The Cerebral Cortex of the PygmyHippopotamus, Hexaprotodon liberiensis

(Cetartiodactyla, Hippopotamidae):MRI, Cytoarchitecture, and Neuronal

MorphologyCAMILLA BUTTI,1* R. EWAN FORDYCE,2 MARY ANN RAGHANTI,3 XIAOSI GU,1,4

CHRISTOPHER J. BONAR,5 BRIDGET A. WICINSKI,1 EDMUND W. WONG,6

JESSICA ROMAN,6 ALANNA BRAKE,1 EMILY EAVES,1,6

MUHAMMAD A. SPOCTER,7 CHEUK Y. TANG,6 BOB JACOBS,8

CHET C. SHERWOOD,9 AND PATRICK R. HOF1

1Fishberg Department of Neuroscience and Friedman Brain Institute, Icahn School ofMedicine at Mount Sinai, New York, New York

2Department of Geology, University of Otago, Dunedin, New Zealand3Department of Anthropology, Kent State University, Kent, Ohio

4Wellcome Trust Centre for Neuroimaging, University College London, London, UK5Dallas Zoo, Dallas, Texas

6Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, New York7Department of Anatomy, Des Moines University, Iowa, Iowa

8Laboratory of Quantitative Neuromorphology, Psychology, Colorado College, ColoradoSprings, Colorado

9Department of Anthropology, The George Washington University, Washington, DC

ABSTRACTThe structure of the hippopotamus brain is virtually unknown because

few studies have examined more than its external morphology. In view oftheir semiaquatic lifestyle and phylogenetic relatedness to cetaceans, thebrain of hippopotamuses represents a unique opportunity for better under-standing the selective pressures that have shaped the organization of thebrain during the evolutionary process of adaptation to an aquatic environ-ment. Here we examined the histology of the cerebral cortex of the pygmyhippopotamus (Hexaprotodon liberiensis) by means of Nissl, Golgi, and cal-retinin (CR) immunostaining, and provide a magnetic resonance imaging(MRI) structural and volumetric dataset of the anatomy of its brain. Wecalculated the corpus callosum area/brain mass ratio (CCA/BM), the gyren-cephalic index (GI), the cerebellar quotient (CQ), and the cerebellar index(CI). Results indicate that the cortex of H. liberiensis shares one featureexclusively with cetaceans (the lack of layer IV across the entire cerebralcortex), other features exclusively with artiodactyls (e.g., the morphologiyof CR-immunoreactive multipolar neurons in deep cortical layers, gyrence-phalic index values, hippocampus and cerebellum volumetrics), and otherswith at least some species of cetartiodactyls (e.g., the presence of a thicklayer I, the pattern of distribution of CR-immunoreactive neurons, thepresence of von Economo neurons, clustering of layer II in the occipital cor-

Grant sponsor: The James S. McDonnell Foundation; Grantnumber: 22002078 (to P.R.H., C.B., and C.C.S.) and 220020293(to C.C.S); Grant sponsor: National Science Foundation; Grantnumber: BCS-0827531 (to C.C.S).

*Correspondence to: Camilla Butti, Fishberg Department ofNeuroscience, Box 1639, Icahn School of Medicine at Mount

Sinai, One Gustave L. Levy Place, New York, NY 10029. Fax:1–646-537–9585. E-mail: butticamilla@gmail.com

Received 01 July 2013; Accepted 04 November 2013.

DOI 10.1002/ar.22875Published online 00 Month 2014 in Wiley Online Library(wileyonlinelibrary.com).

THE ANATOMICAL RECORD 00:00–00 (2014)

VVC 2014 WILEY PERIODICALS, INC.

tex). The present study thus provides a comprehensive dataset of the neu-roanatomy of H. liberiensis that sets the ground for future comparativestudies including the larger Hippopotamus amphibius. Anat Rec, 00:000–000, 2014. VC 2014 Wiley Periodicals, Inc.

Key words: Hippopotamids; cetaceans; artiodactyls; cytoarchi-tecture; neocortex; mammalian evolution

INTRODUCTION

Although all vertebrate brains appear similar at cer-tain stages of embryonic development (Puelles andRubenstein, 2003), developmental differences occurbeyond the embryonic period and their functional sig-nificance is still poorly understood. Vertebrates displaya tremendous variation in brain structure includingbrain size, degree of folding of the cerebral cortex, sizeof cortical fields (the number of which, however,remains rather comparable among species), and corticalorganization (Striedter, 2005). Thus, in the context ofbrain evolutionary studies, it becomes pivotal to under-stand how adaptation to different environments modi-fies brain structure. The clade Cetartiodactyla includesboth terrestrial (artiodactyls) and fully aquatic (ceta-ceans) species. Among artiodactyls, the hippopotamusesare the only species adapted to an amphibious lifestyle.Moreover, recent morphological and molecular evidencesuggests that modern hippopotamuses are the closestextant relatives of cetaceans (Gatesy, 1997; Shimamuraet al., 1997; Milinkovitch, 1998; Nikaido et al., 1999;Agnarsson and May-Collado, 2008; Spaulding et al.,2009; Zhou et al., 2011).

Comparing closely related species (i.e., hippopota-muses and cetaceans) is particularly important insofaras differences in brain structure may account for differ-ences in behavior. Although brain structures tend to bemore similar in close relatives than in distantly-relatedspecies, environmental variables (e.g., whether an ani-mal is terrestrial, amphibious, or fully aquatic) areimportant determinants for brain structure and func-tion. In this regard, the brain of other marine mammals(e.g., manatees and dugongs, and pinnipeds, polar bears,and otters) could certainly help elucidate which neuroa-natomical features might be linked to the aquatic envi-ronment. Nevertheless, it would be difficult todifferentiate between neural characteristics due to envi-ronmental adaptation from those due to phylogenetic dif-ferences. Consequently, hippopotamuses and cetaceansrepresent a key comparison point for the understandingof brain evolution during the transition, for a particulargroup of mammals (the cetaceans), from a terrestrialenvironment to an obligatory aquatic one. Their phyloge-netic relatedness, the differences in brain size and socialbehavior, as well as the diverse niches they occupy, allowfor exploration of the possible relationships betweencortical structure, phylogenetic constraints, and adapta-tion to the aquatic environment.

Although it is relatively clear that cetaceans splitfrom other artiodactyls around 55 million years ago, atthe start of the radiation of archaeocetes (archaic ceta-

ceans) that led to living whales and dolphins (Gingerich,1998, 2005; Thewissen et al., 2007; Uhen, 2010), hippo-potamids have a shorter fossil record of about 23 millionyears (Boisserie et al., 2011) and their evolutionary his-tory is contentious. Several different extinct groups ofearly terrestrial or semi-aquatics artiodactyls, includinganthracotheres and raoellids, have been suggested ashippopotamid ancestors (Boisserie, 2005; Spauldinget al., 2009; Boisserie et al., 2011; O’Leary et al., 2012).Because of uncertainty about the exact closest fossil rel-atives for both hippopotamids and cetaceans, it is notclear whether their most recent common ancestor wasterrestrial or semi-aquatic (Boisserie et al., 2011). Themodern family Hippopotamidae includes only two extantspecies: the river or common hippopotamus (Hippopota-mus amphibius) and the pygmy hippopotamus (Hexapro-todon liberiensis). Five subspecies of riverhippopotamuses are currently recognized based on skullshape and proportions (H. amphibius amphibius, H. a.tschadensis, H. a. kiboko, H. a. constrictis, and H. a.capensis; Lydekker, 1915; Grubb, 1993; Eltringham,1999) but whether these differ genetically or are theresult of intraspecific variability in skull charactersremains unclear (Lydekker, 1915; Eltringham, 1999).Two subspecies of pygmy hippopotamus have been recog-nized in the past, but the only remaining species is H.liberiensis liberiensis, which is restricted to riverineenvironments in Liberia, Guinea, Sierra Leone, andIvory Coast (Corbet, 1969; Eltringham, 1999). Thepygmy hippopotamus’ body size is six to eight timessmaller than that of H. amphibius, with adult H. liber-iensis having a body mass equivalent to a 6-month-oldH. amphibius (Eltringham, 1999). The head of thepygmy hippopotamus is also proportionally smaller andnarrower than that of the river hippopotamus and itsorbits are not raised above the skull roof, a feature thathas been linked to less-aquatic habits than its larger rel-ative (Eltringham, 1993b).

The organization of the brain of cetaceans is welldocumented at the gross anatomical level (Marino et al.2001a,b,c, 2002, 2003a,b, 2004a,b; Montie et al., 2007,2008; Morgane et al. 1980; Hof and Van der Gucht,2007; Oelschl€ager and Oelschl€ager, 2008), as well as thelevel of cyto- and chemoarchitecture (Jacobs et al. 1971,1979, 1984; Morgane et al., 1982, 1988; Glezer and Mor-gane, 1990; Glezer et al., 1992, 1993, 1998; Hof et al.,1999; Hof and Sherwood, 2005; Manger et al., 1998; Hofet al., 2005; Hof and Van der Gucht, 2007; Furutani,2008). In contrast, very little is known about the organi-zation of the brain of hippopotamuses apart from itsgross anatomy (Garrod, 1880; Pilleri, 1962). Such a lack

2 BUTTI ET AL.

of information is probably due to both the restricted geo-graphic distribution of the extant hippopotamuseswithin the African continent, and the small numbers ofindividuals comprising the worldwide population (esti-mated at 157,000 river hippopotamuses and a few thou-sand, at most, pygmy hippopotamuses; Eltringham,1993a,b) that make it difficult to obtain brain speci-mens. This is particularly true for the pygmy hippopota-mus given its “Endangered” status in the InternationalUnion for Conservation of Nature (IUCN, 2013) redlist (IUCN Red List of Threatened Species, Version2013.2).

Here we describe the brain of H. liberiensis usingmagnetic resonance imaging (MRI) and a variety of his-tological approaches, including Nissl and Golgi staining,and immunohistochemistry. Moreover, we provide volu-metric measurements of salient structures from MRimages, as well as the values for the gyrencephalic index

(GI), the cerebellar quotient (CQ), and the cerebellarindex (CI). Finally, we report new findings on the distri-bution of von Economo neurons (VENs) first observed inthis species in the insular cortex (Butti and Hof, 2010).Comparisons with cetaceans and other artiodactyls aremade, when possible, to eludicate the evolutionary sig-nificance of certain neuroanatomical features.

MATERIALS AND METHODS

Specimen and Histological Preparation

The brain of a 33-year-old female pygmy hippopota-mus (brain mass: 262 g; body mass: 284 kg; Fig. 1A,B)that died of natural causes at the Cleveland MetroparksZoo was collected during necropsy, 4 hr after death, andpostfixed by immersion in 4% paraformaldehyde in 0.1M phosphate buffer saline (PBS). The brain was thenhemisected. After MR scanning (see the Magnetic Reso-nance Imaging section for details), the right hemispherewas cut into serial slabs, cryoprotected in graded sucrosesolutions up to 30% in PBS, frozen in dry ice, and proc-essed serially into 80-mm thick sections on a slidingmicrotome (Leica Biosystems, Nussloch, Germany).Every 10th section was mounted on glass slides, stainedwith a solution of 0.2% cresyl violet, and coverslippedusing DPX for examination. An additional adjacentseries of sections was used for immunohistochemistry(see the Immunohistochemistry section for details). Sep-arate blocks including the frontal magnocellular cortex,the cortex of the lateral gyrus and the anterior cingulatecortex (ACC) were sampled from the left hemisphereand stained with a modified rapid Golgi impregnationmethod (see the Golgi Staining section for details). Allphotomicrographs were obtained on a Zeiss Axiophotphotomicroscope equipped with Plan-Neofluar objectives2.53 (N.A. 5 0.075), 403 (N.A. 5 0.75), Plan-Apocromatobjectives 103 (N.A. 5 0.32) and 203 (N.A. 5 0.8) (Zeiss,Thornwood, NY), and an Optronics MicroFire digitalcamera (Optronics, Goleta, CA). Images were adjustedfor luminosity and contrast using Adobe Photoshop anddrawings were made using Adobe Illustrator.

Immunohistochemistry

Sections were washed in phosphate buffer saline(PBS) and treated for antigen retrieval in a solution ofcitrate buffer (pH 8.0–9.0) in a 95�C water bath for 10min. After cooling for 20 min at room temperature, thesections were rinsed in PBS and treated for 20 min in asolution of 0.3% hydrogen peroxide in PBS to eliminateendogenous peroxidase activity. After washes in PBScontaining 0.3% Triton-X100, the blocking step was per-formed for 65 minutes in a solution of 5% normal goatserum, 2% bovine serum albumin, and 0.2% cold waterfish gelatin in PBS containing 0.3% Triton-X100. Thesections were then incubated overnight, at room temper-ature on a shaker, in a solution of the primary monoclo-nal mouse anti-calretinin (CR) antibody (Swant,Bellinzona, Switzerland; dilution 1:6,750; Schwalleret al., 1993) diluted in PBS containing 0.3% Triton-X100.Sections were then washed in PBS containing 0.3%Triton-X100 and incubated in a solution of the secondarypolyclonal goat anti-mouse biotinylated antibody (Dako,Glostrup, Denmark; dilution 1:400) for 60 min. Afterwashes in PBS containing 0.3% Triton-X100 the sections

Fig. 1. Macroscopic view of the brain of the pygmy hippopotamus(H. liberiensis). Lateral (A) and midline (B). Outline of the perimeter ofthe corpus callosum used for the volume estimation (C). Cb, cerebel-lum; cc, corpus callosum; Cing, cingulated gyrus; Cor, coronal gyrus;ES, ectosylvian gyrus; Fr, frontal gyrus; fx, fornix; Lat, lateral gyrus;Ling, lingual gyrus; OB, olfactory bulb, Orb, orbital gyrus; Par, parietalgyrus; Pro, gyrus proreus; PS, perisylvian gyrus; Psplen, posteriorsplenial gyrus; Rec, gyrus rectus; Sant, anterior Sylvian gyrus; Sig,sygmoid gyrus; Spost, posterior Sylvian gyrus; SS, suprasylvian gyrus;Ssplen, suprasplenial gyrus; Subc, subgenual cortex; Th, thalamus;TP, temporal gyrus. Scale bars 5 5 cm.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 3

were processed with the avidin-biotin peroxidase methodusing the Vectastain ABC kit (Vector Laboratories, Bur-lingame, CA) for 60 min. The 3,3’-diaminobenzidine(DAB) peroxidase kit (Vector Laboratories, Burlingame,CA) was then used as a chromogen to visualize the prod-uct of the reaction. The immunoreactivity was intensi-fied using nickel. After washes in a solution of 50% PBSand 50% dH2O, the sections were mounted on glassslides, dehydrated in graded alcohol solutions up to100%, immersed in a clarifying agent (limonene) andcoverslipped using 100% DPX for examination.

The monoclonal anti-CR antibody was produced inmice by immunization with recombinant human CR 22kD (Schwaller et al., 1993; Zimmermann and Schwaller,2002) and reacts specifically with CR in tissue fromhuman and rat with no cross-reaction with othercalcium-binding proteins (Swant Monoclonal anti-CR6B3 product description). Immunoreactivity to CR isobserved mostly in non-pyramidal neurons and is colo-calized with the major inhibitory neurotransmitter g-aminobutyric acid (GABA). Thus, the CR-immunoreactive neurons represent inhibitory interneur-ons (Rogers, 1992; Andressen et al., 1993; DeFelipe,1993; Cond�e et al., 1994; Kubota et al., 1994; DeFelipe,1997; Gonchar and Burkhalter, 1997). Although this CR-antibody reveals specific reactivity with CR in tissuefrom human and rat, the highly conserved sequence ofthe CR protein in evolution, from bacteria to mammals(Parmentier and Lefort, 1991), makes it possible to dem-onstrate immunostaining specificity of this antibodywith interneurons in many other species including birds,reptiles, and mammals (Rogers, 1989; Baimbridge et al.,1992; Martinez-Guijarro and Freund, 1992; R�esibois andRogers, 1992; Glezer et al., 1993, 1998; Hof et al.,1996a,b, 1999; Davila et al., 1997; DeFelipe, 1997; Sher-wood et al., 2009). This antibody has been previouslyused to study GABAergic interneurons of cetaceans andartiodactyls (Glezer et al., 1993, 1998; Hof et al., 1999),revealing staining patterns consistent with thoseobserved in other mammalian species.

Golgi Staining

Tissue blocks from the frontal magnocellular cortex(likely corresponding to motor cortex), from the cortex ofthe lateral gyrus (likely corresponding to the visual cor-tex), and from the ACC, were removed from the lefthemisphere, trimmed to 3 to 5 mm in thickness, andprocessed by a modified rapid Golgi technique (Scheibeland Scheibel, 1978). Briefly, after being immersed in a0.33% osmic acid solution in the absence of light for 72hr at room temperature, the osmicated tissue wasplaced in a 0.75% silver nitrate solution for 24 hr. Theblocks were then cut into 120-mm thick sections using avibratome (752M Vibroslice Campden Instruments, IN)and coverslipped with Permount for examination. Trac-ings of neurons in three dimensions, as previously doneby the authors in the African elephant (Loxodonta afri-cana; Jacobs et al., 2011), could not be performedbecause relatively poor impregnation prohibited tracingdendrites in their entirety. However, the Golgi-impregnated materials allowed for a qualitative descrip-tion of neuronal morphologies based on cortical and lam-inar localization, soma size, and dendritic arborcharacteristics.

Magnetic Resonance Imaging

MRI data were obtained using a 3 T Siemens AllegraMRI system at the Icahn School of Medicine at MountSinai. A high-resolution T2-weighted anatomical volumeof the whole brain was acquired with a turbo spin echopulse sequence. The imaging protocol was as follows: TR(repetition time) 5 6,300 ms, TE (echo time) 5 75 ms,number of signal averages 5 10, slice thickness 5 2 mm,gap 5 0, matrix size 5 256 3 256, and FOV (field ofview) 5 17.9 cm. Raw DICOM images were converted toNIfTI files using the statistical parametric mappingpackage (SPM5; Wellcome Department of Imaging Neu-roscience, London, UK). These were then entered intoChristopher Rorden’s MRIcron program (http://www.cab-iatl.com /mricro/mricro/index.html) to generate the finalimages for display.

Volume and Area Measurements

Manual image segmentation of MRIs was performedon processed coronal T2-weighted images to define thevolume of the whole brain and of the cerebellum, hippo-campus, and corpus callosum. We have chosen to reportthe volume of the corpus callosum, although most stud-ies report only its midsagittal cross-sectional area, asthe volume obtained by means of MRI represents a moreaccurate quantitative measure. Such values can be use-ful for future comparisons, as MR imaging becomesmore widespread in comparative neuroanatomy. Themanual segmentation of the whole brain, cerebellum,hippocampus, and corpus callosum was performed usingImageJ (Version 1.45s, NIH). Matlab (Version 2012a,Mathworks Inc., Natick, MA) code was developed toimport the regions of interest (ROI) into the in-housesoftware of the Department of Radiology, Icahn School ofMedicine at Mount Sinai, and the volumes were com-puted. The area of the ROIs in each MRI slice wasdirectly obtained by manual segmentation using the in-house software. Volumes were then calculated by inte-grating the selected area of each section. Every sectioncontaining the ROIs was used in the segmentation pro-cess. The manual segmentation of each ROI was per-formed three times and the mean volume was taken as afinal value. The percentage of total brain occupied byeach ROI was then calculated by dividing the volume ofeach ROI by the volume of the whole brain multiplied by100 (Table 1).

The midsagittal area of the corpus callosum (CCA)was obtained by manually tracing the callosal perimeterat the midline level in a photograph (Fig. 1C) usingImageJ. This step was necessary, as the quality of theMR images did not allow for a sagittal view of the scanwith an optimal resolution. The callosal perimeter wastraced three times on the same photograph and themean area was taken as a final value (Table 2). Finally,the mid-sagittal area of the corpus callosum/brain massratio (CCA/BM) was calculated both as reported by Tarp-ley and Ridgway (2000), dividing the cross-sectional areaof the corpus callosum (in mm2) by the brain mass (ing), and as reported by Manger et al. (2010), dividing thesquare root of CCA (in mm2) by the cubic root of brainmass (in g) to account for effects of differing unitsof measure (Table 3). We have chosen to calculatethe CCA/BM ratio using both methodologies, as the

4 BUTTI ET AL.

above-mentioned studies report values for a series of dif-ferent species relevant to our study. Using both valuesfacilitates comparisons of our results with previouslypublished data (Tables 2 and 3).

Gyrencephalic Index, Cerebellar Quotient, andCerebellar Index

The gyrencephalic index (GI) is a measure of the fold-ing of the cerebral cortex calculated by dividing theperimeter of the pial contour of the cortex by the perime-ter of its outer contour (Zilles et al., 1989). The pial con-tours and the outer contours of 41 coronal sectionsthroughout the brain of the pygmy hippopotamus weremeasured using the NIH software ImageJ. The GI wascalculated by dividing the sum of the 41 pial contours bythe sum of the 41 outer contours. The tracing of thebrain sections to define pial and outer contours was per-formed three times by the same operator (C.B.) and theaverage value of the three measurements was taken asa final value of GI (Table 4).

The cerebellar quotient (CQ) was calculated using theequation provided by Maseko et al. (2010) (Table 5). Thecerebellar index was calculated by dividing the cerebel-lar volume by the whole brain volume as measured inthe MRI. Moreover, in the present study we have calcu-lated values of CQ and cerebellar index (CI) for somerelevant species of artiodactyls, marine mammals otherthan cetaceans, and eutherian mammals of comparablebrain size for which brain mass, vermal volume andhemispheric volume were reported in Maseko et al.(2010) (Table 5). Particularly, vermal and hemisphericvolumes, reported by Maseko and colleagues, weresummed to obtain an approximation of cerebellar vol-umes. Given that our calculation excluding the cerebel-lar peduncles in the three specimens of elephantsresulted in the same value of cerebellar volume reportedby Maseko et al. (2010) for these specimens, we consid-ered our approximation sufficiently robust to estimatethe total cerebellar volume in relevant species of artio-dactyls and carnivores for which hemisphere volumeand vermal volume were available. Brain volumes wereobtained by multiplying brain mass by the specific massof brain tissue (1.036). CQ and CI were then calculatedas indicated above.

RESULTS

General Morphology

We utilized the nomenclature available for cetaceansand artiodactyls for the identification of gyri and sulci(Gruenberger, 1970; Morgane et al., 1980; Pilleri, 1980;Hof and Van der Gucht, 2007). This nomenclature isapplicable with variations to all species presenting amore or less verticalized (pseudo) Sylvian cleft, as it isthe case in the pygmy hippopotamus. In fact this nomen-clature has been applied in practice, with variants,across a broad number of species from different ordersand extends to carnivores as well as perissodactyls.

The whole fixed brain of the pygmy hippopotamusweighed 262 g and measured 13.5 cm in its anteroposte-rior axis (Fig. 1A,B). The medial aspect of the hemi-sphere was characterized by three main well-definedsulci: rostroventrally, the olfactory sulcus, that dividedthe frontal gyrus from the gyrus rectus; dorsal to the

TA

BL

E1.

Vo

lum

eo

fth

eb

ra

in,

cereb

ell

um

,h

ipp

oca

mp

us,

an

dco

rp

us

ca

llo

sum

inth

ep

yg

my

hip

po

po

tam

us

an

dco

mp

aris

on

wit

ha

va

ila

ble

da

tain

od

on

tocete

s

Sp

ecie

s

Wh

ole

bra

invol

um

e(c

m3)

Cer

ebel

lum

Lef

th

ipp

ocam

pu

sR

igh

th

ipp

ocam

pu

sC

orp

us

call

osu

m

Vol

um

e(c

m3)

%of

bra

inV

olu

me

(cm

3)

%of

bra

inV

olu

me

(cm

3)

%of

bra

inV

olu

me

(cm

3)

%of

bra

in

Hex

ap

roto

don

liber

ien

sis

Mea

n277.3

226.9

59.7

23.8

71.4

4.6

61.6

81.6

20.5

8S

.D.

6.8

60.3

0.2

60.0

50.0

50.1

80.0

80.1

0.0

5L

agen

orh

ynch

us

acu

tusa

Ran

geb

1,0

19.4

–1,4

39.9

141.0

5–172.1

3c

13.7

7–14.9

80.5

44–1.0

43

0.0

69–0.0

95

0.4

62–0.9

67

0.0

58–0.0

99

NA

NA

Del

ph

inu

sd

elp

his

dM

ean

786.2

911

8.5

315.0

75

NA

NA

NA

NA

NA

NA

S.D

.86.9

712.3

3–

––

––

––

Tu

rsio

ps

tru

nca

tusb

Mea

n1,5

41.0

9231.8

915.0

5e

NA

NA

NA

NA

NA

NA

S.D

.135.2

529.0

3–

––

––

––

aD

ata

from

Mon

tie

etal.

(2008).

bR

an

ges

ofvalu

esin

clu

din

gon

lyth

ead

ult

spec

imen

sfr

omM

onti

eet

al.

(2008).

c Su

mof

the

cere

bel

lar

wh

ite

matt

er(W

M)

an

dgra

ym

att

er(G

M)

from

Mon

tie

etal.

(2008).

dD

ata

from

Mari

no

etal.

(2000).

eD

ata

obta

ined

from

mea

nvalu

es.

NA

5d

ata

not

avail

able

.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 5

corpus callosum, the cingulate sulcus, which began at alevel corresponding to the genu of the corpus callosumand ran towards the most caudal part of the brain,defining the cortical belt that lies dorsal to the corpuscallosum (namely, the cingulate gyrus) and separating itfrom the frontal and parietal gyri; and caudally, thesmaller (and almost vertical) calcarine sulcus thatdivided the parasplenial gyrus from the lingual gyrus(Fig. 1B).

The dorsal aspect of the hemisphere was characterizedby the lateral gyrus, which was delimited medially bythe paracingulate sulcus and laterally by the ectolateral

sulcus. The lateral aspect of the hemisphere was charac-terized by three gyri: the large suprasylvian gyrus, thediminutive ectosylvian gyrus, and the perisylvian gyrus(Fig. 1A). Similar to cetaceans, these three gyri wereconcentrically organized around a pronounced andalmost vertical Sylvian fissure that divided the Sylviancortex into anterior and posterior limbs (Gruenberger,1970; Morgane et al., 1980; Pilleri, 1980; Hof and Vander Gucht, 2007). Rostrally was a well-defined cruciatesulcus that divided the coronal gyrus from the sigmoidgyrus, and an orbital sulcus, which divided the anteriorlimb of the Sylvian cortex from the more rostral orbitaland frontal gyri (Fig. 1A). Finally, a large olfactory bulboriginated ventrally to the gyrus proreus (Fig. 1A,B).

MRI, Volumetric Measurements, CCA/BM Ratio,GI, CQ, and CI

Figure 2 shows a sequence of 2-mm thick MR coronalimages originally acquired at an interval of 12 mm. MRIimages show that a portion of the basal forebrain, ven-tral striatum, hypothalamus and brainstem was missingin the left hemisphere following brain removal from theskull. Volumetric measurements performed on this seriesof MR images showed that the corpus callosum meas-ured 1.62 cm3 and occupied about 0.58% of the totalbrain volume. The hippocampus measured 8.53 cm3 (lefthippocampus 5 3.87 cm3 and right hippocampus 5 4.66cm3; Table 1), occupied 3.07% of the total brain volumeand possessed well-developed components (dentate gyrus

TABLE 2. Brain mass (BM), midsagittal area of the corpus callosum (CCA), and CCA/BM ratio in the pygmyhippopotamus and comparison with available data in marine mammals

Family Species Mean BM (g) Mean CCA (mm2) CCA/BM

Hippopotamidae Hexaprotodon liberiensis 262a 193a 0.737a

Delphinidae Cephalorhynchus commersoniib 748 174.4 0.192–0.258Delphinus delphisb 735 150.66 0.140–0.289Feresa attenuatab 1,030 200.23 0.166–0.209Globicephala macrorhynchusb 2,549 410.37 0.151–0.167Grampus griseusb 1,865 291.35 0.122–0.186Lissodelphis borealisb 1,162a 160a 0.138a

Lagenorhynchus obliquidensb 1,122 198.5 0.159–0.198Lagenorhynchus acutusc 1,255 141.71 0.088–0.137Orcinus orcab 6,368 619 0.079–0.119Pseudorca crassidensb 4,249 483.75 0.098–0.130Stenella attenuatab 740 169.9 0.190–0.305Stenella longirostrisb 575 152.3 0.239–0.310Steno bredanensisb 1,539 212.8 0.116–0.162Tursiops truncatusb 1,489 256.36 0.153–0.227

Platanistidae Platanista gangeticab 293a 126.3a 0.431a

Kogiidae Kogia brevicepsb 768 145.62 0.15–0.23Kogia simusb 625a 143.75a 0.23a

Physeteridae Physeter macrocephalusb 3,784a,d 643.28a,d 0.17a,d

Monodontidae Delphinapterus leucasb 2,009 341.53 0.13–0.21Ziphiidae Mesoplodon europaeusb 971a 233.04a 0.24a

Balaenopteridae Balaenoptera acutorostratab 2,217a 334.8a 0.151a

Balaenoptera musculusb 5,678a 850.1 0.150Otariidae Zalophus californianusb 358 160.83 0.348–0.529Odobenidae Odobenus rosmarusb 1,250a 188.7a 0.151a

Trichechidae Trichechus manatusb 302a 250.3a 0.829a

aData based on a single specimen.bData from Tarpley and Ridgway (1994).cData from Montie et al. (2008).dNeonate.

TABLE 3. Brain mass (BM), midsagittal area of thecorpus callosum (CCA), and CCA/BM ratio (calculatedas in Manger et al., 2010) in the pygmy hippopotamus

and comparison with available data in relevanteutherian mammals

BM (g) CCA (mm2) CCA/BM

H. liberiensisa,b 262 193 0.22Cetaceansc 630–6,060 139.3–565.3 0.11–0.15Sireniansa,c 188 89.5 0.17Pinnipedsc 345–1,250 101–188.7 0.13–0.17Artiodactylsc 244–530 117.3–180.1 0.16–0.18Perissodactylsc 531–585 238.5–251.5 0.19–0.20Carnivoresc 8.30–470 7.2–237.0 0.13–0.21

aBased on a single specimen.bPresent study.cData from Manger et al. (2010).

6 BUTTI ET AL.

and CA fields) (Fig. 2). The cerebellum measured 26.95cm3 and occupied 9.72% of the total brain volume.

The CCA/BM ratio, as calculated as reported by Tarp-ley and Ridgway (2000), was 0.737. However, when theCCA/BM ratio was calculated correcting for the effect ofdiffering units of measure (Manger et al., 2010), thevalue obtained was 0.22 (Fig. 2, Tables 1 and 2). Table 3provides comparison points for these measures. Theaverage value of three repeated measurements of GIresulted equal to 1.84. The cerebellar quotient and thecerebellar index resulted 0.80 and 0.97, respectively. Val-ues of CQ and CI of additional relevant species used forcomparison are reported in Table 4.

Cortical Organization

The cortical organization of the gyri bounded by themajor sulci is identified in Fig. 1. The cytoarchitectureof cortical areas was defined according to topographicallocation. Based on coronal Nissl-stained sections, theneocortex of the pygmy hippopotamus was entirely agra-nular—with no identifiable layer IV. Consequently, wedo not mention the absence of layer IV further.

Paleocortex

The paleocortex includes the olfactory cortex, theolfactory tubercle, and the piriform cortex. The piriformcortex, which abuts the anterior insular cortex, repre-sents the most anterior portion of the paleocortex and itwas characterized by a thick layer I, a thick and denselayer II, and a layer III containing sparse and smallpyramidal neurons (Fig. 3A). Ventral to the piriform cor-tex, the cortex of the olfactory tubercle was much thin-ner and was characterized by clusters of neurons inlayer II and a poorly defined transition between layersII and III (Fig. 3B). Posteriorly, the periamygdalar cor-tex, which abuts the archi- and entorhinal cortices, wascharacterized by clusters of small neurons in its most

superficial layers and by a visible lamina dissecans(Fig. 3C).

Archicortex

The hippocampus of the pygmy hippopotamusappeared comparable to that of other terrestrial mam-mals (Fig. 4A). The dentate gyrus was large and con-tained a thick and hyperchromatic granular layerunderlined by a very thick subgranular zone, the poly-morphic layer (Fig. 4A,B). Below the polymorphic layerof the dentate gyrus, the hilar part of CA3 was thick(Fig. 4A,B) and extended laterally into a long CA3 fieldthat contained a thick and loosely organized pyramidallayer (Fig. 4C). CA2 was clearly visible as a regionwhere the pyramidal layer becomes disorganized and itsneurons are very sparse before the transition into thethin and well-organized pyramidal layer of CA1 (Fig.4A,D). The pyramidal neurons of CA2 were comparablein size to those of CA3 and larger than those observedin CA1. CA1 was very well defined and contained apyramidal layer that was organized in two sublayers: adense and intensely stained thin superficial layer and athick and less densely packed lower layer containingpyramidal neurons of smaller size than those of thesuperficial sublayer (Fig. 4D). The transition of CA1 tothe subiculum (Fig. 4A,E) was clearly visible as thepyramidal layer becomes thicker and homogeneous,lacking the subdivision in two sublayers observed inCA1. Laterally, the small presubiculum abutted the par-asubiculum, which displayed a thick superficial pyrami-dal layer that contained small and darkly labeledpyramidal neurons, a thin lamina dissecans, and adeeper pyramidal layer consisting on small neurons(Fig. 4F).

Entorhinal Cortex

The entorhinal cortex of the pygmy hippopotamus,unlike in cetaceans but similar to terrestrial mammals,appeared to be proportionate to the size of the dentategyrus. Anteriorly, the entorhinal cortex was thin with aprominent clustering of layer II neurons in large anddarkly stained islands, a very thick layer III containingmedium size neurons, a broad lamina dissecans, a thinlayer V with an irregular distribution of medium sizeneurons, and a multiform layer VI (Fig. 4G). Moving lat-erally (dorsally) the overall neuronal density decreased,the layer II islands tended to disappear and the laminadissecans became thinner than in the most medial por-tion of this cortex (Fig. 4H). Posteriorly, the caudal ento-rhinal cortex presented an overall increased neuronaldensity compared with more anterior levels, particularlyin layers III, V, and VI (Fig. 4I). Layer II still showedsome clustering but the islands were larger and the neu-rons were less tightly packed than in anterior levels(Fig. 4I). The lamina dissecans was thinner than in theanterior sector, but was still clearly visible. In contrast,layers V and VI were much thicker (Fig. 4I). Laterally,at the caudal level, the transition of the entorhinal cor-tex into the temporal cortex was marked by the disap-pearance of the lamina dissecans, the transition ofclustered layer II into a homogeneous and thinner layerof neurons, and by the disappearance of a marked tran-sition between layer V and VI.

TABLE 4. Brain mass (g), cerebellar volume (cm3),brain volume (cm3), cerebellar quotient (CQ) and cer-

ebellar index (CI) in the pygmy hippopotamus andrelevant species of artiodactyls, cetaceans, and

carnivores

Species Brain mass (g) GI

CetartiodactylsH. liberiensisa 262 1.84Sus scrofab 95.3 2.16Odocoileus virginianusb 160 2.27Lama glamab 200.3 2.70Bos taurus indicusb 474 2.53Kogia simusb 577 5.26Delphinus delphisb 981 5.41Tursiops truncatusb 1,530 5.63Delphinapterus leucasb 1,871 5.23Orcinus orcab 5,617 5.70Megaptera novaeangliaeb 4,600 5.35

CarnivoresPhoca vitulinab 275 2.38Zalophus californianusb 379.1 2.52Panthera leob 258 1.85

aPresent study.bData from Manger et al. (2012).

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 7

Cingulate and Retrosplenial Cortex

The cingulate cortex consisted of one gyrus outlinedventrally by the callosal sulcus and dorsally by the cin-gulate sulcus (Fig. 1). The cingulate and retrosplenialcortices were divided into a subgenual area, a supracal-losal area including the pregenual, anterior and poste-rior cingulate cortices, and the retrosplenial cortexposterior to the splenium of the corpus callosum, includ-ing an anterior and a posterior component, recognizableby their different cortical organization. In coronal sec-tions an additional small sulcus, ventral to the cingulatesulcus and limited to the anterior portion of the cingu-late cortex, could be appreciated. The banks of thesesulci were linear and lacked the convolution oftenobserved in cetaceans (Morgane et al., 1982; Hof andVan der Gucht, 2007).

Subgenual Cortex

The subgenual cortex is the portion of the anteriorcingulate cortex that lies ventral to the genu of the cor-pus callosum. It was characterized by a layer II thatcontained numerous small and irregular neuronal clus-ters (Fig. 5A). These clusters were also observed, to alesser degree, in the adjacent pregenual cortex (Fig. 5B),unlike in cetaceans where clustering of layer II is notpresent in the pregenual cortex (Hof and Van der Gucht,2007). Layer III contained medium-size neurons and itstransition to layer V was not well defined. Layer V wasthick, contained medium-sized neurons. Its boundarywith layer VI was not easily identifiable. Layer VI wasthin and populated by small neurons (Fig. 5A).

Pregenual Cortex

The pregenual cortex is the portion of cortex situateddirectly in front of the genu of the corpus callosum. Itpresented, below a thick layer I, a layer II that wasmore homogeneous than layer II of the adjacent subge-nual cortex and showed an irregular pattern of cluster-ing. Layer III was thick and populated by smallneurons. Layer V was thick and characterized by larger

neurons than in the adjacent subgenual cortex. Layer VIwas thin and contained small neurons with an occa-sional columnar arrangement (Fig. 5B).

Anterior and Posterior Cingulate Cortex

At a level corresponding to the genu of the corpus cal-losum, the anterior cingulate cortex (ACC) was charac-terized, in its ventral bank, below a thick layer I, by alayer II of variable thickness, a thick layer III contain-ing small relatively sparse neurons, a well-developedlayer V consisting of large pyramidal neurons, and athick, multiform layer VI (Fig. 5C). The dorsal bank ofthe ACC, at the same level, was characterized by ahomogeneous and dense layer II, a layer III containingsmall size pyramidal neurons, a layer V containing largepyramidal neurons in large clusters that occasionallyoverlapped with the underlying multiform layer VI(Fig. 5D).

The posterior cingulate cortex (PCC) showed severalvariations from the pattern described in the ACC. In thePCC, in fact, the clustering of layer II was not present.Layer V thickened in the ventral portion of the PCC, incomparison to the ventral aspect of the ACC whereas, inthe dorsal aspect of the PCC, layer V contained largepyramidal neurons organized in vertical modulesthat tended to overlap with the neurons of layer VI (Fig.5E,F).

Anterior and Posterior Retrosplenial Cortex

The retrosplenial cortex showed, in both its anteriorand posterior sectors, an overall increased neuronal den-sity and a thinner and more homogeneous layer II thanthe ACC and PCC (Fig. 5G,H). The anterior sector of theretrosplenial cortex possessed a thick layer III populatedby neurons that were comparable in size to those observedin layer III of the ACC (Figs 5C,G). Layer V was variablein thickness; it contained large pyramidal neurons andlacked the clustering of layer V observed in the dorsal por-tion of the ACC (Fig. 5D,G). Layer VI was thick and con-tained small neurons. The posterior sector of theretrosplenial cortex contained a diminutive layer III,

TABLE 5. Brain mass (g) and gyrencephalic index (GI) in the pygmy hippopotamus and comparison withartiodactyls, cetaceans, and relevant species of other eutherian mammals

Species Brain mass (g) Cerebellar volume (cm3) Brain volume (cm3) CQ CIa10

CetartiodactylsH. liberiensisa 262 26.95 277.32 0.80 0.97Sus scrofab 95.30 9.28 98.73 0.74 0.94Odocoileus virginianusb 160 18.75 165.76 0.90 1.13Llama glamab 200.30 22.46 207.51 0.87 1.08Bos taurus indicusb 520.50 33.71 539.24 0.51 0.62Phocoena phocoenab 486 66.84 503.5 1.08 1.33Phocoena phocoenab 503 74.08 521.10 1.16 1.42Tursiops truncatusb 1,500 258.75 1,554 1.40 1.66

CarnivoresPhoca vitulinab 275 24.01 284.9 0.70 0.84Zalophus californianusb 379.13 58.19 392.78 1.20 1.48Ursus maritimusb 458.60 51.95 475.11 0.90 1.09Pantera leob 258 26.54 267.3 0.80 0.99

Cerebellar volume was calculated adding the vermal volume to the hemispheric volume and brain volume was calculatedmultiplying brain mass by the specific mass of nervous tissue as indicated in the Materials and Methods section.aPresent study.bFrom Manger et al. (2012).

8 BUTTI ET AL.

which was characterized by small-size neurons. Thelarge pyramidal neurons of layer V possessed a verticalorganization that, however, did not continue intolayer VI as observed in the dorsal aspect of the PCC(Fig. 5H,F). Layer VI was thick and contained medium-size neurons.

Insular Cortex

The insular cortex lies in the depth of the Sylvian fis-sure. In H. liberiensis the insula was operculated byparts of the frontal, parietal, and temporal cortices. Its

extent in this species it was limited only to a few gyri(Butti and Hof, 2010),

Anterior Insular Cortex

The anterior portion of the insular cortex was limiteddorsally by what may correspond to the primary somato-sensory cortex (SS; see below), and ventrally by the piri-form cortex. In the most dorsal portion of the anteriorinsular cortex layer II was thin, contained medium sizeneurons, and exhibited only a weak clustering of neu-rons. Layer III was very thick and contained small- and

Fig. 2. Series of coronal sections through the brain of the pygmy hippopotamus acquired by MRI. Scalebar 5 5 cm.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 9

medium-sized neurons. Layer V was thick and populatedby medium-sized and occasionally large pyramidal neu-rons. Layer VI was thick and contained small neurons.Small islands of claustral neurons were present in theunderlying white matter (Fig. 6A).

In the depth of the Sylvian fissure, the cortex becamethinner overall. Layer II was more homogeneous and itsneurons were sparser than dorsally. Layer V was thinand contained neurons comparable in size to thoseobserved in the most ventral portion of the anteriorinsular cortex (Fig. 6B). At this level, layers V and VIwere characterized by a columnar appearance with largeand elongated claustral islands in the underlying whitematter (Fig. 6B). In its most ventral aspect, the anteriorinsular cortex exhibited a layer II characterized by anirregular distribution of neurons, a very thick layer IIIcontaining small-size neurons, and by a reduced thick-ness of layer V compared with the dorsal aspect of theregion (Fig. 6C). Layer V contained occasional groupings

of neurons and layer VI was very thin. Claustral islandsbecame larger where, ventrolaterally, the insular cortexabutted the piriform cortex (Fig. 6C).

Mid and Posterior Insular Cortex

The middle and posterior portions of the insular cor-tex were limited dorsally by what appears to correspondto the auditory cortex (see Parietal Cortex for details)and ventrally by the piriform cortex. In its most dorsalaspect, the midinsular cortex showed a much thinner,less dense and less homogeneous layer II than observedin the anterior insular cortex. Layer III was very thickand contained small-sized neurons. Given the large dif-ference in neuronal size between layer III and layer V,the transition between these two cortical layers was welldefined in the middle portion of the insular cortex (Fig.6D–F). Layer V was organized in columnar structures ofseveral neurons that formed a continuum with layer VI

Fig. 3. Cytoarchitecture of the paleocortex in the pygmy hippopotamus. Piriform cortex (A), cortex ofthe olfactory tubercle (B), and peryamigdalar cortex. Coronal sections in rostro-caudal order showing thelocation of A–C (D, E). Layers are indicated by Roman numerals; ld, lamina dissecans. Scale bar 5 400mm (A–C), and 5 cm (D, E).

10 BUTTI ET AL.

(Fig. 6D), similar to the cortex lying in the depth of theSylvian fissure in the anterior part of the insula (Fig.6B). In the depth of the Sylvian fissure, the midinsularcortex exhibited a pattern of cortical organization com-parable to that in the depth of the Sylvian fissure ante-riorly (Fig. 6B,E). Ventrally the midinsular cortex wascharacterized by an increased density of layer II contain-ing medium-size neurons, by a layer V with a variablethickness, and by a thick layer VI with a columnararrangement of neurons. Large claustral islands werepresent in the underlying white matter (Fig. 6F).

Posteriorly, the organization of the insular cortexincluded a homogeneous layer II that was thicker than

in the anterior and middle sectors of the insula, a thicklayer III including small and medium-sized neurons(Fig. 6G,H) and a columnar organization of layers V andVI that was more evident ventrally (Fig. 6H) than dor-sally (Fig. 6G). Claustral islands were evident at thisposterior level along both the dorsal and the ventralaspect of the insular cortex (Fig. 6G–H).

Frontopolar Cortex and Frontal Convexity

The frontopolar region of H. liberiensis was welldefined and separated from the orbital cortex and thegyrus proreus by two shallow sulci in the frontal tip of

Fig. 4. Cytoarchitecture of the archicortex, and entorhinal cortex inthe pygmy hippopotamus. Panoramic view of the hippocampal forma-tion (A), dentate gyrus (B), CA3 (C), CA1 (D), subiculum (E), presubicu-lum (F), anterior (G, H), and posterior entorhinal cortex (I). Note thewell-developed dentate gyrus including a thick subgranular zone (A,B), CA3 (C), the thick pyramidal layer of CA1 (D), the thick subiculum(E), and presubiculum (F) with an evident lamina dissecans. The ento-rhinal cortex presents both rostrally (G, H) and caudally (I) a clear lam-

ina dissecans, whereas the parvocellular islands of layer II becomemore evident caudally. Coronal sections in rostro-caudal order show-ing the location of panels B–I (J–L). Layers are indicated by Romannumerals. CA1–3, Ammon’s horn fields; DG, dentate gyrus; gl, granu-lar layer; ld, lamina dissecans; ml, molecular layer; pl, polymorphiclayer; so, stratum oriens; sp, stratum pyramidale; sr, stratum radiatum;SUB, subiculum, wm, white matter. Scale bar 5 400 mm (B–I), 1.6 mm(A), and 5 cm (J–L).

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 11

the brain that ran almost perpendicular to each other,(Fig. 1). The orbital cortex, including the gyrus proreus,was defined dorsally by the cortex of the sigmoid gyrusand by the orbital sulcus, which divided it from theanterior limb of the Sylvian cortex, and ventrally by thegyrus proreus (Fig. 1).

The cytoarchitecture of the frontopolar cortex wascharacterized by a thick layer I, and a dense, clusteredlayer II with large spaces of neuropil between adjacentclusters. The depth of such neuropil spaces was limitedto layer II and, occasionally, extended to the most super-ficial part of layer III. Layer III was thick and populated

Fig. 5. Cytoarchitecture of the subgenual (A), pregenual (B), ventral anterior cingulate (C), dorsal anteriorcingulate (D), ventral posterior cingulate (E), dorsal posterior cingulate (F), anterior retrosplenial (G), and pos-terior retrosplenial (H) cortices in the pygmy hippopotamus. Coronal sections in rostro-caudal order showingthe location of panels A–H (I–N). Layers are indicated by Roman numerals. Scale bar 5 400 mm.

12 BUTTI ET AL.

by small neurons; layer V was extremely variable inthickness and contained larger pyramidal neurons thanlayer III. Layer VI was thick with a columnar organiza-tion (Fig. 7A). The peculiarity of this cortical region wasthe presence in layer II of clusters that varied greatly insize, thickness, and length (Fig. 7B–E).

The sigmoid gyrus was characterized by a thin mag-nocellular type of cortex. A well-defined and dense layerII, formed by small neurons, was followed by a thicklayer III containing homogeneously distributedmedium-size pyramidal neurons. Layer V consisted of

large pyramidal neurons that were often grouped inclusters of several cells. Layer VI was thick and con-tained neurons that were arranged into large verticalmodules directly aligned with the clusters of layer V(Fig. 7F). This cortical organization was observed fromthe midline to the most lateral extension of the sigmoidgyrus and continued dorsally until just below the cruci-ate sulcus. At this level the large pyramidal neurons oflayer V progressively disappeared and layer II becameheavily clustered. This magnocellular cortex is in aposition to represent the putative motor cortex in the

Fig. 6. Cytoarchitecture of the insular cortex in the pygmy hippopotamus. Anterior dorsal (A), depth ofSylvian fissure (B), anterior ventral (C), middle dorsal (D), depth of Sylvian fissure (E), middle ventral (F),dorsal posterior (G), and ventral posterior (H) insular domains. Coronal sections in rostro-caudal ordershowing the location of panels A–H (I–K). Layers are indicated by Roman numerals. Scale bar 5 400 mm(A–H), and 5 cm (I–K).

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 13

pygmy hippopotamus and seemed to expand more later-ally than rostrocaudally.

In the dorsal aspect of the sigmoid gyrus, just belowthe cruciate sulcus, the pattern of cortical organizationchanged. Here, the cortex was thicker, a marked cluster-ing of layer II appeared, and the large neurons of layerV tended to disappear. Layer V was less well defined,and contained smaller neurons than the putative motorcortex. Moreover, neurons in the deep layers III and Vwere arranged in columns under the clusters of layer II.The spaces among the large columns were evident illus-trating an extreme columnar appearance to the uppercortical layers. Layer VI was thick with neurons that

seemed to conform to the columnar organization of theupper layers (Fig. 7G).

In the depth of the cruciate sulcus, the pattern of clus-tering of layer II was maintained, but the organizationof the deeper cortical layers became more homogeneous(Fig. 7H). More dorsally in the coronal gyrus the layer IIclusters decreased in size, layer III was thick and con-tained small neurons that were homogeneously distrib-uted under the clusters of layer II. Layer V containedmedium-sized neurons. Layer VI was thick with neuronsorganized in modules that created a columnar appear-ance (Fig. 7I). This parvocellular frontal field, whichextended around the cruciate sulcus and dorsally to the

Fig. 7. Cytoarchitecture of the frontopolar cortex and the frontalconvexity in the pygmy hippopotamus. (A–E) Frontopolar cortex, (F)magnocellular cortex of the sigmoid gyrus corresponding to the puta-tive motor cortex, (G) cortex of the dorsal aspect of the sigmoid gyrus,just below the cruciate sulcus, (H) cortex in the depth of the cruciatesulcus, (I) cortex of the coronal gyrus. Note the clustering pattern of

layer II (A–E; G–I) and the magnocellular elements in layer V of theputative motor cortex (F). Coronal sections in rostro-caudal ordershowing the location of panels A–I. (J–L). Layers are indicated byRoman numerals; wm, white matter. Scale bar 5 400 mm (A–I); 5 cm(J–L).

14 BUTTI ET AL.

magnocellular frontal field of the motor cortex, couldrepresent the putative somatosensory cortex. This pat-tern of cortical organization extended laterally along themost rostral portion of the ectolateral sulcus and dor-sally to the lateral gyrus for two thirds of its extent (seeParietal Cortex below for details).

Temporal Cortex

Although the temporal operculum was well defined inthe pygmy hippopotamus, the concentric disposition ofthe small gyri and sulci around the Sylvian fissure canbe misleading. For this reason we decided to considertemporal cortex, consistent with observations in ceta-ceans (Hof and Van der Gucht, 2007), as the portion ofcortex situated below a horizontal line passing throughthe dorsal tip of the Sylvian fissure. Anteriorly and ven-trally, the temporal cortex abutted the entorhinal cortex.At this level, the temporal cortex presented a thick corti-cal plate characterized by a prominent and thin layer IIwith occasional small clumps of neurons. The transitionbetween deeper layers was not well defined at this leveland the neuronal density was homogeneous from layerIII to VI. Layer V was thick and contained medium-sizeneurons (Fig. 8A).

Medially in the temporal pole the cortex exhibited alow neuronal density but a thick layer II that containedmedium-size neurons. The transition to layer III wasclearly visible given the smaller size of the neurons in

this layer, which made it much less chromatic than theadjacent upper and lower layers. Layer V containedmedium-size neurons distributed homogeneously andlayer VI was thin with small neurons (Fig. 8B). Dorsally,at the tip of the temporal pole towards the parietal cor-tex, the cortex became thicker, with higher neuronaldensity. Layer V contained larger neurons than in themedial aspect and layer VI became thicker (Fig. 8C).Laterally, the cortex had reduced thickness, similar tothe medial aspect. Layer V neurons appeared largerthan dorsally and layer VI showed a progressive colum-nar organization (Fig. 8D). Moving caudally, the tempo-ral cortex maintained the poorly defined transitionbetween adjacent cortical layers, with the exception oflayer II, which was extremely visible as its neuronaldensity increased and a clustering pattern became appa-rent, near the transition with the occipital cortex (Fig.8E).

Parietal Cortex

The parietal cortex was a large domain located poste-riorly to the coronal gyrus extending from the lateralgyrus, at the apex of the brain, down to the perysylviangyrus and posteriorly delimited by the most caudal seg-ment of the ectolateral sulcus (Fig. 1). At the apex of thebrain, the first two-thirds of the lateral gyrus (just pos-terior to the coronal gyrus) were characterized by thesame type of cortex observed in the dorsomost part of

Fig. 8. Cytoarchitecture of the temporal cortex in the pygmy hippopotamus. Anterior ventral (A), medial(B), dorsal temporal cortex towards the parietal cortex (C), lateral (D), and caudomost portion of the tem-poral cortex towards the occipital cortex (E). Coronal sections in rostro-caudal order showing the locationof panels A–E (F, H). Layers are indicated by Roman numerals; wm, white matter. Scale bar 5 400 mm.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 15

the coronal gyrus. This appeared to be the putativesomatosensory cortex. However, in the lateral gyrus, arostrocaudal gradient of clustering of layer II was evi-dent as the clusters became larger and thicker caudally(Fig. 9A–C). Moreover, an evident transition to a colum-nar arrangement of medium-sized neurons in layer VIwas present, caudally, although the upper boundary oflayer V was poorly defined (Fig. 9A).

In the pygmy hippopotamus the somatosensory typeof cortex seemed to extend posteriorly along anextended portion of the lateral gyrus. It was only in themost posterior third of the lateral gyrus that we founda type of cortex with a thin and homogeneous layer II,a thick layer III containing small neurons, a layer V

containing occasional large neurons, and a layer VIorganized in modules (Fig. 9D). The pattern changed inthe most medial aspect of the lateral gyrus where themodular organization of layer VI was evident and thesmall neurons of layer III become sparse (Fig. 9E). Thecortical organization of the lateral aspect of the lateralgyrus expanded laterally into the suprasylvian gyrusand into the depth of the suprasylvian fissure. Suchfields, in view of their position and cytoarchitecture,could correspond to the putative visual cortex asdescribed in cetaceans (Sokolov et al., 1972; Ladyginaand Supin, 1977; Garey et al., 1989; Revishchin andGarey, 1989; Glezer and Morgane, 1990; Hof and Vander Gucht, 2007).

Fig. 9. Cytoarchitecture of the parietal cortex in the pygmy hippo-potamus. Anterior (A), middle (B), and posterior (C) aspects of the lat-eral gyrus (la) showing the putative somatosensory cortex; caudalmostaspect of the lateral gyrus (E) towards the occipital cortex and dorsalmost aspect of the suprasylvian cortex (D) showing the putative visual

cortex; ectosylvian (F) and perisylvian cortex (G) showing the putativeauditory cortex. Coronal sections in rostro-caudal order showing thelocation of panels A–G (H–K). Layers are indicated by Roman numer-als; wm, white matter. Scale bar 5 400 mm.

16 BUTTI ET AL.

Laterally, in the ectosylvian cortex, the cytoarchitec-ture was characterized by a thin layer II, a high neuro-nal density in layer III, and by layers V and VIcontaining small neurons disposed in vertical modules(Fig. 9F). More laterally, layer II thickened into the peri-sylvian cortex. Layers V and VI were characterized by alower cellularity than the ectosylvian cortex, but stilldisplayed a high degree of columnar organization (Fig.9G). These regions are in a position to represent audi-tory cortices as described in cetaceans (Sokolov et al.,1972; Popov and Supin, 1976, 1986; Ladygina and Supin,1977; Ladygina et al., 1978).

Occipital Cortex

The occipital cortex is the posterior polar cortex and issituated posterior to the most caudal part of the ectolat-eral sulcus and includes the inferior portion of temporalcortex, which covers the anterior surface of the cerebel-lum. Most of the occipital cortex was characterized byclustering of layer II neurons. Medially, the occipital cor-tex was characterized by high neuronal density, variablethickness of layer II, large neurons in layer III, medium-sized neurons in a thick layer V, and a thick layer VIthat displayed cellular columns (Fig. 10A). More dorsally

the cortex was less dense and the layer II clusters wereapparent. Layer II contained medium-sized neurons andcould be divided into two separate sublayers, the mostsuperficial arranged in clusters, and the deeper onebeing more homogeneous. Layer III was thick and con-tained sparse neurons. Layer V was formed by small-sized neurons and layer VI was organized in shortcolumnar arrangements (Fig. 10B). Dorsally, towardsthe lateral aspect of the hemisphere, the organization ofthe cortex was reminiscent of the architecture of theputative visual cortex described above (Fig. 10C).

Laterally, the pattern of cortical organization changed.Layer II was organized into striking clusters, inter-spersed with large neuropil spaces. Layer III was thinand the very small size of its neurons contrasted withthe size of neurons of layer II and V. Layer V was thickand formed by medium- to large-sized neurons, andlayer VI was thin and contained medium-sized neurons(Fig. 10D). Ventrally, the clustering pattern of layer IIdisappeared gradually (Fig. 10E); however, occasionalneuropil spaces interrupted the homogeneity of layer II,which was here divided in two sublayers, as in the dor-solmost aspect of the occipital cortex (see above and Fig.10B). However, at this ventral level, the difference indensity of the two sublayers was more evident due to

Fig. 10. Cytoarchitecture of the occipital cortex in the pygmy hippopotamus. (A) medial, (B, C) dorsaland dorsolateral, (D) lateral, (E) ventral, (F) caudal aspects of the occipital cortex. Coronal sections inrostro-caudal order showing the location of panels A–F (G, H). Layers are indicated by Roman numerals;wm, white matter. Scale bar 5 400 mm.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 17

the intense staining of the superficial portion of layer II.Layer III was thin and characterized by very small neu-rons as observed in the lateral aspect of the occipital cor-tex (Fig. 10D,E). Layer V was thick and consisted ofsmall neurons that often had an elongated morphology.Layer VI was thick and characterized by small-sizedneurons. Clustering of layer II was maintained in themost caudal portion of the occipital cortex. However, thesize of these clusters decreased and layer III was charac-terized by medium-sized neurons (Fig. 10F).

Distribution and Morphology of Calretinin-Immunoreactive Neurons

In the pygmy hippopotamus, CR-immunoreactive neu-rons were distributed homogeneously across corticalareas and displayed their highest densities in layers I,II, and superficial III, but were only sparsely visible indeeper layers (Fig. 11A-J). CR-immunoreactive neuronsof layers I, II, and superficial III, were large fusiform,bipolar, or multipolar neurons possessing elongated den-dritic arbors that spanned into adjacent cortical layers(Figs. 11A–J and 12A,B). Giant CR-immunoreactive mul-tipolar neurons were observed in layer III of the tempo-ral cortex (Fig. 12C). Fusiform and horizontally orientedneurons of the Cajal-Retzius type (Ram�on y Cajal, 1911;Marin-Padilla, 1990) were observed in layer I of thefrontopolar cortex (Fig. 12D,E). Occasionally, large bipo-lar and multipolar neurons were noted in layer VI inanterior and posterior lateral gyrus (Fig. 12F,G), repre-senting the putative somatosensory and visual cortices,respectively.

Golgi Preparations

Golgi preparations of the three cortical regions stud-ied (e.g. lateral gyrus, putative visual cortex; frontalmagnocellular field, putative motor cortex; and ACC)revealed a large variety of spiny neuronal morphologies.However, it is important to note that absence in Golgipreparations of particularly neuron types does notmeans these neurons were not present in the corticalregion analyzed. Golgi staining results can vary consid-erably among regions with only a small percentage ofneurons stained in each preparation. In the presentstudy, we obtained a greater number of well-stained neu-rons in the cortex of the lateral gyrus and in the magno-cellular cortex than in the ACC. As a consequence, ourGolgi samples did not allow for a comprehensive deter-mination of all neuronal morphologies belonging to aspecific cortical area, but represent instead a qualitativedescription of a sample of neurons and the characteris-tics of their dendritic arbors from each analyzed block.

Typical pyramidal neurons of small and medium sizewith triangular or round somata were observed in allcortical layers in the putative visual and motor cortices,and in the ACC. These neurons possessed prominentapical dendrites, often bifurcating at some distance fromthe soma as they ascend toward the pial surface, thusproviding two distinct apical shafts. The basilar skirtwas relatively simple radiating in many directions for ashort distance from the soma (Figs. 13Aa,Ac,Ba,F,G and14B,E).

Magnopyramidal neurons were observed in layer V ofthe magnocellular frontal field, the putative motor cor-

tex. They projected very long ascending apical dendritesthat often bifurcated, at varying distances from themostly triangular soma, into two branches (Fig. 15A).The basilar skirt spread in multiple directions (Fig.14A,D).

Extraverted pyramidal neurons exhibiting the typicalpredominant subpial dendritic arborization (Sanides andSanides, 1972) were observed in both superficial anddeep layers (II, III, and V) of the putative visual andmotor cortices. These neurons of small, and sometimesmedium soma size, were characterized by a relativelylarge apical branch system compared with the basilardendritic arbor. Interestingly, this neuronal type wasalso observed in deeper cortical layers (Fig. 13Ab,H andFig. 14Cb,G,H).

Multiapical pyramidal neurons were mainly located inlayer V and observed in both the putative visual andmotor cortices. These neurons were characterized by atriangular or spherical soma, from which prominent andsymmetric apical dendrites extended toward the pialsurface, and by a simple, and often diminutive, basilarskirt with dendrites descending towards the white mat-ter or projecting radially for a short distance from thesoma (Figs. 13Bc,E and 14Ca,F).

Von Economo neurons (VENs) exhibiting the typicalmorphological features described in other species (Nim-chinsky et al., 1995, 1999; Butti et al., 2009; Hakeemet al., 2009; Allman et al., 2010; Butti and Hof, 2010)were observed in layer V of the putative visual cortex(with a few in deep layer III; Fig. 13Bb). However, it isworth noting that VENs were observed in all corticalregions of the pygmy hippopotamus in Nissl-stained sec-tions (see von Economo Neurons below, and Fig. 16 forfurther details on VEN morphologies and distribution).

Fork neurons, previously described in the cortex ofhumans and great apes (Ngowyang, 1932; Allman et al.,2010; Seeley et al., 2012), as well as elephants (Jacobset al., 2011) and cetaceans (Hof and Van der Gucht,2007), were observed in layer V of the putative visualcortex of the pygmy hippopotamus and were character-ized by a thick soma that split into two thick ascendingapical dendrites and by the presence of thin basal den-drites running laterally (Fig. 13D).

Large pyramidal neurons with a descending taprootwere observed in layer V of the visual cortex. They pre-sented a large fusiform soma from which a thick apicaldendrite originated. Their basilar skirt was character-ized by a central thick dendrite (the taproot; Jacobset al., 2011) descending for a long distance towards thewhite matter and from which a dense array of small lat-eral dendrites extended radially (Fig.13C).

Von Economo Neurons

Vens were observed in layer V of all cortical regions ofthe cerebral cortex of H. liberiensis including the typicalregions of distribution of VENs (ACC, anterior insula,and frontopolar cortex), but also parietal, temporal, andoccipital cortices. Even though VENs were abundantthroughout all cortical areas, they were still restricted tolayer V and, occasionally deep layer III, as seen in thespecies in which VENs have been described to date(Nimchinsky et al., 1995, 1999; Hof and Van der Gucht,2007; Fajardo et al., 2008; Butti et al., 2009; Hakeemet al., 2009; Allman et al., 2010; Butti and Hof, 2010;

18 BUTTI ET AL.

Evrard et al., 2012). VENs in the pygmy hippopotamusshowed the typical set of morphological featuresdescribed in hominids, macaque monkeys, cetaceans,and elephants (von Economo, 1926; Nimchinsky et al.,

1995, 1999; Butti et al., 2009; Hakeem et al., 2009; All-man et al., 2010; Evrard et al., 2012) (Fig. 16A–F). Themorphology of VENs observed in the cortex of H. liber-iensis resembled closely those that we described in

Fig. 11. Photomicrographs showing the distribution of CR-immunoreactive neurons in several cortical regions of the pygmy hip-popotamus. Frontopolar cortex (A), cortex of the sigmoid gyrus (puta-tive motor cortex) (B), cortex of the anterior lateral gyrus (putativesomatosensory cortex) (C), suprasylvian cortex (putative auditory cor-

tex) (D), anterior insular cortex (E), perisylvian cortex (F), posteriorinsular cortex (G), temporal cortex (H), anterior retrosplenial cortex (I),cortex of the posterior lateral gyrus (putative visual cortex) (J). Notethe distribution of the CR-immunoreactive neurons mostly in thesuperficial layers I, II, and III. Scale bar 5 400 mm.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 19

Fig. 12. Photomicrographs showing the typical bipolar and fusiformmorphology as well as large atypical morphologies of CR-immunoreactive neurons in the cortex of the pygmy hippopotamus.Bipolar and fusiform neurons in layer III of the posterior lateral gyrus(putative visual cortex; A); multipolar neuron in layer II of the frontopo-lar cortex (B); giant multipolar neuron in layer III of the temporal cortex

(C); Cajal-Retzius neurons in layer I of the frontopolar cortex (D, E);large fusiform neuron in layer VI of the cortex of the posterior lateralgyrus (putative visual cortex; F); multipolar neuron in layer VI of thecortex of the anterior lateral gyrus (putative somatosensory cortex; G).Coronal sections in rostro-caudal order showing the location of panels11A–11J, and 12A–12G (12H–12M). Scale bar 5 100 mm.

20 BUTTI ET AL.

several cetacean species (Hof and Van der Gucht, 2007;Butti et al., 2009). The most frequently observed mor-phology was characterized by a stout cell body with api-cal and basal dendrites almost as thick as the cell body(Fig. 16A,C,D,F). We also observed VENs with extremelyslender cell bodies (Fig. 16B), resembling the morpholo-gies mostly described in hominids (Nimchinsky et al.,

1995, 1999; Allman et al., 2010), and fork neurons char-acterized by a soma divided into two thick apical den-drites and by a single and thick basal dendrite (Fig.16E). Occasionally, VENs possessed a basal dendritedivided in two branches and a hint of a basilar skirt(Fig. 15B,C), as previously described in humans (Nim-chinsky et al., 1995).

Fig. 13. Photomicrographs of Golgi-impregnated neurons observedin the lateral gyrus of the pygmy hippopotamus. (A) a, pyramidal neu-ron; b, extraverted pyramidal neuron; c, pyramidal neuron); (B) a,pyramidal neuron; b, von Economo neuron; c, multiapical pyramidalneuron; (C) large deep pyramidal neuron with a descending taproot;

(D) fork neuron; (E) multiapical pyramidal neuron; (F–G) pyramidal neu-rons; (Ga) higher magnification insert of basilar dendrite showing spinedensity; (H) extraverted pyramidal neuron. Scale bar 5 100 mm (A–H),and 50 mm (Ga).

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 21

DISCUSSION

The present study is the first to document the corticalorganization of the brain of a hippopotamid, the pygmyhippopotamus, H. liberiensis. It is known that there are

cytoarchitectural and neurochemical similarities betweencetaceans and artiodactyls (Hof et al., 1999), and manyrecent phylogenetic analyses have supported a sister-group relationship between cetaceans and hippopotamids.

Fig. 14. Photomicrographs of Golgi-impregnated neurons observedin the frontal magnocellular cortex (A–D, F–H) and in the ACC (E) ofthe pygmy hippopotamus. (A, D) magnopyramidal neuron; (Aa) highermagnification inset of apical dendrite showing spine density. (B, E)

pyramidal neurons; (C) a, multiapical pyramidal neuron; b, extravertedpyramidal neuron; (F) multiapical pyramidal neuron; (G, H) extravertedpyramidal neurons. Note the presence of bifurcating apical dendrites.Scale bars 5 80 mm (A–C, F–H), 200 mm (D, E), and 50 mm (Aa).

22 BUTTI ET AL.

In comparing brain structure between the pygmy hippo-potamus and cetaceans, some features are primitive (ple-siomorphic) and shared widely with other cetartiodactylsand other mammals. Other features are specialized (apo-morphic, derived) relative to other cetartiodactyls. Forthese, a feature present in only one group, either hippo-potamus or cetacean, might reflect the amphibious orwholly aquatic lifestyle in that group, whereas a derivedfeature present in both groups could be a sharedamphibious-aquatic specialization. Alternatively, such afeature might have evolved convergently in the twogroups. Some judgment is needed in interpreting a fea-ture as being an adaptation to an amphibious-aquatichabitat, because fossil evidence does not indicate whetherthe most recent common ancestor of hippopotamids andcetaceans was terrestrial or amphibious-aquatic.

GI and CCA/BM Ratio

For its size and degree of convolution, the externalanatomy of the pygmy hippopotamus brain resemblesthat of other terrestrial artiodactyls, diverging extremelyfrom the large size and high degree of convolution of thebrain of small and large cetaceans (Kojima, 1951;Breathnach, 1955; Jacobs et al., 1971, 1979; Morganeet al., 1980; Marino et al., 2001a,c, 2002, 2003a,b,2004a,b; Hof et al., 2005; Hof and Van der Gucht, 2007;Manger et al., 2012; Fig. 1). Based on visual inspection,this reduced gyrification relative to cetaceans appeared

in all regions of the cortex, particularly in the insularcortex and in the temporal lobe (Fig. 2; Morgane et al.,1980; Manger et al., 1998; Marino et al., 2004a; Hof andVan der Gucht, 2007; Butti and Hof, 2010). This is con-firmed by the calculation of the GI of the pygmy hippo-potamus (1.84) that fits within the range of GIpreviously calculated for artiodactyls (Manger et al.,2012). Such value is considerably lower than the GI cal-culated by Manger et al. (2012) in cetaceans (average GIof six species 5 5.43) and in pinnipeds (average GI of twospecies 5 2.45). Comparison of the pygmy hippopotamusGI value to the log-transformed data indicating the rela-tionship between brain mass and GI calculated by Man-ger et al. (2012), indicates that the brain of the pygmyhippopotamus appears to have the value of GI expectedfor an artiodactyl brain of its size. In their extensiveanalysis of GI, Manger et al. (2012) have found that thethree groups of marine mammals that they have studied(cetaceans, pinnipeds, and manatees) possess GI valuesthat differ consistently from those of terrestrial mam-mals. In this respect, the pygmy hippopotamus fits wellinto the group of artiodactyls and differs from cetaceans(Table 4).

Quantitatively, the values of the CCA/BM ratioobtained with both the method of Tarpley and Ridgway(2000) (CCA/BM 5 0.737; Table 2) and the method ofManger et al. (2010) (CCA/BM 5 0.22; Table 3) showedthat the corpus callosum was larger in comparison tothe mass of the hemispheres than in cetaceans. The

Fig. 15. Photomicrographs of a Golgi-impregnated large magnopyramidal neuron of the frontal magno-cellular cortex, (A) and von Economo neurons (VENs) of the cortex of the lateral gyrus (B, C). Note thebifurcating apical dendrite in the magnopyramidal neuron and the presence of a diminutive basilar dendri-tic skirt in the VENs. Scale bar 5 100 mm.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 23

latter are, in fact, characterized by a relatively smalland thin corpus callosum (with CCA/BM ratios rangingfrom 0.079 in the killer whale, Orcinus orca, to 0.431 inthe Ganges river dolphin, Platanista gangetica as inTarpley and Ridgway, 2000, and from 0.11 in the North-ern right whale dolphin, Lissodelphis borealis to 0.15 inthe bottlenose dolphin, Tursiops truncatus, and othersmall odontocetes as in Manger et al., 2010; Tables 2and 3). The same conclusion can be drawn when compar-ing the CCA/BM value of the pygmy hippopotamus,obtained as in Manger et al. (2010), with the valuesreported by these authors for a group of artiodactyls

(CCA/BM ratios spanning from 0.16 in the dromedary,Camelus dromedarius, to 0.18 in the reindeer, Rangifertarandus. including a value of 0.17 in the river hippopot-amus, H. amphibius) or other eutherian mammals ofcomparable brain size (see Manger et al., 2010; Table 3).As such, the CCA/BM ratio for the pygmy hippopotamusfell above the range found by Manger and colleagues(2010) for cetaceans (0.11–0.15), for other aquatic speciesincluding pinnipeds and sirenians (0.13–0.17), and forartiodactyls (0.16–0.18), including artiodactyls of compa-rable brain mass (e.g., reindeer, R. tarandus, andscimitar-horned Oryx, Oryx dammah). The pygmy

Fig. 16. Photomicrographs showing different morphologies of vonEconomo neurons (VENs) in Nissl-stained sections through all corticalregions of the pygmy hippopotamus. Note the presence of VENs withstout cell bodies (A, C, D, F) similar to what has been described incetaceans (Butti et al., 2009) and the more occasional presence of

very elongated and slender VENs (B), comparable to those seen inhominids. Note also the presence of fork neurons (A, E). VENs areindicated by arrows and fork neurons by arrowheads. Scale bar 5 50mm.

24 BUTTI ET AL.

hippopotamus has the highest CCA/BM ratio amongartiodactyls examined to date and such value was com-parable only to the value obtained in certain perissodac-tyls (Rhinoceros, Diceros bicornis), carnivores (brownbear, Ursus arctos), numerous primates, and two femalesIndian elephants, Elephas maximus (Manger et al.,2010). Our results, together with those reported by Man-ger et al. (2010) and Tarpley and Ridgway (2000), indi-cate that the brain of the pygmy hippopotamus is verydifferent, in terms of mid-sagittal cross-sectional area ofthe corpus callosum, from that of cetaceans. It thusappears that cetaceans, as previously suggested, mayhave evolved a different morphology from other mam-mals (Manger et al. 2010). Interestingly, the CCA/BMvalue found in our specimen using the method of Tarp-ley and Ridgway (2000) is comparable to the valuesreported by these authors for the Florida manatee, Tri-chechus manatus manatus (CCA/BM 5 0.829). However,the CCA/BM ratio reported by Manger et al. (2010) forthe Amazonian manatee, Trichechus inunguis (CCA/BM 5 0.17), the smallest species of manatees, is lowerthan the value calculated with the same method for thepygmy hippopotamus (0.22). These contrasting resultscould be due to a specific feature of this particular spe-cies of manatees or, more likely, to lack of methodologicalprecision, as most of these studies are based on a singlespecimen due to their rarity. As such, differences foundbetween the pygmy hippopotamus and other eutherianmammals of comparable brain size, and aquatic mam-mals in general, deserve additional research. This couldbe done ideally by increasing the sample size of a givenspecies and by comparing the pygmy hippopotamus tothe larger river hippopotamus, H. amphibius, to deter-mine if the CCA/BM results are due to an outlier speci-men or whether they represent a general patterndisplayed by the pygmy hippopotamus or hippopota-muses in general.

Volumetric Measurements, CQ, and CI

In the pygmy hippopotamus the cerebellum was notas large relative to the cerebral hemisphere as in ceta-ceans (occupying 9.72% of the total brain volume in thepygmy hippopotamus, a range of 13.77% to 14.98% inthe white-sided dolphin, Lagenorhynchus acutus, 15.07%in the common dolphin, Delphinus delphis, and 15.05%in the bottlenose dolphin, T. truncatus; Fig. 2, Table 1;Breathnach, 1960; Marino et al., 2000). This conclusionis reflected in the calculation of the cerebellar quotient(CQ) of the pygmy hippopotamus (CQ 5 0.80; Table 4)based on the equation given by Maseko et al. (2012).The cerebellar volume of the pygmy hippopotamus was0.8 times the size that would be expected based on themammalian baseline calculated by Maseko et al.; adirect comparison with a cetacean of similar brain size,the franciscana (Pontoporia blainvillei; Maseko et al.,2012) shows that the relative amount of neural tissuededicated to the cerebellum is much reduced in thepygmy hippopotamus compared with the cetacean. Cal-culation of the cerebellar index (cerebellar volume/brainvolume) for the two species (pygmy hippopotamusCI 5 0.97; Pontoporia CI 5 1.52) supports this conclusion.The CQ of the pygmy hippopotamus is lower than thatof each species of cetaceans examined to date (Masekoet al., 2012) and it is consistent with an enlarged cere-

bellum being, within cetartiodactyls, a characteristicthat evolved after the transition from a terrestrial to afully aquatic environment. Our calculation of CQ and CIin additional species of artiodactyls and carnivores (bothaquatic and terrestrial; Table 4) shows that these valuesfor the pygmy hippopotamus fit well into the range ofCQs and CIs of artiodactyls and that they are also com-parable to CQs and CIs of carnivores of comparablebrain mass (harbor seal, Phoca vitulina, and Africanlion, Panthera leo; Table 4).

The hippocampus was larger, in both absolute sizeand size relative to the whole brain volume, in compari-son to what has been reported in cetaceans (e.g., white-sided dolphin: left hippocampus volume range 5 0.544 to1.043 cm3, right hippocampus volume range 5 0.462 to0.967 cm3; Table 1; Breathnach and Goldby, 1954;Breathnach, 1960; Filimonoff, 1965; Gruenberger, 1970;Jacobs et al., 1979; Morgane et al., 1982; Schwerdtfegeret al., 1984; Marino et al., 2003b, 2004a,b; Hof and Vander Gucht, 2007; Montie et al., 2008). In addition, theentorhinal cortex in the pygmy hippopotamus did notappear as enlarged as in cetaceans. When comparing thevolume of the hippocampus measured in the presentstudy to that of the few artiodactyls for which data areavailable (range 2,667.84–8,577.12 mm3; Reep et al.,2007), it becomes apparent that hippocampal volume inthe pygmy hippopotamus (8,530 cm3) is among the high-est values reported for artiodactyls. More hippopotamusspecimens are necessary to confirm such a conclusion.

All cetacean species reported to date have exhibited arelatively small hippocampus, which putatively indicatesan adaptation to the aquatic environment. This uniquefeature of the cetacean brain has been suggested toreflect a different organization of the circuits supportingmemory processing, compared with terrestrial mammals(Squire and Zola-Morgan, 1991; Zola-Morgan, 1993; Hofand Van der Gucht, 2007). For other aquatic mammals,the sirenians, discordant results have been reported onthe size of the hippocampus. Some authors havereported a reduced hippocampus in sirenians as in ceta-ceans (Oelschl€ager and Oelschl€ager, 2002; Marino,2007), whereas others have noted the hippocampal vol-ume in sirenians to be comparable to that of many ter-restrial mammals (Reep et al 2007). Larger sample sizesand comparison with the structure of the hippocampalformation of amphibious carnivores (pinnipeds: seals,walruses, and sea lions; and otters) could help elucidatethis matter further.

Differences in the development of the hippocampalformation between cetaceans and their closest relative,the pygmy hippopotamus (as well as other artiodactyls),could be related specifically to differences in spatialmemory and aquatic navigational demands. The anatom-ical structure of the hippocampal formation in aquaticmammals might reflect migratory patterns insofar asmigration is considered a major determinant of the sizeof the hippocampus in birds (Pravosudov et al., 2006). Toinvestigate this matter further, marine mammals clearlyrepresent an excellent group of study because of therange of variation in their migratory behavior (forreview see Stern, 2009). Our observation of an enlargedhippocampus in the pygmy hippopotamus, even com-pared with other artiodactyls (Table 1), is consistentwith a diminutive hippocampus being an adaptation tothe obligatory aquatic lifestyle of cetaceans.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 25

When considering the volumetrics of the cerebellumand hippocampus the brain of H. liberiensis can be con-sidered an artiodactyl-like brain. However, a differentsituation arises when considering the corpus callosum,as the CCA/BM ratio is higher than any artiodactyl (andcetacean) analyzed to date.

Cytoarchitecture and Chemoarchitecture

The pygmy hippopotamus cortex lacks an internalgranular layer (layer IV: Figs. 3–10) as cetaceans (Mor-gane et al., 1988; Glezer and Morgane, 1990; Hof et al.,2005; Hof and Van der Gucht, 2007; Oelschl€ager andOelschl€ager, 2008). There has been a long debate aboutthe presence or absence of an internal granular layer(layer IV) in artiodactyls and cetaceans and, to date, anincipient layer IV has been described in the primary vis-ual cortex of certain artiodactyls (Rose, 1942), but it isgenerally recognized that the cetacean neocortex lacksthis layer. In this respect, the cortex of the pygmy hippo-potamus is cetacean-like and very different from the cor-tex of other artiodactyls. The absence of a layer IV mayindicate a specific wiring pattern (Hof and Van derGucht, 2007), possibly associated with a shift of thalamicinputs to layer I rather than deeper layers (Ferrer andPerera, 1988; Glezer and Morgane, 1990). Second, thecerebral cortex of the pygmy hippopotamus containsextensive clustering of neurons in layer II, not only inthe anterior insular cortex, but also in large portions ofthe occipital cortex (Fig. 10). Although neuronal cluster-ing of layer II is observed in the anterior insula of bothodontocetes and mysticetes (Jacobs et al., 1984; Mangeret al., 1998; Hof and Van der Gucht, 2007; Butti andHof, 2010), this finding is important because, to date,clustering of layer II in the occipital cortex has onlybeen formally documented in two balaenopterids, namelythe humpback whale and the fin whale (Balaenopteraphysalus) (Hof and Van der Gucht, 2007). Such a patternwas suggested to be a rorqual specialization and to rep-resent a specific type of cortical connectivity in the occi-pital cortex, possibly shaped by thalamic afferents in theabsence of layer IV (Hof and Van der Gucht, 2007). Ourresults suggest that, although the clustering pattern inlayer II of the anterior insula has been retained by allcetaceans during the adaptation to the aquatic life, themodularity of layer II in the occipital cortex, and itspotentially related cortical connectivity, may have beenretained in certain species of cetaceans (rorquals) butnot in others (odontocetes). Alternatively this modulararchitecture of the occipital cortex might have evolvedindependently in the pygmy hippopotamus and rorquals.Observations in a limited number of species support thepresence of clustering of layer II in the occipital cortexof other artiodactyls (Hof, personal communication).However, the degree to which such clustering occurs andthe distribution of this feature among cetartiodactylshas not yet been formally documented. This issuerequires an in depth study which is beyond the aim ofthis article.

We also noted that the cortex of the pygmy hippopota-mus showed layer II clustering beyond the anterior insu-lar and occipital cortices. The putative somatosensorycortex, in this species, is located consistently with whathas been defined as somatosensory cortex in cetaceans(Lende and Welker, 1972; Hof and Van der Gucht, 2007)

and is characterized by a high degree of layer II cluster-ing throughout its entire extent. This trait can be easilyused to define its cytoarchitectural boundaries fromadjacent cortical fields (see above, Frontopolar Cortexand Frontal Convexity and Parietal Cortex sections fordetails). Such an extensive clustering could be a corticalspecialization of hippopotamids, partially retained dur-ing the adaptation to the aquatic environment by somecetacean groups. In this context, the analysis of thepresence and distribution of layer II islands in the cortexof the larger river hippopotamus, and other artiodactyls,in comparison to H. liberiensis, should elucidate the pos-sible function and evolutionary significance of suchpattern.

In terms of distribution of neocortical fields on thesurface of the hemisphere, the brain of the pygmy hippo-potamus is probably comparable to that of cetaceans andartiodactyls, but the distribution of the visual and soma-tosensory cortices deserves some additional comments.In fact, although the putative motor and somatosensorycortices are arranged consistently with their localizationin cetaceans (Lende, 1968; Lende and Welker, 1972), theputative somatosensory cortex of the pygmy hippopota-mus extends caudally for a greater length of the lateralgyrus, confining the putative visual cortex to only itsmost caudal part, as seen in artiodactyls, and in contrastto cetaceans where the somatosensory cortex is limitedto a rather small area in the frontal convexity of thehemisphere.

The putative motor cortex of the pygmy hippopotamuswas very similar to that recognized both electrophysio-logically and cytoarchitecturally as the motor cortex incetaceans (Kojima, 1951; Lende, 1968; Kesarev andMalofeeva, 1969; Ladygina and Supin, 1977; Hof andVan der Gucht, 2007). It is interesting to note that incetaceans, the belt of motor cortex in the frontal convex-ity extends rostro-caudally and is separated laterallyfrom the somatosensory cortex by the cruciate sulcus(Kesarev and Malofeeva, 1969; Lende and Welker, 1972;Sokolov et al., 1972; Ladygina and Supin, 1977; Glezer,2002). However, although the cruciate sulcus develops ina rostrocaudal direction in cetaceans, possibly as a con-sequence of the rostroventral rotation that the cetaceanbrain undergoes during development, the cruciate sulcusdevelops laterally in the pygmy hippopotamus, reachingthe most dorsal aspect of the anterior Sylvian cortex.

While in cetaceans almost the totality of the lateralgyrus is occupied by visual cortex (Sokolov et al., 1972;Ladygina and Supin, 1977; Revishchin and Garey, 1989;Hof and Van der Gucht, 2007), in the pygmy hippopota-mus the putative visual cortex occupied only the mostposterior aspect of the lateral gyrus. This location isreminiscent of the location of the visual cortex in manyterrestrial mammals such as carnivores (Sanides andHoffmann, 1969; Sherk, 1986; Payne, 1993; Innocentiet al., 2002; Manger et al., 2004) and other artiodactyls(Campbell, 1905; Alouf, 1929; Rose, 1942).

The distribution of CR-immunoreactive neurons in thecortex of the pygmy hippopotamus is relatively uniformthroughout cortical regions and comparable to what hasbeen observed in the cortex of cetaceans and several speciesof artiodactyls (Glezer et al., 1993, 1998; Hof et al., 1996a,b,1999; Fig. 11). In addition, the morphologies of CR-immunoreactive neurons in the pygmy hippopotamus aresimilar to what has been observed in cetaceans (Fig. 12).

26 BUTTI ET AL.

However, the frequent presence of giant CR-immunoreactive multipolar neurons in deep layers isuniquely shared with artiodactyls, to the exclusion of ceta-ceans. Additionally, the CR- immunoreactive multipolarneurons in layer III of the temporal cortex (Fig. 12C)resembled those observed in layer III and V of the primarymotor cortex of carnivores (for review see Hof et al., 1999).The large CR-bipolar and multipolar neurons of layer VI(Fig. 12F,G) of the putative somatosensory and visual corti-ces, were reminiscent of the large multipolar neuronsdescribed in layer VI of the giraffe (Giraffa cameloparda-lis), llama (Lama glama), and Bactrian camel (Camelusdromedarius) (Hof et al., 2000). Such pattern of distributionand morphologies of CR-immunoreactive neurons is inter-esting in light of the recent morphological and molecularevidence of phylogenetic relatedness between hippopota-muses and cetaceans and are consistent with the occur-rence of shared evolutionary traits between cetaceans andartiodactyls (Glezer et al., 1992, 1993, 1998; Hof et al.,1996a, 1999, 2000). In fact, the CR immunoreactivity in thepygmy hippopotamus cerebral cortex reveals patternsshared with artiodactyls, as well as morphologies sharedwith cetaceans (Figs. 11 and 12).

Calcium-binding proteins, including CR, are foundmostly in inhibitory interneurons (Gilbert, 1983; DeFe-lipe, 1993; Markram et al., 2004). Thus, the predominantdistribution of CR-immunoreactive neurons in the upperlayers of the cetacean cortex, together with the findingthat the majority of the neocortical synapses in ceta-ceans are contained in layer I (Glezer and Morgane,1990), suggests that these superficial layers may be therecipient of thalamocortical afferents that require exten-sive modulatory and inhibitory microcircuits (Glezeret al., 1992). Without direct evidence, however, itremains impossible to extend such interpretation to thepygmy hippopotamus and ungulates in general. It isnevertheless intriguing to think that, as suggested byHof et al. (2000), calcium-binding protein-immunoreac-tive interneurons could subserve similar functions acrossmammalian taxa.

Von Economo Neurons and Other NeuronalMorphologies

Compared with hominids, elephants, and cetaceans, inwhich VENs occur selectively in layer V of the ACC,frontoinsular cortex, and frontopolar and dorsolateralprefrontal cortex, the pygmy hippopotamus representsan interesting case characterized by the presence of avery large number of VENs that have a radically differ-ent cortical distribution which deserves some discussion.Whereas VENs have been proposed to be involved insocial, cognitive, and emotional networks in humans,great apes, macaque monkeys, elephants, and cetaceans(Hof and Van der Gucht, 2007; Hakeem et al., 2009; All-man et al., 2010; Evrard et al., 2012), there is no firmevidence that they subserve in fact such functions, andthat they would do so in all species in which they occur.VENs are a distinct neuron type that we showed to havedifferent distribution in different species. Such differen-tial distribution across the neocortex indicates that theymight also be involved in other functions.

In the neocortex of the pygmy hippopotamus, VENsexhibit a ubiquitous distribution pattern across regionsthat differs markedly from their selective high density in

ACC, frontoinsular cortex, and frontopolar cortex in ceta-ceans, elephants, humans and great apes (Nimchinskyet al., 1995, 1999; Hof and Van der Gucht, 2007; Buttiet al., 2009, 2011; Hakeem et al., 2009; Allman et al.,2010; Evrard et al., 2012). There is evidence of the pres-ence of VENs in the frontopolar cortex of certain artiodac-tyls including the domesticated pig (Sus scrofadomesticus), sheep (Ovis aries), cow (Bos taurus), andwhite-tailed deer (Odocoileus virginianus) (Raghantiet al., in press). However, the lack of information on thecortical distribution of VENs beyond the frontopolar cor-tex makes it impossible to compare the distribution ofVENs between the pygmy hippopotamus and these spe-cies. VENs are present also in layer II of the frontopolarcortex in the few artiodactyl species described so far.Such observation is consistent with a laminar distributionof VENs in the pygmy hippopotamus comparable to whatdescribed in hominids, elephants, and cetaceans, but dif-ferent from artiodactyls (at least at a laminar level andin the species observed so far).

Sparse and rare VENs have also been observed acrossthe cortical mantle in another aquatic mammal, theFlorida manatee (T. manatus manatus) (Butti and Hof,2010), which is closely related to elephants (Kellogget al., 2007). Such differences are consistent with a reor-ganization of the distribution of VENs into a regionallyspecific pattern. This distribution, which perhaps trans-lates into a more specific projection pattern and func-tion, may have taken place during the evolution ofcetaceans, but not in hippopotamuses. The presentresults, together with those of Hakeem et al. (2009), sug-gest that, although VENs were present before the transi-tion to an aquatic environment, their distribution wasdrastically modified in cetaceans, compared with theirclosest terrestrial relatives, in the direction of a morerefined and specific function in cetaceans, in contrast totheir closest terrestrial relatives. Moreover, such an evo-lutionary transition between a “general” pattern of dis-tribution to a more specific one happened at least twicein the descent of elephants and cetaceans.

Another intriguing observation is the presence, inpygmy hippopotamus cortex, of a neuronal morphologythat is reminiscent of the taproot neuron (called the“matriarch neuron” by Jacobs et al., 2011; Fig. 13)observed in the frontal cortex of the African elephant(Jacobs et al., 2011). These neurons, possessing a com-plex and extensive dendritic system may representextreme adaptations of pyramidal neurons, much thesame as other large cortical neurons in primates (e.g.,Betz, Meynert, and von Economo’s spindle cells) (Jacobset al., 2011; p 292). Based on their peculiar morphology,Jacobs et al. (2011; p 292) suggested that these neuronsmight sample a wide range of cortical inputs and exertmodulatory field effect in the surrounding neuropil,potentially contributing to associative cognitive processes.Such neuronal morphology has not been reported incetaceans, and its presence in the visual cortex of thepygmy hippopotamus, although difficult to interpret,points to the need for a detailed study of neuronal mor-phology in artiodactyls and cetaceans.

In addition, the presence of extraverted neurons inlayers II, III, and V in the cortex of the pygmy hippopot-amus (Figs. 13 and 14) is worth noting. Such neuronswith apical dendrites spreading into layer I have beenreported in the cortex of cetaceans, but exclusively in

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 27

layer II (Morgane et al., 1985; Ferrer and Perera, 1988;Morgane et al., 1988), and are generally considered aconservative feature of the cetacean neocortex (Mor-gane et al., 1985, 1988; Glezer, 1988) as a comparableneuronal morphology was previously observed in mam-mals such as bats, hedgehogs, opossums, xenarthansand afrotherians (Sanides and Sanides, 1972; Sherwoodet al., 2009). In cetaceans extraverted neurons havebeen proposed by Glezer and Morgane (1990) to repre-sent strong neuronal inputs into layer I of the axoden-dritic type which, thus, resemble the pattern of inputseen in allocortical formations (Glezer and Morgane,1990; p 418).

Our results show that such neuronal morphology ispresent in hippopotamids as well (Fig. 13Ab,H; Fig.14Cb,G,H). As such, this morphology appears to havedeveloped in cetartiodactyls before their transition to anaquatic environment. However, the distribution of extra-verted pyramidal neurons is different in the hippopota-mus and cetaceans, being restricted to layer II in thelatter, but similar to xenarthrans and afrotherians(Sherwood et al., 2009). The functional significance ofthe different distribution of this neuronal type is difficultto assess and the analysis of neuronal morphologies inthe river hippopotamus, as well as other terrestrialartiodactyls, could elucidate this matter further.

Methodological Limitations and FutureDirections

Methodological limitations regarding the Golgi stainingof postfixed tissue have been discussed elsewhere (Jacobset al., 2011). The major limitation in the present studywas the impossibility to trace completely dendritic sys-tems and, thus, to provide a quantitative description ofthe neuronal morphologies identified. For this reason ouranalysis of neuronal morphologies has focused solely onqualitative aspects. Another limitation of the presentstudy was the availability of only one specimen of H. lib-eriensis due to its relatively rarity and “Endangered” sta-tus according to the IUCN 2013 red list (IUCN Red Listof Threatened Species, Version 2013.2). Consequently, wecould not assess intraspecific and sex variability in theorganization of the cerebral cortex. Moreover, the pygmyhippopotamus represents a distinct lineage from otherhippopotamids (Boisserie, 2005). As such, our results canbe tentatively extended to other hippopotamids only ingeneral terms. Our results provide an initial qualitativedescription of the cortical organization in one hippopota-mid that can be used as a reference in future studiesattempting to define differences in cortical organization,and specializations, between H. liberiensis and the largerH. amphibius. In this context, the study of the brain ofH. amphibius would be necessary to assess if differencesin brain size and lifestyle are reflected at a neuroanatom-ical level within hippopotamids.

ACKNOWLEDGEMENTS

The authors thank the Cleveland Metroparks Zoo fordonation of the brain specimen of H. liberiensis, Drs.J.M. Allman, A.D. Craig, W.W. Seeley, and K.F. Manayefor helpful discussion, and W.G.M. Janssen for experttechnical assistance.

LITERATURE CITED

Agnarsson I, May-Collado LJ. 2008. The phylogeny of Cetartiodac-tyla: the importance of dense taxon sampling, missing data, andthe remarkable promise of cytochrome b to provide reliablespecies-level phylogenies. Mol Phyl Evol 48:964–985.

Allman JM, Tetreault NA, Hakeem AY, Manaye KF, Semendeferi K,Erwin JM, Goubert V, Hof PR. 2010. The von Economo neuronsin frontoinsular and anterior cingulate cortex of great apes andhumans. Brain Struct Funct 214:495–517.

Alouf J. 1929. Die vergleichlende Cytoarchitektonik der Areastriata. J Psychol Neurol 38:5–41.

Andressen C, Bl€umcke I, Celio MR. 1993. Calcium-binding proteins:selective markers of nerve cells. Cell Tissue Res 271:181–208.

Baimbridge KG, Celio MR, Rogers JH. 1992. Calcium-binding pro-teins in the nervous system. Trends Neurosci 15:303–308.

Boisserie J. 2005. The phylogeny and taxonomy of Hippopotamidae(Mammalia: Artiodactyla): a review based on morphology andcladistic analysis. Zool J Linn Soc 143:1–26.

Boisserie JR, Fisher RE, Lihoreau F, Weston EM. 2011. Evolvingbetween land and water: key questions on the emergence and his-tory of the Hippopotamidae (Hippopotamoidea, Cetancodonta,Cetartiodactyla). Biol Rev 86:975–975.

Breathnach AS. 1955. The surface features of the brain of thehumpback whale (Megaptera novaeangliae). J Anat 89:343–354.

Breathnach AS. 1960. The cetacean central nervous system. BiolRev 35:187–230.

Breathnach AS, Goldby F. 1954. The amygdaloid nuclei, hippocam-pus and other parts of the rhinencephalon in the porpoise (Pho-caena phocaena). J Anat 88:267–291.

Butti C, Hof PR. 2010. The insular cortex: a comparative perspec-tive. Brain Struct Funct 214:477–493.

Butti C, Raghanti MA, Sherwood CC, Hof PR. 2011. The neocortexof cetaceans: cytoarchitecture and comparison with other aquaticand terrestrial species. Ann NY Acad Sci 1225:47–58.

Butti C, Sherwood CC, Hakeem AY, Allman JM, Hof PR. 2009. Totalnumber and volume of Von Economo neurons in the cerebral cor-tex of cetaceans. J Comp Neurol 515:243–259.

Campbell A. 1905. Histological studies on the localisation of cere-bral function. Cambridge: University Press.

Cond�e F, Lund JS, Jacobowitz DM, Baimbridge KG, Lewis DA.1994. Local circuit neurons immunoreactive for calretinin, calbin-din D-28k or parvalbumin in monkey prefrontal cortex: distribu-tion and morphology. J Comp Neurol 341:95–116.

Corbet GB. 1969. The taxonomic status of the pygmy hippopotamus,Choeropsis liberiensis, from the Niger Delta. J Zool 158:387–394.

Davila JC, Padial J, Andreu MJ, Real MA, Guirado S. 1997. Calre-tinin immunoreactivity in the cerebral cortex of the lizard Psam-modromus algirus: a light and electron microscopic study. J CompNeurol 382:382–393.

DeFelipe J. 1993. Neocortical neuronal diversity: chemical heteroge-neity revealed by colocalization studies of classic neurotrans-mitters, neuropeptides, calcium-binding proteins, and cell surfacemolecules. Cereb Cortex 3:273–289.

DeFelipe J. 1997. Types of neurons, synaptic connections and chem-ical characteristics of cells immunoreactive for calbindin-D28K,parvalbumin and calretinin in the neocortex. J Chem Neuroanat14:1–19.

Eltringham S. 1993a. The common hippopotamus (Hippopotamusamphibius). In: Oliver WLR, editor. Pigs, peccaries and hippos:status survey and action plan. IUCN, The World ConservationUnion, Gland, Switzerland. p 71–87.

Eltringham S. 1993b. The pygmy hippopotamus (Hexaprotodon lib-eriensis). In: Oliver WLR, editor. Pigs, peccaries and hippos: sta-tus survey and action plan. IUCN, The World ConservationUnion, Gland, Switzerland. p 87–94.

Eltringham S. 1999. The hippos: natural history and conservation.London: Academic Press.

Evrard HC, Forro T, Logothetis NK. 2012. Von Economo neurons inthe anterior insula of the macaque monkey. Neuron 74:482–489.

Fajardo C, Escobar MI, Buritica E, Arteaga G, Umbarila J,Casanova MF, Pimienta H. 2008. Von Economo neurons are

28 BUTTI ET AL.

present in the dorsolateral (dysgranular) prefrontal cortex ofhumans. Neurosci Lett 435:215–218.

Fay JM. 2004. In the land of the surfing hippos. Natl Geogr Mag206:100.

Ferrer I, Perera M. 1988. Structure and nerve cell organisation inthe cerebral cortex of the dolphin Stenella coeruleoalba: a golgistudy. With special attention to the primary auditory area. AnatEmbryol 178:161–173.

Filimonoff IN. 1965. On the so-called rhinencephalon in the dol-phin. J Hirnforsch 8:1–23.

Furutani R. 2008. Laminar and cytoarchitectonic features of thecerebral cortex in the Risso’s dolphin (Grampus griseus), stripeddolphin (Stenella coeruleoalba), and bottlenose dolphin (Tursiopstruncatus). J Anat 213:241–248.

Garey LJ, Takacs J, Revishchin AV, Hamori J. 1989. Quantitativedistribution of GABA-immunoreactive neurons in cetacean visualcortex is similar to that in land mammals. Brain Res 485:278–284.

Garrod AH. 1880. On the brain and other parts of the hippopota-mus (H. amphibius). Trans Zool Soc Lond 11:11–17.

Gatesy J. 1997. More DNA support for a Cetacea/Hippopotamidaeclade: the blood-clotting protein gene gamma-fibrinogen. Mol BiolEvol 14:537–543.

Gilbert CD. 1983. Microcircuitry of the visual cortex. Annu RevNeurosci 6:217–247.

Gingerich PD. 2005. Cetacea. In: Rose KD, Archibald JD, editors. Pla-cental mammals: origin, timing, and relationships of the major extantclades. Baltimore: Johns Hopkins University Press. p 234–252.

Gingerich PD, Uhen MD. 1998. Likelihood estimation of the time oforigin of cetacean and the time of divergence of Cetacea andArtiodactyla. Paleo Electron 2:1–47.

Glezer, II, Hof PR, Leranth C, Morgane PJ. 1993. Calcium-bindingprotein-containing neuronal populations in mammalian visualcortex: a comparative study in whales, insectivores, bats, rodents,and primates. Cereb Cortex 3:249–272.

Glezer, II, Hof PR, Morgane PJ. 1992. Calretinin-immunoreactiveneurons in the primary visual cortex of dolphin and humanbrains. Brain Res 595:181–188.

Glezer, II, Hof PR, Morgane PJ. 1998. Comparative analysis ofcalcium-binding protein-immunoreactive neuronal populations inthe auditory and visual systems of the bottlenose dolphin (Tur-siops truncatus) and the macaque monkey (Macaca fascicularis).J Chem Neuroanat 15:203–237.

Glezer, II, Morgane PJ. 1990. Ultrastructure of synapses and golgianalysis of neurons in neocortex of the lateral gyrus (visual cor-tex) of the dolphin and pilot whale. Brain Res Bull 24:401–427.

Glezer I. 2002. Neural morphology. In: Hoelzel AR, editor. Marinemammal biology: an evolutionary approach: Oxford: BlackwellScience. p 98–115.

Glezer I, Jacobs, MS, Morgane, PJ. 1988. Implications of the "initialbrain" concept for brain evolution in Cetacea. Behav Brain Sci 11:75–116.

Gonchar Y, Burkhalter A. 1997. Three distinct families of GABAer-gic neurons in rat visual cortex. Cereb Cortex 7:347–358.

Grubb P. 1993. The Afro-tropical hippopotamuses—hippopotami-dae—Hippopotamus and Hexaprotodon. In: Oliver WLR, editor.Pigs, peccaries and hippos: status survey and action plan. IUCN,The World Conservation Union, Gland, Switzerland. p 68–71.

Gruenberger H. 1970. On the cerebral anatomy of the Amazon dol-phin Inia geoffrensis. Invest Cetacea 2:129–146.

Hakeem AY, Sherwood CC, Bonar CJ, Butti C, Hof PR, Allman JM.2009. Von Economo neurons in the elephant brain. Anat Rec 292:242–248.

Hof PR, Bogaert YE, Rosenthal RE, Fiskum G. 1996a. Distributionof neuronal populations containing neurofilament protein andcalcium-binding proteins in the canine neocortex: regional analy-sis and cell typology. J Chem Neuroanat 11:81–98.

Hof PR, Chanis R, Marino L. 2005. Cortical complexity in cetaceanbrains. Anat Rec 287:1142–1152.

Hof PR, Glezer, II, Cond�e F, Flagg RA, Rubin MB, Nimchinsky EA,Vogt Weisenhorn DM. 1999. Cellular distribution of the calcium-binding proteins parvalbumin, calbindin, and calretinin in the

neocortex of mammals: phylogenetic and developmental patterns.J Chem Neuroanat 16:77–116.

Hof PR, Glezer, II, Nimchinsky EA, Erwin JM. 2000. Neurochemicaland cellular specializations in the mammalian neocortex reflectphylogenetic relationships: evidence from primates, cetaceans,and artiodactyls. Brain Behav Evol 55:300–310.

Hof PR, Rosenthal RE, Fiskum G. 1996b. Distribution of neurofila-ment protein and calcium-binding proteins parvalbumin, calbin-din, and calretinin in the canine hippocampus. J ChemNeuroanat 11:1–12.

Hof PR, Sherwood CC. 2005. Morphomolecular neuronal phenotypesin the neocortex reflect phylogenetic relationships among certainmammalian orders. Anat Rec 287:1153–1163.

Hof PR, Van der Gucht E. 2007. Structure of the cerebral cortex ofthe humpback whale, Megaptera novaeangliae (Cetacea, Mysti-ceti, Balaenopteridae). Anat Rec 290:1–31.

Innocenti GM, Manger PR, Masiello I, Colin I, Tettoni L. 2002.Architecture and callosal connections of visual areas 17, 18, 19and 21 in the ferret (Mustela putorius). Cereb Cortex 12:411–422.

IUCN Red List of Threatened Species. 2013 (Version 2013.2).www.iucnredlist.org. Accessed 23 January 2014.

Jacobs B, Lubs J, Hannan M, Anderson K, Butti C, Sherwood CC,Hof PR, Manger PR. 2011. Neuronal morphology in the Africanelephant (Loxodonta africana) neocortex. Brain Struct Funct 215:273–298.

Jacobs MS, Galaburda AM, McFarland WL, Morgane PJ. 1984. Theinsular formations of the dolphin brain: quantitative cytoarchitec-tonic studies of the insular component of the limbic lobe. J CompNeurol 225:396–432.

Jacobs MS, McFarland WL, Morgane PJ. 1979. The anatomy of thebrain of the bottlenose dolphin (Tursiops truncatus). Rhinic lobe(Rhinencephalon): The archicortex. Brain Res Bull 4(Suppl 1):1–108.

Jacobs MS, Morgane PJ, McFarland WL. 1971. The anatomy of thebrain of the bottlenose dolphin (Tursiops truncatus). Rhinic lobe(rhinencephalon). I. The paleocortex. J Comp Neurol 141:205–271.

Kellogg ME, Burkett S, Dennis TR, Stone G, Gray BA, McGuirePM, Zori RT, Stanyon R. 2007. Chromosome painting in the man-atee supports Afrotheria and Paenungulata. BMC Evol Biol 7:6.

Kesarev VS, Malofeeva LI. 1969. Structural organization of themotor zone of the cerebral cortex in dolphins. Arkh Anat GistolEmbriol 56:48–55.

Kojima T. 1951. On the brain of the sperm whale (Physeter catodonL). Sci Rep Whales Res Inst Tokyo 6:49–72.

Kubota Y, Hattori R, Yui Y. 1994. Three distinct subpopulations ofGABAergic neurons in rat frontal agranular cortex. Brain Res649:159–173.

Ladygina TF, Mass AM, Supin A. 1978. Multiple sensory projectionsin the dolphin cerebral cortex. Zh Vyssh Nerv Deiat Im I P Pav-lova 28:1047–1053.

Ladygina TF, Supin A. 1977. Localization of the sensory projectionareas in the cerebral cortex of the dolphin, Tursiops truncatus. ZhEvol Biokhim Fiziol 13:712–718.

Lende R, Akdikmen,S. 1968. Motor field in cerebral cortex of thebottlenose dolphin. J Neurosurg 29:495–499.

Lende RA, Welker WI. 1972. An unusual sensory area in the cere-bral neocortex of the bottlenose dolphin, Tursiops truncatus.Brain Res 45:555–560.

Lydekker R. 1915. Catalogue of the ungulate mammals in the Brit-ish museum (Natural History). London: Order of Trustees.

Manger PR, Hemingway J, Haagensen M, Gilissen E. 2010. Cross-sectional area of the elephant corpus callosum: comparison toother eutherian mammals. Neuroscience 167:815–824.

Manger PR, Nakamura H, Valentiniene S, Innocenti GM. 2004. Vis-ual areas in the lateral temporal cortex of the ferret (Mustelaputorius). Cereb Cortex 14:676–689.

Manger PR, Prowse M, Haagensen M, Hemingway J. 2012. Quanti-tative analysis of neocortical gyrencephaly in African elephants(Loxodonta africana) and six species of cetaceans: comparisonwith other mammals. J Comp Neurol 520: 2430–2439.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 29

Manger PR, Sum M, Szymanski M, Ridgway SH, Krubitzer L. 1998.Modular subdivisions of dolphin insular cortex: does evolutionaryhistory repeat itself? J Cogn Neurosci 10:153–166.

Marin-Padilla M. 1990. Three dimensional organization of layer I of thehuman cerebral cortex: a Golgi study. J Comp Neurol 299:89–105.

Marino L. 2007. Cetacean brains: how aquatic are they? Anat Rec290:694–700.

Marino L, Murphy TL, Deweerd AL, Morris JA, Fobbs AJ, HumblotN, Ridgway SH, Johnson JI. 2001a. Anatomy and three-dimensional reconstructions of the brain of the white whale (Del-phinapterus leucas) from magnetic resonance images. Anat Rec262:429–439.

Marino L, Murphy TL, Gozal L, Johnson JI. 2001b. Magnetic reso-nance imaging and three-dimensional reconstructions of the brainof a fetal common dolphin, Delphinus delphis. Anat Embryol 203:393–402.

Marino L, Rilling JK, Lin SK, Ridgway SH. 2000. Relative volumeof the cerebellum in dolphins and comparison with anthropoidprimates. Brain Behav Evol 56:204–211.

Marino L, Sherwood CC, Delman BN, Tang CY, Naidich TP, HofPR. 2004a. Neuroanatomy of the killer whale (Orcinus orca) frommagnetic resonance images. Anat Rec 281:1256–1263.

Marino L, Sudheimer K, McLellan WA, Johnson JI. 2004b. Neuroa-natomical structure of the spinner dolphin (Stenella longirostrisorientalis) brain from magnetic resonance images. Anat Rec 279:601–610.

Marino L, Sudheimer K, Pabst DA, McLellan WA, Johnson JI.2003a. Magnetic resonance images of the brain of a dwarf spermwhale (Kogia simus). J Anat 203:57–76.

Marino L, Sudheimer K, Sarko D, Sirpenski G, Johnson JI. 2003b.Neuroanatomy of the harbor porpoise (Phocoena phocoena) frommagnetic resonance images. J Morphol 257:308–347.

Marino L, Sudheimer KD, Murphy TL, Davis KK, Pabst DA,McLellan WA, Rilling JK, Johnson JI. 2001c. Anatomy and three-dimensional reconstructions of the brain of a bottlenose dolphin(Tursiops truncatus) from magnetic resonance images. Anat Rec264:397–414.

Marino L, Sudheimer KD, Pabst DA, McLellan WA, Filsoof D,Johnson JI. 2002. Neuroanatomy of the common dolphin (Delphi-nus delphis) as revealed by magnetic resonance imaging (MRI).Anat Rec 268:411–429.

Markram H, Toledo-Rodriguez M, Wang Y, Gupta A, Silberberg G,Wu C. 2004. Interneurons of the neocortical inhibitory system.Nat Rev Neurosci 5:793–807.

Martinez-Guijarro FJ, Freund TF. 1992. Distribution of GABAergicinterneurons immunoreactive for calretinin, calbindin D28K, andparvalbumin in the cerebral cortex of the lizard Podarcis hispan-ica. J Comp Neurol 322:449–460.

Maseko BC, Spocter MA, Haagensen M, Manger PR. 2012. Ele-phants have relatively the largest cerebellum size for mammals.Anat Rec 295: 661–672.

Milinkovitch M, B�erub�e, M, Palsboll, PJ. 1998. Cetaceans arehighly derived artiodactyls. In: JGM Thewissen, editor. The emer-gence of whales. New York: Plenum Press. p 113–131.

Montie EW, Schneider GE, Ketten DR, Marino L, Touhey KE, HahnME. 2007. Neuroanatomy of the subadult and fetal brain of theAtlantic white-sided dolphin (Lagenorhynchus acutus) from insitu magnetic resonance images. Anat Rec 290:1459–1479.

Montie EW, Schneider G, Ketten DR, Marino L, Touhey KE, HahnME. 2008. Volumetric neuroimaging of the atlantic white-sideddolphin (Lagenorhynchus acutus) brain from in situ magnetic res-onance images. Anat Rec 291:263–282.

Morgane P, Jacobs M, McFarland W. 1980. The anatomy of thebrain of the bottlenose dolphin (Tursiops truncatus). Surface con-figuration of the telencephalon of the bottlenose dolphin withcomparative anatomical observations in four other cetaceans spe-cies. Brain Res Bull 5:1–107.

Morgane PJ, Glezer, II, Jacobs MS. 1988. Visual cortex of the dol-phin: an image analysis study. J Comp Neurol 273:3–25.

Morgane PJ, Jacobs MS, Galaburda A. 1985. Conservative featuresof neocortical evolution in dolphin brain. Brain Behav Evol 26:176–184.

Morgane PJ, McFarland WL, Jacobs MS. 1982. The limbic lobe ofthe dolphin brain: a quantitative cytoarchitectonic study. J Hirn-forsch 23:465–552.

Ngowyang G. 1932. Beschreibung einer Art von Spezialzellen in derInselrinde - zugleich Bemerkungen €uber die v. EconomoschenSpezialzellen. J Psychol Neurol 44:671–674.

Nikaido M, Rooney AP, Okada N. 1999. Phylogenetic relationshipsamong cetartiodactyls based on insertions of short and long inter-persed elements: hippopotamuses are the closest extant relativesof whales. Proc Natl Acad Sci USA 96:10261–10266.

Nimchinsky EA, Gilissen E, Allman JM, Perl DP, Erwin JM, HofPR. 1999. A neuronal morphologic type unique to humans andgreat apes. Proc Natl Acad Sci USA 96:5268–5273.

Nimchinsky EA, Vogt BA, Morrison JH, Hof PR. 1995. Spindle neu-rons of the human anterior cingulate cortex. J Comp Neurol 355:27–37.

O’Leary MA, Patel BA, Coleman MN. 2012. Endocranial petrosalanatomy of Bothriogenys (Mammalia, Artiodactyla, Anthracotherii-dae), and petrosal volume and density comparisons among aquaticand terrestrial artiodactyls and outgroups. J Paleontol 86:44–50.

Oelschl€ager HA, Oelschl€ager J. 2008. Brain. In: Perrin WF, W€ursigB, Thewissen JGM, editors. Encyclopedia of marine mammals.San Diego, CA: Academic Press. p 134–149.

Oelschl€ager HHA, Oelschl€ager JS. 2002. Brain. In: Perrin WF,W€ursig B, Thewissen JGM, editors. Encyclopedia of marine mam-mals. San Diego: Academic Press. p133–158.

Parmentier M, Lefort A. 1991. Structure of the human braincalcium-binding protein calretinin and its expression in bacteria.Eur J Biochem 196:79–85.

Payne BR. 1993. Evidence for visual cortical area homologs in catand macaque monkey. Cereb Cortex 3:1–25.

Pilleri G. 1962. Zur Anatomie des Gehirnes von Choeropsis liberien-sis Morton (Mammalia, Artiodactyla). Acta Zool 43:229–245.

Pilleri G, Peixun, C, Zuohua, S. 1980. Concise macroscopical atlasof the brain of the common dolphin (Delphinus delphis Linnaeus,1758). Bern-Waldau, Switzerland: Brain Anatomy Institute.

Popov VV, Supin A. 1976. Detemination of the hearing characteris-tics of the dolphin by the evoked potential method. Fiziol ZhSSSR Im I M Sechenova 62:550–558.

Popov VV, Supin A. 1986. Evoked potentials of the dolphin auditorycortex recorded from the body surface. Dokl Akad Nauk SSSR288:756–759.

Pravosudov VV, Kitaysky AS, Omanska A. 2006. The relationshipbetween migratory behaviour, memory and the hippocampus: anintraspecific comparison. Proc Biol Sci 273:2641–2649.

Puelles L and Rubenstein JL. 2003. Forebrain gene expressiondomains and the evolving prosomeric model. Trends Neurosci 26:469–476.

Raghanti MA, Spurlock LB, Uppal N, Sherwood CC, Butti C, Hof PR.In press. Von Economo neurons. In: Toga AW, Zilles K, Amunts K, edI-tors. Brain mapping: an encyclopedic reference. Elsevier, Amsterdam.

Ram�on y Cajal S. 1911. Histologie du systeme nerveux de l’hommeet des vert�ebr�es. Paris: Maloine.

Reep RL, Finlay BL, Darlington RB. 2007. The limbic system inmammalian brain evolution. Brain Behav Evol 70:57–70.

R�esibois A, Rogers JH. 1992. Calretinin in rat brain: an immunohis-tochemical study. Neuroscience 46:101–134.

Revishchin AV, Garey LJ. 1989. Sources of thalamic afferent neu-rons, projecting into the suprasylvian gyrus of the dolphin cere-bral cortex. Neirofiziol 21:529–539.

Rogers JH. 1989. Two calcium-binding proteins mark many chicksensory neurons. Neuroscience 31:697–709.

Rogers JH. 1992. Immunohistochemical markers in rat cortex: co-localization of calretinin and calbindin-D28k with neuropeptidesand GABA. Brain Res 587:147–157.

Rose J. 1942. A cytoarchitectural study of the sheep cortex. J CompNeurol 76:1–55.

Sanides F, Hoffmann J. 1969. Cyto- and myeloarchitecture of thevisual cortex of the cat and of the surrounding integration corti-ces. J Hirnforsch 11:79–104.

Sanides F, Sanides D. 1972. The "extraverted neurons" of the mam-malian cerebral cortex. Z Anat Entwickl Gesch 136:272–293.

30 BUTTI ET AL.

Scheibel ME, Scheibel AB. 1978. The methods of Golgi. In: RT Rob-ertson, editor. Neuroanatomical research techniques. New York:Academic Press. p 89–114.

Schwaller B, Buchwald P, Blumcke I, Celio MR, Hunziker W. 1993.Characterization of a polyclonal antiserum against the purifiedhuman recombinant calcium binding protein calretinin. Cell Cal-cium 14:639–648.

Schwerdtfeger WK, Oelschl€ager HA, Stephan H. 1984. Quantitativeneuroanatomy of the brain of the La Plata dolphin, Pontoporiablainvillei. Anat Embryol 170:11–19.

Seeley WW, Merkle FT, Gaus SE, Craig AD, Allman JM, HofPR. 2012. Distinctive neurons of the anterior cingulate andfrontoinsular cortex: a historical perspective. Cereb Cortex 22:245–250.

Sherk H. 1986. Location and connections of visual cortical areas inthe cat’s suprasylvian sulcus. J Comp Neurol 247:1–31.

Sherwood CC, Stimpson CD, Butti C, Bonar CJ, Newton AL,Allman JM, Hof PR. 2009. Neocortical neuron types in Xenarthraand Afrotheria: implications for brain evolution in mammals.Brain Struct Funct 213:301–328.

Shimamura M, Yasue H, Ohshima K, Abe H, Kato H, Kishiro T,Goto M, Munechika I, Okada N. 1997. Molecular evidence fromretroposons that whales form a clade within even-toed ungulates.Nature 388:666–670.

Sokolov VE, Ladygina TF, Supin A. 1972. Localization of sensoryzones in the dolphin cerebral cortex. Dokl Akad Nauk SSSR 202:490–493.

Spaulding M, O’Leary MA, Gatesy J. 2009. Relationships of Cetacea(Artiodactyla) among mammals: increased taxon sampling alters

interpretations of key fossils and character evolution. PLoS One4:e7062.

Squire LR, Zola-Morgan S. 1991. The medial temporal lobe memorysystem. Science 253:1380–1386.

Stern S. 2009. Migrations and movement patterns. In: Perrin WF,W€ursig B, Thewissen JGM, editors. Encyclopedia of marine mam-mals: Academic Press. p 726–730.

Striedter GF. 2005. Principles of Brain Evolution. Sunderland, MA:Sinauer Associates.

Tarpley RJ, Ridgway SH. 1994. Corpus callosum size in delphinidcetaceans. Brain Behav Evol 44:156–165.

Thewissen JGM, Cooper LN, Clementz MT, Bajpai S, Tiwari BN.2007. Whales originated from aquatic artiodactyls in the Eoceneepoch of India. Nature 450:1190–1194.

Uhen M. 2010. The origin(s) of whales. Annu Rev Earth Planet Sci38:189–219.

von Economo C. 1926. Eine neue Art Spezialzellen des Lobus cin-guli and Lobus insulae. Zschr Ges Neurol Psychiatr 100:706–712.

Zhou X, Xu S, Yang Y, Zhou K, Yang G. 2011. Phylogenomic analy-ses and improved resolution of Cetartiodactyla. Mol Phyl Evol 61:255–264.

Zilles K, Armstrong E, Moser KH, Schleicher A, Stephan H.1989.Gyrification in the cerebral cortex of primates. Brain Behav Evol34:143–150.

Zimmermann L, Schwaller B. 2002. Monoclonal antibodies recogniz-ing epitopes of calretinins: dependence on Ca21-binding statusand differences in antigen accessibility in colon cancer cells. CellCalcium 31:13–25.

Zola-Morgan S, Squire, LR. 1993. Neuroanatomy of memory. AnnuRev Neurosci 16:547–563.

STRUCTURE OF THE HIPPOPOTAMUS BRAIN 31