In contrast to the cerebral cortex, the cerebellum of all carnivorans in the dataset conforms to the neuronal scaling rule that applies to the ensemble of afrotherians (minus the elephant), glires, and artiodactyls – with the sole exception, again, of the raccoon (Figure 4C). The relationship between cerebellar mass and number of cerebellar neurons of carnivorans (excluding the raccoon) can be described by a power law of exponent 1.100 ± 0.084 (r 2 = 0.971, p < 0.0001) that is not significantly different from linearity but is significantly different from the exponent of 1.283 ± 0.035 that applies to the ensemble of afrotherians (minus the elephant), glires and artiodactyls, which we have proposed to represent the ancestral neuronal scaling rule for the mammalian cerebellum (Herculano-Houzel et al., 2014b). In contrast, the raccoon cerebellum has nearly two times more neurons than predicted for a mammalian species belonging to those non-primate orders, conforming instead to the number of neurons found in the cerebellum of a primate of similar cerebellar mass. As expected from these relationships, neuronal densities in the cerebellum of carnivorans, again with the exception of the raccoon, conform to the relationship that applies to the ensemble of afrotherians (minus the elephant), glires and artiodactyls (Figure 4D), even though the power function relating neuronal densities in the cerebellum of carnivorans to the number of neurons in the cerebellum does not reach significance (p = 0.2918 without the raccoon; Figure 4D).
The mass of the carnivoran rest of brain scales with the number of neurons in the structure raised to an exponent of 1.875 ± 0.134 (without the raccoon). This exponent is not significantly different from the exponent of 2.041 ± 0.143 that applies to the ensemble of artiodactyls, marsupials and eulipotyphlans (r 2 = 0.928, p < 0.0001; Figure 4E; Herculano-Houzel, 2017), and indeed the 95% confidence interval for carnivorans includes these species (Figure 4E).
The discrepancies between expected and observed numbers of neurons in some carnivoran species, most notably the brown bear, could in principle be due to aberrant immunoreactivity to NeuN, which might fail to label all neurons in these species. Similarly, the aberrantly large numbers of neurons found in raccoon brain structures might in principle be due to non-specific labeling of non-neuronal cells with the anti-NeuN antibody. In these scenarios, any unlabeled neurons in the bear cerebral cortex would be classified as non-neuronal cells and cause aberrantly high numbers of non-neuronal cells in brain structures for their mass; conversely, any labeled glial cells would be mistakenly classified as neurons and lead to aberrantly low numbers of non-neuronal cells in raccoon brain structures. These aberrations would be particularly easy to spot since all major brain structures (cerebral cortex, cerebellum and rest of brain) of all mammalian species examined so far exhibit a relationship between structure mass and number of other (non-neuronal) cells that can be described by a single power function of near-linear exponent 1.051 ± 0.014 (r 2 = 0.974, p < 0.0001; Figure 5A; Herculano-Houzel, 2017).
FIGURE 5. All carnivoran species and brain structures conform to the scaling of brain structure mass with numbers of other cells that applies universally across other mammalian species. Cerebral cortex is shown in circles, cerebellum in squares, rest of brain in triangles. (A) Brain structure mass scales universally as a power function of the number of non-neuronal (other) cells in the structures across non-carnivoran species (plotted function; exponent 1.051 ± 0.014, r 2 = 0.974, p < 0.0001), and all carnivorans conform to that relationship. (B) The density of other cells in the different structures of carnivoran brains overlaps with densities in the same structures in other mammalian species, which scales very slowly as a power function of structure mass of exponent -0.075 ± 0.012 (r 2 = 0.200, p < 0.0001). (C) The ratio between numbers of other cells (which approximates the number of glial cells) and numbers of neurons in each structure is not a universal function of structure mass across mammalian species and structures. (D) The ratio between numbers of other cells and neurons in each structure does vary universally with average neuronal density in the structure across non-carnivoran species (plotted function, exponent -0.942 ± 0.019, r 2 = 0.946, p < 0.0001, n = 146), and all carnivoran species and brain structures conform to that relationship.