It has been consistently noted, in both a qualitative and quantitative sense, that elephants have large cerebellums, larger than in both humans and cetaceans (Beddard,1893; Precechtel,1925; Haug,1970; Lange,1971; Cozzi et al.,2001), (Fig. 1). Shoshani et al. (2006) described the gross anatomical appearance of the elephant cerebellum in great detail, and while there are some specific characteristics, such as complex foliation and an enlarged lateral recess of the fourth ventricle, the general appearance is not unusual apart from some asymmetries in fissure patterns between the two cerebellar hemispheres (Beddard,1893). Further to this, Shoshani et al. (2006), measured cerebellar mass from four elephants (one African, three Asian) and found that the weight of the cerebellum contributes to 18.6% of the total brain mass, making the elephant cerebellum 1.8 times larger, relative to brain mass, than that observed for humans and 1.9 times larger than that observed in sheep. Precechtel (1925) reported that the corpus restiformes of the inferior cerebellar peduncle and the brachium pontis (or middle cerebellar peduncle) are large, and their nuclei are more prominent than those in humans, which is in contrast to the statement made by Shoshani et al. (2006) indicating that the pons appears small leading to the possibility that the corticopontine input to the cerebellum is reduced in the elephant compared to other mammals.
The work of Finlay and Darlington (1995) argues that there is an apparent pattern, or linked regularity, in the sizes of mammalian brain structures and the brains in which they are found, so much so that the mass or volume of a given structure within a brain can be reliably predicted when the total brain mass or volume is known. This implies that the cerebellum will show a consistent proportionality in mass/volume to the total mass/volume of the brain across all mammals. Finlay and Darlington (1995) demonstrated this consistency in size of the cerebellum relative to total brain mass when examining a series of insectivore and primate brains; however, a shortcoming of their study, and subsequent interpretation, was the lack of a broader comparative approach by including data available from a larger range of species. Two exceptions to the linked allometric regularity demonstrated by Finlay and Darlington (1995) have been documented for cerebellar size in relation to brain mass, this being the relative cerebellar volume of microchiropteran and odontocete cetacean brains (Baron et al.,1996; Marino et al.,2000), where the cerebellum has been recorded to be larger in relative volume than expected for mammals of a similar brain size.
Given that elephants are consistently reported, in mostly qualitative comparisons, to have a large cerebellum, what is the size of the cerebellum of the elephant—larger, smaller, or as expected for total brain mass when compared to a broad range of mammalian species? To date, quantification of the size of the elephant cerebellum has not been undertaken (Fig. 1), and the potential significance of the seemingly larger size has not been discussed in detail. Moreover, given that microchiropterans and odontocete cetaceans are exceptions to the “rule” of linked regularities for mammalian cerebellar volumes, how does the relative size of the elephant cerebellum relate to these species? If certain mammalian species do have a relatively large cerebellum, is this formed of hemispheric or vermal cerebellum? Lastly, if there are changes in the relative total cerebellar size or components of the cerebellum, do these have potential functional significances that can be highlighted through volumetric comparisons? Thus, the objective of the current study was to quantify these potential regularities and irregularities across as many mammalian species as possible, including elephants, and to determine if potential functional correlates may be revealed.
MATERIALS AND METHODS
In the current study, data on brain mass and cerebellar volume were obtained from previously published literature and analyses based on MRI undertaken specifically for the current study. Data on microchiropterans (N = 222 species) and megachiropterans (N = 47 species) were taken from Baron et al. (1996). Data for insectivores (N = 28 species) and primates (N = 46 species) were taken from Stephan et al. (1981). Data for odontocete cetaceans were taken from Marino et al. (2000) and Schwerdtfeger et al. (1984) or generated from specimens available for the current study; however, instead of only taking species averages, every adult individual from this database was included for analysis, resulting in 39 individuals from 6 species being sampled (Delphinus delphis—10 individuals; Tursiops truncatus—21; Pontoporia blainvillei—1; Inia geoffrensis—1; Lagenorhynchus acutus—4; Phocoena phocoena—2). Data for four elephants, one African and three Asian, were taken from Shoshani et al. (2006). In addition to this, total cerebellar volume was determined for three adult male African elephants (Loxodonta africana, Manger et al.,2009) (Table 1). Obviously, with gathering such a large dataset, where volumes have been calculated in different ways and fixation and storage protocols of the brains have been different, there may be some mismatch in the data, however, our analysis (see below) appears to indicate that these errors would be of a minor scale and would not change the overall conclusions of the current study.
Table 1. Brain mass, cerebellar volumes, and cerebellar quotients of the species analyzed in the current study
Brain mass (g)
Cerebellar volume (mL)
For insectivores, microchiropterans, megachiropterans, and primates ranges are provided. Sources of data: (1) this study; (2) Shoshani et al. (2006); (3) Stephan et al. (1981); (4) Baron et al. (1996); (5) Schwerdtfeger et al. (1984); (6) Marino et al. (2000).
Loxodonta africana (LA1)
Loxodonta africana (LA2)
Loxodonta africana (LA3)
Insectivores (28 species)
Microchiropterans (222 species)
Megachiropterans (47 species)
Primates (46 species)
In addition to this comparison of total brain mass to total cerebellar volume, data on cerebellar component volume (hemispheric and vermal volumes) were determined for 51 eutherian mammals (Table 2) with measurements made using MRI (three African elephants and two harbor porpoises), freely available histological sections (www.brainmuseum.org) and previously published data for primates (Smaers et al.,2011).
Table 2. Brain mass, vermal, and hemispheric volume of a range of eutherian mammals
Brain mass (g)
Vermal vol (mL)
Hemispheric vol (mL)
Sources of data: (1) this study; (2) this study using sections available at www.brainmuseum.org; (3) Smaers et al. (2011).
African elephant (LA1)
African elephant (LA2)
African elephant (LA3)
Common woolly monkey
Geoffroy's spider monkey
Red howler monkey
Western red colobus
California sea lion
White tailed deer
Bos taurus indicus
Magnetic Resonance Imaging (MRI)
Brains from three African elephants (Loxodonta africana, LA1, LA2, LA3, Manger et al.,2009) and two harbor porpoises (Phocoena phocoena) underwent MR imaging to obtain measurements of total cerebellar volume and vermis and hemispheric volumes. The brains were fixed and stored as described in Manger et al. (2009) and then scanned in coronal, sagittal, and horizontal planes. The specimens were scanned on a Philips 1.5 Tesla Intera System (Eindhoven, The Netherlands), using all three elements of the head and neck coil. The brains were removed from their storage containers, drained of excess fluid, and placed in the head coil wrapped in a dry sheet, thus being exposed directly to air, which also partly entered the ventricles. After testing different scan parameters the following sequence was selected as giving the best detail and the least artifact (especially at air–fluid interfaces). The selected T1 weighted inversion recovery sequence consisting of 2 mm slices without a gap, had a TR (time to repeat) of between 6.5 and 10.9 sec depending on the number of slices, a TE (time to echo) of 10 msec and a TI (time to invert) of 300 msec. The number of signal averages varied between 3 and 4 with a flip angle of 90 degrees and an echo train of 10. The scan times varied between 15 and 25 min. The antifreeze liquid in which the brains were stored showed high signal on both T1 and T2 weighted sequences and the routine clinical T1 and T2 sequences produced very similar T2 like images of the brain specimens. This is possibly related to the lack of water in the tissues of the specimen secondary to the fixation and storage process. The images were processed using the freely available open source software program OsiriX (Rosset et al.,2004; www.osirix-viewer.com), which allowed us to also measure the volume of the cerebellums and cerebellar regions in all animals scanned.
Calculating Cerebellar Volume From MR and Other Images
By outlining the cerebellum in each coronal slice along its clearly visible borders, the total volume was calculated, that is, by summing each of the areas occupied by the cerebellum on each slice from the most rostral to the most caudal slices in which the cerebellum appeared, and multiplying this by slice thickness (2 mm for MRI and 40 μm for images from www.brainmuseum.org). For total cerebellar volume, the cerebellar peduncles in the appropriate regions, the hemispheres, and the vermis were included as contributing to the cerebellar volume (Fig. 2) as described by Baron et al. (1996) and Stephan et al. (1981).
Cerebellar Quotients (CQ)
The calculation of quotients for the size of the cerebellum in the different species studied followed the methodology of Jerison (1973). A baseline allometric equation was determined from the relationship between brain mass and cerebellar volume for the clustered data from primates, megachiropterans, and insectivores (which as individual groups were statistically indistinguishable). This provided our “mammalian” baseline used for calculation of CQ. The allometric equation used was (Fig. 3):
where Cbvol is cerebellar volume and Mb is brain mass. This equation was used to calculate CQ where:
Cerebellar quotients were calculated for all species (or individuals) analyzed in the current study (Table 1).
Calculating Vermal and Hemispheric Components of the Cerebellum
The volumes of the cerebellar hemispheres and the vermis were measured to quantify the relationship between the two, if it varied and how it related to brain mass. This was done for the elephants and several other eutherian mammals (Table 1). For the elephants and harbor porpoises, MR images were used to outline each individual hemisphere and the vermis (Fig. 2) throughout the cerebellum, using the software program Osirix. The volume of the vermis consisted of the vermal cortex and underlying white matter, while the hemispheric volumes consisted of the hemispheric cortex, underlying white matter and the cerebellar peduncles (Fig. 2).
For the other mammals for which hemispheric and vermal volumes were calculated, the database Comparative Mammalian Brains (www.brainmuseum.org) was used to download coronal slices of Nissl-stained brains of 27 mammals (primates, carnivores and Cetartiodactyls, Table 2), and taken from the publication by Smaers et al (2011) (19 primates). From the sections taken from the Comparative Mammalian Brains website, it was clearly visible where the borders of the cerebellar hemispheres (including the cortex, white matter and peduncles) and those of the vermis (including the vermal cortex and underlying white matter) were located. Using the freely available software program Image J, the downloaded images were opened and the polygon tool used to trace around the hemispheres and vermis in all sections where these structures appeared. The areas were summed and multiplied by slice thicknesses to give the total volume of each of the structures (Table 2).
The data were logarithmically transformed to the base ten for allometric analysis. We then divided the data into four groups, these being elephants, primates/megachiropterans/insectivores, odontocete cetaceans, and microchiropterans. The clustered primate/megachiropteran/insectivore groups showed no statistically significant allometric differences and were thus used as a baseline upon which to compare the size of the cerebellum of the other species. The microchiropterans have a significantly raised elevation when compared to the megachiropterans and thus these two groups were treated separately, making cerebellar volume yet another neuroanatomical criterion that separates the two suborders of Chiroptera. We then undertook three separate analyses of the transformed data. First, standardized major axis, or reduced major axis (see Warton et al.,2006 for discussion of the terminology), was used to describe and compare the bivariate relationship of the different groups. The software SMATR ver. 2.0 (Warton et al.,2006; www.bio.mq.edu.au/ecology/SMATR) was used to test for common slopes between the groups, and where present, test for shifts in elevation or along the common axis. Second, ordinary least squares regressions were calculated for the primate/megachiroptera/insectivore, odontocete cetacean, and microchiropteran groups. The probability of using these regressions to correctly predict cerebellar volume for individual observations in the elephant was assessed (Sokal and Rohlf,1995, p 469), and prediction intervals drawn to aid description.
The volume of the cerebellum in elephants, including its individual components (the vermis and hemispheres), was measured in the current study using MR imaging. The absolute size of the elephant cerebellum was found to be large, ranging from 761.46 to 1036 mL (Table 1). It was found that the volume of the cerebellum in the elephants was large relative to brain mass, with cerebellar quotients for elephants ranging between 1.66 and 2 (Table 1), as were both the vermal and hemispheric components of the elephant cerebellum, when compared to a range of other mammals (Table 2).
Relative Cerebellar Size of Primates, Megachiroptera, and Insectivores
The data for primates, megabats, and insectivores were grouped, as a test for commonality of slopes and elevations indicated, with strong statistical significance (Fig. 3), that their slopes and elevations were similar, allowing for the data from these three groups to be treated as one. This allowed the formation of a “mammalian” baseline to compare the data of the odontocete cetaceans, microchiropterans, and elephants against (Fig. 3). The allometric relationship between brain mass and cerebellar volume for this group was slightly negative (slope = 0.978), but very close to isometry. Given the strength of the correlation between brain mass and cerebellar volume (r2 = 0.994, P = 4.2 × 10−134) this baseline was considered to be sufficient for comparative purposes. This grouping was also confirmed by the calculation of cerebellar quotients (CQ), where each of these three groups had average CQs close to 1 (Fig. 4, Table 1).
Relative Cerebellar Size of Odontocete Cetaceans and Microchiropterans
Both these mammalian groups appeared to have larger than expected cerebellar volumes for brain mass when compared to the baseline determined using the primate/megachiroptera/insectivore cluster (Fig. 3). Both odontocete cetaceans and microchiropterans exhibited slight positive allometry in the relationship between brain mass and cerebellar volume, but this was very close to isometry (for odontocete cetaceans the slope = 1.003, r2 = 0.957, P = 3.5 × 10−28; and for the microchiropterans the slope = 1.006, r2 = 0.969, P = 6.4 × 10−168). The larger relative cerebellar volume was reflected in the calculation of CQ (Fig. 4), where odontocete cetaceans had an average CQ of 1.38 (having a cerebellar volume 1.38 times larger than expected for mammals of equal brain mass, range, 1.13–1.70) and microchiropterans 1.44 (range, 1.09–2.15). The apparent grade shift between these two orders and the “mammalian” group as assessed visually is supported when tested for significance in shifts of elevation; however, both orders share a common slope with the “mammal” group.
Elephant Total Cerebellar Volume Compared to Other Mammals
Both species of elephant analyzed (African and Asian elephants) had large cerebellar volumes for their brain sizes when compared to the mammalian baseline (see above) and to odontocete cetaceans (Table 1). While both species also showed large cerebellar volumes for their brain sizes when compared to the microchiropterans, the microchiropterans showed overlap in relative cerebellum size with the elephants. To compare the relative size of the cerebellar volume in elephants to the other mammals, we calculated the 95% confidence intervals for each group (Fig. 3). The cerebellar volume of the elephants fell above the 95% confidence intervals (CI) calculated for primates + megachiropterans + insectivores and for primates + megachiropterans only. When compared to odontocete cetaceans, some of the data points for the elephants were within the 95% CIs, but some were above. When compared to the microchiropterans, all data points for the elephants were within the 95% CI. Thus, for the majority of data points, it is clear the elephants have absolutely and relatively large cerebella.
The cerebellar quotients that were calculated in the current study indicate that, on average, elephants have the largest relative cerebellar size for mammals (Fig. 4), but some microchiropterans do posses cerebella that have relative volumes larger than that of the elephants. For example, the range of CQs for elephants was 1.66–2.00, while that of the microchiropterans was 1.09–2.15. Of the 222 microchiropteran species analyzed, 6 species had larger CQs than the largest elephant CQ, while 202 species had lower CQs than the smallest elephant CQ. Thus, on average, the CQ of the elephants studied is larger than the microchiropterans studied (Fig. 4).
A Comparison of Hemispheric and Vermal Volumes Amongst Species
As the cerebellum is readily divisible into hemispheric and vermal portions, and that each of these portions are related to different functional aspects related to connectivity, we calculated the volumes of each of these portions in 51 eutherian mammalians for comparison (Table 2). A reliable correlation was observed between the brain mass and hemispheric volume for mammals (excluding the elephant and odontocete cetacean data) (slope = 1.11, r2 = 0.97, P = 2.38 × 10−33), (Fig. 5). Volumes of the cerebellar hemispheres for the three male African elephants used in this analysis fell on or just above the 95% CI, therefore having larger than expected cerebellar hemispheres in comparison to other mammals of a similar brain mass. The volumes of the cerebral hemispheres for the three odontocete cetaceans used in this analysis fell above the 95% CI, therefore having larger than expected hemispheres in comparison to other mammals of a similar brain mass.
For the volume of the vermis, a reliable correlation was found between the brain mass and vermal volume for mammals (excluding elephants and odontocete cetaceans) (slope = 0.70, r2 = 0.86, P = 7.87 × 10−20), (Fig. 5). The volume of the vermis for the three elephants was found to lie above the 95% CI, therefore the vermal volume of the elephant appears to be larger than would be expected compared to mammals of a similar brain size. On the other hand, the volume of the vermis for two odontocete cetaceans (harbor porpoises) fell below the 95% CI, indicating that the vermal volume in harbor porpoises is smaller than would be expected for mammals of a similar brain mass, but that of the bottlenose dolphin fell within the 95% CIs indicating that for this species the vermal volume is what would be expected for a mammal with a similar brain mass.
The analyses undertaken in the current study revealed several findings of interest. The first finding, that primates, insectivores, and megachiropterans all appear to follow a standard mammalian relationship between the size of the brain and the size of the cerebellum confirms, and extends through the addition of the megachiropterans, the findings reported by Finlay and Darlington (1995) of a linked regularity between brain volume and cerebellar volume across a wide range of mammalian species. Despite this, in terms of relative cerebellar size, three groups investigated, the elephants, odontocete cetaceans, and microchiropterans, were seen to have larger than expected total cerebellar volumes, with the elephants having both the largest absolute and relative cerebellar volumes of any mammal studied to date. Our analysis of the vermal and hemispheric components of the cerebellum also revealed that for the most part across mammals there is a predictable relationship between the volume of these components and the volume of the brain. In contrast, the elephants and odontocete cetaceans appeared to differ from this baseline, with the elephants having large vermal and hemispheric components, matching the overall large size of the elephant cerebellum, but the odontocete cetaceans showed small or standard vermal and large hemispheric components relative to brain size. The cerebellum is known to play a major role in sensorimotor integration, using sensory feedback to control the temporal patterns of the force, extent and duration of muscular contractions during movement (Yarom and Cohen,2002; Butler and Hodos,2005; Molinari et al.,2007; Bower,2011). The variances from mammalian cerebellar volumetric baselines observed in the current study are discussed in terms of the life histories and specific anatomies of these species.
Relatively Large Cerebellums in Elephants, Odontocete Cetaceans, and Microchiropterans
When examining total cerebellar size, the elephants, odontocete cetaceans, and microchiropterans were seen to have cerebellums that were clearly larger than the baseline determined from the analysis of primates, insectivores, and megachiropterans. From this comparison it appears that in general, we can conclude that the elephants have the largest relative cerebellar size, and that the odontocete cetaceans and microchiropterans have cerebellums that are enlarged to a similar extent. In all three cases there are similarities in terms of life histories that may be proposed to explain this observation. All three groups have unusual systems of vocalization (infrasound in elephants, Garstang,2004; echolocation in cetaceans and microchiropterans, Rendell et al.,1999; Speakman,2001). In both the elephants and odontocete cetaceans, this specialized vocalization appears to be associated with an expansion of the periaqueductal grey matter into a specific nucleus ellipticus (Cozzi et al.,2001; Manger,2006; Shoshani et al.,2006). Associated with the production of these vocalizations is a complex air sac anatomy within the blowhole of odontocete cetaceans (Reidenberg and Laitman,2008), an association between flapping flight and laryngeal vocalization in microchiropterans (Speakman,2001), and an unusual unpaired appendage, the trunk, in elephants (Endo et al.,2001) although the mechanism for the production of infrasound in elephants is currently unknown. Add to this the fact that each of these groups have other specific motor demands not generally observed in other mammals, adaptation to an aquatic (swimming—odontocete cetaceans) or aerial (flying—microchiropterans) environment and the control of the trunk in elephants (the trunk being a very complex muscular appendage, Endo et al.,2001). That both unusual vocalization and motor abilities are apparent in these species may form a basis for explanation of the relatively enlarged cerebellum in these three groups.
Elephant Cerebellar Size
Elephants have, on average, the largest relative cerebellar size of all mammals studied to date. Associated with this large size are the relatively large cerebellar hemispheres and vermis, that is, both components of the cerebellum are large in elephants. But what is the significance of having a large cerebellum with large components? As outlined above, the muscular control required to produce infrasound and control the anatomically complex trunk may play a role in the overall enlargement of the cerebellum relative to brain mass in the elephant. Elephants cannot lower their heads significantly, and therefore make use of the trunk to take food and drink, rendering the trunk as a refined manipulator of the environment (Endo et al.,2001), with the trunk also being involved in social interaction (Lee and Moss,1999), which are all crucial for survival and therefore need fine motor control that is based on sensory feedback. However, in addition to this, the requirements of the control of the trunk may be involved in the enlargement of both the vermal and hemispheric portions of the elephant cerebellum. The vermal portion of the cerebellum generally receives input from the spinal cord, including the spinal trigeminal system (e.g., Andersson and Eriksson,1981), and given that the tip of the elephant trunk is highly sensitive (Rasmussen and Munger,1996), is moved extensively, often delicately, as the elephant explores and manipulates its immediate environment, and that the majority of the representation of the body surface in the vermis is that of the face in other mammals (e.g., Welker et al,1988; Bower,2011), the finding that the vermal portion of the cerebellum is large in elephants, is congruent with these observations.
In addition to the large vermal component, is the large size of the hemispheric portions of the elephant cerebellum. The hemispheric portion of the cerebellum has a strong association with the dorsal thalamus and cerebral cortex and is involved in complex goal-directed limb movements, where it regulates movements requiring the integration of motor behaviors with cortical functions requiring learning and also with the control of eye movements (Stein and Glickstein,1992). The large size of the cerebellar hemispheres appears to be coordinated with an increase in size of the inferior olivary complex of elephants (Cozzi et al.,2001; Shoshani et al.,2006). The complexity of the trunk musculature (Endo et al.,2001), the learning of the use of the trunk in the infant elephant (Lee and Moss,1999), the complex use of the trunk for environmental manipulation in the adult elephant, the observation that the organization of the elephant retina reflects that they use trunk-eye coordination extensively (Pettigrew et al.,2010), and that the majority of the representation of the body surface in the hemispheric cerebellar cortex is that of the face in other mammals (e.g., Welker et al,1988; Bower,2011), all speak towards the control of the trunk, especially in cortically controlled learning of complex movements, as the main driving factor behind hemispheric enlargement in the elephant cerebellum.
Odontocete Cetacean Cerebellar Size
In odontocete cetaceans, the relative size of the total cerebellum has been found to be large compared to the total brain size (Marino et al.,2000), a finding confirmed and extended in the current study through the addition of more species. The cerebellar components (vermal and hemispheric) of the odontocete ceteaceans reveal an interesting story, with the vermis of one species being small, the other normal is size, and in both species studied the cerebellar hemispheres are large when compared to other mammals of similar brain size. The relatively small size of the vermal component of the harbor porpoise cerebellum and the standard size of the bottlenose dolphin vermis may be related to the lack of limbs and associated small size of the spinal cord (Ridgway et al.,1966). It would appear that the reduction in input and control of limb musculature may have led to the overall reduction in relative size of the vermal component of the odontocete cetacean cerebellum.
In contrast to the small vermal portion of the cerebellum, the hemispheric portions of the cerebellum are large in size in the odontocete cetaceans compared to the other mammals studied. As mentioned above, the cerebellar hemispheres are connected to the cerebral cortex via the pons and this circuit is thought to facilitate motor learning and online processing (Laforce and Doyon,2001). Thus, the increase in the size of the cerebellar hemispheres would suggest that odontocete cetaceans perform more motor learning than other mammals. Observations that contradict this conclusion include the lack of limbs and other mobile appendages, that odontocete cetaceans have mostly gross movements of the trunk and head (Madsen and Herman,1980) and a much reduced facial musculature (Caldwell and Caldwell,1972). The evidence that supports enhanced motor learning is that odontocete cetaceans control blowholes for respiration, vocalizations, and echolocation (Reidenberg and Laitman,2008), which requires extensive motor learning during development (McCowan and Reiss,1995; Fripp et al.,2005; Vergara and Barrett-Lennard,2008). It would be of interest to further the studies of cetacean cerebellar volumes across a broader range of species, specifically including some of the larger brained odontecete and mysticete species.
The authors thank Dr. Theo Nel and staff at Wits-DGMC, for the kind use of the MR scanner and their help during this project, Dr. Hilary Madzikanda of the Zimbabwe Parks and Wildlife Management Authority, and Dr. Bruce Fivaz and the team at the Malilangwe Trust, Zimbabwe.