Total neocortical cell number in the mysticete brain

Authors

  • Nina Eriksen,

    Corresponding author
    1. Research Laboratory for Stereology and Neuroscience, University of Copenhagen Bispebjerg Hospital, Copenhagen, Denmark
    • Research Laboratory for Stereology and Neuroscience, University of Copenhagen Bispebjerg Hospital, Copenhagen, Denmark
    Search for more papers by this author
  • Bente Pakkenberg

    1. Research Laboratory for Stereology and Neuroscience, University of Copenhagen Bispebjerg Hospital, Copenhagen, Denmark
    Search for more papers by this author

Abstract

The cetacean brain has long been of scientific interest, not only because of its large size – the largest in the animal kingdom – but also because of its high gyrification. It shows several adaptations to the aquatic environment, especially in the cortical arrangements of functional areas. To study structural aspects of the mysticete brain we estimated neocortical features in the common minke whale using stereological methods. The neocortex was surprisingly thick, equal to that in humans. The total neocortical neuron number was 12.8 × 109, and the total neocortical glia number 98.2 × 109. Total cell numbers in the auditory and visual cortex were also estimated, and showed that the auditory cortex contained more cells than the visual cortex. In this small sample, no sexual dimorphism was seen within the neocortex of the common minke whale. Our aim was to estimate the total cell number, cortical volume and cell density in the entire mysticete neocortex and compare the total cell number in the auditory cortex with that of the visual cortex using stereological methods. Here, we used the common minke whale as a model of all mysticetes. We wanted to compare these neocortical features to those of other mammals to forward understanding of the evolution of the mammalian brain. Anat Rec, 290:83–95, 2007. © 2006 Wiley-Liss, Inc.

The number of neurons and their relative abundance in different parts of the brain is at least partly a determinant of neural functions and consequently also a determinant of behavior. The neurons and glia in the neocortex in particular are involved in important brain processes, and it is assumed that neurons are somehow responsible for the evolution of intelligence, since phyla with larger brains and more neurons more easily adapt to environmental change with a greater range and versatility of behavior (Jerison, 1985). Cetaceans possess complex behaviors and social patterns, such as cultural transmission of vocal and behavioral traits (Payne and Payne, 1985; Rendall and Whitehead, 2001), and bottlenose dolphins have shown that they are capable of self-recognition (Reiss and Marino, 2001), cooperation, and tool use (Krutzen et al., 2005). Moreover, they also have the largest brain in the animal kingdom in absolute size, but not in relation to body size [here expressed as encephalization quotient, EQ (Jerison, 1973)]. Odontocetes (toothed whales) possess above-average encephalization levels compared to other mammals, second only to humans, whereas the encephalization level of the mysticete (baleen whales) is much lower, because of nonlinearities in EQ for very large animals (Marino, 2002, 2004). However, the large absolute size and high degree of cortical gyrification show that the mysticete brain has undergone substantial enlargement and elaboration during the course of evolution (Oelschläger and Oelschläger, 2002).

The cetacean brain differs from the terrestrial mammalian brain in many ways. It has a very high level of gyrification, and, as the rest of the whale body, it possesses several traits that show adaptation to water. The blowhole has migrated from the frontal to the parietal region, which makes it easier for whales to breath at the surface. This migration has changed the shape of the brain and the distribution of the cranial nerves (Oelschläger and Oelschläger, 2002). Cytoarchitectural organization does not resemble that of terrestrial mammals; cortical layer IV is absent or very poorly developed essentially guarantees that inputs and outputs and interneuronal connections are very different from other mammals (Glezer et al., 1988; Morgane and Glezer, 1990). Furthermore, layer I is far more cellular, layer II contains atypical neurons, and layer III contains very large pyramidal neurons (Glezer, 2002; Hof et al., 2005). The cortical arrangement of functional areas has also changed compared to terrestrial mammals. Whereas the primate brain contains large frontal lobes, the frontal region in the dolphin brain is very modest (Morgane et al., 1980). Compared to primates, it displays its own unique pattern of differentiation, but is distinctly laminated and comprises several cortical fields as in other lobes (Hof et al., 2005). Electrophysiological mapping studies have placed both the auditory and the visual cortices in the parietal region in dolphins and porpoises (Supin et al., 1978), whereas they are located in the temporal (auditory) and occipital (visual) regions in terrestrial mammals. The auditory region occupies broad areas of the suprasylvian gyrus, which adjoins the visual areas in the lateral gyrus and dorsal parietal regions. There is no intervening cortex between the auditory areas and the visual areas, or between visual-auditory areas and the sensorimotor areas (Glezer et al., 1988). Whether this is because cetaceans use sound as their primary communication channel is unknown. The auditory cortex is involved in elementary sound processing in mammals, and mammals therefore expectedly need a certain amount of neurons in the auditory cortex to be able to discriminate sound properly (Talwar et al., 2001). The auditory systems are smaller in mysticetes than in odontocetes, but the mysticete visual system contains more axons than that of the odontocete (Oelschläger and Oelschläger, 2002).

Cetacea have a completely different evolutionary history compared to terrestrial mammals: they returned to the aquatic environment about 50 million years ago (Gingerich et al., 1983). Both evolutionary lines have culminated in large complex central nervous systems with highly developed neocortices, and comparison should help to unravel some of the basic mechanisms essential for such convergent development. This investigation studies the mysticete neocortex; here we use the common Minke whale (Balaenoptera acutorostrata) as a model of all mysticetes. The Minke whale is a small baleen whale and grows to about 7–9 m long, weighing about 6–7.5 tonnes. Females are about 0.6 meter longer than males.

Some measurements of cetacean neuronal density have been reported (Tower, 1954; Oelschläger and Oelschläger, 2002), but no actual total cell numbers have previously been estimated. Thus, our aim was to estimate the total neocortical cell number, cortical volume, and cell density in the common Minke whale in the entire neocortex and to estimate and compare the total cell number in the auditory cortex with that of the visual cortex using stereological methods.

MATERIALS AND METHODS

Five brains from common Minke whales were used in this study. Due to their length, they were all estimated to be mature animals (Christensen, 1980, 1981). They were immersion-fixed in 10% formalin and kept in our brain bank for 1–2 years with origin from the Barents Sea in Norway. We used a single hemisphere from each brain, taking the best preserved from each specimen covering both sexes (Table 1).

Table 1. The material
 Animals
Total length (cm)802780771756678
SexFemaleFemaleMaleMaleMale
Hemisphere intactRightLeftLeftRightLeft
Hemisphere weight (g)1142107911191033974

Ethics

The brains were originally collected during commercial whaling, and CITES (Convention on International Trade in Endangered Species) guidelines for importing animal material were followed.

Cortical Mapping

It was assumed that the auditory and visual cortices of mysticetes follow the same structure as in odontocetes. Therefore, primary auditory and visual cortices (referred to as auditory and visual cortices) were delineated according to the mapping of primary cortices in bottlenose dolphins (Tursiops truncatus) (Morgane et al., 1990; Glezer, 2002; Oelschläger and Oelschläger, 2002) and the morphology of the balaenopterid brain (Breathnach, 1955; Pilleri, 1966a, 1966b). In our definition, the auditory cortex is located at the suprasylvian gyrus, while the visual cortex is placed next to the auditory cortex on the lateral gyrus, as are the primary cortices in odontocetes. The deep end of the gyrus was used to delineate the anterior and posterior cortical borders. The two subregions were delineated on the pial surface in different colors to distinguish the different regions after embedding (Fig. 1A).

Figure 1.

Tissue processing. A: On one hemisphere, the functional cortical areas were delineated on the pial surface with ink, where blue indicates the auditory cortex, and red the visual cortex. B: The hemisphere was cut into 2.5 cm thick blocks that were C: embedded in paraffin and D: cut exhaustively into 40 μm thick coronal sections on a Leica microtome using a wetted filter paper to allow collection of the section without the use of water baths. E: By gently pressing a printing roller on the paper side with the section side down F: the section was then mounted on a microscope slide.

Stereological Design

A series of parallel sections was cut at 2.5 cm intervals through the formalin-fixed hemisphere (Fig. 1B). The unbiased Cavalieri principle of point counting was used to estimate the shrinkage of the tissue area from formalin fixation to paraffin. The areas of the cut surfaces of the formalin-fixed sections (∑area) were estimated using a point grid, which has a known area associated with each point, a(p). The sum of points, Pi, on each section falling on the cut surface of the cerebrum was counted, and the area of the cerebrum at that position estimated from

equation image

The blocks were then dehydrated and embedded in paraffin in a LeicaASP300 tissue processor, and due to their large size the blocks were cut in half (Fig. 1C). The brain was cut coronally into parallel sections of equal thickness of 40 μm using a Leica microtome, resulting in a total number of about 3,000 sections from each hemisphere. A wetted filter paper was placed on the paraffin block allowing collection of the section without using water baths (Fig. 1D). When the section stuck to the filter paper, it was mounted on a double-silane-coated microscope slide by placing the paper with the section down on the slide, pressing gently on the paper side with a printing roller (Fig. 1E and F). This resulted in a complete absence of artifacts stemming from tissue deformation usually taking place in water baths, one consequence being that tissue section height is ∼ 40 μm postprocessing. Using a random start within the sampling period, a predetermined fraction of these sections was systematically sampled, dried for a minimum of 1 hr at 40°C, and subsequently stained with a modified Giemsa, which stains neuron and glial cell bodies (Fig. 2A).

Figure 2.

The optical fractionator. A: Sections were sampled systematically with a predetermined fraction, the section sampling fraction (ssf), with a random start and stained with a modified Giemsa stain. B: Optical disectors were positioned systematically random on each of the sections C: and cells were counted at final magnification of 1,900×. Cells were counted when inside the counting frame and when hitting the green inclusion line, providing they did not touch the red exclusion line. The large counting frame was used for neurons and the small counting frame for glial cells. The area of the counting frame of the disector represents the area sampling fraction (asf) for the x,y steps. D: Cell nuclei were counted by moving the counting frame through the tissue in an optical plane inside the section. The height of the disector, h, to the tq-weighted section thickness represents the height sampling fraction (hsf).

Areas of the paraffin sections of the selected positions were estimated again also using point counting, and the shrinkage was calculated from

equation image

The shrinkage of the surface was less than the shrinkage of volume, since volume shrinks in three directions, whereas surface shrinks in only two directions. Surface shrinkage was calculated as

equation image

Counting Procedure

A combination of two stereological principles, the optical disector and the fractionator sampling design, the so-called optical fractionator method was applied (Gundersen, 1986). This technique was ideal for this study because it is less affected by the considerable and unpredictable shrinkage during fixation of the brain. The optical fractionator method involves counting particles with optical disectors in a systematic, uniform, random sample that constitutes a known fraction of the structure to be analyzed. This is achieved by counting cells on a known fraction of sections (ssf; Fig. 2A), under a known fraction of the sectional area of the region (asf; Fig. 2B and C) in a known fraction of the thickness of a section (hsf; Fig. 2D). The total neocortical cell number was found by multiplying the total number of counted particles, ∑Q, by the reciprocal sampling fractions, adding a multiplication of 2 for bilateral number:

equation image

The optical disector, a 3D probe, was placed randomly on the sections generated with the aid of a microscope and an oil immersion objective, in which it was possible to observe thin focal planes inside thick sections (Fig. 2B–D). A counting frame with exclusion and inclusion lines was superimposed on the magnified image of the tissue on a monitor (Fig. 2C), and the orientation in the z-axis was measured with a digital microcator with a precision of 0.5 μm. The purpose of the exclusion and inclusion lines of the counting frame is to exclude edge effects arising from subsampling (Gundersen, 1977, 1978). Similarly, upper and lower guard zones protect the counting frame to prevent bias as a consequence of loss of cells close to the section surfaces. All cells that came into focus within the counting frame were counted as the focal plane was moved through the section in a predetermined height, h (Fig. 2D). All cells were counted in a microscope array with a BX-50 Olympus microscope with a 60× oil-immersion objective (1,900× magnification), a motorized stage, a Heidenhain microcator, and a corresponding camera and computer running CAST-GRID software (Visiopharm, Denmark).

Counting Criteria

Cells were identified as neurons if they had a clearly defined nucleus with a pale cytoplasm and a dark centrally located nucleolus, including a relatively large size of the cell body (Fig. 3). The nucleolus was used as the counting item, and the cell was counted only if the nucleolus was in focus inside the counting frame and the disector height. Glial cells were defined as much smaller than neurons, without a clear identifiable nucleolus and cytoplasm (Fig. 3). No differentiation was made between astrocytes, oligodendrocytes, or microglia. Endothelial cells were easily recognized by their dark and elongated nucleus and excluded from all counts. Cells were counted in all layers of cortex from the pial surface to the gray/white transition (Fig. 4). The neocortical boundary from gray/white transition was determined by the following method: presence of at least 15 neurons at magnification = 65× was considered to belong to neocortex. Layer I was always included in neocortex even though fewer than 15 neurons were present.

Figure 3.

Minke whale brain cells. Note the large size difference between neurons (black arrow) and glial cells (red arrow), and the distinct nucleolus in the neuron. Magnification = 1,900×.

Figure 4.

Cytoarchitecture in Minke whale neocortex. A: Suprasylvian gyrus corresponding to primary auditory cortex. B: Lateral gyrus corresponding to primary visual cortex. Roman letters indicate layers; wm, white matter. Magnification = 65×.

Sampling Fractions

To ensure that all parts of the neocortex were uniformly sampled, the design was based on systematic uniform random sampling (SURS). Based on a pilot study, the ssf, asf, and hsf in the z-axis were estimated and optimized for statistical reasons to obtain a count of approximately 150–200 neurons and 150–200 glial cells per sample, which previously has provided an optimal precision compared to the biological variation (Pakkenberg and Gundersen, 1988, 1997; Korbo et al., 1990).

Section sampling fraction

Depending on the size of the brain, the ssf was chosen to be between 1/260th to 1/290th to obtain 12–13 sections per brain. The first section was selected randomly using a random number from 1–260 or 1–290, and the following sections were sampled systematically. In case of tissue artifacts due to preparation or cutting, the next section was sampled without altering the systematic sampling in the following sections.

Area sampling fraction

In each sampled section, neuron or glial cell counts were made with optical disectors at regular predetermined x,y positions in the neocortex. The first disector was positioned randomly over cortex within the x,y step pattern by the CAST-GRID software. The area of the counting frame, a(frame), of the disector was known relative to the area associated with each step in the x,y direction, a(x,y step). The area sampling fraction, asf, was calculated as

equation image

A small counting frame was used for glial cell counting, because of the high number of cells.

Height sampling fraction

The height of the disector should be known relative to the section thickness. In this study, we applied guard zones of 5 μm at the top and around 14 μm at the bottom, and the fixed height, h, of the disector was 21 μm. To compensate for deformation of the section thickness, hsf is equal to

equation image

for

equation image

where ti is the local section thickness centrally in the ith counting frame with a disector count of qi (Dorph-Petersen et al., 2001). The stereological parameters used in this study are shown in Table 2.

Table 2. Stereological parameters
Stepsize
 BA (μm)1/ssf∑secta(frame) (μm2)x,y (μm)1/asfdishdis (μm)tq− (μm)1/hsfQ
Neurons
 Neocortex403061144931500050074.41255–2541.42.0215
 Auditory cortex4027011449336002663.7965–2540.51.9167
 Visual cortex402708449324001553.51065–2540.21.9177
Glial cells
 Neocortex403061167015000335820.91255–2541.42.0247
 Auditory cortex4027011670360019343.3965–2540.51.9192
 Visual cortex402708670240085971065–2540.21.9210

Volume and Surface Estimation

The systematic uniform random placements of the disectors were also used to estimate neocortical volumes in accordance to the Cavalieri estimator:

equation image

for ∑P being the number of upper-corner points hitting the neocortical tissue, a(p) is the x,y step area, t is the thickness of the section, and k is the inverse section sampling fraction. The volume was corrected for shrinkage (as previously described) to obtain the fixed volume of the neocortex.

Cortical surface area was estimated using a line-point grid. The intersections between grid lines and the inner and outer surface of neocortex, ∑I, were counted as well as the number of grid points on each section overlying neocortical tissue, ∑P.

The surface of neocortex was then calculated from

equation image

where 2 is a constant, and l(p) is the unit test line length per point.

It was assumed that there was isotropy and no differential shrinkage in the surface; accordingly, the estimations of cortical thickness and cortical surface area are not unbiased estimates.

The thickness of cortex is estimated from the volume and surface estimates of neocortex:

equation image

Statistical Analysis

The precision of the estimate of the total volume from each sample was estimated as the coefficient of error (CE). CE is a function of the noise effect, also known as the point counting variance, and the SURS variance for sums of areas, ∑a (Gundersen et al., 1999). The noise effect is the uncertainty that comes from point counting, ∑P:

equation image

where n is the number of sections, b/√a is the average profile shape (found by eyeballing the nomogram from Gundersen and Jensen, 1987).

SURS variance for the sum of areas is the uncertainty of sampling between sections, because repeated estimates based on different sections may vary. It was calculated from

equation image

where Pi is the number of points counted on one section, and Pi+1 is the number of points counted on the next section (Gundersen et al., 1999). The total sampling variance, CE(∑P), is estimated from

equation image

The sampling is considered optimal when CE is about half the coefficient of variation (CV = SD/mean), since CV2 = biological CV2 + estimated CE2. Then the real biological variability (biological CV2) contributes most to the observed relative variance. In the estimation of CE in the optical fractionator counting particles, noise effect is equal to ∑Q.

A Student's t-test was used to test for differences between sexes in the neocortex, auditory and visual cortices, and for differences between auditory and visual cortex, applied with a significant level of 0.05. One-way ANOVA test was performed to test for differences in density of cells between the different cortical regions. Spearman's correlation coefficient was used to test for correlation between sex and brain size, and between total body length and brain size.

RESULTS

The neocortex of the Minke whale was thick (Table 3) and characterized by a thick layer I and a dense layer II. Layers III–VI are less distinguishable with large pyramidal neurons (Fig. 4). Overall, the Minke neocortex resembles that of other cetaceans. The gyrification level (= surface area/brain weight) of the Minke whale was 2.2.

Table 3. Major structural components of the Minke whale brain
 NeocortexCE (CV)Auditory cortexCE (CV)Visual cortexCE (CV)
  • *All total quantities are bilateral. P-values are from the sexual dimorphism tests.

  • *

    Statistically significant.

Mean neuron number, 109
 F14.00.06 (0.12)0.520.08 (0.20)0.170.08 (0.15)
 M12.00.07 (0.22)0.430.08 (0.41)0.270.08 (0.10)
 All12.80.07 (0.10)0.470.08 (0.21)0.230.08 (0.27)
 p-value0.42 0.58 0.46 
Mean glial cell number, 109
 F97.70.06 (0.27)4.810.07 (0.08)1.590.07 (0.17)
 M98.60.07 (0.20)3.120.08 (0.05)1.840.09 (0.06)
 All98.20.07 (0.17)3.880.07 (0.25)1.740.08 (0.14)
 p-value0.68 *0.05 0.23 
Mean density of neurons, 106/cm3
 F7.60.07 (0.19)6.700.08 (0.20)6.830.08 (0.16)
 M8.30.07 (0.07)9.630.08 (0.18)11.080.08 (0.12)
 All8.00.07 (0.11)8.450.08 (0.27)9.370.08 (0.26)
 p-value0.45 0.20 *0.04 
Mean density of glial cells, 106/cm3
 F53.30.06 (0.33)6.280.07 (0.10)6.310.07 (0.18)
 M68.80.07 (0.16)7.440.08 (0.21)7.640.09 (0.03)
 All62.60.07 (0.24)6.980.07 (0.09)7.110.08 (0.14)
 p-value0.68 0.40 0.20 
Mean volume, cm3
 F18570.04 (0.07)76.960.04 (0.17)25.170.04 (0.01)
 M14660.05 (0.28)43.420.07 (0.24)24.070.06 (0.08)
 All16220.05 (0.11)56.900.06 (0.33)24.510.05 (0.01)
 p-value0.30 *0.05 0.50 
Mean surface area, cm2
 F65630.05 (0.01)  
 M54790.04 (0.14)  
 All59120.05 (0.08)  
 p-value0.50     
Mean cortical thickness, mm
 F2.830.05 (0.01)  
 M2.750.04 (0.14)  
 All2.780.05 (0.08)  
 p-value0.80     
Glial cell/neuron ratio
 All7.7/1 8.2/1 7.6/1 

Cell Numbers

The mean neocortical neuron number in the common Minke whale was 12.8 × 109 and the glial cell number was 98.2 × 109. For the auditory cortex, the total neuron number was 4.65 × 108 and the total glial cell number was 3.80 × 109; 2.28 × 108 neurons were found in the visual cortex, whereas the total glial cell number was equal to 1.74 × 109 (Table 3). The total number of cells in each region (entire neocortex, auditory and visual cortices) for each individual as a function of its hemispherical weight and total body length is shown in Figure 5. Other measurements of the mysticete neocortex, such as total neocortical volume and cell density, are shown in Table 3. The mean percent shrinkage from formalin-fixed brain tissue to paraffin was 50.5% and was used to correct for cortical volume and surface area shrinkage.

Figure 5.

Total neocortical cell number in the Minke whale. A: Neuron and glial cell number as a function of hemisphere weight and total body length of the animal in the entire neocortex. B: Neuron and glial cell number as a function of hemisphere weight and total body length in the auditory cortex. C: Neuron and glial cell number as a function of hemisphere weight and total body length in the visual cortex.

Figure 6 shows comparison of different mammals and their neocortical features, reporting only total number or density obtained using stereological methods. Estimates have been obtained for the following: rat (Rattus norvegicus) (Mooney and Napper, 2005), pig (Sus scrofa domisticus), Göttingen minipig (Sus scrofa göttingen) (Jelsing et al., 2006), rhesus monkey (Macaca mulatto; data not shown), and human (Homo sapiens) (Pakkenberg and Gundersen, 1997; Pelvig et al., 2003). Minke whales have more neurons than small mammals, but less than humans (Fig. 6A), whereas they have more glial cells (Fig. 6B). The density of neurons is lower in the common Minke whale (Fig. 6C), but the glial cell density is about the same as in humans, though lower than other mammals (Fig. 6D). The Minke neocortical volume is much higher than any of the other mammals listed above (Fig. 6E). Figure 7 shows neocortical features of cetaceans, including fin whale (Balaenoptera physalus) (Tower, 1954) and humpback whale (Megaptera novaeanligae) (Oelschläger and Oelschläger, 2002), bottlenose dolphin (Tursiops truncatus), harbor porpoise (Phocoena phocoena), common dolphin (Delphinus delphis) (Ridgway and Brownson, 1984; Haug, 1987; Oelschläger and Oelschläger, 2002), and Stenella genus (saddleback and spinner dolphins) (Ridgway and Brownson, 1984). It should be noted that cetacean data are not directly comparable with our data because of differences in tissue processing and measurements. The neocortical neuron density is about the same for mysticetes, but higher in small-toothed whales (Fig. 7A), whereas the glial cell density is about the same (Fig. 7B). The surface area increases with increasing brain weight (Fig. 7C), and the gyrification level of neocortex in cetaceans is about the same, whereas it is lower in humans (Fig. 7D).

Figure 6.

Comparison of neocortex in mammalian brain. A: Neocortical neuron number. B: Neocortical glial cell number. Note that the larger the brain, the more glial cells. C: Neuron density. D: Glial cell density of different mammals. E: Volume of neocortex.

Figure 7.

Comparison of neocortex in cetacean brain. A: Neuron density. B: Glial cell density. C: Surface area. D: Gyrification level in cetaceans and humans. Note that gyrification level is about the same in cetaceans, but lower in humans.

Sexual Differences

Data were tested for sex differences, and brain weight was correlated to sex (r = 0.99), with females having the larger brain, and total body length (r = 0.7). All values from the following tests in the different cortical regions are shown in Table 3.

Entire neocortex

No significant differences were found in the entire neocortex, neither in cell number nor in cell density or volume of neocortex.

Auditory cortex

Sexual differences were not found in neuron number, but females had significantly more glial cells than males. No significant differences were found in cell density, but females had a significantly larger volume of the auditory cortex than males.

Visual cortex

No significant differences were found between sexes in the visual cortex in either total neuron number or glial cell number. Males had a significantly different higher neuron density than females, but no differences were found in glial cell density or volume of cortex.

Functional Cortices

No significant difference was seen in cell density in the different cortices (neurons: P = 0.68, glial cells: P = 0.94). However, the auditory cortex had significantly more cells than the visual cortex (neurons: P = 0.01, glial cells: P = 0.001). The auditory cortex also had the largest glial cell-to-neuron cell ratio (Table 3).

DISCUSSION

The Minke whale neocortex has the same cytoarchitecture as other cetaceans (Glezer et al., 1988, 2002; Morgane and Glezer, 1990; Hof et al., 2005), and there was no obvious difference in cytoarchitecture between primary auditory and visual cortices. Not only is the Minke whale neocortex highly gyrified, it is also thicker than most mammals, probably because the neocortical cell number is high, especially the glial cell number. The auditory cortex contains more cells than the visual cortex, but in this relatively small sample, there seems to be no obvious sexual dimorphism in the neocortex.

The total neocortical cell number in the common Minke whale was estimated to be 12.8 × 109 neurons and 98.2 × 109 glia. It is approximately 2/3 of the human neuron number, but 13 times that of rhesus monkeys and around 500 times that of rats.

The Minke whale has the highest number of glial cells in the neocortex seen in any mammal studied to date. The ratio of neocortical glial cells to neocortical neurons is species-specific and therefore varies among the mammalian groups and during ontogenesis due to changes in neuron density (Oelschläger and Oelschläger, 2002; Nishiyama et al., 2005). The glia/neuron ratio is 7.7/1 in Minke whales, 1.4/1 in humans, and 2/1 in rats. Few ratios have been reported for cetaceans: Tursiops ∼ 3/1, fin whales ∼ 5/1 (Oelschläger and Oelschläger, 2002), which shows that there is a tendency for glia/neuron ratio to increase with increasing brain mass. Consequently, the glia/neuron ratio signals the importance of glia for growing neurons, and thus for neocortical function, and they probably play a crucial role in cetaceans seen in the high ratio of glia. Since the Minke neocortical neurons seem to be very large, it is also possible that large neurons may need a higher number of glia than smaller ones. Though the Minke whale possesses fewer neurons than humans, it still has more than other mammals. This, along with the very high number of glia, means that whales evolved a substantially larger number of cells in their neocortex than other nonhuman mammals, even larger than lower primates such as the rhesus monkey. This could possibly be an adaptation to maintenance and control of large bodies as well as a consequence of a complex social system or due to the aquatic environment, as suggested for odontocetes by Marino et al. (2005).

The Minke whale density of neurons in the neocortex is lower than in other mammals, and neuron density declines with increasing brain size, as was earlier hypothesized (Tower, 1954; Machail, 1982; Haug, 1987). Since neurons need axonal connectivity to neurons in other parts of the brain, neuron density cannot exceed below a certain value. In Figure 7A, it is shown that the neuron density in Minke whales is about the same as in fin whales, which are twice as large. Therefore, it is possible that the lower limit for neuron density is around 60–85 × 106/cm3, at least for larger cetaceans. The glial cell density, however, was earlier thought to be independent of brain mass (Haug, 1987), but as is shown in Figures 6D and 7B, it seems to be fairly stable around 60–80 × 106/cm3, disregarding the rat.

Prothero and Sundsten (1984) hypothesized that there is an upper limit for brain mass, and as the brain mass increases, so does the cortical surface, resulting in an increase in gyrification (gyral-window hypothesis). Hence, an increase in brain mass and cortical surface is only possible by increasing the cortical fissures and thickening the neocortex. The Minke neocortex is thicker (2.78 mm) than most mammalians', including that of odontocetes (Tursiops 1.2–1.76 mm) (Morgane et al., 1980; Ridgway and Brownson, 1984), but it is about the same thickness as the human neocortex (2.63 mm) (Pakkenberg and Gundersen, 1997). This does not support the gyral-window hypothesis, as a large brain would be expected to have a thicker cortex than a smaller brain. It is seemingly important not to increase cortical thickness above a certain limit. If volume grows and cortical thickness is kept stable, cortical surface must then increase in area. Therefore, not surprisingly, as brain size increases, so does neocortical volume, with increase of surface area and cortical fissures in the large cetacean brain. The Minke neocortex is highly gyrified; the level of gyrification is about the same in cetaceans, with a tendency for dolphins to be a bit larger, but it is higher than in humans. That part of the gyral-window hypothesis holds for the cetacean brain; nevertheless, it might also be an aquatic trait instead of only a “large brain” trait.

The auditory cortex contained more cells and had a larger volume than the visual, and the highest glia/neuron ratio was seen in the auditory cortex. This is not surprising as cetaceans are much more dependent on sound than on vision, as the aquatic environment has poor visibility. They use sound as their primary communication channel; the aquatic environment is very conductive to sound: sound travels about five times faster in water than in air (Urick, 1973). Minke whales, like all mysticetes, are assumed to use sound for communication (Winn and Perkins, 1971; Mellinger et al., 2000). Together with other balaenopterids (fin, blue, and humpback whales), they use sequences of sounds in a hierarchical organization in which individual utterances are produced in a regular pattern that is repeated at regular intervals. Possibly, the mysticete auditory structure is relatively complex compared to other mammals, which are less dependent on sound communication. However, we only measured the primary auditory cortex, which is only 3.5% of the entire neocortex. Possibly there are other neocortical areas of higher-level integrative processing that employ information from the primary auditory cortex. Since Minke whales do not echolocate, as do odontocetes, their auditory cortex is most likely less developed than would be expected of the odontocete auditory cortex. Mysticetes do use their vision to a greater extent than odontocetes (Oelschläger and Oelschläger, 2002), which means that they may have a more complex visual system. However, in a recent study, Poth et al. (2005) found decreasing neuron number per neocortex unit (the number of perikarya below a defined area of cortex) with increasing brain mass in sensory neocortices of delphinids, which seem to be equally reliable on audition. More knowledge on visual and auditory structures is needed to address this topic.

As mysticetes have reverse sexual dimorphism, it is unsurprising that the female brain is also larger; however, it is surprising that female Minkes do not possess more brain cells than males despite their larger brain size. This is in contrast to humans, where males have larger brains and more neurons than females (Pakkenberg and Gundersen, 1997). Additionally, there was no sexual dimorphism in cell density, volume, or thickness of the Minke neocortex. There was, however, some sexual dimorphism in the functional areas of the neocortex, since females had a larger volume of auditory cortex and more glia, and in the visual cortex males had a larger neuron density. The larger volume of auditory cortex in females is most likely because of the larger brain size. The higher neocortical neuron number in males in the visual cortex compared to females is most probably a result of the low sample size. Because of the high biological variation, all of these findings need to be replicated in a larger sample.

In conclusion, the Minke whale neocortex is complex; it has more neocortical neurons than other mammals, but fewer than humans, despite its much larger body size. Minkes possess many more glial cells in their neocortex than do other mammals. They have a large neocortical volume and surface area, with a neuron density as that of larger mysticetes, but less than other mammals, including small odontocetes. There was no sexual dimorphism in either the neocortex or the functional areas. The auditory cortex is larger than the visual cortex and contains more cells, as expected from the acoustic behavior of the Minke, possibly because of an adaptation to the aquatic environment. The evidence of considerable complexity in the mysticetes' neocortex is consistent with their behavioral and social complexity.

Ancillary