The helmeted hornbill casque is reinforced by a bundle of exceptionally thick, rod‐like trabeculae

Abstract

Among hornbill birds, the critically endangered helmeted hornbill (Rhinoplax vigil) is notable for its casque (a bulbous beak protrusion) being filled with trabeculae and fronted by a very thick keratin layer. Casque function is debated but appears central to aerial jousting, where birds (typically males) collide casques at high speeds in a mid‐flight display that is audible for more than 100 m. We characterized the structural relationship between the skull and casque anatomy using X‐ray microtomography and quantitative trabecular network analysis to examine how the casque sustains extreme impact. The casque comprises a keratin veneer (rhamphotheca, ∼8× thicker than beak keratin), which slots over the internal bony casque like a tight‐fitting sheath. The bony casque's central cavity contains a network of trabeculae—heavily aligned and predominantly rod‐like, among the thickest described in vertebrates—forming a massive rostrocaudal strut spanning the casque's length, bridging rostral (impact), and caudal (braincase) surfaces. Quantitative network characterizations indicate no differences between male and female trabecular architectures. This suggests that females may also joust or that casques play other roles. Our results argue that the casque's impact loading demands and shapes a high‐safety‐factor construction that involves extreme trabecular morphologies among vertebrates, architectures that also have the potential for informing the design of collision‐resistant materials.

Critically endangered helmeted hornbills (Rhinoplax vigil) use their casques for aerial jousting, where birds collide mid‐air. X‐ray microtomography revealed that the casque consists of a thick keratin veneer and a bony core with unusually large, rod‐like trabeculae forming a massive strut that sustains impact. The casque's impressive trabeculae, with similar thickness to those of Asian elephant femurs, offer insights into the natural history of this elusive species and inspiration for collision‐resistant materials.

INTRODUCTION

Impact‐resistant materials are in high demand for diverse applications, from wearable protection devices, sports, or combat armor to structures and materials for naval and rail transportation. In real‐world loading scenarios, impacts and the resultant material responses (e.g., from collisions of two objects) do not just take place under static loads but rather often result from loads that are dynamic, and the stresses, strains, and deformations vary with time in complex ways. ¹ In the development of new impact‐resistant engineering materials, nature can act as a valuable inspiration where tissues have evolved and been refined over millennia to manage the biomechanics of life and activity. ¹ , ² Indeed, animals exhibit a huge range of architectural materials associated with managing the potential damage from biological impacts associated with the noted struggles between predator and prey, but also in intraspecies interactions. ³ , ⁴ Combat associated with mating‐ or territory‐related displays can be especially violent. Rams, musk oxen, and others engage in high‐velocity head‐butting collisions at speeds of ∼20–50 km/h ³ , ⁴ —impact velocities that are akin to traffic accidents and potentially devastating for traumatic brain and skull injury. ⁵

Our study focuses on an unexplored biological model for impact resistance. The helmeted hornbill (Rhinoplax vigil) is a large species of hornbill bird living across primary rainforests in Southeast Asia. As in many of the 62 hornbill species, the upper jaw of R. vigil bears a prominent casque—a vaulted oblong structure dorsal and anterior to the eyes—formed from an outer keratin sheath (the rhamphotheca) and internal bony scaffolding (Figure 1). ⁶ The casques of most hornbill species are largely hollow with very thin keratin shells and only sparse struts of bone (trabeculae) in regions near the eyes (e.g., Figure 1E). ⁷ , ⁸ , ⁹ In contrast, the helmeted hornbill's casque comprises ∼10% of the bird's body weight. Not only does the bony portion of the casque extend far rostrally (unlike those of other hornbills), matching the shape of its keratin sheath, but it is also nearly filled with a structurally complex network of dense trabecular bone. ⁶ , ⁹ Moreover, the rostral portion of the keratin sheath is exceptionally thick (∼25 mm), forming a layered slab fronting the cranial vault. ¹⁰ , ¹¹ , ¹² The yellow‐orange front of the rhamphotheca has historically been a target for collectors and sellers of arts and antiques for carving into ornamental wildlife products. ¹³ Poaching to supply this market has threatened wild populations ¹⁰ and driven R. vigil to Critically Endangered status on the IUCN Red List of Threatened Species. ¹⁴ Conservation efforts, however, are hampered by a lack of data on the natural history of this species, which is vital for understanding population viability and guiding management.

FIGURE 1.

Anatomy of the helmeted hornbill casque. (A) The helmeted hornbill (Rhinoplax vigil) carries a characteristic huge vaulted casque on its upper jaw, anterior to the orbit. (B) The casque consists of an outer keratin sheath (rhamphotheca), overlying a bony portion (see Figure 2). The rhamphotheca features an extremely thick yellow slab (∼25 mm) at the rostral end of the casque and is used as an impact surface during aerial jousting, a behavior primarily observed in males. Caudal from the impact surface, the rhamphotheca forms a thin red cladding (∼1–2 mm) laterally and a layer of intermediate thickness dorsally. Long notched lateral ridges (∼5 mm thick) run rostrocaudally down the casque on both sides, terminating near the upper orbit. (C) A semi‐transparent volume rendering of an X‐ray microtomography (XMT) scan reveals the dense bundle of trabecular bone within the bony casque, extending through the casque to the braincase. (D) The thick keratin impact surface of the helmeted hornbill rhamphotheca has long been prized for carving (as in this specimen, shown in rostral view); associated illegal poaching has contributing to the species’ decline and its current Critically Endangered status. (E) In contrast to the heavy reinforcement of the helmeted hornbill casque, most other hornbills have largely hollow casques with thin keratin coverings and only scant bony trabeculae (as shown in a volume rendering of an XMT scan of Bycanistes brevis). Scale bar = 1 cm in (B–E). Source: Species image in A courtesy of Justin Grubb/Planet Indonesia.

The helmeted hornbill's casque has been proposed to function in attracting mates at sexual maturity ¹⁵ and as a tool for feeding, digging, and nest‐building. ¹⁶ However, it is its apparent role as a weapon or shield that is most striking. Helmeted hornbills perform an impressive combat display—a high‐impact aerial jousting behavior rarely observed and scantily described—where individuals collide mid‐air at speeds up to ∼36 km/h (10 m/s) after a 50 m approach flight. The collisions are dynamic and repeated, often performed continuously over a 1‐h period, generating sounds audible from at least 100 m away. ¹⁶ , ¹⁷ Helmeted hornbill jousting likely represents one of the fastest known biological impacts, rivaling collision speeds of a galloping horse's hoof and pecking woodpecker's beak. ¹ , ¹⁸ However, although the small size of woodpeckers minimizes dangers of brain or cranial damage, the helmeted hornbill weighs up to 3 kg with a 1–1.5 m wingspan, putting considerable mass behind its high‐speed impact. Therefore, as helmeted hornbills do not shed and regrow their casques (like the antlers of some head‐butting mammals ¹⁹ , ²⁰ ), the bone–keratin composite likely possesses anatomical adaptations to sustain repetitive high impact loads without permanent damage to the skeleton or brain. ³ , ⁴

Despite the casque being a dominant feature of the helmeted hornbill, its anatomy and mechanical performance have barely been explored. ²¹ Manger Cats‐Kuenen ⁹ provided an impressively detailed description of hornbill casque anatomy, naming the cranial structures and describing their general arrangements, but only included one cross‐sectional image of the described features and the internal trabeculation, thus challenging efforts to understand complex tissue relationships. Moreover, characterization of the three‐dimensional arrangements of the bony trabeculae (similar to a building's structural girders) is key to understanding casque mechanical function. To date, neither material nor mechanical analyses of hornbill casques have been performed. The combination of bone and keratin has evolved multiple times in animals where impact resistance is vital (e.g., horse's hooves, turtle shells, and bovid horns), and the helmeted hornbill casque surely capitalizes on the combined fracture resistance of keratin and stiffness of bone. ²² However, as previous calculations indicate that hornbill bone and keratin have material densities (specific gravity) similar to those of other animals, ⁹ this suggests that it is the particular architectural arrangements of these materials in the casque that impart exceptional damage tolerance against high‐impact repetitive dynamic loading.

Helmeted hornbill jousting behavior has been observed mainly in males, yet one record exists of a male jousting a perched female. ¹⁷ Although smaller in females than males, the casque's external morphology is similar in both sexes, ²³ suggesting that either females also engage in combat or that the casque has multiple functions. In this study, we provide an updated and quantitative description of helmeted hornbill casque anatomy and compare male and female casques using X‐ray microtomography (XMT) imaging. This allowed for a refined high‐resolution and three‐dimensional structural characterization of the helmeted hornbill casque trabeculae bone network, thereby facilitating potential identification of anatomical features with sexual dimorphism and/or relation to impact resistance. We hypothesized that (1) as both males and females likely perform aerial jousting behavior and have similar casques externally, the casques of both sexes would have dense casque trabecular networks, and (2) as male helmeted hornbills have been observed performing aerial jousting more than females, the bony casque of males will have thicker trabeculae and a greater bone volumetric density than that of females. A better understanding of the casque's internal structure and whether it varies between sexes will help clarify the casque's role in this elusive species’ ecology, while potentially offering a valuable conservation tool for discriminating sexes from wildlife products/parts (e.g., via plain radiograph/X‐ray imaging) when external gross morphology and genetic analyses are not available.

MATERIALS AND METHODS

Specimen collection

To analyze differences between male and female helmeted hornbill casques (including skull and rhamphotheca), we selected five male and five female specimens for analysis. Skulls were severed from the rest of the birds’ bodies pre‐seizure, typically at the occipital region of the braincase. Hatten et al. ²³ previously demonstrated that the sex of specimens could be identified by genetic techniques or morphology, with females having blue skin and black beak markings and males having red skin and no black beak markings. ²³ For the current study, we examined four male and four female casques that had been confiscated by the Agriculture, Fisheries, and Conservation Department (AFCD) of the Hong Kong SAR government in 2013 and donated to City University of Hong Kong for this research (AFCD ref: L/M 608/2022 in AF GR CON 07/13). Additional casques from one male (ZMB 2000/4555, Museum für Naturkunde, Berlin) and one female (private collection; Melbourne, Australia) were also examined.

Data acquisition and analysis

XMT scans were performed of all specimens to examine and characterize internal morphology using scanners available local to where specimens were housed. Hong Kong specimens were scanned with a Comet Yxlon FF35 CT scanner at Hong Kong Polytechnic University (80 kV voltage, 60 µA current, 0.5 mm Al filter, 0.33 s exposure, 360° rotation, and 1 projection per angular step). Reconstruction was conducted by CERA (version 2206.4.0) using a filtered back projection algorithm and beam hardening correction with the primary material set to bone. The reconstruction produced a stack of 3600 images with isotropic pixel spacing of 51.7 µm in all specimens but one, which was scanned at 140 kV (rather than 80 kV), producing 2860 images with isotropic spacing of 74.9 µm. This sampling density was determined from preliminary scans to be sufficient for resolving the structure and dimensions of trabeculae, while allowing imaging of the entire trabecular network. The two scans at other locations were performed with settings to result in similar reconstructed resolutions.

The Berlin specimen was scanned with a Comet Yxlon FF20 CT scanner at the Museum für Naturkunde, Berlin (95 kV voltage, 45 µA current, 0.5 s exposure, 1 mm Al filter, 360° rotation, and 1 projection per angular step). Reconstruction was performed in a similar manner to the Hong Kong scans, producing a stack of 1800 images with voxel resolution of 45.2 µm. The Melbourne specimen was scanned with a phoenix nanotom m (Waygate Technologies) operated using XS control and the phoenix datos|x acquisition software at the TrACEES platform of the University of Melbourne (120 kV voltage, 300 µA current, 0.5 s exposure, 0.1 mm Cu filter and tungsten target, 360° rotation, and 1 projection per angular step). Two 15‐min scans at different vertical positions were conducted in a multiscan mode to capture the full length of the specimen with 54.9 µm isotropic pixel spacing. Reconstruction of the XMT data was performed using phoenix datos|x reconstruction software applying inline median and region of interest (ROI) filters during reconstruction.

For general comparison of casque gross morphology, skulls of two related species were also investigated. An intact, dried head of a black‐and‐white‐casqued hornbill (Bucerotidae: Bycanistes subcylindricus) was photographed in the zoology collection, University of Montpellier (UM‐ZOOL‐609VO; housed in the Institut de Botanique, Montpellier), and scanned at the Montpellier MRI imaging platform (84.1 µm voxel resolution). Additionally, a skeletonized skull of a great hornbill (Bucerotidae: Buceros bicornis), with rhamphotheca and braincase separated, was photographed in the collection of the Zoological Institute at Kiel University (ZIK‐21533).

Image slices from all helmeted hornbill scans were imported into Dragonfly software. ²⁴ As our analysis focused on the bony trabecular network, an attention U‐Net‐based model was trained for segmentation of the rhamphotheca relative to bone using the machine learning toolkit in Dragonfly. This outputted a mask of the rhamphotheca. Cranial anatomy was examined and rendered in 2D and 3D using Amira software. ²⁵ The rhamphotheca mask generated by the Dragonfly segmentation was then imported into Amira as a label field and used to remove the rhamphotheca, with background (non‐bone) pixels removed using the interactive thresholding module. The resultant bone‐only volume was rotated in Amira to a standard anatomical orientation, then cropped according to anatomical landmarks into two separate volumes of interest (VOIs): a larger volume (VOI1) including the majority of the bony portion of the casque (i.e., the pre‐cranial vault, except the upper jaw) and a smaller subvolume (VOI2) comprising the densest zone of trabeculation (i.e., occupying the center of the pre‐cranial vault space; see below and Figure S1). The binary label fields (bone vs. background) from both VOIs were exported as image stacks for quantitative analysis.

Quantitative and statistical analyses

To assess and quantify trabecular bone thickness (Tb.Th), volume fraction (BV/TV), connectivity density (Conn.D), ellipsoid factor (EF), ²⁶ and degree of anisotropy (DA) measurements were performed on the segmented volumes for casques in (Fiji Is Just) ImageJ (2.14.0/1.54f, Schneider et al.), ²⁷ with BoneJ (v 7.0.17). ²⁸ For all variables, a single parameter value per individual was obtained for inter‐individual comparison. To further understand the local differences in the morphology of trabeculae across specimens, intensity maps of Tb.Th and EF were extracted for all slices in the transverse and frontal planes from the segmented volumes using BoneJ (v 7.0.17). ²⁸ Histograms were then calculated from three slices roughly equidistantly spaced in the dorsal–ventral and rostral–caudal directions in all 10 specimens. Tb.Th and EF histograms were plotted with ranges from 0 to 4.0 mm and from −1.0 to +1.0, respectively, divided into 200 bins. Plotting was performed using Python software (version 3.12). The pandas module was used for data cleaning and filtering. Scatterplots (bin vs. count) and violin plots of Tb.Th and EF were generated for each of the segmented volumes using matplotlib and seaborn modules, to compare the distributions of Tb.Th and EF among slices and individuals, and between sexes.

We used R version 4.4.1 ²⁹ to statistically test the effect of sex on the different trabecular parameters. To compare BV/TV, Conn.D, and DA between sexes, we conducted nonparametric Mann–Whitney U tests using the wilcox.test function. Tb.Th and EF were compared using the binned histogram data. The frequency n of each bin was transformed into n data points of the same value, using the lower limit of each bin interval. The trabecular data within an individual are expected to be more similar on average than the data between individuals, thus representing pseudo‐replicated (i.e., non‐independent) data points in inferential statistical analyses. The same can be assumed for the data within a specific slice (i.e., casque region) as compared to other slices of an individual. Hence, we used a mixed effects model design. Sex was defined as a fixed effect, with individual number and slice number as random effects, and “slice” being nested within “individual”. For computational reasons, 1000 samples were drawn at random from the more than 100,000 values contained in each image slice (six slices per individual = 60 XMT slices sampled overall).

The violin plots of Tb.Th measurements were distributed exponentially (see “Results” section), which we accommodated by using a generalized linear model with a Gamma distribution and a log link function. We used the glmer function of the R package lme4 ³⁰ to construct the model. The distribution of the EF measurements resembled a mirrored exponential distribution. Moreover, as this variable is bounded by values −1 and 1, we considered a generalized linear model with a beta distribution and a logit link function as the most sensible model for this data distribution. However, as the beta distribution is only defined for values between 0 and 1, we transformed the raw data by multiplying all values by 0.5 and then adding 0.5. The glmmTMB function from the glmmTMB package ³¹ was used to build this model.

To test the significance of both random and fixed effects, we used the model‐building procedure recommended by Zuur et al. ³² First, we established the full model including all effects. From this, we removed the effects one at a time and compared the more complex model with the simpler model using a likelihood ratio test (LRT). The p‐values were used as a guideline for model‐building without using a fixed significance level. For both trabecular parameters, the same procedure was used. First, the random effects were removed by removing “slice” but keeping “individual”, and then removing also “individual”. As the glmer function cannot be defined without random effects, we used the glm function. As both random effects (“individual” and “slice”) were significant, we kept them and removed sex from the model to test its significance against the full model (see “Results” section).

RESULTS

For the purpose of this study, we refer to the casque as the composite structure comprised of an outer keratin sheath (rhamphotheca) and the underlying contribution from the skull (bony casque), rostral to the craniofacial joint (at the rostral face of the braincase, just rostral to the orbit) and excluding the rostral upper jaw (Figures 1 and 2). The casque is a tall arch extruded in the rostrocaudal direction, like an architectural barrel vault, with a sloping rostral face. Although jousting incidents are fleeting and observations of them extremely rare, anecdotal accounts in the literature ⁶ , ¹⁷ , ³³ and a video of a single event we have seen suggest individuals collide head‐on; we therefore refer to the rostral surface of the casque as the impact surface (Figure 1B). All helmeted hornbill specimens we examined had casques with this gross morphology. Our study focuses on the structural mechanics of the skull relating to the jousting behavior of this species. Our anatomical description therefore centers in particular on the features of the bony casque associated with the region of impact, with brief descriptions of relevant aspects of the rhamphotheca. Anatomical terminologies used in our descriptions (e.g., slice orientations) are illustrated in Figure S1. For a more detailed description of the cranial anatomy, see Manger Cats‐Kuenen. ⁹

FIGURE 2.

The inner architectures of the helmeted hornbill casque. (A) Unlike other hornbills (e.g., Buceros, in the inset image), the shape of the internal bony portion of the helmeted hornbill's casque matches that of the rhamphotheca, resulting in an intimate structural fit with bone and keratin associated on all casque surfaces. (B) Sections of volume renderings of X‐ray microtomography (XMT) scans demonstrate the variation in rhamphotheca thickness in different casque regions and show the trabecular bundle's location within the bony casque. Although the beak houses only sparse bony trabecular bone, the casque's trabeculae are localized in a central bundle, with large struts aligned between the rostral face and the craniofacial joint. Note the extremely thin bone cortex at the interface of rhamphotheca and bony casque. (C) When slotted together, the caudal face of the rhamphotheca (left) is apposed to the rostral face of the bony casque (right). Although the caudal face of the rhamphotheca and the lateral faces of the bony casque are smooth, (D) the bony casque's rostral face contains a network of pits and holes of various sizes surrounded by circuitous furrows. The patterns formed by the furrows correspond to impressions in the associated rhamphotheca face (inset image, horizontally reflected) and the holes are continuous with the spaces between trabeculae within the bony casque's cavity (see also Figure 3C).

The impact surface of the rhamphotheca is visibly and structurally distinct. The rhamphotheca is thickest (∼25 mm) directly caudal to the impact surface and bright yellow in color, continuous with and a similar color as the portion of the rhamphotheca that covers the bill (Figures 1A,B,D and 2A,B). On the sides of the casque, the rhamphotheca tapers to a thin veneer (∼1–2 mm), deep red in color (Figures 1A,B and 2A), and long notched ridges run caudally down the casque at approximately the height of the upper orbit (Figures 1B,C and 2A,B). Our XMT data verify that the ridges (∼5 mm thick) are almost entirely composed of keratin, with only shallow corresponding elevations on the bony casque contributing to the ridges' overall structure (Figure 3). The rhamphotheca exhibited intermediate thicknesses on the dorsal side of the casque and ensheathing the bill (Figure 3B,C). Overall, the rhamphotheca's gross morphology matched that of the bony casque, slotting over the top of it like a sword in its sheath (Figure 2). In two of our specimens, the casque could be smoothly removed as a single intact element by loosening it at its caudal edge (e.g., Figure 2A); however, in most of our dried specimens, the rhamphotheca had adhered to the bone and could not be removed.

FIGURE 3.

Cross‐sectional morphology of the trabecular bone bundle from multiple orientations. (A–C) Color inset images in the upper left of each panel show slice location, with gray arrows in A–B indicating slice order; in all slices, the casque and rhamphotheca can be distinguished by the narrow space between them and their different radio‐opacities. For visual descriptions of anatomical terminologies used (e.g., slice orientations), see Figure S1. (A) Transverse slices show the enormous central strut spanning the casque's length, bridging the rostral impact face (apposing the rhamphotheca) and the craniofacial joint, which borders the braincase. Thinner and shorter trabeculae at the rostral and caudal ends of the bundle converge on the largest trabecular struts in the center of the casque, which pass between and around the nasal passages (indicated by *). Note that in this specimen, the impact surface of the rhamphotheca has been carved (see Figure 1D), disrupting what should be a smoother natural profile in the section. (B) Three frontal slices show the clustering of the most robust trabecular bone in the center of the casque, with the nasal passages (*) traversing the bundle. The middle slice exhibits centrally‐clustered large trabeculae and their links to sparser trabecular bone at the casque surface, while the rostral most section displays part of the distinct porous trabeculation swath that apposes the rhamphotheca rostrally (see Figure 2C,D). Note also the tight association of rhamphotheca and bony casque on all surfaces; the elevated and largely keratinous lateral ridge; and the thin cortex of the bony casque. (C) A midsagittal slice illustrates the rostrocaudal continuity of the central bundle, spanning the impact surface and the craniofacial joint (between the two arrows). The inset image illustrates the hierarchical structure of the bundle, where smaller trabeculae merge into larger trabeculae more caudally. The gaps in the rostral surface of the bony casque correspond to the pores in Figure 2D.

In the two specimens where the rhamphotheca could easily be detached, the bony casque also showed disparate morphologies on its lateral and impact faces. Although the lateral walls were relatively smooth (Figure 2A), the rostral impact face was pockmarked with a network of pits and holes of various sizes (∼0.5–1.0 mm), encircled by circuitous furrows (Figure 2C,D). This texturing continued a short distance onto the lateral faces of the bony casque (Figure 2A), and also left matching mirror‐image impressions on the corresponding interior (caudal) face of the rhamphotheca's impact zone (Figure 2D inset). In virtual sections, we verified Manger Cats‐Kuenen's ⁹ descriptions that the cortex was extremely thin on all faces of the bony casque and enclosed an interior space with two dominant populations of bony trabeculae (Figures 2B and 3). Most prominently, a forest of extremely thick and highly oriented trabeculae (see below) was observed bridging the impact face and the bony casque's contribution to the craniofacial joint (Figures 2B and 3). This dense bundle of trabeculae occupies the upper central portion of the casque, like a fixed architectural beam running caudoventrally from the impact plate. Surrounding the group of dense trabeculae was a second population of threadlike trabeculae, which were considerably thinner (∼0.3 mm), arranged far less densely, and in diverse orientations (Figure 3). These sparse trabeculae occupy nearly all spaces surrounding the dense trabecular group, including the interior of the upper and lower jaws (Figure 3B,C). The density of sparse trabeculae varied locally. For instance, the comparatively low density immediately dorsal and rostroventral to the dense trabecular stand resulted in near‐empty voids in XMT (Figure 3C). Given the obvious association with the impact face and the likely impact direction, we focus the remainder of our analyses on the dense trabeculae.

We provide a quantitative and synthesized description of trabeculae populations below, but several general qualitative features also bear mentioning. In virtual sections of the impact face, the cortex of the bony casque cannot be clearly distinguished from the trabeculae; rather, the fanning rostral ends of the trabeculae appear to merge to form the bony casque's rostral surface (Figure 3A,C, and inset). The impact plate relative to the associated dense trabeculae, therefore, is like the canopy relative to the trunks of a group of trees. From XMT sections, we verified that the holes on the casque's impact face (Figure 2D) communicate with the internal spaces between trabeculae and not with spaces inside the trabeculae themselves, which are apparently solid rods (Figure 3C). Immediately caudal to the impact surface, there is first a distinct narrow swath of trabeculation (∼1.8 mm thick), with short and comparatively narrow trabecular branches (Figure 3A,C). Caudally, these shorter trabeculae merge into fewer larger trunks that extend the majority of the casque, with thinner bridging trabeculae linking adjacent trunks obliquely. Toward the caudal end of the casque, the trabeculae diverge around and between the internal paired nasal passages. These passages run parasagittally inside the casque immediately rostral to the orbit, as near‐vertical tubes that then exit laterally dorsal to the eye and the casque's lateral ridge (Figure 3). After passing around and between the nasal passages inside the casque, the dense trabeculae terminate in a broad zone at the caudal cortical wall of the casque (Figures 2B and 3A,C). Unlike the porous rostral end of the bony casque, the caudal end has a true (albeit thin) cortex, visible in virtual sections (Figure 3A,C).

Quantitative comparisons of male versus female anatomy

Despite superficial differences previously noted in the appearances of males and females, ²³ internal cranial skeletal anatomy did not differ noticeably between sexes in our dataset (Figure 4). This was also reflected in the quantitative statistical comparisons (e.g., from both VOI1 and VOI2; Figures S2–S4, Table 1). The Wilcoxon statistic of the Mann–Whitney U tests conducted on BV/TV, Conn.D, and DA was never significant (VOI2; p = 0.68, p = 0.53, and p = 1, respectively, Table S1). Moreover, sex contributed negligibly to the observed variation in comparisons made among individual slices from roughly similar anatomical positions across all specimens (Figures S3 and S4). The final mixed effects models that we constructed with Tb.Th and EF included “individual” as well as “slice” nested within “individual” as random effects, as the probability that the variability on these levels in our sample was observed by chance was very low (i.e., the p‐values of the LRT were always far below 0.001, Table S2). In these final models, Tb.Th and EF differed on average only negligibly between sexes, the p‐value of the LRT being 0.67 for the former, and 0.93 for the latter (Table S2).

FIGURE 4.

Lack of effect of sex on helmeted hornbill trabecular thickness (Tb.Th; A and B, E and F) and ellipsoid factor (ellipsoid factor [EF]; C and D, G and H). Scatterplots (B, C, F, G) show pooled data from all individuals for each sex (n = 3 slices per individual), summarized by overlain violin plots. Note the negligible variation between sexes regardless of slice orientation. Exemplar transverse (A and D) and frontal (E and H) sections are colored according to trabecular thickness (Tb.Th; A and E) and ellipsoid factor (EF; D and H). In the transverse section from a male individual (A and D) Tb.Th ranges from 0.14 to 2.33 mm and EF from −0.88 to 0.92 (where dark blue = minimum and yellow = maximum). Similarly, in the frontal section (E and H), Tb.Th ranges from 0.14 to 1.92 mm and EF from −0.81 to 0.91.

TABLE 1.

Quantitative measurements of trabecular parameters.

Animal [CityU HK specimen reference number]	Sex	Mean trabecular thickness (Tb.Th, VOI1)	Volume fraction (BV/TV, VOI2)	Connectivity (ConnD, VOI2)	Median ellipsoid factor (EF, VOI1)	Anisotropy (DA, VOI2)
Hong Kong 1 [HH11]	Female	0.90666	0.3562	0.1703	0.2597	0.6364
Hong Kong 2 [HH15]	Female	1.1764	0.3331	0.1551	0.3998	0.4903
Hong Kong 3 [HH16]	Female	0.6665	0.2852	0.1894	0.2178	0.6665
Hong Kong 4 [HH21]	Female	0.7070	0.3135	0.2220	0.2832	0.6448
Melbourne	Female	0.8186	0.2976	0.1878	0.3417	0.6403
Hong Kong 5 [HH13]	Male	0.8887	0.4656	0.3761	0.2939	0.5647
Hong Kong 6 [HH14]	Male	0.9581	0.3022	0.0966	0.1919	0.6539
Hong Kong 7 [HH24]	Male	0.9295	0.3617	0.0876	0.2767	0.7207
Hong Kong 8 [HH25]	Male	0.7564	0.3222	0.1940	0.2364	0.6301
Berlin	Male	1.3923	0.2836	0.0708	0.4551	0.5151

Abbreviations: CityU HK, City University of Hong Kong; VOI1, volume of interest 1; VOI2, volume of interest 2.

DISCUSSION

Casques exhibit interesting shape diversity within the hornbill family (Bucerotidae)—from rugose ridges to swollen oblong or vaulted chambers surmounting the upper jaw—and are thought to be diversely employed: in communication, mating displays, ³⁴ foraging, ³⁵ as acoustic resonating chambers, ⁷ and as nest‐building tools. ³⁶ Casque–beak and casque–casque interactions, either in play or combat, have been variously described ³⁷ , ³⁸ , ³⁹ as beak grappling, barging, beak‐wrestling, and jousting, and observed in the great hornbill (B. bicornis), ³⁷ the Indian gray hornbill (Ocyceros birostris), ³⁹ the Malabar pied hornbill (Anthracoceros coronatus), ³⁸ and silvery‐cheeked hornbills (Bycanistes brevis). ⁴⁰ In these species, casque‐butting behaviors are comparatively lower impact, involving either one flying individual clashing casques with a perched individual, or two individuals colliding mid‐air, but softening the impact by grasping/locking bills and circling downwards acrobatically immediately afterward. ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ Regardless of the intensity of the combat behavior, however, in all of these species (even Buceros, the sister genus to Rhinoplax), the casque is effectively hollow and devoid of trabeculation. ⁴¹

In contrast, the jousting behavior of the helmeted hornbill is far more violent, involving direct mid‐air casque–casque collisions with long approach flights. ¹⁷ , ³³ The ecological role of head‐butting in this species remains unclear but could be involved in competition over food resources (e.g., figs, their primary diet) rather than status or territories, as this behavior was mainly observed near fruiting trees. ¹⁷ Establishing dominance over suitable naturally‐formed tree cavities for nesting, which are an important resource for helmeted hornbills, ⁴² , ⁴³ could be another reason. Observational data for aerial jousting between helmeted hornbills are rare, but recorded instances typically involve males colliding. ¹⁷

This collected evidence raises an interesting conundrum: Although cancellous bone biology supports the idea that this species’ robust trabecular architecture is linked to an extreme loading behavior, our data indicate that the trabecular networks of males and females are indistinguishable in their morphologies. It is difficult to imagine the casque's trabeculation as a benign character, as its substantial contribution to the bird's body mass ⁹ should represent a trade‐off with flight efficiency. Extensive trabeculation in both sexes could suggest that females also regularly perform the head‐butting behavior. There is only a single observation of female helmeted hornbill head‐butting, involving a male colliding with a perched female. ¹⁷ Strong female–female aggression in other monogamous, cavity‐nesting bird species, however, argues that aerial jousting between female helmeted hornbills is plausible. ⁴⁴ , ⁴⁵ , ⁴⁶ Alternatively, the casque could have functions beyond jousting and/or might be used differently by the sexes. During nesting season, for example, helmeted hornbills use their casques and beaks as tools to seal the females into their nests in hollow trees ⁴² ; however, it is difficult to imagine this behavior causes loading of the casque sufficient to stimulate formation of the extensive casque reinforcements we describe (especially as other, less‐trabeculated hornbills also use their beaks and casques in nest‐building ⁶ ). As posited for other hornbills, the casque could be involved in amplifying sound across a loud environment; helmeted hornbills exhibit one of the loudest calls in the forest, audible over huge distances (∼100 m). ⁴⁷ Acoustic properties, however, are doubtful as primary shapers of the helmeted hornbill's casque trabeculation, as hollow casques would provide greater resonance (instead, their calls likely resonate in their gular neck pouches ⁷ ). As the helmeted hornbill casque (and, likely, the trabecular bone it contains) develops at sexual maturity when juveniles initiate head‐butting, ⁴⁸ aerial jousting remains the most likely selective pressure shaping casque anatomy.

Our quantitative, XMT‐based analysis of the trabecular network of helmeted hornbills demonstrates several striking aspects of this species’ casque trabeculation relative to other hornbills, but also with regard to vertebrate trabecular bone in general. Trabeculae are common but variable internal structural features in vertebrate bones, typically forming foam‐like latticeworks, especially bridging the cortices of narrow or necked regions, and filling the ends of long bones where they associate with joints. ⁴⁹ , ⁵⁰ Trabecular contributions to bone strength are relatively macroscopic: not via local variations in their tissue material properties (e.g., increased mineral density) but rather through their architectural arrangement. Where trabecular bone is positioned and how it is shaped is orchestrated locally by bone cells, depositing and resorbing bone to build, remove, and refine trabeculae in ways that minimize regional tissue stresses. ⁵¹ , ⁵² , ⁵³ , ⁵⁴ As a result, the alignment and orientation of trabecular networks may reflect to some degree the trajectories of principal stress in loaded bones ⁵⁵ , ⁵⁶ , ⁵⁷ and the thickness of trabeculae (at least in hips, heads, and condyles of weight‐bearing femurs) scales with the size of animals. ⁵⁶ , ⁵⁸ , ⁵⁹ The internal trabeculation of bones can therefore act as a physiological fingerprint, proffering clues as to how a bone is used and the size and habits of its owner.

Our data show that the casques of helmeted hornbills, in comparison with other vertebrates, possess exceptionally thick trabecular bone (Figure 5A). The helmeted hornbill's trabecular thickness is nearly five times greater than that of the occipital trabecular bone of woodpeckers (Picidae), birds known to experience extreme blunt trauma. ⁶⁰ , ⁶¹ Doube et al. ⁵⁸ defined an allometric relationship between trabecular thickness and femoral head radius (a proxy for body size), derived from femoral cancellous bone from 90 terrestrial avian and mammalian species, a six order of magnitude variation in body mass. Among those diverse vertebrate species, trabeculae were shown to vary roughly an order of magnitude in their thickness, from ∼0.045 to ∼0.50 mm in the femurs of Etruscan shrews and Asian elephants, respectively. In contrast, the central casque trabeculae of the helmeted hornbill are on average 0.92 ± 0.50 mm thick, with a maximum of 4.34 mm. Helmeted hornbill trabeculae therefore represent a significant departure from the allometric relationship defined by Doube et al. ⁵⁸ Contrary to this apparently high safety factor construction, the sparse trabeculae populating the margins of the casque and upper jaw (∼0.2 mm thick; Figures 2B and 3) are more in keeping with expectations of trabecular thickness given the size of the animal.

FIGURE 5.

Helmeted hornbill trabecular bone in context. Comparison of helmeted hornbill trabecular parameters relative to those of various birds and mammals (Doube et al. ⁵⁸ )—(A) trabecular thickness (Tb.Th), (B) bone volume fraction (BV/TV) and (C) connectivity density (Conn.D)—revealing the exceptional structural features of hornbill trabeculae. (A) The trabecular thickness of the helmeted hornbill's bony casque (0.92 ± 0.50 mm) is unusually high for a bird, and comparable to or much larger than that of Asian elephant femurs. (B) In contrast, the bone volume fraction of helmeted hornbill trabecular bone is similar to that of the femurs of other vertebrates, whereas (C) the Conn.D is much lower, indicating less interlinking among trabeculae. Note, however, that whereas Conn.D values are largely as predicted by the allometric relationship described for bird bone by Doube et al. ⁵⁸ , Tb.Th is much higher than expected. To allow comparison with the data from Doube et al. ⁵⁸ (plotted relative to femoral head radius), helmeted hornbill casque parameters are plotted against hypothetical femoral head radius values suggested by their body mass, derived using the scaling relationship r = 3.08M ^0.42, where M is approximated as 2.7 kg for females and 3.1 kg for males. ⁶²  Yellow swaths are approximated convex hulls around helmeted hornbill data to emphasize the data visually. Source: The plots, modified from Doube et al., ⁵⁸ are used with permission from the Royal Society, UK.

In vertebrates, very thick trabeculae typically indicate an adaptive tissue response to high dynamic loads, the struts thickened by apposition of new bone to reduce high tissue strains in directions relevant to loading. For example, in bighorn sheep, the thick (0.65–2.87 mm) plates of foam‐like velar bone in the horn core help absorb a significant amount of strain energy, dissipating impact stresses during ramming. ^{22 , 63} The higher surface‐to‐volume ratio of the trabecular structure, relative to a similar volume of cortical bone, may further enhance bone remodeling processes in response to impact. ⁶⁴ , ⁶⁵ Our observations of exceptionally thick trabeculae in helmeted hornbill (and the near lack of trabeculae in other hornbill casques; Schindler et al. ⁴¹ ) are therefore in keeping with the hypothesis that helmeted hornbill casque trabeculation is linked to the high‐speed collisions they endure during aerial jousting behaviors. The apparent role of trabeculae in resisting sudden impact in helmeted hornbill casques would explain their significant thickness relative to the femoral and condylar trabeculae of other avian and mammalian species studied by Doube et al., ⁵⁸ where compression resistance is the primary function (Figure 5). Additionally, the high degree of alignment of trabecular bone in the central bundle (VOI2; DA = 0.62 ± 0.07; Figures S1 and S2), bridging the rostral impact surface of the casque and the craniofacial joint, suggests that predominantly rostrocaudally‐directed loads are withstood and dispersed by the trabecular network. The convergence of the many smaller trabeculae nearest to the impact face into larger bundles caudally (e.g., Figure 3C) further suggests that stress is concentrated from across the impact surface into the largest trabeculae, like a building's hierarchical system of trusses and structural girders.

Given the peculiarly large trabeculae in the helmeted hornbill casque, it is worth considering whether the resolution of our scans resulted in some over‐estimation of trabecular thickness (e.g., by failing to resolve small separation between adjacent trabeculae). We imagine this type of error to be relatively rare due to the low volume fraction (BV/TV) and high trabecular separation (Tb.Sp) typical of birds. ⁵⁸ Given the limited resolution we could achieve due to the large diameter of specimens, it is likely that small nutrient foramina supplying casque trabeculae as intra‐trabecular osteons were also not resolved. ⁶⁶ In human bone, the maximum distance from an exchange surface to each osteocyte is limited by diffusion to around 230 µm. ⁶⁷ , ⁶⁸ Resolving these putative foramina could cause Tb.Th measurements to drop to around 0.5 mm; these would still be impressive Tb.Th values for an animal of this size, and regardless, the overall structural arrangement of robust, merged bundles of rods in the helmeted hornbill casque remains remarkable. Higher resolution, tissue‐level investigations of helmeted hornbill trabeculae will be valuable for understanding how this network grows and perhaps repairs, but also how bone as a material can attain such comparatively massive trabecular thicknesses.

Despite the comparatively massive trabeculae in the central bundle of the helmeted hornbill casque, the bone volume fraction of the central bundle (BV/TV = 0.332 ± 0.051; not including peripheral regions with little or no trabeculation) is within the range of values reported for other vertebrates (BV/TV = 0.198 ± 0.099 in avian femora and 0.367 ± 0.094 in mammal femurs ⁵⁸ ), albeit somewhat high (Figure 5B). However, it should be noted that this is only half the bone volume fraction of woodpecker skull trabecular bone. ⁶⁰ , ⁶¹ More striking, however, is the form of helmeted hornbill trabecular bone, which is especially rod‐like (EF = 0.296, average median; Figure 3C), whereas most other vertebrates studied to date have a mix of rod‐ and plate‐like morphologies. ⁵¹ , ⁶⁹ , ⁷⁰ These observations further support the idea that improved mechanical properties in trabecular bone in general are achieved through variations in trabecular architecture (e.g., trabecular density and/or morphology), rather than the amount or quality of the bone material. The nearly unidirectional and rod‐like nature of helmeted hornbill trabeculae may also speak to a relatively limited loading regime for the casque, ⁵¹ , ⁵⁶ , ⁷¹   just as the relative lack of trabeculae in other hornbill casques (Figure 1E) suggests they have little or no mechanical function.

Observations in other species and anatomical regions support the idea that bone under high and directed load has a tendency to form oriented plate‐like trabeculae. ⁷² , ⁷³ , ⁷⁴ Indeed, this is largely the appearance of trabeculae in avian femoral heads and condyles, ⁵⁸ , ⁷⁵ but particularly true for the velar bone core of bighorn sheep horns. Rather than a network of struts, velar bone comprises interconnected, broad “sail‐like” bony plates, forming a cellular architecture with high resistance to twisting and bending loads during ramming. ²² , ⁷⁶ In contrast, we posit that helmeted hornbills face impact loads primarily normal to the impact surface, directed rostrocaudally toward the brain during aerial jousting. The anisotropic arrangement, rod‐like form, and extreme thickness of trabeculae in hornbill casques may therefore represent an evolutionary adaptation for impulse‐resistant architecture, where bending loads are relatively minimal. Answering how the trabecular rod‐bundle arrangement in helmeted hornbill casques develops—for example, as axial elongation, radial dilatation, and merging of individual rod‐like elements—would benefit studies on trabecular bone biomechanical response behaviors and require an ontogenic series of skulls to answer.

CONCLUSION

Our findings underline that the highly aligned and exceptionally robust trabeculation can be considered a diagnostic feature of helmeted hornbill skulls, perhaps even identifiable in small VOIs relative to any other vertebrate trabeculations. The trabecular geometry itself, however, cannot be used to help discriminate between males and females, meaning external features remain the most accurate method for determining the sex of individual specimens and products circulating in the illegal wildlife trade. Not only is sex discrimination of traded specimens particularly important for this critically endangered species for improved understanding of aspects of their natural history, but it is also important for determining if poachers are specifically or inadvertently targeting one member of the mate pair. Targeting of foraging males, for example, could have a profound effect on family unit survival during the long periods when females and fledglings are sealed in nests. Quantitative tissue and biomaterial investigations of endangered species’ products therefore have the capacity to define diagnostic anatomical features with insights into the ecologies and biomechanics of rare or elusive species, while also offering the potential for the development of fast‐detection tools to support conservation efforts.

Most studies of trabecular bone focus on low‐intensity loading scenarios (e.g., related to hip joint behaviors during locomotion). The helmeted hornbill's use of bone for high‐velocity impact may, therefore, offer novel structural solutions for creating energy‐absorbing architectural, sport, and biomedical materials (e.g., as in Aguirre et al. ⁶³ ). The multiscale structural differences among trabeculated bio‐architectures (e.g., avian/mammal long bones, bighorn sheep horn, and helmeted hornbill casque) and their associated and various loading conditions illustrate the scope of mechanical material optimization achievable through a tailorable natural building strategy, while also providing a valuable template for functional bioinspired composite design. ⁶³ The trabecular buttressing of helmeted hornbill casques, for example, could prove an important model for collision‐resistant structures for preventing traumatic brain or collision injury. Understanding the “manufacturing process” of natural, impact‐resistant architectures during animal growth and associated mechanisms of damage tolerance frames a valuable intersection point for tissue biology, ecology, industrial design, and architectural materials.

AUTHOR CONTRIBUTIONS

Venkata A. Surapaneni: Conceptualization; data curation; formal analysis; investigation; software; project administration; visualization; writing—original draft; writing—review and editing. Benjamin Flaum: Data curation; formal analysis; investigation; software; visualization; writing—original draft; writing—review and editing. Mike Schindler: Formal analysis; investigation; software; visualization; writing—original draft; writing—review and editing. Khizar Hayat: Formal analysis; investigation; software; visualization; writing—review and editing. Jan Wölfer: Formal analysis; software; visualization; writing—review and editing. Daniel Baum: Formal analysis; methodology; resources; software; visualization; writing—review and editing. Ruien Hu: Investigation; methodology; resources; visualization. Ting Fai Kong: Investigation; methodology; resources; visualization. Michael Doube: Conceptualization; formal analysis; methodology; project administration; resources; software; supervision; visualization; writing—review and editing. Mason N. Dean: Conceptualization; formal analysis; funding acquisition; methodology; project administration; resources; supervision; visualization; writing—original draft; writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

PEER REVIEW

The peer review history for this article is available at: https://publons.com/publon/10.1111/nyas.15254

Supporting information

Supporting Information

Surapaneni, V. A. , Flaum, B. , Schindler, M. , Hayat, K. , Wölfer, J. , Baum, D. , Hu, R. , Kong, T. F. , Doube, M. , & Dean, M. N. (2025). The helmeted hornbill casque is reinforced by a bundle of exceptionally thick, rod‐like trabeculae. Ann NY Acad Sci., 1544, 78–91. 10.1111/nyas.15254

DATA AVAILABILITY STATEMENT

All data and code that support the findings of this study are publicly available in an online repository FigShare at https://doi.org/10.6084/m9.figshare.27636192.