A new study published in the Swiss Journal of Palaeontology tackles this gap head-on. Researchers developed a methodology designed to quantify the degree of pachyostosis in ribs and vertebrae, using three-dimensional surface models of bones scanned from museum specimens. The approach works on both complete skeletal elements and fragmentary fossils, a crucial advantage because the fossil record rarely delivers complete bones. The team applied the method to sirenians, the clade of manatees, dugongs, and their extinct relatives, whose skeletons display a documented disparity in pachyostosis degrees, especially in their ribs. The results confirm not only that the method can detect this disparity but also that pachyostosis affects sirenian vertebrae, a finding that had previously escaped quantitative documentation.

The sample comprised seven specimens. Four species of extant sirenians were included: the West Indian manatee Trichechus manatus, the African manatee Trichechus senegalensis, the strictly freshwater Amazonian manatee Trichechus inunguis, and the dugong Dugong dugon. One fossil ascribed to Kaupitherium bronni, an Oligocene sirenian also referred to in the literature as Halitherium schinzii, represented the extinct condition. Two terrestrial species served as comparisons: the rock hyrax Procavia capensis, a fellow paenungulate and thus a phylogenetic relative of sirenians, and the common eland Taurotragus oryx, chosen for its roughly similar body size. Proboscideans, the sirenians’ closest extant relatives, were deliberately excluded because their graviportality and massive heads might confound the analysis. Most bones were scanned with handheld structured-light scanners, and the eland scans were retrieved from the MorphoSource digital repository. Notably, the Amazonian manatee specimen now exhibited in the Galerie d’Anatomie comparée of the Muséum national d’Histoire naturelle in Paris was brought back from Portugal during the Napoleonic campaign of 1807–1808 and studied by Cuvier himself in 1809.

The technical core of the method lies in segmenting each bone model into regions that could plausibly be found as fragmentary fossils. For ribs, a custom Python script, implemented as an extension of the Blender 3D software, aligns each rib automatically in three-dimensional space, with the proximal end positioned at the origin and the distal tip rotated onto the X-axis. The rib is then divided into a user-defined number of equal-length chunks, with cutting planes perpendicular to the bone’s longest axis. For vertebrae, the researchers aligned the transverse processes and vertebral canal to establish orientation, then manually placed rectangular cuboids to enclose three regions: the centrum, the spinous process, and what they term the pedicle region, which encompasses the pedicles, transverse processes, and costal facets. Each region was defined by a Boolean intersection between the cuboid and the vertebra model, and the resulting meshes were verified as watertight before their volumes were measured.

The pachyostosis index itself is elegantly simple: a dimensionless ratio between the cube root of a region’s volume and a region-specific size proxy. For ribs, the proxy is the proximodistal length of the chunk. For vertebrae, different proxies apply to different regions, including centrum length, pedicle width, and spinous process length. Statistical differences among the resulting index distributions were assessed with pairwise Wilcoxon rank-sum tests corrected for multiple testing using the Holm–Bonferroni method, applied after exclusion of outlying regions. The outlying values came from the first and last ribs, the two proximal-most chunks containing the head and tubercle, and the distal-most tapered chunk, all of which depart morphologically from the rest of the rib series. Importantly, the analysis demonstrated that even diaphyseal rib fragments representing as little as ten percent of the total bone length can yield meaningful results.

The results are unambiguous for the ribs. On average, the rib chunks yielded pachyostosis index values of 0.70 for Dugong dugon, 0.79 for Trichechus manatus, 0.83 for Trichechus senegalensis, 0.79 for Trichechus inunguis, and 0.91 for the fossil Kaupitherium bronni, while the terrestrial comparison taxa scored 0.38 for the hyrax and 0.43 for the eland. The overlap between sirenian and non-sirenian distributions is limited, and among sirenians the dugong stands out with distinctly lower values, particularly in its caudal ribs, where robusticity decreases drastically until roughly the fifth rib and then holds steady at approximately 0.65. Trichechus species maintain more constant and higher values throughout the series. The fossil Kaupitherium bronni consistently shows the highest values, bulkier in both relative and absolute terms than any extant sirenian sampled.

The vertebral results are arguably the study’s most novel contribution. Average total vertebral pachyostosis index values were 0.92 for Dugong dugon, 1.01 for Trichechus manatus, 1.02 for Trichechus senegalensis, and 1.07 for Kaupitherium bronni, compared with 0.80 for the hyrax and 0.66 for the eland. Centrum values were also higher in sirenians, ranging from 0.78 to 0.93 against 0.62 to 0.67 in the outgroups. The clearest vertebral pattern emerged in the spinous processes, which show increasingly high values from the dugong to the manatees. The authors conclude that the absence of previous documentation of vertebral pachyostosis in sirenians reflects a methodological impairment rather than a biological signal, and that bone mass increase can be expected throughout the sirenian axial postcranium. Vertebral pachyostosis has been documented in cetaceans and reptiles such as mesosaurs, pachypleurosaurs, and pachyophiid snakes, and the strongest degree recorded to date belongs to the basilosaurid whale Perucetus colossus, whose vertebrae represent an estimated 4.5 times the volume of a non-pachyostotic relative scaled to the same size. By comparison, the mid-thoracic vertebra of Kaupitherium bronni entails a volume increase factor of roughly 1.8.

Despite the confirmation of vertebral pachyostosis, the study demonstrates that rib pachyostosis preponderantly affects overall body composition. Scaling the mid-series rib of Kaupitherium bronni against a terrestrial relative yields a volume increase factor of 11.5, far exceeding the vertebral factor. The whole rib series volume represents 1.6 to 2.5 times that of the vertebrae in extant sirenians, and up to 3.9 times in Kaupitherium bronni, while the ratio is only 0.7 in the hyrax. This effect is amplified by the strong osteosclerosis documented in sirenian ribs, with even the less pachyostotic dugong featuring a compactness value of 97.7 percent, while mammalian vertebrae always retain greater porosity, especially in the centrum. Sirenians also possess a relatively large number of ribs, from 14 to 21 pairs in extant taxa, further concentrating the ballast effect in the thorax.

The lifestyle implications are compelling. The dugong, equipped with a high aspect ratio caudal fluke practicing caudal oscillation, is seen as more active than manatees, which rely on generalized axial undulation. The dugong is also fully marine and capable of deeper dives, over 30 meters, where substantial hydrostatic pressure lowers overall buoyancy and additional ballast becomes detrimental when animals turn negatively buoyant at depth. Swimming style and diving behavior thus appear to have a preponderant effect on the degree of pachyostosis observed in extant sirenians. The slender ribs of the dugong have been interpreted as a reversal from a more pachyostotic ancestral condition, and the high index values of Kaupitherium bronni sustain this hypothesis. The authors argue that the dugong’s milder expression of bone mass increase should be recognized as present rather than absent, likely reflecting further specialization to its specific lifestyle. Among the three manatee species, no clear difference in pachyostosis degree was detected, though the caudal-most ribs of the Amazonian manatee are clearly more slender than in its relatives.

The craniocaudal distribution of the index values offers a further refinement. In Trichechus, rib volume increases up to the tenth or eleventh rib, while in the dugong the most voluminous rib is the sixth, with average chunk volume remaining more uniform throughout the thorax. The stronger relative pachyostosis in the cranial ribs of the dugong seems to compensate for their smaller absolute size. The skeletal center of gravity for an adult Amazonian manatee has been estimated ventral to the eighth thoracic vertebra, consistent with a ballast and trim control function maintaining horizontal body position. The dugong, an obligate bottom feeder, may either locate its centers of gravity and buoyancy slightly more cranially or require excess bone mass in the cranial thorax to maintain the required trim. The authors suggest that the particularly elevated values of Kaupitherium probably reflect environmental conditions rather than swimming style, since salinity deviations are recognized near the Mainz Basin in the early Oligocene, and particularly saline, denser waters have already been linked to acute bone mass increase in mammals and teleost fishes. Alternatively, cooler waters requiring more insulating blubber, which is less dense than water, would demand additional bone mass. A combination of precise quantification and palaeoenvironmental reconstruction will be necessary to explain the full disparity of pachyostosis degrees in the sirenian fossil record.

The method’s caveats are candidly acknowledged. The choice of outgroup critically influences the analysis, and the old divergence between hyraces and sirenians, probably in the late Mesozoic or early Cenozoic, means their skeletal morphology differs drastically. It cannot be excluded that the terrestrial representatives carry their own specializations affecting relative axial bone volume, such as the expanded ribs seen in some xenarthrans, primates, and early turtles. Body size is another potential caveat, since although the index is dimensionless, it could scale allometrically, and future sampling of taxa with similar ecologies but disparate sizes, such as bovids or cetaceans, should assess this correlation. The methodology may also prove valuable for fossil identification, since pachyostosis has historically been used as a criterion to ascribe specimens to Sirenia, and quantifying the degree of pachyostosis in co-occurring clades such as cetaceans and pinnipeds will be necessary to assess the metric’s taxonomic utility. Beyond sirenians, the approach can be applied to the myriad other tetrapods that secondarily invaded water, and even to cases of pachyostosis unlinked to aquatic adaptation, such as the swollen mandibles of some Pleistocene cervids or the hyperostotic bones of over 150 species of actinopterygian fishes, which might help clarify their poorly understood biological significance. Given that the sirenian bauplan has not drastically changed since the late Eocene, the great disparity in pachyostosis degrees across extinct sirenians stands out as a rich record of lifestyle and environmental variation, one that this new quantitative framework is now positioned to decode.

Subject of Research: Sirenian pachyostosis revisited

Article Title: Sirenian pachyostosis revisited

Article References: Wagner, N., Horn, J., von Kleist, J., Martinez, Q., Ferreira, G. S., & Amson, E. (2026). Sirenian pachyostosis revisited. Swiss Journal of Palaeontology, 145, 687-702. https://doi.org/10.3897/sjp.145.191327

Image Credits: AI Generated

DOI: 10.3897/sjp.145.191327

Keywords: Sirenian, pachyostosis, revisited, scientific research, peer-reviewed research, research findings