits in captive and wild coyotes and summary statistics from Levene’s tests testing for significant differences in variance between captive and wild coyotes. Codes for linear measurements are given in Table 1. REFERENCES Akaike H. 1973. Maximum likelihood identification of Gaussian autoregressive moving average models. Biometrika  60: 255– 265. Google Scholar CrossRef Search ADS   Anderson MJ. 2006. Distance-based tests for homogeneity of multivariate dispersions. Biometrics  62: 245– 253. Google Scholar CrossRef Search ADS PubMed  Anderson MJ, Ellingsen KE, McArdle BH. 2006. Multivariate dispersion as a measure of beta diversity. Ecology Letters  9: 683– 693. Google Scholar CrossRef Search ADS PubMed  Bartel RA, Knowlton FF. 2005. Functional feeding responses of coyotes, Canis latrans, to fluctuating prey abundance in the Curlew Valley, Utah, 1977–1993. Canadian Journal of Zoology  83: 569– 578. Google Scholar CrossRef Search ADS   Binder WJ, Van Valkenburgh B. 2000. Development of bite strength and feeding behavior in juvenile spotted hyneas (Crocuta crocuta). Journal of Zoology  252: 273– 283. Google Scholar CrossRef Search ADS   Blood BR, Matson JO, Patten DR. 1985. Multivariate analysis of allometry in a single population of coyotes (Canidae: Canis latrans Say). Australian Mammalogy  8: 221– 231. Cardini A. 2017. Left, right or both? estimating and improving accuracy of one-side-only geometric morphometric analyses of cranial variation. Journal of Zoological Systematics and Evolutionary Research  55: 1– 10. Google Scholar CrossRef Search ADS   Curtis AA, Lai G, Wei F, Van Valkenburgh B. 2015. Repeated loss of frontal sinuses in arctoid carnivorans. Journal of Morphology  276: 22– 32. Google Scholar CrossRef Search ADS PubMed  Curtis AA, Van Valkenburgh B. 2014. Beyond the sniffer: frontal sinuses in Carnivora. The Anatomical Record  297: 2047– 2064. Google Scholar CrossRef Search ADS PubMed  Cypher BL, Spencer KA, Scrivner JH. 1994. Food-item use by coyotes at the Naval Petroleum Reserves in California. Southwestern Naturalist  39: 91– 95. Google Scholar CrossRef Search ADS   Davis DD. 1964. The giant panda: a morphological study of evolutionary mechanisms. Fieldiana  3: 1– 339. Dumont ER, Dávalos LM, Goldberg A, Santana SE, Rex K, Voigt CC. 2012. Morphological innovation, diversification and invasion of a new adaptive zone. Proceedings of the Royal Society B: Biological Sciences  279: 1797– 1805. Google Scholar CrossRef Search ADS   Farke AA. 2007. Morphology, constraints, and scaling of frontal sinuses in the hartebeest, Alcelaphus buselaphus (Mammalia: Artiodactyla, Bovidae). Journal of Morphology  268: 243– 253. Google Scholar CrossRef Search ADS PubMed  Farke AA. 2008. Frontal sinuses and head-butting in goats: a finite element analysis. The Journal of Experimental Biology  211: 3085– 3094. Google Scholar CrossRef Search ADS PubMed  Farke AA. 2010. Evolution and functional morphology of the frontal sinuses in Bovidae (Mammalia: Artiodactyla), and implications for the evolution of cranial pneumaticity. Zoological Journal of the Linnean Society  159: 988– 1014. Google Scholar CrossRef Search ADS   Fatu C, Puisoru M, Rotaru M, Truta AM. 2006. Morphometric evaluation of the frontal sinus in relation to age. Annals of Anatomy = Anatomischer Anzeiger: official organ of the Anatomische Gesellschaft  188: 275– 280. Google Scholar CrossRef Search ADS PubMed  Fitch HM, Fagan DA. 1982. Focal palatine erosion associated with dental malocclusion in captive cheetahs. Zoo Biology  1: 295– 310. Google Scholar CrossRef Search ADS   Forbes-Harper JL, Crawford HM, Dundas SJ, Warburton NM, Adams PJ, Bateman PW, Calver MC, Fleming PA. 2017. Diet and bite force in red foxes: ontogenetic and sex differences in an invasive carnivore. Journal of Zoology  303: 54– 63. Google Scholar CrossRef Search ADS   Fox J, Weisberg S. 2011. An {R} companion to applied regression , 2nd edn. Thousand Oaks: Sage. Frontier S. 1976. Étude de la decroissance des valeurs propres dans une analyze en composantes principals: comparison avec le modèle de baton brisé. Journal of Experimental Marine Biology and Ecology  25: 67– 75. Google Scholar CrossRef Search ADS   Gay SW, Best TL. 1996. Age-related variation in skulls of the puma (Puma concolor). Journal of Mammalogy  77: 191– 198. Google Scholar CrossRef Search ADS   Hartstone-Rose A, Selvey H, Villari JR, Atwell M, Schmidt T. 2014. The three-dimensional morphological effects of captivity. PLoS ONE  9: e111347. PubMed  Herring SW. 1993. Formation of the vertebrate face: epigenetic and functional influences. American Zoologist  33: 472– 483. Google Scholar CrossRef Search ADS   Hollister N. 1917. Some effects of environment and habit on captive lions. Proceeding of the United States Natural History Museum  53: 177– 193. Google Scholar CrossRef Search ADS   Huxley TH. 1880. On the cranial and dental characters of the Canidae. Journal of Zoology  48: 238– 288. Jackson DA. 1993. Stopping rules in principal components analysis: a comparison of heuristical and statistical approaches. Ecology  74: 2204– 2214. Google Scholar CrossRef Search ADS   Joeckel RM. 1998. Unique frontal sinuses in fossil and living Hyaenidae (Mammalia, Carnivora): description and interpretation. Journal of Vertebrate Paleontology  18: 627– 637. Google Scholar CrossRef Search ADS   Johnson MK. 1978. Food habits of coyotes in south central Idaho . Unpublished octoral dissertation, Colorado State University. Klingenberg CP. 2011. MorphoJ: an integrated software package for geometric morphometrics. Molecular Ecology Resources  11: 353– 357. Google Scholar CrossRef Search ADS PubMed  Klingenberg CP, Monteiro LR. 2005. Distances and directions in multidimensional shape spaces: implications for morphometric applications. Systematic Biology  54: 678– 688. Google Scholar CrossRef Search ADS PubMed  Klugh DO. 2010. Principles of equine dentistry . London: Manson Publishing. Google Scholar CrossRef Search ADS   La Croix S, Holekamp KE, Shivik JA, Lundrigan BL, Zelditch ML. 2011a. Ontogenetic relationships between cranium and mandible in coyotes and hyenas. Journal of Morphology  272: 662– 674. Google Scholar CrossRef Search ADS   La Croix S, Zelditch ML, Shivik JA, Lundrigan BL, Holekamp KE. 2011b. Ontogeny of feeding performance and biomechanics in coyotes. Journal of Zoology  285: 301– 315. Google Scholar CrossRef Search ADS   Levene H. 1960. Robust tests for equality of variances. In: Olkin I, Ghurye SG, Hoeffding W, Madow WG, Mann HB, eds. Contributions to probability and statistics . Stanford: Stanford Universtiy Press, 278– 292. Lingle S. 2002. Coyote predation and habitat segregation of white-tailed deer and mule deer. Ecology  87: 2037– 2048. Google Scholar CrossRef Search ADS   Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E, Wagner H. 2017. Vegan: community ecology package, R package version 2.4–4 . Available at: https://cran.r-project.org/web/packages/vegan (last accessed 15 January 2018). O’Regan HJ. 2001. Morphological effects of captivity in big cat skulls. In: Wehnelt S, Hudson C, eds. Proceedings of the 3rd Zoo research symposium . Chester: Chester Zoo, 18– 22. O’Regan HJ, Kitchener AC. 2005. The effects of captivity on the morphology of captive, domesticated and feral mammals. Mammal Review  35: 215– 230. Google Scholar CrossRef Search ADS   O’Regan HJ, Turner A. 2004. The interface between conservation biology, palaeontology and archaeozoology: morphometrics and population viability analysis. In: Lauwerier RCGM, Plug I, eds. The future from the past: archaeozoology in wildlife conservation and heritage management . Oxford: Oxbow Books, 90– 96. Paulli S 1900. Über die Pneumaticität bei den Säugerthieren. Ein morphologische Studie. III. Über die Morphologie des Siebbens und die der Pneumaticität bie den Insectivoren, Hyracoidean, Chiropteran, Carnivoren, Pinnipedien, Edentaten, Rodentiern, Prosimiern und Primaten. Morphologisches Jahrbuch  28: 483– 564. R Core Team 2017. R: A language and environment for statistical computing . Vienna: R Foundation for Statistical Computing. Ripple WJ, Van Valkenburgh B. 2016. Linking top-down forces to the Pleistocene megafaunal extictions. BioScience  60: 516– 526. Google Scholar CrossRef Search ADS   Samuels JX. 2009. Cranial morphology and dietary habits of rodents. Zoological Journal of the Linnean Society  156: 864– 888. Google Scholar CrossRef Search ADS   Shen L, Farid H, McPeek MA. 2009. Modeling three-dimensional morphological structures using spherical harmonics. Evolution  63: 1003– 1016. Google Scholar CrossRef Search ADS PubMed  Shen L, Makedon F. 2006. Spherical mapping for processing of 3D closed surfaces. Image and Vision Computing  24: 743– 761. Google Scholar CrossRef Search ADS   Slater GJ, Dumont ER, Van Valkenburgh B. 2009. Implications of predatory specialization for cranial form and function in canids. Journal of Zoology  278: 181– 188. Google Scholar CrossRef Search ADS   Smith RJ. 2009. Use and misuse of the reduced major axis for line-fitting. American Journal of Physical Anthropology  140: 476– 486. Google Scholar CrossRef Search ADS PubMed  Tanner JB, Dumont ER, Sakai ST, Lundrigan BL, Holekamp KE. 2008. Of arcs and vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta). Biological Journal of the Linnean Society . 95: 246– 255. Google Scholar CrossRef Search ADS   Tseng ZJ, Wang X. 2010. Cranial functional morphology of fossil dogs and adaptations for durophagy in Borophagus and Epicyon (Carnivora, Mammalia). Journal of Morphology  271: 1386– 1398. Google Scholar CrossRef Search ADS PubMed  Tseng ZJ, Wang X. 2011. Do convergent ecomorphs evolve through convergent morphological pathways? cranial shape evolution in fossil hyaenids and borophagine canids (Carnivora, Mammalia). Paleobiology  37: 470– 489. Google Scholar CrossRef Search ADS   Van Valkenburgh B. 1988. Incidence of tooth breakage among large, predatory mammals. The American Naturalist  131: 291– 302. Google Scholar CrossRef Search ADS   Van Valkenburgh B. 1999. Major patterns in the history of carnivorous mammals. Annual Review of Earth and Planetary Sciences  27: 463– 493. Google Scholar CrossRef Search ADS   Van Valkenburgh B. 2009. Costs of carnivory: tooth fracture in Pleistocene and Recent carnivores. Biological Journal of the Linnean Society  96: 68– 81. Google Scholar CrossRef Search ADS   Van Valkenburgh B, Koepfli K-P. 1993. Cranial and dental adaptations to predation in canids. Symposium of the Zoological Society of London  65: 15– 37. Wang X, Tedford RH, Taylor BE. 1999. Phylogenetic systematics of the Borophaginae (Carnivora, Canidae). Bulletin of the American Museum of Natural History  243: 1– 391. Warton DI, Duursma RA, Falster DS, Taskinen S. 2012. Smatr 3- an R package for estimation and inference about allometric lines. Methods in Ecology and Evolution  3: 257– 259. Google Scholar CrossRef Search ADS   Warton DI, Wright IJ, Falster DS, Westoby M. 2006. Bivariate line-fitting methods for allometry. Biological reviews of the Cambridge Philosophical Society  81: 259– 291. Google Scholar CrossRef Search ADS PubMed  West-Eberhard MJ. 2003. Developmental plasticity and evolution . New York: Oxford University Press. Witmer LM. 1997. The evolution of the antorbital cavity of archosaurs: a study in soft tissue reconstruction in the fossil record with an analysis of the function of pneumaticity. Journal of Vertebrate Paleontology  17: 1– 77. Google Scholar CrossRef Search ADS   Witmer LM. 1999. The phylogenetic history of paranasal air sinuses. In: Koppe R, Nagai H, Alt KW, eds. The paranasal sinuses of higher primates: development, function, and evolution . Chicago: Quintesence, 21– 34. Wolff JD. 1892. Das Gesetz der Transformation der Knochen . Berlin: A. Hirschwald. Zuccarelli MD. 2004. Comparative morphometric analysis of captive vs. wild African lion (Panthera leo) skulls. Bios  75: 131– 138. Google Scholar CrossRef Search ADS   © 2018 The Linnean Society of London, Biological Journal of the Linnean Society http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Biological Journal of the Linnean Society Oxford University Press

Diet-related differences in craniodental morphology between captive-reared and wild coyotes, Canis latrans (Carnivora: Canidae)

Loading next page...
 
/lp/ou_press/diet-related-differences-in-craniodental-morphology-between-captive-HfD0htHIl6
Publisher
The Linnean Society of London
Copyright
© 2018 The Linnean Society of London, Biological Journal of the Linnean Society
ISSN
0024-4066
eISSN
1095-8312
D.O.I.
10.1093/biolinnean/blx161
Publisher site
See Article on Publisher Site

Abstract

Abstract In mammals, skull loading incurred during feeding and prey capture shapes craniodental morphology as a result of bone plasticity, but the degree to which skull form is influenced by mechanical loading is poorly understood. Here, we tested how craniodental morphology is affected by differences in diet and age between wild and captive-reared coyotes (Canis latrans) raised on a relatively soft diet lacking bone. For both samples, we quantified tooth wear, external skull dimensions and frontal sinus morphology. Skull dimensions were similar between both populations, but captive coyotes had significantly broader zygomatic widths and palates, and lesser tooth wear and tooth fracture. However, differences observed between captive and wild coyotes are less than those observed in larger, more hypercarnivorous carnivorans. Frontal sinus volume did not differ significantly between populations, but was significantly correlated with external cranial dimensions in wild, but not captive coyotes. We hypothesize that greater mechanical demands of a wild diet, as evidenced by greater tooth wear, resulted in a stronger correlation between frontal sinus form and external skull form in wild coyotes. In captive coyotes, reduced skull loading because of a softer diet reduces constraints on cranial morphology and results in less covariation between internal and external cranial morphology. INTRODUCTION Bone is a highly plastic tissue that is remodelled throughout an organism’s life in response to how it is loaded. This results in skeletons seemingly optimized to meet the mechanical demands of diet, locomotion and support (Wolff, 1892). In mammals, skull morphology is influenced by the mechanical properties of food, as well as prey capture and feeding behaviour (Herring, 1993; Van Valkenburgh, 1999; Samuels, 2009; Dumont et al., 2012; Santana et al., 2012). However, the degree to which skull shape within a species is affected by differences in food texture is poorly understood. The impact of the mechanical properties of food on mammal skull morphology is perhaps best understood from a limited number of studies comparing wild and captive-reared conspecifics (see review by O’Regan & Kitchener, 2005). Captive individuals included in these studies were typically reared on commercially produced foods with softer textures that presumably load the skull to a lesser degree than the wild-type diet, and captive individuals often differ in skull morphology from their wild counterparts. For captive lions (Panthera leo), leopards (Panthera pardus) and tigers (Panthera tigris), the consumption of a softer diet appears to result in atypical development of the zygomatic arches and palate, lesser doming of the dorsal roof of the skull, and a reduced size of crests that serve as attachment sites for the muscles that close the jaws and control head movement during prey capture and feeding (Hollister, 1917; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014). Cheetahs (Acinonyx jubatus) fed softer foods in captivity are more likely to suffer from malocclusion of the teeth attributable to minimal use and consequent atrophy of the masticatory muscles (Fitch & Fagan, 1982). However, in these studies, precise age and diet information are often unknown and could bias or obscure patterns in the data. Although the aforementioned studies suggest that external skull dimensions differ within species owing to differences in the mechanical properties of food, there is little information on how differences in skull loading impact the internal architecture of the skull. In mammal skulls, pneumatic cavities called paranasal sinuses are commonly found in the bones surrounding the nasal chamber, including the frontal, maxilla, ethmoid and sphenoid bones. Paranasal sinuses appear to develop in areas where bone is mechanically unnecessary and promote even dissipation of stress across the skull during use (Witmer, 1997, 1999; Farke, 2007, 2008, 2010; Tanner et al., 2008). In agreement with this, the morphology of the frontal bone and frontal sinus in Carnivora is correlated with diet and ecology across species (Curtis and Van Valkenburgh, 2014; Curtis et al., 2015). The frontal sinuses of carnivorans best adapted to produce high bite forces are typically large and underlie dome-shaped frontal bones, which appear to aid in uniform distribution of stress across the skull during feeding (Tanner et al., 2008; Tseng & Wang, 2010, 2011). The most extreme examples of this are seen in bone-cracking hyenas (Crocuta crocuta), bamboo-eating panda bears (Ailuropoda melanoleuca) and extinct bone-cracking canids (Borophaginae), in which the frontal sinuses pneumatize the sagittal crest (Davis, 1964; Wang, Tedford & Taylor, 1999; Tanner et al., 2008; Tseng & Wang, 2010, 2011; Curtis and Van Valkenburgh, 2014; Curtis et al., 2015). However, it remains unclear whether the link between frontal sinus morphology and cranial shape differences related to dietary differences persists at the intraspecific level and what role age plays in this relationship. Witmer (1997) hypothesized that development and maintenance of pneumatic cavities, including frontal sinuses, represents a compromise between pneumatic epithelia that opportunistically resorb bone and bone deposition in response to biomechanical loading. If this is the case, then similarity in sinus form among individuals within a species should be consistent with similar loading regimes on the skull. Correspondingly, differences in the mechanical properties of the diet among individuals should result in intraspecific variation in sinus form. In a study examining frontal sinus morphology in Carnivora that included 31 wild-caught and two captive-reared animals (one bush dog, Speothos venaticus and one snow leopard, Uncia uncia), Curtis & Van Valkenburgh (2014) found evidence suggesting that sinus morphology is affected by intraspecific differences in skull loading. Huxley (1880) reported that bush dogs have large frontal sinuses, but the skull of a captive bush dog (Speothos venaticus) included in the study by Curtis & Van Valkenburgh (2014) had no frontal sinuses. Bush dog skulls are adapted for subduing relatively large prey (Van Valkenburgh & Koepfli, 1993), and a softer diet in captivity might have affected development of the feeding musculature and skull in ways that resulted in a lack of space for a frontal sinus to develop in that particular individual. In addition, a wild-caught male raccoon dog (Nyctereutes procyonoides) examined by Curtis & Van Valkenburgh (2014) had small frontal sinuses, whereas a wild-caught female lacked frontal sinuses. The skull of the male showed greater development of muscle attachment sites for the jaw-closing muscles than the female, which also suggests that frontal sinus development might be impeded or unnecessary when the skull is subjected to lesser loading. However, with such small intraspecific sample sizes, it is not clear whether these differences are genuinely a result of differences in food texture and consequent skull loading. Other authors have reported that captive adult spotted hyenas (Crocuta crocuta) retain a cub-like appearance and lack the domed frontal bones, large sagittal crests and well-developed zygomatic arches typically present in their bone-eating wild counterparts (Binder & Van Valkenburgh, 2000; West-Eberhard, 2003: see caption for fig. 3.3). The domed frontal bones of wild spotted hyenas cover impressively large frontal sinuses that completely fill the sagittal crest, and it is likely that captive individuals fed on softer diets retain a cub-like appearance owing to a lack of development of the frontal sinus and sagittal crest, although this has yet to be quantified. To gain a better understanding of how external and internal skull shape are affected by differences in the mechanical properties of food, we investigated intraspecific variation in skull and frontal sinus morphology in the coyote (Canis latrans). We used a sample of captive-reared individuals with associated age and diet information. These individuals were raised in controlled conditions and fed soft, commercially produced food for a study of cranial growth and development (La Croix et al., 2011a, b). We compared the skull and sinus morphology of our captive-reared sample with that of a sample of wild coyotes with associated age and diet data (Johnson, 1978; Blood, Matson & Patten, 1985; Bartel & Knowlton, 2005). This allowed us to test whether skull dimensions and frontal sinus morphology vary in response to dietary differences within species, and whether frontal sinus morphology is remodelled with age. The diet of the wild coyotes (largely lagomorphs and rodents) is expected to have demanded higher bite forces on average and subjected their skulls to greater stress and strain than the soft commercial food fed to the captive coyotes. Skull morphology in captive coyotes fed a soft diet might be less functionally constrained than when subjected to the mechanical challenges of a wild diet. As a result, we expect skull shape to be more variable in captive coyotes and to show lesser development of attachment sites for the feeding musculature, as seen in similar comparisons in other taxa (Hollister, 1917; Fitch & Fagan, 1982; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014). We also expect to observe more pronounced doming of the frontal bones and larger frontal sinuses in wild coyotes that would indicate that their skulls are better shaped to cope with a more mechanically challenging diet (Davis, 1964; Wang et al., 1999; Tseng & Wang, 2010, 2011; Curtis and Van Valkenburgh, 2014; Curtis et al., 2015). With reduced skull loading in captive coyotes, we expect frontal sinus size and shape to be more variable and less strongly correlated with external skull shape than in wild coyotes, reflecting relaxed constraints on skull performance. This is the first study to test for intraspecific differences in sinus morphology related to differences in diet and is the first to look at the impact of age on sinus morphology in a non-human species. The results should improve our understanding of the role of mechanical loading in shaping mammal skull morphology. MATERIAL AND METHODS Specimens We sampled specimens from two populations: wild-caught and captive-reared C. latrans (Supporting information, Table S1). The wild-caught population consisted of 12 skulls of male C. latrans collected in Kern County, CA, USA between the years 1969 and 1973 that were aged by counts of cementum annuli (Blood et al., 1985). We selected our wild sample to include only those skulls that were scored as having negligible error in age estimation by Blood et al. (1985). The diet of this population consisted predominantly of lagomorphs and rodents with some livestock supplementation, and bone consumption was common (Cypher, Spencer & Scrivner, 1994). The captive-reared sample consisted of 12 skulls from male C. latrans drawn from a captive population maintained at the Logan Field Station in Millville, UT, USA (La Croix et al., 2011a, b). This population was established using wild coyotes trapped as pups in Idaho and Utah, where their natural diet is predominantly lagomorphs and rodents (Johnson, 1978; Bartel & Knowlton, 2005). The diet of the captive-reared sample consisted of commercially produced wet food for fur-bearing mammals, and never included bones, other foods or chew toys (La Croix et al., 2011a, b). Specimens ranged in age from 6 months (by which time the skull size and shape have matured; La Croix et al., 2011a, b) to 13 years. We age matched our captive and wild samples as closely as possible, and populations did not differ significantly in age (Student’s paired t-test, t = 1.19, d.f. = 10, P > 0.05). We included only males because male and female coyotes mature at different rates and are sexually dimorphic (La Croix et al., 2011a, b;,Blood et al., 1985). Although a sample size of 12 individuals per population is not particularly large, previous studies including similar sample sizes have shown differences in captive and wild individuals (e.g. Hartstone-Rose et al., 2014), and the time required to collect and process computed tomography (CT) data in non-trivial. One specimen (MSU 36583) in the captive sample was an outlier in frontal sinus volume, with considerably larger frontal sinuses than all other skulls in our sample, and showed a prominently domed frontal bone that distinguished it qualitatively from other skulls in our sample. Consequently, this specimen was excluded from all quantitative analyses (Supporting information, Table S1). To make sample sizes even for paired analyses, we randomly selected LACM 43391 (one of two wild specimens aged 13 years old) using a coin-flip for exclusion. LACM 43391 was included in all other analyses. Dental indicators of food texture The degree of tooth wear and tooth fracture, as well as differences in wear and fracture among tooth types (e.g. canines vs. molars), are indicators of food texture and feeding behaviour in mammalian carnivores (Van Valkenburgh, 1988, 2009). We quantified the amount of tooth wear in our captive and wild coyote samples using five categories: slight, slight–moderate, moderate, moderate–heavy and heavy (Van Valkenburgh, 1988, 2009). We documented the tooth wear stage for each tooth type [incisors, canines, premolars (P1–P3), carnassials (P4) and molars] and an overall tooth wear stage for each individual. We assessed tooth fracture frequency by counting the number of teeth broken in life per individual. A tooth was counted as broken if it showed wear on the broken surface indicating that the fracture occurred antemortem. We used pairwise Wilcoxon signed rank tests and a Bonferroni correction to compare tooth wear stage and Pearson’s χ2 test with a continuity correction to test for differences in tooth fracture frequency between wild and captive coyotes in R v. 3.4.1 (R Core Team, 2017). We expected to observe significantly greater amounts of tooth wear and tooth fracture in wild coyotes than in captive coyotes, owing to a more mechanically challenging diet in wild coyotes and consequent differences in skull loading between groups. We also expected to see tooth wear and tooth fracture increase at a greater rate with age in wild than in captive coyotes. To test how tooth wear and tooth fracture vary with age, we used ordinary least squares regressions to estimate the relationship between these variables and age and to test the strength of the relationships. To test for differences in the rate of tooth wear and tooth fracture accumulation between populations, we performed a slope test using the ‘smatr’ package in R (Warton et al., 2012; R Core Team, 2017). Computed tomography scanning We CT scanned all skulls in the UCLA School of Dentistry using a dental Cone-Beam CT scanner, with slice thickness of 0.03 mm for all scans. We segmented skulls from CT scans by manually thresholding bone from air using Mimics v. 17 (http://biomedical.materialise.com, last accessed 15 January 2018. Materialise, 2014) specialized imaging software, and generated volumetric models of skulls for use in morphometric analyses. The CT scan parameters are available from the corresponding author. Skull size and shape We quantified skull dimensions in captive and wild coyotes in two ways. First, we took a set of 16 linear measurements (Table 1; raw data provided in Supporting information, Table S1, Fig. S1) from volumetric models of each skull to the nearest 0.01 mm in Mimics (Materialise, 2014). Measurements included those used by previous authors to compare differences in skull dimensions between captive and wild individuals (Hollister, 1917; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004), as well as additional measures that captured the dimensions of the frontal bone, rostrum and braincase. Thus, we were able to test how our results fit within the context of prior work on differences in skull morphology between captive and wild conspecifics. Table 1. Descriptions of linear measurements taken from coyote crania Measurement  Description  WPOP  Width across postorbital processes of the frontal bone  OOL  Occiput to orbit length  CBL  Condylobasal length  ZW  Zygomatic width  JWC  Palate width between medial borders of canine alveoli  JWCn  Palate width between anterolateral borders of carnassial alveoli  PL  Palate length from anteriormost midsagittal point on incisive to posterior margin of palatine  M1O  Depth from anterolateral alveolus of M1 to the orbit  HJ  Dorsoventral height of the jugal  WFL  Width between the fronto-lacrimal sutures  POC  Width of postorbital constriction  FL  Mid-sagittal length of frontal bone  MW  Mastoid width  LCM  Length from tip of canine to mandibular fossa  WMF  Width across mandibular fossae  BCL  Basicranial length  Measurement  Description  WPOP  Width across postorbital processes of the frontal bone  OOL  Occiput to orbit length  CBL  Condylobasal length  ZW  Zygomatic width  JWC  Palate width between medial borders of canine alveoli  JWCn  Palate width between anterolateral borders of carnassial alveoli  PL  Palate length from anteriormost midsagittal point on incisive to posterior margin of palatine  M1O  Depth from anterolateral alveolus of M1 to the orbit  HJ  Dorsoventral height of the jugal  WFL  Width between the fronto-lacrimal sutures  POC  Width of postorbital constriction  FL  Mid-sagittal length of frontal bone  MW  Mastoid width  LCM  Length from tip of canine to mandibular fossa  WMF  Width across mandibular fossae  BCL  Basicranial length  For tooth fracture: total = mean raw number of broken teeth; percentage broken = mean (number of broken teeth/total teeth present). View Large As linear morphometrics cannot reveal changes in the relative positions of anatomical structures, we also used geometric morphometric techniques as a second method to investigate skull shape. We placed 20 landmarks (Fig. 1; Supporting information, Table S2) on the left sides of volumetric models of crania using the MedCAD module in Mimics (Materialise, 2014). Unilateral landmarking is time efficient, produces accurate size estimates for bilateral structures, and results in fewer redundant data points than bilateral landmark configurations. However, recent work suggests that landmarking only one side may result in slightly lower accuracy in estimating shape (Cardini, 2017). We divided landmarks into two sets; one that delimited frontal bone dimensions and the second that described overall skull shape minus the frontal bone (Supporting information, Table S2). Dividing the landmark sets in this way allowed us to test whether frontal sinus morphology is more strongly correlated with the morphology of the frontal bone than the overall skull, as has been shown before in interspecific studies (Farke, 2007, 2010; Curtis & Van Vakenburgh, 2014). Figure 1. View large Download slide Landmarks used for geometric morphometric analyses of skull and frontal bone shape, respectively. Landmarks included in the ‘skull’ dataset are in black; landmarks included in the ‘frontal bone’ dataset are in white. Landmark descriptions are given in the Supporting information (Table S2). Figure 1. View large Download slide Landmarks used for geometric morphometric analyses of skull and frontal bone shape, respectively. Landmarks included in the ‘skull’ dataset are in black; landmarks included in the ‘frontal bone’ dataset are in white. Landmark descriptions are given in the Supporting information (Table S2). We performed a Procrustes fit to align, rotate and scale landmark configurations to a common centroid size in MorphoJ v. 1.06b (Klingenberg, 2011). We extracted centroid sizes for the frontal bone and skull, respectively, for each specimen as a proxy for frontal bone and skull size in subsequent analyses (raw data provided in Supporting information, Table S1). We used Student’s paired t-tests and Bonferroni post hoc tests to test for significant differences among our 16 linear measurements, as well as skull and frontal bone centroid sizes between captive and wild coyotes in R (R Core Team, 2017). We used Levene’s test to test for significant differences in variance among these same variables in R using the ‘levene.test’ function in the package ‘car’ (Levene, 1960; Fox & Weisberg, 2011; R Core Team, 2017). We examined skull shape variation in captive and wild coyotes using principal components analysis (PCA) on a covariance matrix generated from Procrustes-fitted landmark data from the entire skull and the frontal bone, respectively, in MorphoJ (Klingenberg, 2011). We used a broken stick test to determine significant principal components (PCs) using the ‘bstick’ function in the ‘vegan’ package in R (Frontier, 1976; Jackson, 1993; Oksanen et al., 2017). To test for differences between skull and frontal bone shape, respectively, between captive and wild coyotes, we used a MANOVA with a Pilllai approximation to estimate an F-value on all significant PCs in R (R Core Team, 2017). To test for differences in variance in skull and frontal bone shape between captive and wild coyotes, we used an analysis of multivariate homogeneity of group dispersions on a distance matrix computed from Procrustes coordinates for skull and frontal bone shape, respectively, using the ‘betadisper’ function in the package ‘vegan’ in R (Anderson, 2006; Anderson, Ellingsen & McArdle, 2006; Oksanen et al., 2017). This method is a multivariate analogue of a Levene’s test that uses dissimilarity measures (Anderson, 2006; Anderson et al., 2006). In both linear morphometric and three-dimensional geometric morphometric analyses, we expected wild coyotes to show less skull shape disparity than captive coyotes because of the greater functional demands of a harder diet. We also expected wild coyotes to show greater development of attachment sites for feeding musculature, such as the zygomatic arches and sagittal crest, as well as enhanced doming of the frontal bone. Sinus size and shape We segmented left and right frontal sinuses from CT scans and estimated total frontal sinus volume (in cubic millimetres) using Mimics (Materialise, 2014). We used Student’s paired t-tests to test for significant differences in sinus size between populations and Levene’s tests to test for differences in variance of frontal sinus volume in R (R Core Team, 2017). Given that wild coyote skulls are subjected to more diet-related loading than captive coyote skulls, we expect to observe larger frontal sinuses in wild coyotes, as well as less variability in sinus size in wild coyotes owing to greater constraints on skull function. We quantified frontal sinus shape using spherical harmonics software (SPHARM v. 2, http://www.enallagma.com/SPHARM.php, last accessed 15 January 2018; Shen & Makedon, 2006; Shen, Fared & McPeek, 2009), following the methods of Curtis and Van Valkenburgh (2014). We performed SPHARM analyses on right frontal sinuses only, and rotated, aligned and scaled all frontal sinuses to a common centroid size in SPHARM before shape analysis. SPHARM approximates the shapes of three-dimensional surfaces by building upon the equation for a unit sphere, and is particularly useful for quantifying the shape of structures, such as frontal sinuses, that lack definitive and homologous landmarks. SPHARM generates a series of coefficients, on which multivariate statistics can be performed to quantify shape disparity among a set of surfaces, as we do here for frontal sinuses (Shen & Makedon, 2006; Shen et al., 2009). SPHARM coefficients are provided in the Supporting information (Table S1). To explore shape variation in frontal sinus shape in captive and wild coyotes, we performed a PCA on sinus SPHARM coefficients using the functions available in SPHARM software. We used a broken stick test to determine significant PCs using the ‘bstick’ function in the ‘vegan’ package in R (Frontier, 1976; Jackson, 1993; Oksanen et al., 2017). To test for frontal sinus shape differences between populations, we conducted a MANOVA with a Pilllai approximation to estimate an F-value on all significant PCs in R. As an additional test of shape differences between captive and wild coyotes, we conducted a Mahalanobis distance test with a permutation test (10000 rounds) to estimate a P-value in MorphoJ (Klingenberg & Monteiro, 2005; Klingenberg, 2011). If shape showed significant differences between populations, we conducted a discriminant function analyses on Procrustes coordinates and used a ‘leave-one-out’ cross-validation to test the reliability of the discrimination using the functions available in MorphoJ (Klingenberg, 2011). To test for differences in skull and frontal bone shape variance between captive and wild coyotes, we performed an analysis of multivariate homogeneity of group dispersions with a distance matrix based on SPHARM coordinates using the ‘betadisper’ function in the package ‘vegan’ in R (Anderson, 2006; Anderson et al., 2006; Oksanen et al., 2017). Similar to our expectations for skull shape, we expected frontal sinus shape to show less variation in wild coyotes than in captive coyotes. This is because we presume that wild coyotes have greater functional demands on skull performance. Changes in frontal sinus size with age and tooth wear To test whether frontal sinus size is correlated with age or tooth wear, we plotted the cube root of total frontal sinus volume against age and tooth wear, respectively. Preliminary scatter plots of frontal sinus size against age and tooth wear, respectively, suggested that frontal sinus size initially increases then decreases with age and tooth wear, and suggested that these data might covary in a non-linear fashion. Therefore, we fitted these relationships with three separate regression models (linear, quadratic with a linear term, and quadratic without a linear term) in R (R Core Team, 2017). We used Akaike information criterion (AIC; Akaike, 1973) scores to rank which model best fitted the data for both captive and wild coyotes, with the lowest AIC score representing the model that best fitted the data. We examined the relationship with both age and tooth wear, despite the fact that they covary, because sinus morphology might show different relationships with each. Covariation between frontal sinus size and external skull dimensions We examined the relationship between log10-transformed frontal sinus volume and linear cranial measurements using reduced major axis (RMA) regression. We performed RMA regressions using the ‘smatr’ package in R (Warton et al., 2012; R Core Team, 2017). RMA regression analysis is preferred over ordinary least squares when both variables are measured with error, and the allometric relationship between two variables is the desired outcome rather than the ability to predict one variable from another (Warton et al., 2006; Smith, 2009). If frontal sinus morphology is affected by diet-related skull use, we expected sinus volume to be correlated more strongly with cranial measurements in wild coyotes than in captive coyotes. RESULTS Dental indicators of food texture As expected, tooth wear and tooth fracture patterns were significantly different between wild and captive coyotes, supporting our assumption that food texture and consequent levels of skull loading differed between captive and wild individuals (Table 2). Although mean overall tooth wear did not differ significantly between wild and captive coyotes, wild coyotes had significantly greater wear for all individual tooth types (pairwise Wilcoxon signed-rank tests, P < 0.05), except the incisors (pairwise Wilcoxon signed-rank test, P > 0.05; Table 2). Total tooth fracture (Pearson’s χ2 test, χ2 = 10.51, P = 0.001) was significantly higher in wild coyotes. Table 2. Summary statistics from paired Wilcoxon signed rank tests comparing tooth wear stage in captive vs. wild coyotes Tooth wear  Captive mean  Wild mean  P-value  Total  2.36  2.82  NS  Incisors  2.45  2.91  NS  Canines  2.36  3.00  0.03  Premolars  2.36  3.00  0.04  Carnassials  2.18  2.82  0.03  Molars  2.18  2.82  0.03  Tooth wear  Captive mean  Wild mean  P-value  Total  2.36  2.82  NS  Incisors  2.45  2.91  NS  Canines  2.36  3.00  0.03  Premolars  2.36  3.00  0.04  Carnassials  2.18  2.82  0.03  Molars  2.18  2.82  0.03  View Large Tooth wear was positively and significantly correlated with age in both wild and captive coyotes (Fig. 2A, Table 3). Tooth wear was more strongly correlated with age in wild than captive coyotes (R2 = 0.93, P < 0.0001 for wild; R2 = 0.72, P = 0.0008 for captive). Although the slope of this relationship was higher in wild than captive coyotes (wild slope = 0.27, captive slope = 0.19), this difference was not significant at the 0.05 level. Figure 2. View largeDownload slide Ordinary least squares regressions of (A) tooth wear stage against age and (B) raw number of broken teeth against age. Regression statistics are given in Table 3. Figure 2. View largeDownload slide Ordinary least squares regressions of (A) tooth wear stage against age and (B) raw number of broken teeth against age. Regression statistics are given in Table 3. Table 3. Regression statistics for total tooth wear against age and total tooth fracture against age Regression  Slope  y-intercept  R2  P-value  Tooth wear vs. age          Captive  0.19 (0.10–0.28)  1.28 (0.65–1.92)  0.72  0.0008  Wild  0.27 (0.22–0.33)  1.31 (0.91–1.70)  0.93  < 0.0001  Tooth fracture vs. age          Captive  0.22 (−0.05 to 0.50)  0.85 (−1.08 to 2.79)  0.27  NS  Wild  0.67 (0.32–1.03)  0.85 (−1.78 to 3.64)  0.64  0.003  Regression  Slope  y-intercept  R2  P-value  Tooth wear vs. age          Captive  0.19 (0.10–0.28)  1.28 (0.65–1.92)  0.72  0.0008  Wild  0.27 (0.22–0.33)  1.31 (0.91–1.70)  0.93  < 0.0001  Tooth fracture vs. age          Captive  0.22 (−0.05 to 0.50)  0.85 (−1.08 to 2.79)  0.27  NS  Wild  0.67 (0.32–1.03)  0.85 (−1.78 to 3.64)  0.64  0.003  Tooth fracture represents total number of broken teeth. The 95% confidence intervals are given for slope and y-intercept. P-values are from a test of the null expectation that slope = 0. View Large Tooth fracture was significantly correlated with age in wild but not in captive coyotes (Fig. 2B; R2 = 0.64, P = 0.003 for wild; R2 = 0.27, P > 0.05 for captive). All captive individuals > 7 years of age had fewer broken teeth (fewer than five broken) than age-matched wild individuals in the same age cohort (seven to 15 broken). We also observed that the pulp cavities of the teeth in the oldest individuals from our wild sample, most of which exhibited the greatest amount of tooth wear and fracture, were almost completely or completely filled with secondary dentin, whereas in the captive sample, pulp cavities were still clearly visible in CT scans (Fig. 3). Figure 3. View largeDownload slide Comparison of pulp cavity of the first upper molar (M1) taken at the paracone between captive and wild coyotes, showing that the pulp cavities of the oldest wild coyotes in our samples are filled almost completely with secondary dentin in comparison to those of captive coyotes, in which there are visible pulp cavities even at a similar age and tooth wear stage. Figure 3. View largeDownload slide Comparison of pulp cavity of the first upper molar (M1) taken at the paracone between captive and wild coyotes, showing that the pulp cavities of the oldest wild coyotes in our samples are filled almost completely with secondary dentin in comparison to those of captive coyotes, in which there are visible pulp cavities even at a similar age and tooth wear stage. Skull size and shape We found that skull and frontal bone centroid sizes did not differ significantly between captive and wild coyotes (Table 4; Student’s paired t-tests, P > 0.05), nor did they differ significantly in variance (Supporting information, Table S3; Levene’s tests, P > 0.05). Nevertheless, we did observe significant differences in the size of three of 16 linear measurements between captive and wild coyotes. Zygomatic width (Table 4; Student’s paired t-test, t = 2.25, d.f. = 10, P = 0.05), occiput to orbit length (Table 4; Student’s paired t-test, t = 2.63, d.f. = 10, P = 0.03) and palate width at the carnassial (Table 4; Student’s paired t-test, t = 5.69, d.f. = 10, P = 0.0002) were all significantly larger in captive than in wild coyotes. Contrary to expectation, the variance in linear cranial measurements did not differ significantly between captive and wild coyotes (Supporting information, Table S3; Levene’s test, all variables P > 0.05). Table 4. Summary statistics from Student’s paired t-tests comparing age, frontal sinus volume, centroid size and linear measurements in captive vs. wild coyotes Trait  Captive mean  Wild mean  d.f.  t  P-value  Age  5.74  5.54  10  1.19  NS  Frontal sinus volume  2712.76  2779.94  10  −0.32  NS  Skull centroid size  240.32  237.85  10  1.13  NS  WPOP  46.99  46.55  10  0.23  NS  OOL  108.64  106.22  10  2.63  0.03  CBL  184.03  183.08  10  0.52  NS  ZW  98.24  96.62  10  2.25  0.05  JWC  30.58  31.13  10  −1.02  NS  JWCn  53.84  51.54  10  5.69  0.0002  PL  95.44  96.19  10  −0.48  NS  M1O  26.39  27.21  10  −1.26  NS  HJ  11.11  11.86  10  −1.63  NS  WFL  39.01  38.53  10  0.52  NS  POC  35.43  34.52  10  1.05  NS  FL  52.65  52.45  10  0.26  NS  MW  55.52  54.73  10  1.00  NS  LCM  124.11  123.13  10  0.79  NS  WMF  85.30  85.04  10  0.39  NS  BCL  46.32  46.36  10  −0.06  NS  Trait  Captive mean  Wild mean  d.f.  t  P-value  Age  5.74  5.54  10  1.19  NS  Frontal sinus volume  2712.76  2779.94  10  −0.32  NS  Skull centroid size  240.32  237.85  10  1.13  NS  WPOP  46.99  46.55  10  0.23  NS  OOL  108.64  106.22  10  2.63  0.03  CBL  184.03  183.08  10  0.52  NS  ZW  98.24  96.62  10  2.25  0.05  JWC  30.58  31.13  10  −1.02  NS  JWCn  53.84  51.54  10  5.69  0.0002  PL  95.44  96.19  10  −0.48  NS  M1O  26.39  27.21  10  −1.26  NS  HJ  11.11  11.86  10  −1.63  NS  WFL  39.01  38.53  10  0.52  NS  POC  35.43  34.52  10  1.05  NS  FL  52.65  52.45  10  0.26  NS  MW  55.52  54.73  10  1.00  NS  LCM  124.11  123.13  10  0.79  NS  WMF  85.30  85.04  10  0.39  NS  BCL  46.32  46.36  10  −0.06  NS  For measurement abbreviations, see Table 1. View Large A PCA on the covariance matrix generated using Procrustes coordinates for skull shape excluding the frontal bone revealed seven significant PCs. A plot of PC2 against PC1 (together explaining 35.91% of cumulative shape variation) showed that captive and wild coyotes shared similar shape space (Fig. 4A). PC1 (19.60% of total skull shape variation) separated individuals with dorsoventrally shallow, mediolaterally slender and downturned rostra, slender zygomatic arches, poorly developed sagittal crests and small mastoid processes (positive scores) from individuals with deep, broad upturned rostra, broad zygomatic arches, well-developed sagittal crests and large mastoid processes (negative scores). PC2 (16.31% of total skull shape variation) separated individuals with proportionally short and dorsoventrally deep rostra, broad zygomatic arches and a well-developed sagittal crest (positive scores) from individuals with proportionally long, dorsoventrally shallow rostra and poorly developed sagittal crests (negative scores). A MANOVA on the first seven PCs showed that skull shape is significantly different between populations (Pilllai approximation, d.f. = 7,15, F = 3.39, P = 0.02). A Mahalanobis distance test also revealed that skull shape differs significantly between captive and wild coyotes in our sample (Mahalanobis distance = 5.42, T2 = 168.70, P = 0.01). The single discriminant function successfully separated captive and wild coyotes. However, in leave-one-out cross-validation tests, the discriminant function was able to classify skull shape correctly only 57% of the time, which suggests that separation of wild and captive coyotes by the discriminant function is likely to be an artefact of low sample sizes and a large number of variables in our dataset (Table 5). Figure 4. View largeDownload slide (A) Principal components analysis on Procrustes coordinates for skull shape. Principal component (PC) 1 accounted for 19.60% of total shape variation, and PC2 accounted for 16.31% of total shape variation. (B) Principal components analysis on Procrustes coordinates for frontal bone shape. PC1 accounted for 36.77% of total shape variation, and PC2 accounted for 23.96% of total shape variation. Wild and captive coyotes are denoted as in Figure 2. Figure 4. View largeDownload slide (A) Principal components analysis on Procrustes coordinates for skull shape. Principal component (PC) 1 accounted for 19.60% of total shape variation, and PC2 accounted for 16.31% of total shape variation. (B) Principal components analysis on Procrustes coordinates for frontal bone shape. PC1 accounted for 36.77% of total shape variation, and PC2 accounted for 23.96% of total shape variation. Wild and captive coyotes are denoted as in Figure 2. Table 5. Summary of results from a discriminant function analysis of skull shape with a leave-one-out cross-validation (classification analysis) Group  Correct  Misclassified  Total  Percentage correct  Discriminant function          Wild coyotes  12  0  12  100  Captive coyotes  11  0  11  100  Total  23  0  23  100  Classification analysis  Wild coyotes  9  3  12  75  Captive coyotes  4  7  11  36  Total  13  10  23  57  Group  Correct  Misclassified  Total  Percentage correct  Discriminant function          Wild coyotes  12  0  12  100  Captive coyotes  11  0  11  100  Total  23  0  23  100  Classification analysis  Wild coyotes  9  3  12  75  Captive coyotes  4  7  11  36  Total  13  10  23  57  View Large A PCA on a covariance matrix generated using Procrustes coordinates for frontal bone shape alone revealed three significant PCs. Captive coyotes occupied a greater amount of shape space in a plot of PC2 against PC1 (together explaining 60.73% of cumulative shape variation), but this difference in shape variance was not significant (Fig 4B; multivariate homogeneity of group dispersions test, P > 0.05). PC1 (36.77% of total frontal bone shape variation) distinguished frontal bones with slender postorbital processes that were dorsoventrally shallow with flat dorsal profiles (positive scores) from those with wide postorbital processes that were dorsoventrally deep with more domed profiles (negative values). PC2 (23.96% of total frontal bone shape variation) distinguished frontal bones with wider postorbital processes that were dorsally convex (positive scores) from those with slender postorbital processes that were dorsally flat (negative scores). Despite falling within the same shape space as captive coyotes, wild coyote frontal bone shape tended to plot slightly more positively along PC2, but a MANOVA of the first three PCs showed no significant differences in frontal bone shape between populations (Pilllai approximation, F = 1.85, d.f. = 3, 19, P > 0.05). A Mahalanobis distance test also showed no significant differences in frontal bone shape between captive and wild coyotes in our sample (Mahalanobis distance = 3.71, T2 = 78.86, P = 0.014. Sinus size and shape Contrary to expectations, frontal sinus volume did not differ significantly between captive and wild populations (Table 4; Student’s paired t-test, t = −0.42, d.f. = 10, P > 0.05), and variance in sinus size also did not differ between populations (Supporting information, Table S3; Levene’s test, d.f. = 1, 20, P > 0.05). Qualitatively, frontal sinus morphology was broadly similar among all specimens examined (Figs 5, 6). In general, the frontal sinuses were centred at the postorbital processes, and never reached the margin of the fronto-parietal suture as seen in larger hypercarnivorous canids, such as the gray wolf (Canis lupus; Curtis & Van Valkenburgh, 2014). However, in the wild-caught coyotes, the frontal sinuses tended to pneumatize further posteriorly relative to the postorbital processes than those of the captive animals. In both groups, the posterior margin of the frontal sinuses on each side of the midline typically showed what appeared to be three lobes of pneumatization separated by two small bony trabeculae. Anteriorly, the frontal sinuses on each side of the midline usually had two lobes that differed little within and between captive and wild coyotes; both displayed a medial lobe where the sinus was connected to the nasal chamber via an ostium located posterior to frontoturbinal 2 (sensuPaulli, 1900), and a lateral lobe that expanded anteriorly from the postorbital process. Figure 5. View largeDownload slide Dorsal views of sinuses within the skulls of wild coyotes. Age of animal and tooth wear stage are listed beneath each skull. LACM 43391 is shown with only a right frontal sinus because the left frontal bone had a gunshot wound. Figure 5. View largeDownload slide Dorsal views of sinuses within the skulls of wild coyotes. Age of animal and tooth wear stage are listed beneath each skull. LACM 43391 is shown with only a right frontal sinus because the left frontal bone had a gunshot wound. Figure 6. View largeDownload slide Dorsal view of frontal sinuses within the skulls of captive-reared coyotes. Age of animal and tooth wear stage are listed beneath each skull. Figure 6. View largeDownload slide Dorsal view of frontal sinuses within the skulls of captive-reared coyotes. Age of animal and tooth wear stage are listed beneath each skull. A PCA on SPHARM coefficients describing frontal sinus shape revealed four significant PCs. Both groups overlapped widely in morphospace (Fig. 7). PC1 (21.2% of total frontal sinus shape variation) contrasted individuals with dorsoventrally deep, dorsally convex frontal sinuses (positive values) from those with dorsoventrally shallow, less dorsally convex (negative values), and showed a weak but significant relationship with frontal sinus volume (R2 = 0.30, P = 0.007). PC2 accounted for 13.9% of total frontal sinus shape variation, and again separated individuals based on dorsoventral depth and complexity, and also showed a weak but significant correlation with frontal sinus volume (R2 = 0.20, P = 0.03). In this case, sinuses that plotted negatively were dorsoventrally deep and pneumatized further posteriorly across the braincase. Wild coyotes tended to plot more positively along PC1 and ore negatively along PC2 than captive coyotes, corroborating our qualitative observations that the frontal sinuses tend to pneumatize further posteriorly and to be deeper dorsoventrally and more convex dorsally in our wild sample. A MANOVA of the first four PCs detected no differences in frontal sinus shape between captive and wild coyotes (Pilllai approximation, d.f. = 4, 18, F = 1.79, P > 0.05), and frontal sinus shape variance also did not differ between both groups in our sample (multivariate homogeneity of group dispersions test, P > 0.05). Changes in frontal sinus size with age and tooth wear There was no significant relationship between frontal sinus volume and either age or tooth wear in our captive sample. However, within the wild sample, there was a significant non-linear relationship between frontal sinus volume and tooth wear (P = 0.04; Table 6, Fig. 8). The relationship between frontal sinus volume and age in our wild sample was similar in shape to that for tooth wear, but did not quite achieve significance (P = 0.09; Table 6, Fig 8). In the wild sample, the significant relationship between frontal sinus volume and tooth wear was best modelled as a quadratic function including a linear term (Fig. 8). Frontal sinus volume increased as tooth wear stage increased from slight to moderate, but then frontal sinus volume declined as wear stage progressed. Table 6. Comparison of regression models of frontal sinus volume against tooth wear and age for the wild sample (N = 12)   R2  P-value  AIC score  TSV⅓ vs. tooth wear        Quadratic with linear term  0.51  0.04  44.36  Quadratic without linear term  0  NS  50.79  Ordinary least squares  0.04  NS  50.31  TSV⅓ vs. age        Quadratic with linear term  0.42  NS  46.36  Quadratic without linear term  0  NS  50.82  Ordinary least squares  0.03  NS  50.48    R2  P-value  AIC score  TSV⅓ vs. tooth wear        Quadratic with linear term  0.51  0.04  44.36  Quadratic without linear term  0  NS  50.79  Ordinary least squares  0.04  NS  50.31  TSV⅓ vs. age        Quadratic with linear term  0.42  NS  46.36  Quadratic without linear term  0  NS  50.82  Ordinary least squares  0.03  NS  50.48  Lowest Akaike information criterion score represents the model that best fits the data. TSV, total frontal sinus volume. View Large Figure 7. View largeDownload slide Principal components analysis on SPHARM coefficients. Principal component (PC) 1 accounted for 21.22% of total shape variation, and PC2 accounted for 13.93% of total shape variation. Dorsal and medial views of skulls showing sinus shapes at the extremes for each PC along each axis. Wild and captive coyotes are coded as in Figure 2. Figure 7. View largeDownload slide Principal components analysis on SPHARM coefficients. Principal component (PC) 1 accounted for 21.22% of total shape variation, and PC2 accounted for 13.93% of total shape variation. Dorsal and medial views of skulls showing sinus shapes at the extremes for each PC along each axis. Wild and captive coyotes are coded as in Figure 2. Figure 8. View largeDownload slide Plot of frontal sinus volume against tooth wear stage. The curve represents a quadratic equation with a linear term that best fitted this relationship in wild coyotes. The relationship between sinus size and tooth wear was not significant in captive coyotes. Regression statistics are given in Table 6. Data for wild and captive coyotes are coded as in Figure 2. Figure 8. View largeDownload slide Plot of frontal sinus volume against tooth wear stage. The curve represents a quadratic equation with a linear term that best fitted this relationship in wild coyotes. The relationship between sinus size and tooth wear was not significant in captive coyotes. Regression statistics are given in Table 6. Data for wild and captive coyotes are coded as in Figure 2. Covariation between frontal sinus size and external skull dimensions In our wild sample, frontal sinus volume was positively correlated with width across the postorbital processes, zygomatic width, width between the fronto-lacrimal sutures on the orbit, width across the mandibular fossae, and basicranial length (Table 7). Among these five cranial measurements, frontal sinus volume scaled with positive allometry (slope > 3) in all but width across the postorbital processes, where the scaling coefficient was not significantly different from isometry (slope = 3; Table 7). We found no significant correlations between frontal sinus volume and cranial measurements in captive coyotes. Table 7. Regression statistics for log10/log10 reduced major axis regression of total frontal sinus volume against individual cranial measurements that showed a statistically significant relationship in wild coyotes TSV vs.  Slope  y-intercept  R2  WPOP  3.38 (2.05–5.58)  −1.88 (−4.82 to 1.07)  0.52  ZW  9.68 (6.72–13.93)  −15.46 (−22.60 to −8.30)  0.76  WFL  6.63 (3.91–11.24)  −6.76 (−12.57 to −0.94)  0.46  WMF  8.86 (6.35–12.37)  −13.33 (−19.15 to −7.53)  0.80  BCL  6.88 (4.04–11.70)  −7.70 (−14.08 to −1.33)  0.46  TSV vs.  Slope  y-intercept  R2  WPOP  3.38 (2.05–5.58)  −1.88 (−4.82 to 1.07)  0.52  ZW  9.68 (6.72–13.93)  −15.46 (−22.60 to −8.30)  0.76  WFL  6.63 (3.91–11.24)  −6.76 (−12.57 to −0.94)  0.46  WMF  8.86 (6.35–12.37)  −13.33 (−19.15 to −7.53)  0.80  BCL  6.88 (4.04–11.70)  −7.70 (−14.08 to −1.33)  0.46  The 95% confidence intervals are given for slope and y-intercept. TSV, total frontal sinus volume (in cubic millimetres). View Large DISCUSSION Previous explorations of the impact of differences in diet and associated feeding behaviour on skull shape within mammal species have been limited by a lack of age and diet information for specimens (however, see Forbes-Harper et al., 2017), as well as limited information on source populations for captive individuals (Hollister, 1917; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014). Without these data, it is difficult to tease apart the relative effects of age, material properties of food, and geographical variation on cranial shape. Here, we were able to use a unique sample of wild and captive-reared coyotes obtained from a geographically similar source population. Both populations also had associated age and diet information, allowing us to quantify the effects of food texture (relatively soft vs. relatively hard) on tooth condition and cranial dimensions, including the size and shape of the frontal sinuses. We found evidence that craniodental morphology was affected by differences in loading, but differences were minimal. This indicates that coyote skull morphology, and probably biomechanical performance, are not greatly affected by reduced mechanical loading. However, there were marked differences between the two samples in tooth wear and fracture that were consistent with more regular use of higher bite forces to eat more mechanically challenging food in the wild sample than the captive animals, all of which were fed processed commercial pet food without bones. As expected, we observed a greater degree of tooth wear and tooth fracture frequency in wild coyotes than in captive coyotes, supporting the assumption that our wild sample consumed a more mechanically challenging diet. Wild coyotes showed a higher frequency of tooth fracture, and all teeth, except the incisors, showed much greater wear than those of captive coyotes. Increased tooth wear and fracture in other wild carnivorans are thought to reflect more bone consumption (greater carcass utilization), as well as more rapid consumption of food as a consequence of competition for food access (Van Valkenburgh, 1988, 2009). Tooth fracture, especially of the canine teeth, could also occur during prey capture or intraspecific and interspecific combat. However, the presence of fractured cheek teeth (premolars, molars) as well as the co-occurrence of tooth fracture with increased wear on non-fractured teeth suggests that much of the breakage occurs during feeding. Although we have no data on levels of intraspecific competition in our wild coyote sample, it is clear that their diet of rodents and lagomorphs included more bone than the processed wet food consumed by the captive individuals. Wild coyotes also had a greater amount of secondary dentin filling the pulp cavities (sometimes completely), and this is likely to reflect repeated trauma to the teeth, given that trauma induces deposition of secondary dentin by odontoblasts along the periphery of the tooth pulp cavity (Klugh, 2010). This observation further supports a more mechanically challenging diet in wild coyotes. We observed a few subtle, yet significant, differences in skull dimensions between captive and wild coyotes. Similar to previous studies that used linear measurements to compare skull morphology between captive and wild conspecifics (e.g. O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004), as well as one three-dimensional geometric morphometric study (Hartstone-Rose et al., 2014), the two groups differed in zygomatic width and palate width, as well as occiput to orbit length. As in lions (O’Regan, 2001; Zuccareli, 2004; Hartstone-Rose et al., 2014), zygomatic width was broader in captive individuals, and like lions, the absence of normal muscle forces experienced by wild counterparts resulted in the zygomatic arches being somewhat circular in cross-section, rather than mediolaterally compressed, as in typical wild lions and coyotes. However, these differences were less pronounced in coyotes. We found no significant differences in variance for any traits between captive and wild coyotes in our sample, which does not support the hypothesis that the skulls of captive coyotes are less morphologically constrained than wild conspecifics. However, with a larger sample size, it is possible that results would show that frontal bone shape is more variable in captive individuals (see Fig. 4B). Although we expected to find that wild coyotes had larger frontal sinuses than captive coyotes, the two samples did not differ significantly in sinus size, suggesting that differences in food texture do not impact frontal sinus morphology. However, there were slight qualitative differences in frontal sinus shape between the coyote samples, such as the tendency for the frontal sinuses of wild individuals to be deeper dorsoventrally and to extend further posteriorly. We also observed the greatest frontal sinus volume among all coyotes sampled in a captive individual (MSU 36583), which was an outlier in our sample and excluded from all statistical analyses. This suggests that patterns of bone deposition and removal might alter spatially in response to differences in diet-related loading of the skull, and that aberrant morphologies might be more common in captive individuals. We suspect that a larger sample size could reveal significant differences in sinus shape or size not visible in our sample. Frontal sinus volume increased with age and tooth wear up to a point in our wild sample but then declined as they aged further. The decrease in frontal sinus volume associated with heavier tooth wear in the oldest wild coyotes in our sample might reflect recruitment of proportionally greater bite forces to apprehend prey and/or process food with heavily worn or missing teeth. This hypothesis is supported by qualitative observations that sagittal crest size was largest in the oldest wild individuals (Fig. 9). An enlarged sagittal crest reflects greater development of the temporalis muscle (the primary jaw-closing muscle in carnivores). Production of higher bite forces is likely to subject the skull to higher strain and could induce bone deposition in the frontal region, resulting in a decrease in sinus volume. Sagittal crest size and tooth wear are also reported to be greater in older adult wild pumas (Puma concolor), but whether their sinuses are affected is unknown (Gay & Best, 1996). This pattern contrasts with our hypothesis that we should observe larger frontal sinuses in individuals with more mechanically challenging diets. However, one species of canid (the African wild dog, Lycaon pictus) has relatively small sinuses inside relatively thick frontal bones, and their skulls are well adapted to bring down large, struggling prey (Slater, Dumont & Van Valkenburgh, 2009; Curtis & Van Valkenburgh, 2014). Thus, there might be other ways in which sinus morphology can be remodelled to improve skull function. Given the limits of our sample size, it is possible that the apparent increase and decrease of frontal sinus volume with age might not be real. The initial upward slopes in the relationship between sinus volume and age or tooth wear, respectively, are driven by the youngest coyotes in our sample (between 6 months and 2 years age) and could be linked to ontogenetic changes rather than changes because of skull loading. Likewise, the downward slope in older individuals might be an artefact of over-fitting models to our relatively small dataset. A better understanding of the relationships among skull shape, frontal sinus size and muscle development would require a larger sample that also includes more landmarks along the sagittal and lamboidal crests. Figure 9. View largeDownload slide Differences in the development of the sagittal crest in (A) captive vs. (B) wild coyotes. Dorsal views (top), with red bracket denoting the length of the sagittal crest, and curve extending to the postorbital process delimiting the temporal line. Lateral views (bottom), with red line delimiting the base of the sagittal crest. Note the greater development of the sagittal crest in the wild coyote in both views. Figure 9. View largeDownload slide Differences in the development of the sagittal crest in (A) captive vs. (B) wild coyotes. Dorsal views (top), with red bracket denoting the length of the sagittal crest, and curve extending to the postorbital process delimiting the temporal line. Lateral views (bottom), with red line delimiting the base of the sagittal crest. Note the greater development of the sagittal crest in the wild coyote in both views. The significant correlation we observed between frontal sinus size and external cranial morphology in wild coyotes, but not captive-reared coyotes, suggests that the development and maintenance of internal and external skull morphology might be more integrated in wild individuals. This suggests that the forces experienced by wild coyote skulls could induce remodelling of frontal sinus morphology to improve mechanical efficiency. This could be tested by comparing patterns of stress and strain across the frontal region in captive and wild individuals using finite element methods. Overall, our study suggests that coyote skull and frontal sinus morphology is not greatly impacted by reduced skull loading. Given two coyote skulls of similar age, one from our captive sample and one from our wild sample, it would be difficult to determine their origin from their external or internal appearance, especially if they lacked any teeth. However, the greater mechanical demands of a wild diet, as evidenced by heavier tooth wear and tooth fracture, may have resulted in a stronger relationship between sinus form and external skull morphology in wild coyotes. In the captive sample, a softer diet and the consequent reductions in skull loading could have relaxed the relationship between internal and external cranial morphology. However, it is unlikely that these differences translate to drastic differences in biomechanical performance of the skull. It would be interesting to explore how skull and sinus morphology is affected under greater, rather than lesser, loading conditions. For example, some coyote populations frequently take prey much larger than lagomorphs, including white-tailed deer, which undoubtedly subject their skulls to much greater stress and strain (Lingle, 2002). The impact of mechanical loading on skull form might also be more pronounced in larger carnivore species with more specialized diets, such as spotted hyenas that crack open bones with their teeth, or lions that can bring down buffalo (Syncerus caffer) and even elephants (Loxodonta africana). Prior comparisons of the skulls of captive and wild large carnivorans reported more pronounced differences in cranial dimensions (e.g. Fitch & Fagan, 1982; O’Regan, 2001; O’Regan & Turner, 2004; Zuccareli, 2004; Hartstone-Rose et al., 2014), which suggests that skull loading plays a larger role in shaping skull form in these taxa. As a smaller-bodied generalist, wild coyotes probably do not subject their skulls to loads that vary in magnitude to the degree experienced by species that regularly hunt large, struggling prey or consume very hard foods, which might explain the smaller differences in cranial form we observed between captive and wild coyotes. This suggests that the effects of a non-wild diet lacking in bones or other tough foods probably has much greater negative impacts on craniodental morphology and function in large than small carnivores in captivity. This could also have important implications in wild individuals if their typical prey becomes scarce, as is believed to have occurred for many large carnivorans towards the end of the Pleistocene as large prey species went extinct (see review by Ripple & Van Valkenburgh, 2016). Might atypical skull development in large carnivorans occur if their large herbivorous prey becomes scarce and, consequentially, skull loading is reduced? This would be likely to make it more difficult for large carnivorans to switch successfully to smaller prey and compete with smaller carnivorans that are better adapted to subdue and eat smaller prey. Similar studies in a broader sample of taxa, especially among wild populations, would greatly improve our understanding of how and to what degree phenotype is shaped by environmental conditions in mammals and how this might affect feeding performance and, consequently, fitness. ACKNOWLEDGEMENTS The authors are grateful to L. Parker in the UCLA School of Dentistry for her help with scanning all skulls included in this study. We also thank L. Abraczinskas at Michigan State University, Lansing and J. Dines at the Los Angeles County Museum of Natural History for their help with specimens and associated data. We thank D. Bird, J. Wolf and J. Pajoli for insightful discussions and many helpful suggestions, and we thank the three anonymous referees and N. Warburton for their constructive feedback that greatly improved the quality of this manuscript. The authors declare no conflicts of interest. Funding was provided by National Science Foundation (NSF) IOB-0517748 to B.V.V. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Figure S1. Linear measurements taken from skulls. Descriptions and codes for linear measurements are given in Table 1. Table S1. (Attached file: Coyote_Skull_Raw_Data.xls). Raw data collected from wild (LACM) and captive (MSU) specimens included in this study. Specimen, specimen number. Museum: MSU, Michigan State University; LACM, Los Angeles County Museum of Natural History. Total tooth wear: ib (broken incisors), cb (broken canines), pb (broken premolars), cnb (broken carnassials), mb (broken molars), tb (total broken teeth), tt (total teeth), tw (total tooth wear stage), iw (incisors tooth wear stage), cw (canine wear stage), pw (premolar wear stage), cnw (carnassial wear stage), mw (molar wear stage), TSV (total frontal sinus volume, in mm3), SKCS (skull centroid size), FCS (frontal bone centroid size). Remaining columns represent raw data for linear measurements, followed by landmark coordinates (x1, y1, z1,…), and SPHARM coordinates (coordx1, coordyy1, coordz1, …). Codes for linear measurements are given in Table 1. Table S2. Descriptions of landmarks used for geometric morphometric analysis of skull shape. Table S3. Variance for traits in captive and wild coyotes and summary statistics from Levene’s tests testing for significant differences in variance between captive and wild coyotes. Codes for linear measurements are given in Table 1. REFERENCES Akaike H. 1973. Maximum likelihood identification of Gaussian autoregressive moving average models. Biometrika  60: 255– 265. Google Scholar CrossRef Search ADS   Anderson MJ. 2006. Distance-based tests for homogeneity of multivariate dispersions. Biometrics  62: 245– 253. Google Scholar CrossRef Search ADS PubMed  Anderson MJ, Ellingsen KE, McArdle BH. 2006. Multivariate dispersion as a measure of beta diversity. Ecology Letters  9: 683– 693. Google Scholar CrossRef Search ADS PubMed  Bartel RA, Knowlton FF. 2005. Functional feeding responses of coyotes, Canis latrans, to fluctuating prey abundance in the Curlew Valley, Utah, 1977–1993. Canadian Journal of Zoology  83: 569– 578. Google Scholar CrossRef Search ADS   Binder WJ, Van Valkenburgh B. 2000. Development of bite strength and feeding behavior in juvenile spotted hyneas (Crocuta crocuta). Journal of Zoology  252: 273– 283. Google Scholar CrossRef Search ADS   Blood BR, Matson JO, Patten DR. 1985. Multivariate analysis of allometry in a single population of coyotes (Canidae: Canis latrans Say). Australian Mammalogy  8: 221– 231. Cardini A. 2017. Left, right or both? estimating and improving accuracy of one-side-only geometric morphometric analyses of cranial variation. Journal of Zoological Systematics and Evolutionary Research  55: 1– 10. Google Scholar CrossRef Search ADS   Curtis AA, Lai G, Wei F, Van Valkenburgh B. 2015. Repeated loss of frontal sinuses in arctoid carnivorans. Journal of Morphology  276: 22– 32. Google Scholar CrossRef Search ADS PubMed  Curtis AA, Van Valkenburgh B. 2014. Beyond the sniffer: frontal sinuses in Carnivora. The Anatomical Record  297: 2047– 2064. Google Scholar CrossRef Search ADS PubMed  Cypher BL, Spencer KA, Scrivner JH. 1994. Food-item use by coyotes at the Naval Petroleum Reserves in California. Southwestern Naturalist  39: 91– 95. Google Scholar CrossRef Search ADS   Davis DD. 1964. The giant panda: a morphological study of evolutionary mechanisms. Fieldiana  3: 1– 339. Dumont ER, Dávalos LM, Goldberg A, Santana SE, Rex K, Voigt CC. 2012. Morphological innovation, diversification and invasion of a new adaptive zone. Proceedings of the Royal Society B: Biological Sciences  279: 1797– 1805. Google Scholar CrossRef Search ADS   Farke AA. 2007. Morphology, constraints, and scaling of frontal sinuses in the hartebeest, Alcelaphus buselaphus (Mammalia: Artiodactyla, Bovidae). Journal of Morphology  268: 243– 253. Google Scholar CrossRef Search ADS PubMed  Farke AA. 2008. Frontal sinuses and head-butting in goats: a finite element analysis. The Journal of Experimental Biology  211: 3085– 3094. Google Scholar CrossRef Search ADS PubMed  Farke AA. 2010. Evolution and functional morphology of the frontal sinuses in Bovidae (Mammalia: Artiodactyla), and implications for the evolution of cranial pneumaticity. Zoological Journal of the Linnean Society  159: 988– 1014. Google Scholar CrossRef Search ADS   Fatu C, Puisoru M, Rotaru M, Truta AM. 2006. Morphometric evaluation of the frontal sinus in relation to age. Annals of Anatomy = Anatomischer Anzeiger: official organ of the Anatomische Gesellschaft  188: 275– 280. Google Scholar CrossRef Search ADS PubMed  Fitch HM, Fagan DA. 1982. Focal palatine erosion associated with dental malocclusion in captive cheetahs. Zoo Biology  1: 295– 310. Google Scholar CrossRef Search ADS   Forbes-Harper JL, Crawford HM, Dundas SJ, Warburton NM, Adams PJ, Bateman PW, Calver MC, Fleming PA. 2017. Diet and bite force in red foxes: ontogenetic and sex differences in an invasive carnivore. Journal of Zoology  303: 54– 63. Google Scholar CrossRef Search ADS   Fox J, Weisberg S. 2011. An {R} companion to applied regression , 2nd edn. Thousand Oaks: Sage. Frontier S. 1976. Étude de la decroissance des valeurs propres dans une analyze en composantes principals: comparison avec le modèle de baton brisé. Journal of Experimental Marine Biology and Ecology  25: 67– 75. Google Scholar CrossRef Search ADS   Gay SW, Best TL. 1996. Age-related variation in skulls of the puma (Puma concolor). Journal of Mammalogy  77: 191– 198. Google Scholar CrossRef Search ADS   Hartstone-Rose A, Selvey H, Villari JR, Atwell M, Schmidt T. 2014. The three-dimensional morphological effects of captivity. PLoS ONE  9: e111347. PubMed  Herring SW. 1993. Formation of the vertebrate face: epigenetic and functional influences. American Zoologist  33: 472– 483. Google Scholar CrossRef Search ADS   Hollister N. 1917. Some effects of environment and habit on captive lions. Proceeding of the United States Natural History Museum  53: 177– 193. Google Scholar CrossRef Search ADS   Huxley TH. 1880. On the cranial and dental characters of the Canidae. Journal of Zoology  48: 238– 288. Jackson DA. 1993. Stopping rules in principal components analysis: a comparison of heuristical and statistical approaches. Ecology  74: 2204– 2214. Google Scholar CrossRef Search ADS   Joeckel RM. 1998. Unique frontal sinuses in fossil and living Hyaenidae (Mammalia, Carnivora): description and interpretation. Journal of Vertebrate Paleontology  18: 627– 637. Google Scholar CrossRef Search ADS   Johnson MK. 1978. Food habits of coyotes in south central Idaho . Unpublished octoral dissertation, Colorado State University. Klingenberg CP. 2011. MorphoJ: an integrated software package for geometric morphometrics. Molecular Ecology Resources  11: 353– 357. Google Scholar CrossRef Search ADS PubMed  Klingenberg CP, Monteiro LR. 2005. Distances and directions in multidimensional shape spaces: implications for morphometric applications. Systematic Biology  54: 678– 688. Google Scholar CrossRef Search ADS PubMed  Klugh DO. 2010. Principles of equine dentistry . London: Manson Publishing. Google Scholar CrossRef Search ADS   La Croix S, Holekamp KE, Shivik JA, Lundrigan BL, Zelditch ML. 2011a. Ontogenetic relationships between cranium and mandible in coyotes and hyenas. Journal of Morphology  272: 662– 674. Google Scholar CrossRef Search ADS   La Croix S, Zelditch ML, Shivik JA, Lundrigan BL, Holekamp KE. 2011b. Ontogeny of feeding performance and biomechanics in coyotes. Journal of Zoology  285: 301– 315. Google Scholar CrossRef Search ADS   Levene H. 1960. Robust tests for equality of variances. In: Olkin I, Ghurye SG, Hoeffding W, Madow WG, Mann HB, eds. Contributions to probability and statistics . Stanford: Stanford Universtiy Press, 278– 292. Lingle S. 2002. Coyote predation and habitat segregation of white-tailed deer and mule deer. Ecology  87: 2037– 2048. Google Scholar CrossRef Search ADS   Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E, Wagner H. 2017. Vegan: community ecology package, R package version 2.4–4 . Available at: https://cran.r-project.org/web/packages/vegan (last accessed 15 January 2018). O’Regan HJ. 2001. Morphological effects of captivity in big cat skulls. In: Wehnelt S, Hudson C, eds. Proceedings of the 3rd Zoo research symposium . Chester: Chester Zoo, 18– 22. O’Regan HJ, Kitchener AC. 2005. The effects of captivity on the morphology of captive, domesticated and feral mammals. Mammal Review  35: 215– 230. Google Scholar CrossRef Search ADS   O’Regan HJ, Turner A. 2004. The interface between conservation biology, palaeontology and archaeozoology: morphometrics and population viability analysis. In: Lauwerier RCGM, Plug I, eds. The future from the past: archaeozoology in wildlife conservation and heritage management . Oxford: Oxbow Books, 90– 96. Paulli S 1900. Über die Pneumaticität bei den Säugerthieren. Ein morphologische Studie. III. Über die Morphologie des Siebbens und die der Pneumaticität bie den Insectivoren, Hyracoidean, Chiropteran, Carnivoren, Pinnipedien, Edentaten, Rodentiern, Prosimiern und Primaten. Morphologisches Jahrbuch  28: 483– 564. R Core Team 2017. R: A language and environment for statistical computing . Vienna: R Foundation for Statistical Computing. Ripple WJ, Van Valkenburgh B. 2016. Linking top-down forces to the Pleistocene megafaunal extictions. BioScience  60: 516– 526. Google Scholar CrossRef Search ADS   Samuels JX. 2009. Cranial morphology and dietary habits of rodents. Zoological Journal of the Linnean Society  156: 864– 888. Google Scholar CrossRef Search ADS   Shen L, Farid H, McPeek MA. 2009. Modeling three-dimensional morphological structures using spherical harmonics. Evolution  63: 1003– 1016. Google Scholar CrossRef Search ADS PubMed  Shen L, Makedon F. 2006. Spherical mapping for processing of 3D closed surfaces. Image and Vision Computing  24: 743– 761. Google Scholar CrossRef Search ADS   Slater GJ, Dumont ER, Van Valkenburgh B. 2009. Implications of predatory specialization for cranial form and function in canids. Journal of Zoology  278: 181– 188. Google Scholar CrossRef Search ADS   Smith RJ. 2009. Use and misuse of the reduced major axis for line-fitting. American Journal of Physical Anthropology  140: 476– 486. Google Scholar CrossRef Search ADS PubMed  Tanner JB, Dumont ER, Sakai ST, Lundrigan BL, Holekamp KE. 2008. Of arcs and vaults: the biomechanics of bone-cracking in spotted hyenas (Crocuta crocuta). Biological Journal of the Linnean Society . 95: 246– 255. Google Scholar CrossRef Search ADS   Tseng ZJ, Wang X. 2010. Cranial functional morphology of fossil dogs and adaptations for durophagy in Borophagus and Epicyon (Carnivora, Mammalia). Journal of Morphology  271: 1386– 1398. Google Scholar CrossRef Search ADS PubMed  Tseng ZJ, Wang X. 2011. Do convergent ecomorphs evolve through convergent morphological pathways? cranial shape evolution in fossil hyaenids and borophagine canids (Carnivora, Mammalia). Paleobiology  37: 470– 489. Google Scholar CrossRef Search ADS   Van Valkenburgh B. 1988. Incidence of tooth breakage among large, predatory mammals. The American Naturalist  131: 291– 302. Google Scholar CrossRef Search ADS   Van Valkenburgh B. 1999. Major patterns in the history of carnivorous mammals. Annual Review of Earth and Planetary Sciences  27: 463– 493. Google Scholar CrossRef Search ADS   Van Valkenburgh B. 2009. Costs of carnivory: tooth fracture in Pleistocene and Recent carnivores. Biological Journal of the Linnean Society  96: 68– 81. Google Scholar CrossRef Search ADS   Van Valkenburgh B, Koepfli K-P. 1993. Cranial and dental adaptations to predation in canids. Symposium of the Zoological Society of London  65: 15– 37. Wang X, Tedford RH, Taylor BE. 1999. Phylogenetic systematics of the Borophaginae (Carnivora, Canidae). Bulletin of the American Museum of Natural History  243: 1– 391. Warton DI, Duursma RA, Falster DS, Taskinen S. 2012. Smatr 3- an R package for estimation and inference about allometric lines. Methods in Ecology and Evolution  3: 257– 259. Google Scholar CrossRef Search ADS   Warton DI, Wright IJ, Falster DS, Westoby M. 2006. Bivariate line-fitting methods for allometry. Biological reviews of the Cambridge Philosophical Society  81: 259– 291. Google Scholar CrossRef Search ADS PubMed  West-Eberhard MJ. 2003. Developmental plasticity and evolution . New York: Oxford University Press. Witmer LM. 1997. The evolution of the antorbital cavity of archosaurs: a study in soft tissue reconstruction in the fossil record with an analysis of the function of pneumaticity. Journal of Vertebrate Paleontology  17: 1– 77. Google Scholar CrossRef Search ADS   Witmer LM. 1999. The phylogenetic history of paranasal air sinuses. In: Koppe R, Nagai H, Alt KW, eds. The paranasal sinuses of higher primates: development, function, and evolution . Chicago: Quintesence, 21– 34. Wolff JD. 1892. Das Gesetz der Transformation der Knochen . Berlin: A. Hirschwald. Zuccarelli MD. 2004. Comparative morphometric analysis of captive vs. wild African lion (Panthera leo) skulls. Bios  75: 131– 138. Google Scholar CrossRef Search ADS   © 2018 The Linnean Society of London, Biological Journal of the Linnean Society

Journal

Biological Journal of the Linnean SocietyOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial