BackgroundMammals as a rule have seven cervical vertebrae, except for sloths and manatees. Bateson proposed that the change in the number of cervical vertebrae in sloths is due to homeotic transformations. A recent hypothesis proposes that the number of cervical vertebrae in sloths is unchanged and that instead the derived pattern is due to abnormal primaxial/abaxial patterning.ResultsWe test the detailed predictions derived from both hypotheses for the skeletal patterns in sloths and manatees for both hypotheses. We find strong support for Bateson's homeosis hypothesis. The observed vertebral and rib patterns cannot be explained by changes in primaxial/abaxial patterning. Vertebral patterns in sloths and manatees are similar to those in mice and humans with abnormal numbers of cervical vertebrae: incomplete and asymmetric homeotic transformations are common and associated with skeletal abnormalities. In sloths the homeotic vertebral shift involves a large part of the vertebral column. As such, similarity is greatest with mice mutant for genes upstream of Hox.ConclusionsWe found no skeletal abnormalities in specimens of sister taxa with a normal number of cervical vertebrae. However, we always found such abnormalities in conspecifics with an abnormal number, as in many of the investigated dugongs. These findings strongly support the hypothesis that the evolutionary constraints on changes of the number of cervical vertebrae in mammals is due to deleterious pleitropic effects. We hypothesize that in sloths and manatees low metabolic and activity rates severely reduce the usual stabilizing selection, allowing the breaking of the pleiotropic constraints. This probably also applies to dugongs, although to a lesser extent.

Homeotic transformations of vertebrae are particularly common in humans and tend to come associated with malformations in a wide variety of organ systems. In a dataset of 1,389 deceased human foetuses and infants a majority had cervical ribs and approximately half of these individuals also had missing twelfth ribs or lumbar ribs. In ~10 % of all cases there was an additional shift of the lumbo-sacral boundary and, hence, homeotic transformations resulted in shifts of at least three vertebral boundaries. We found a strong coupling between the abnormality of the vertebral patterns and the amount and strength of associated malformations, i.e., the longer the disturbance of the vertebral patterning has lasted, the more associated malformations have developed and the more organ systems are affected. The germ layer of origin of the malformations was not significantly associated with the frequency of vertebral patterns. In contrast, we find significant associations with the different developmental mechanisms that are involved in the causation of the malformations, that is, segmentation, neural crest development, left-right patterning, etc. Our results, thus, suggest that locally perceived developmental signals are more important for the developmental outcome than the origin of the cells. The low robustness of vertebral A-P patterning apparent from the large number of homeotic transformations is probably caused by the strong interactivity of developmental processes and the low redundancy of involved morphogens during early organogenesis. Additionally, the early irreversibility of the specification of the A-P identity of vertebrae probably adds to the vulnerability of the process by limiting the possibility for recovery from developmental disturbances. The low developmental robustness of vertebral A-P patterning contrasts with a high robustness of the A-P patterning of the vertebral regions. Not only the order is invariable, also the variation in the number of vertebrae per region is small. This robustness is in agreement with the evolutionary stability of vertebral regions in tetrapods. Finally, we propose a new hypothesis regarding the constancy of the presacral number of vertebrae in mammals.

No abstract