Heritability of Brain Size and Surface Featuresin Rhesus Macaques (Macaca mulatta)J. M. Cheverud, D. Falk, M. Vannier, L. Konigsberg, R. C. Helmkamp, andC. Hildebolt

The extent of heritability for overall brain size and regional cortical surface featuressuch as sulcus lengths is important for demonstrating a genetic component to theobserved phenotypic differences among individuals and for evaluating the potentialfor evolutionary change in response to selection. Although the genetics of brain sizehas been extensively considered, the detailed morphology of the cortical surfacehas not previously been subjected to genetic analysis. We estimated the heritabilityof brain size and cortical sulcus lengths using 438 endocranial casts taken fromskeletons of rhesus macaques (Macaca mulatta) from the Cayo Santiago population.Estimates were obtained both by mother-offspring regression and symmetric-dif-ferences-squared (SDS) methods. Brain size, measured as cranial capacity, washighly and significantly heritable in this population, confirming results of previousstudies with laboratory mice. Overall, cortical sulcus lengths were also heritable,with 35% of the sulci significantly heritable at the 5% level in the mother-offspringanalysis. The average mother-offspring heritability estimate, 0.31, was the same asthe average heritability obtained previously from a series of 56 cranial metric char-acters. The SDS analyses generally corresponded to the findings based on mother-offspring regressions, although the significance test appeared more conservative.Both gross and detailed morphology of the brain are heritable.

From the Departments of Anthropology and Cell Bi-ology and Anatomy, Northwestern University, Evans-ton, Illinois (Cheverud and Konigsberg); the Depart-ment of Anthropology, State University of New Yorkat Albany (Falk); the Mallinkradt Institute of Radiol-ogy, Washington University School of Medicine, StLouis, Missouri (Vannier and Hildebolt); and the De-partment of Sociology and Anthropology, Purdue Uni-versity, West Lafayette, Indiana (Helmkamp). Dr.Cheverud is now at the Department of Anatomy andNeurobiology, Washington University School of Med-icine. This research was supported by PHS grant 7 RO1NS24904. The authors thank the University of PuertoRico for free access to the Cayo Santiago skeletal col-lection. They also thank Allen Moore and Cashell Jac-quish for their help with the analysis. Address reprintrequests to Dr. Cheverud, Department of Anatomy andNeurobiology, Washington University School of Med-icine, Box 8108,660 S. Eudid Ave, SL Louis, MO 63110.

Journal of Heredity 1990:81:51-57; 0022-1503/90/12.00

Heritability for brain size in laboratory ro-dents has been estimated frequent-Iy2̂ 9j7j8.42-+i o w jng to interest in the evo-lution of brain size itself and brain-bodyscaling in mammals.2-56 These studies re-ported a relatively high heritability forbrain size, usually 0.60-0.70, althoughLeamy38 is an exception. Leamy estimateda heritability of only about 0.20 for adultbrain size using father-offspring regres-sions and suggested that this low estimatemay be due to the relatively recent foun-dation of his randomly bred CV1 strain (52generations of random mating after deri-vation from inbred strain 101) and/or theinclusion of significant prenatal maternaleffects in estimates obtained through sibanalyses by previous workers but not con-founded with heritability in Leamy's38

analysis. Even so, the selection experi-ment of Roderick et al.43 yielded a realizedheritability of 0.64, supporting Leamy's38

suggestion that his estimate is low becauseof the special character of his population.This realized heritability is perhaps thebest guide to the level of heritable varia-tion for evolutionary analyses. Thus, brain

size is usually portrayed as a relativelyhighly heritable feature in mammals.

To consider brain evolution solely withregard to size is undoubtedly an oversim-plification. 2iJ2.34.36.4i jpe focus o n brain sizein previous analyses has been due largelyto its relative ease of measurement com-pared with more detailed aspects of brainmorphology, not to its status as the singleneural morphological feature of interest.In particular, Geschwind and Galaburda25"27

suggested a major role for genes in localneural development and lateralization. Weextend previous analyses in two importantways. First, in addition to brain size, weconsider the detailed morphology of thebrain's cortical surface to determinewhether differences in regional brain mor-phologies are inherited. Second, we esti-mate brain size heritability in a free-rang-ing primate colony (rhesus macaques,Macaca mulatto) rather than in a labora-tory rodent population. Specifically, we testthe hypothesis that brain size and corticalsulcus lengths are inherited in the free-ranging rhesus macaques from Cayo San-tiago, Puerto Rico.

61

Figure 1. Sulci digitized on rhesus monkey endocasts, left lateral (left) and orbital (right) views. Abbreviations of sulcl: r - rectus; arc - arcuate; c - central; syl -syhian; its — rostral superior temporal; cts — caudal superior temporal; rtm — rostral middle temporal; of — orbitofrontal; mo •» medial orbital; f — fork medial orbital;lo - lateral orbital. Other abbreviations: spt - sphenotemporal suture; tpet = traditional frontal petalia; ntpet = nontraditional frontal petalia. From D. Falk, personalcommunication; see Appendix for details of how sulci were defined on endocasts.

Materials and Methods

Population and MeasurementsSkeletons collected from the free-rangingrhesus macaques on Cayo Santiago, a 40-acre island off the southeast coast of Puer-to Rico, formed the sample for our study.The colony of rhesus macaques wasfounded in 1939 with wild-trapped animalsimported from India/6 Although early ob-servations were made by Carpenter,8 thecolony was not intensively observed untilAltmann1 began his sociobiological stud-ies in 1956. At that time most animals wereindividually marked, and regular provi-sioning with monkey chow was institut-ed.146 New births were noted and the ge-nealogical relationships of animalsrecorded.1 •« The study by Sade et al.46 con-tains an excellent comprehensive historyof the island population and its manage-ment

Beginning in 1970, an earlier practice ofopportunistically saving the skulls of mon-keys found dead on the island was system-atized so that the complete skeleton of anyanimal found dead or moribund was pre-pared and curated. Also, in 1972, skeletonswere obtained from Group K, an intact so-cial group that was being removed from

the island for management purposes.46 Thisintact social group contained several ma-ternal families and forms the core of theskeletal collection. The practice of pre-paring skeletons from animals found deadon the island has continued to the present

Demographic information, including sex,age, birth date, and genealogical relation-ships, was obtained from records provid-ed in Sad_e_e_t_al.46 and by C. Busse (per-sonal communication). Animals' ages atdeath and birth dates typically are knownto within one week. Because paternity isunknown in this colony, genealogical rec-ords only include relationships throughthe maternal line. This results in an overallunderestimate of genetic relationships inthe colony using demographic recordsalone. Data from electrophoretic markersindicate that paternal relationships aremost likely random with respect to ma-ternal ones.12

For our study, brain size was denned asendocranial capacity and was measuredby filling the cranium with mustard seedand pouring the enclosed seed into a grad-uated cylinder. The cube root of cranialcapacity was used in statistical analysis tobetter fit the data to statistical models.

In rhesus macaques and many other pri-

Table 1. HerltabUity of brain »lze and corticalsnlcns lengths estimated by mother-offspringregression, probability of /r1 > 0, and sample sizeof pairs

Trait N

Cranial capacityLeft rectusRight rectusLeft arcuateRight arcuateLeft lateral orbitalRight lateral orbitalLeft fork medial orbitalRight fork medial orbitalLeft medial orbitalRight medial orbitalLeft sylvianRight sytvianLeft centralRight centralLeft caudal superior temporalRight caudal superior temporalLeft rostral superior temporalRight rostral superior temporalLeft rostral middle temporalRight rostral middle temporal

0.600.690.680.770.240.34

-0.380.360.360.22

-0.24-0.07

0.380.65

-0.49- 0 40

1.600.370.340.76

-0.06

.002

.024

.032

.034

.270

.118

.898

.074

.106

.384

.772

.560

.274

.038

.766

.616

.002

.054

.042

.002

.608

1792726272356496661212526273329202096898370

mates, the inner table of the neurocraniumaccurately reproduces details of externalbrain morphology, including sulci andbrain shape features. The lengths of 10sulci on the right and left sides of latexendocranial casts (endocasts) were re-corded using a three-dimensional digitizer

52 The Journal of Hejecfity 199ft81(1)

(Table 1 and Figure 1). D.F. prepared en-docasts from rhesus braincases usingstandard procedures1940 and sprayed themwith gold paint to accentuate morpholog-ical detail and maintain cast stiffness.Points were placed along the length of thesulcus at about 2 mm intervals by R.C.H.to aid in digitizing sulcus lengths. C.H. dig-itized the series of points along each sul-cus with a 3Space electromagneticdigitizer31 and strung them together to ob-tain total sulcus lengths using softwarecontributed by M.V. Repeatability mea-surements for these sulci ranged from 84%up to 98%, with four of five values greaterthan 90% (average repeatability about92%).35 Error from digitizing was minimal(average about 2%), with slightly moreerror attributable to sulcus endpoint iden-tification (average about 6%). Detailed de-scriptions of sulcus measurements are giv-en in the Appendix.

We corrected all data for variation re-lated to age, sex, and date of birth whenpreliminary analyses indicated these vari-ables were important. There was no sig-nificant secular trend for these measure-ments, and so we did not include date ofbirth in the analysis. For all sulci, we cor-rected for sex differences in sulcal length35

by adding the difference in male/femalemeans to each individual female's scores.After sex correction, small but statisticallysignificant age-related effects were foundfor the left and right rectus, left and rightarcuate, and left medial orbital sulci. Allthese sulci became very slightly shorterwith increased age, perhaps indicating apoorer-quality cast of the brain on the in-ner table of the neurocranium for olderanimals. For these sulci, we used the re-siduals of sulcal length on age regressionsas input to the genetic analysis. The gen-eral lack of age effects is not surprisinggiven the relatively early growth of therhesus cortex. Because these age effectswere of only minor importance, heritabil-ity estimates were minimally affected bythe correction.

Cranial capacity grows after birth in rhe-sus macaques, and its growth pattern var-ies by sex.35 Therefore, we removed age-related variation separately in each sex byregressing the cube root of cranial capac-ity on age for all animals less than six yearsold. For these subadults, we used the re-siduals of the sex-specific regressions asindividual scores, adding in the adult malemean to equalize means across age cate-gories, and input them into the geneticanalysis. Also, the difference in male/fe-male adult means was added to each in-

dividual adult female cranial capacity tocorrect for sex differences among adults.

In a quantitative genetic analysis, it isassumed that the population is in Hardy-Weinberg and linkage equilibrium at therelevant loci.18 The Cayo Santiago popu-lation conforms to these assumptions rea-sonably well.6-717 Mating appears randomwith respect to phenotype, and no immi-gration has occurred since the colony wasformed.4* Emigration due to removal of an-imals has not greatly altered allele fre-quencies at single loci.67 The populationof the colony probably has been below 100at various times, but the colony was found-ed only six generations ago and showsconsiderable single-locus1017-39 and poly-genic variation.911 The pattern of matingand migration among social groups on theisland militates against inbreeding, be-cause most males leave their natal groupsbefore mating.45 It is also assumed that therelevant environmental factors are ran-domly distributed with respect to familymembership. Owing to the small size ofthe island and ample provisioning, nutri-tion and climate are likely to vary random-ly across families, largely eliminating en-vironmental covariance among relatives.However, the social environment and itsconsequent psychological effects, espe-cially those related to lineage dominancerank,47 may not be randomly distributedwith respect to family membership.

Genetic and Statistical MethodsA basic quantitative genetic model is usedhere in which the phenotypic value (F) isthe sum of the additive genetic value (.4)and an environmental deviation (£) [P =A + £].18 The additive genetic value is thatpart of the phenotype which is inherited,while the environmental deviation ac-counts for all other genetic and environ-mental influences. Using this model, thephenotypic variance (s2^) is simply the sumof the additive genetic (s2^) and environ-mental (s2,) variances, assuming no ge-notype-environment covariance, as above.

We estimated heritabilities by standardmother-offspring regression'8 and a slightmodification of the symmetric-differ-ences-squared (SDS) method.4-5-28 Moth-er-offspring heritability estimates wereobtained by pairing each individual off-spring with its mother. Because individualmother-offspring pairs are not indepen-dent of one another, as mothers are re-peated in the sample and are geneticallyrelated to one another, we estimated thestatistical significance of the heritabilitylevels using a randomization procedure.

We randomly assigned offspring to moth-ers and recalculated the mother-offspringregression coefficient 500 times in orderto estimate the distribution of heritabilityestimates expected with this sample underthe null hypothesis of no heritability (i.e.,ff = 0). The proportion of randomizedregression coefficients exceeding the ob-served value is the probability of obtainingthe observed coefficient, or one even moreextreme, when there is no heritability.

The SDS method was also used to esti-mate heritabilities. With this method in-formation from all kinds of relatives canbe incorporated in a single heritability es-timate; thus, the available sample is uti-lized to its fullest advantage. This is par-ticularly important in nonexperimentalstudies such as ours in which sample sizesare limited and genetic relationships donot fit any single particular statistical de-sign.

The SDS method takes advantage of themathematical relationship between a sam-ple's variance (s2) and the squared phe-notypic difference between pairs of indi-viduals:

2 (y< - y - 0)0)

where y is the character of interest mea-sured in individuals i and j and N is thesample size. Because the expected valueof the pairwise squared differences amongindividuals is twice the sample variance,one could estimate the sample varianceby using an ordinary least-squares solu-tion to a regression equation.28 This pro-cedure can be extended to the estimationof variance components. We used a rela-tively simple model including only addi-tive genetic effects,

- yjy = 2(1 - (2)

where rtf is Wright's coefficient of relation-ship and s2,, and s2, are the additive geneticand environmental variances, respective-ly. In this equation, (y, — yjf is the de-pendent variable, (1 — ro~) is the indepen-dent variable, 2s2,, is the regressioncoefficient, and 2s2, is the regression con-stant. Heritability is then estimated as &J(s2

Q + s2,). The equation essentially de-scribes a matrix regression of a phenotyp-ic distance matrix between all pairs of in-dividuals on their genetic relatedness. Notethat in Equation 2 when rtf is zero, the ex-pected value of the dependent variable istwice the phenotypic variance (2s2,,).

Computation time was cut drastically bymeans of a slight modification of this

Cheverud et al • Brain See and Surface Features in Rhesus Macaques 53

method, i.e., by directly considering onlyrelated pairs of individuals. The regres-sion was calculated by forcing the regres-sion line through the phenotypic varianceestimate obtained in the ordinary fashion(rather than through the mean) and oth-erwise using only pairs of related individ-uals to estimate the regression coefficient.This modification circumvented the needto calculate the squared differences for un-related pairs, which make up well over 95%of the pairs in this sample but may pro-duce a slight bias towards lower herita-bility estimates.

Statistical significance of genetic vari-ances and heritabilities estimated usingSDS was determined by a randomizationtest based on Mantel's test.1648 We ran-domized individuals with respect to ge-nealogical relationship by randomly re-cording the rows and columns of the matrixof squared differences between individu-als while leaving the matrix of genetic re-lationship in its original configuration. Therandomization procedure was repeated 500times, and the estimates obtained wereused to approximate the distribution of ge-netic variance estimates expected underthe null hypothesis of no heritability. Theproportion of random estimates greaterthan or equal to the observed estimate isthe probability of obtaining the observedresults, or results even more extreme, giv-en no heritability. This test is often lesspowerful than parametric tests but alsoavoids the unrealistic assumptions ofparametric tests.1648

Sample sizes for the SDS analysis wererelatively small and varied from trait totrait owing to missing data. The reliabilityof SDS estimates depends on the numberof related pairs of individuals and theirdegree of relatedness. Some idea of theeffective sample size (N^ for each anal-ysis can be obtained by deriving the sam-ple size of pairs weighted by their degreeof relatedness (Table 2):

^r=S»-/> (3)where /) is a particular degree of relation-ship (such as 0.50, 0.25, or 0.125) and Fjis the number of pairs of individuals withthat particular degree of relationshipsummed over levels of relationship.

Results

In the mother-offspring analysis, brain sizewas found to be highly and significantlyheritable (/i2 = 0.60, Table 1). The averageheritability of the sulcus lengths was lower

Table 2. Heritability of brain size and cortical ralco* length* estimated by the symmetric-dlflerence*-sqnared method, probability of A* > 0, sample size of pair*, and sample (ize of pairs weighted byrelatednes*

Trait N

Cranial capacityLeft rectusRight rectusLeft arcuateRight arcuateLeft lateral orbitalRight lateral orbitalLeft fork medial orbitalRight fork medial orbitalLeft medial orbitalRight medial orbitalLeft sylvianRight sylvianLeft centralRight centralLeft caudal superior temporalRight caudal superior temporalLeft rostral superior temporalRight rostra] superior temporalLeft rostral middle temporalRight rostral middle temporal

0.750.511.341.140.850.040.420.070.62

-0.570.030.230.590.570.700.410.51

-0.020.310.740.39

.002

.248

.002

.010

.064

.480

.106

.456

.026

.804

.510

.396

.170

.080

.078

.262

.264

.542

.178

.002

.128

2,969502538410482991894

1,2431,049

544513538556542506357385

1,4091,3851,3061,198

37061726069

1361221661426864697279744750

206197188171

" N, — sum of relatedness for all pairs.

(average /i2 = 0.31). Thirty-five percent ofthese sulcus lengths were significantlyheritable at the .05 level. Heritabilitiesabove 0.50 tended to be significant, al-though differences in sample size acrosssulci make generalization difficult here.Sample sizes for the mother-offspringregressions tended to be low for sulcallengths owing to missing data. Sulci show-ing a significant heritability included theright and left rectus, left arcuate, left cen-tral, right caudal superior temporal, rightrostral superior temporal (the left rostralsuperior temporal was nearly significant),and the left rostral middle temporal.

In the SDS analysis, the heritability ofbrain size was quite high (/l2 = 0.75) andstatistically significant. Twenty percent ofthe sulci were significantly heritable at the.05 level, including the right rectus, leftarcuate, left rostral middle temporal, andthe right fork medial orbital. Heritabilitiesabove 0.65 tended to be significant, butlarge variations in sample size make thisa weak generalization. The lower propor-tion of significant results for the SDS meth-od as compared with mother-offspringregressions may reflect the conservativenature of the accompanying significancetest Sulci with heritabilities significant onlyat the .10 level, including the right arcuateand the left and right central, may be plau-sibly considered heritable because of thisconservative tendency. The average SDSheritability estimate for the sulci was 0.44.Other sulci that showed some evidence forheritability (probability less than 10% ineither analysis) included the right lateral

orbital, left fork medial orbital, and leftrostral superior temporal.

Heritability estimates outside the validrange (below zero and above one) wereobtained in both mother-offspring and SDSanalyses. None of these estimates differedsignificantly from permissible values. Theyoccurred because of sampling error and,perhaps, the influence of factors not in-cluded in the estimation model (Equation2).

Discussion

Overall, the results provide solid evidencefor heritable variation in brain size andregional neural morphology. The brain sizeheritability estimates presented here arequite high and are similar to those re-ported in previous work on laboratory ro-dents.2-2337-3842"" Our findings extend pre-vious results on high brain size heritabilityto a primate species and to a nonlabora-tory population and thereby suggest thatthis result may be a general one. If so, thisindicates the great potential for responseto selection on brain size in mammals.

Furthermore, we found significant her-itability for the detailed morphology of thebrain's surface. This supports Geschwindand Galaburda's25"27 hypothesis that genesplay an important role in the developmentof local neural morphology. However, thelevel of heritability for sulci generally wasmoderate and thus was lower than thatfound for brain size. In particular, someevidence points to an additive geneticcomponent to variation in the rectus, ar-

6 4 The Journal of Heredrty 1990-81(1)

cuate, fork medial orbital, central, supe-rior temporal (both rostral and caudalcomponents), and rostral middle tempo-ral sulci. Heritabilities of cortical sulci fellin the same range as did those previouslyestimated for cranial metric and nonmet-ric skeletal characters in this same pop-ulation of macaques." The studies of cra-nial skeletal trait heritability wereperformed on a largely overlapping sam-ple of individuals and included mother-offspring regression analyses. The aver-age mother-offspring heritability estimatefor these 20 sulci was the same (average/i2 = 0.31) as for 56 cranial metric traits,"indicating that regional neural morpho-logical features, as with most other mor-phological features, are heritable.

Patterns evident in sulcal length herita-bility estimates may relate to brain devel-opment, function, and lateralization. Ex-cept for the rectus, significant heritabilityappeared only on one side or the other forany given sulcus (three on the left and twoon the right). Two of these sulci, the cen-tral and superior temporal, appear earlierin the right than in the left hemisphereduring fetal development in humans, in-dicating an asymmetry in developmenttime.13 However, in general, the centralsulcus develops before the superior tem-poral sulcus, so that the right superiortemporal and left central sulci appear atnearly the same time in humans, at ap-proximately 22 weeks gestation.13 If thesesulci develop in rhesus monkeys as theydo in humans, the left central and rightsuperior temporal sulci likely would besubject to the same prenatal hormone-me-diated neurohumoral influences.22-25"27 Sig-nificant heritability for these sulci may re-flect genetic variation in hormone levelsand/or responsiveness of the neural tis-sues to hormonal influences.

We also found that the major sulci onthe lateral surface of the frontal lobe (rec-tus and arcuate) are heritable. The rectussulcus of macaques appears to be homol-ogous with the inferior frontal sulcus ofhumans, whereas the inferior portion ofthe macaque arcuate sulcus is homologouswith the ascending limb of the human syl-vian fissure, a sulcus that forms the rostralboundary of area 44 (part of Broca's speecharea on the left side in humans).23 Thesesulci form relatively late and rapidly dur-ing human fetal development, between 28and 31 weeks gestation, and appear si-multaneously.13 Connolly15 suggested thatthe left side may develop before the rightThe contemporaneity of sulcus develop-ment may result in their common herita-

bility, again owing to potential geneticvariation in hormone levels and/or tissuereceptivity.22-24-27 Findings of significantheritabilities on the left side for these fron-tal lobe features is particularly interestingbecause the homologous sulci in humansare related to the development of the leftfrontal operculum, which is associated withlanguage function and lateralization.22-24-27

In addition to being heritable, these twosulci were also found to be directionallyasymmetric in these macaques, as was thefrontal lobe as a whole (frontal petalia)(D. Falk, personal communication). Fron-tal lobe asymmetry, referred to as frontalpetalia, is also known in humans.22-24-27

Heritabilities are reported from bothmother-offspring and SDS analyses. Theresults from the two analyses were gen-erally similar (Spearman rank order cor-relation between sulcal heritability esti-mates = 0.41; P < .05), although manydifferences also can be noted despite theinclusion of mother-offspring pairs in theSDS analysis. However, significant herita-bilities were detected by the SDS methodeven with relatively restricted samples. Ithas been noted that the significance testapplied to the SDS estimates is conser-vative. This approach to quantitative ge-netic estimation deserves closer attentionand may prove invaluable in future re-search on relatively small samples in non-experimental situations. Both Grimes andHarvey28 and Bruckner and Slanger4 foundthat the method provided relatively un-biased estimates of genetic variance intheir simulation analyses. We plan furthersimulation analyses of the SDS method andthe associated significance test to deter-mine their ability to detect heritable vari-ation in morphological features.

A disadvantage of the SDS method inthis application is that genetic relationshipthrough the paternal line is unknown, sothat animals presumed to be half-sibs mayindeed be full-sibs and seemingly unre-lated animals may in fact be related. Thus,genetic relationship is only known witherror. This may reduce the heritability es-timates obtained from the SDS analysis butwould be unlikely to indicate heritabilitywhere it does not exist.

The results of this analysis also indicatethe reliability of our method of measuringexternal morphological features by meansof endocasts. Our approach to endocastand brain measurement is likely to be muchmore accurate than traditional ap-proaches such as using flexible measuringtape to take linear measurements of sulcidirectly by hand,30 taking linear measure-

ments and their ratios from photographsof the brain,20 or using stereoplotting tomeasure positions of cortical features.33

Regrettably, repeatability estimates forthese measurement techniques have notbeen reported.

Some have questioned in general thecredibility of brain cortical surface mea-surements derived from latex casts of theinner table of the neurocranium.3-14 How-ever, taking care with trait definition andmeasurement (especially by using accu-rate digitizing equipment) has allowed usto extract highly repeatable measure-ments.35 Furthermore, these measure-ments tend to run in families, somethingthat is unlikely to be attributable to castingartifacts. Thus, genuine cortical sulcuslengths must be very highly correlated, ifnot nearly identical, with their traces mea-sured on endocasts.

Appendix

Identification of Sulci on EndocaatsWell-defined measures of 10 sulci shownin Figure 1 were digitized on endocastsprepared from skulls of rhesus monkeys.Dots were placed at the endpoints of eachsulcus and approximately 2 mm apart alongtheir intervening lengths by R.C.H. Thedots were verified by D.F. for each sulcuson each endocast, and any questionablesulci were discarded from the study. Thus,only portions of sulci whose entire lengthswere clearly and identically identified bytwo workers were digitized and analyzed.The criteria for defining each sulcus orportion of sulcus were as follows:

1. Rectus. Its endpoints and the con-figuration of the intervening sulcal lengthwere usually clearly represented on theendocasts. The entire sulcus was dottedand digitized.

2. Arcuate. The medial end of the ar-cuate sometimes converged with a trans-verse sulcus located dorsomedially in thefrontal lobe (i.e., somewhat parallel to thecentral sulcus). When this occurred, themedial end of the arcuate would appear tobe arched caudally. However, the inter-section of the two sulci was usually clear,and we defined the medial endpoint of thearcuate as the point of intersection in thesecases. The other end of the arcuate andthe intervening pathway were well de-fined.

3. Lateral orbital. This was a difficultsulcus to define because, unlike the fork,there was a good deal of variation in itspattern. Sometimes, this sulcus extendedonto the lateral surface of the frontal lobe.

Cheverud et al • Brain Size and Surface Features in Rhesus Macaques 55

In these instances, the entire length wascoded as the lateral orbital sulcus (i.e.,none of it was denned as the orbitofrontalsulcus). In the "H" orbital pattern, the lat-eral arm of the "H" was defined as thelateral orbital sulcus.

4. Fork of the medial orbital. This isone of the clearest sulci on rhesus monkeyendocasts. If the orbital pattern was an"H" configuration, the fork was denned asthe crossbar on the "H." If the configu-ration was a "Y" pattern, the fork was thelateral arm of the "Y." Endpoints and theshort intervening lengths were alwaysclear. The rostral endpoint of the arm ofthe "Y" configuration was also clear.

5. Medial orbital. This sulcus was prob-lematic at both ends. Sometimes the cau-dal end was crossed by a suture that ob-scured its endpoint. The rostral endfrequently tapered off in such a way thatan endpoint was not clearly defined. Theintervening portion usually reproducedclearly.

6. Sylvian. This fissure was problematicat both ends. The rostral endpoint had tobe arbitrarily defined, because it is not cleareven on actual brains exactly where thisfissure ends. We chose to define the an-terior endpoint as the most rostrally lo-cated point along the transitional curvefrom the lateral to the basal surface of theendocast. This point is located somewhatmore medially than is the traditional end-point defined from a strictly lateral pointof view, i.e., the sylvian as we define itcurves around a little underneath the fron-tal lobe. The caudal end of the sylvian usu-ally does not reproduce well on rhesusmonkey endocasts. We therefore mea-sured only those sylvian fissures thatmerged with superior temporal sulci andconsidered the point of intersection as thecaudal end of both sulci.

7. Central. The lateral end of the centralsulcus was frequently obscured by a smallvessel, and its medial end simply did notreproduce well in numerous cases. Theintervening portion of the sulcus was usu-ally clear, and only central sulci whoseendpoints were clearly visible were digi-tized.

8. Rostral superior temporal. This isthe portion of the superior temporal sul-cus rostral to the sphenotemporal suture.Both endpoints and the length of this shortsulcus were extremely clear on most en-docasts.

9. Caudal superior temporal. The ros-tral endpoint of this portion of the superiortemporal sulcus was the same as the cau-dal endpoint of the medial orbital sulcus

and was clearly defined on most of thecasts. The caudal endpoint was consid-ered to be the point where this sulcusmerged with the sylvian fissure. Becausethe caudal endpoint was difficult to see inother casts, the caudal superior temporalsulcus was analyzed only in endocasts inwhich this merging was observed. The in-tervening portion of the caudal superiortemporal sulcus was usually quite clear.

10. Rostral middle temporal. Because thecaudal portion of the middle temporal sul-cus was extremely variable and hard todefine in many endocasts, we chose toanalyze only the clearly defined portionrostral to the sphenotemporal suture. Bothendpoints and the short intervening lengthwere extremely clear.

References

I. Altmann S. A field study of the socioblology of rhesusmonkeys, Macaca mulatto. Ann NY Acad Scl 1962; 102:338-435.

2 Atchley W, Riska B, Kohn L, Plummer A, and Rut-ledge J. A quantitative genetic analysis of brain andbody size associations, their origin and ontogeny; datafrom mice. Evolution 1984; 38:1165-1179.

3. Bauchot R, and Stephan H. Encephales et moulagesendocraniens de quelques insectivores et primate* ac-tuels Problemes actuels de paleontologle. Colloq IntCentre Natl Rech Scl 1967; 163:575-587.

4. Bruckner C, and Slanger W. Symmetric difference*squared and analysis of variance procedures for es-timating genetic and environmental variances and co-variances for beef cattle weaning weight I. Comparisonvia simulation. J Anim Scl 1986; 63:1779-1793.

5. Bruckner C, and Slanger W. Symmetric differencessquared and analysis of variance procedures for es-timating genetic and environmental variances and co-variances for beef cattle weaning weight II. Estimatesfrom a data se t J Anim Sci 1986; 63:1794-1803.

6. Buettner-Janusch J, Dame L, Mason G, and Sade D.Primate red cell enzymes: glucose-6-phosphate de-hydrogenase and 6-phosphogIuconate dehydroge-nase. Am J Phys Anthropol 1974; 41:7-14.

7. Buettner-Janusch J, Mason G, Dame L, Buettner-Janusch V, and Sade D. Genetic studies of serum trans-ferring of free-ranging rhesus macaques of Cayo San-tiago, Macaca mulatto (Zimmerman 1780). Am J PhysAnthropol 1974; 41 217-231.

8. Carpenter C. Sexual behavior of free-ranging rhesusmonkeys. 1. Specimens, procedures, and behavioralcharacteristics of estrus. J Comp Psych 1942; 33:113-142.

9. Cheverud J. Variation in highly and lowly heritablemorphological traits among social groups of rhesusmacaques {Macaca mulatto) on Cayo Santiago. Evo-lution 1981:35:75-83.

10. Cheverud J, Buettner-Janusch J, and Sade D. Socialgroup fission and the origin of intergroup genetic dif-ferentiation among the rhesus monkeys of Cayo San-tiago. Am J Phys Anthropol 1978; 49:449-456.

II. Cheverud J, and Buikstra J. Quantitative geneticsof skeletal nonmetric traits in the rhesus macaques ofCayo Santiago. 111. Relative heritabllity of skeletal non-metric and metric traits. Am J Phys Anthropol 1982;59:151-156.

12. Cheverud J, Chepko-Sade BD, Dow M, and Sade D.Group selection models with trait groups based onsocial interaction networks. Am J Phys Anthropol 1988;77:427-434.

13. Chi JG, Dooling EC, and Gilles FH. Gyral devel-opment of the human brain. Ann Neurol 1977; 1:86-93.

14. CUrk WE Le Gros, Cooper DM, and Zuckerman S.The endocranial cast of the chimpanzee. J R AnthropolInst Great Britain 1936; 66:249-268.

15. Connolly CJ. External morphology of the primatebrain. Springfield, Illinois: CC Thomas; 1980.

16. Dow M, and Cheverud J. Comparison of distancematrices in studies of population structure and geneticmicrodifferentiation: quadratic assignment Am J PhysAnthropol 1985; 68.367-374.

17. Duggleby C. Blood group antigens and the popu-lation genetics of Macaca mulatto on Cayo Santiago. 1.Genetic differentiation of social groups. Am J PhysAnthropol 1978; 48:35-40.

18. Falconer D. Introduction to quantitative genetics,2nd ed. London: Longman Press; 1981. 340 p.

19. Falk D. External neuroanatomy of Old World mon-keys (Cercoplthecoidea) Contrib Primatol 1978; 15:1-95.

20. Falk D. The Taung endocast a reply to Holloway.Am J Phys Anthropol 1983; 60:479-489.

21. Falk D. Hominid paleoneurology. Ann Rev Anthro-pol 1987; 16:13-30.

22. Flor-Henry P. Functional hemispheric asymmetryand psychopathology. Integr Psychiatry 1983; 1:46-52.

23. Galaburda AM, and Pandya DN. Role of archltec-tronics and connections. In: Primate brain evolution:methods and concepts (Armstrong E, and Falk D, eds).New York: Plenum; 1982:203-216.

24. Galaburda AM, LeMay M, Kemper T, and Gesch-wind N. Right-left asymmetries in the brain. Science1978; 199:852-856.

25 Geschwind N, and Galaburda AM Cerebral later-alization: biological mechanisms, associations, and pa-thology: I. A hypothesis and a program for research.Arch Neurol 1985; 42.428-459.

26. Geschwind N, and Galaburda AM. Cerebral later-al ization: biological mechanisms, associations, and pa-thology: II. A hypothesis and a program for research.Arch Neurol 1985; 42:521-552.

27. Geschwind N, and Galaburda AM. Cerebral later-alization: biological mechanisms, associations, and pa-thology III. A hypothesis and a program for research.Arch Neurol 1985; 42:634-654.

28. Grimes L, and Harvey W. Estimation of geneticvariances and covariances using symmetric differ-ences squared. J Anim Sci 1980; 50*34-644.

29. Hahn R. and Haber S. A diallel analysis of brainand body weight In male inbred laboratory mice (Musmuscutus). Behav Genet 1978; 8:251-260.

30. Hellbroner PL, and Holloway RL. Anatomical brainasymmetries in New World and Old World monkeys:stages of temporal lobe development In primate evo-lution. Am J Phys Anthropol 1988; 76:39-48.

31. Hildebolt C, and Vannier M. Three-dimensionalmeasurement accuracy of skull surface landmarks AmJ Phys Anthropol 1988; 76:497-504.

32. Holloway RL. The role of human social behaviorin the evolution of the brain. Forty-third James ArthurLecture. New York: The American Museum of NaturalHistory, 1973.

33. Holloway RL. Exploring the dorsal surface of hom-inoid brain endocasts by stereoplotter and discrimi-nant analysis. Philos Trans R Soc Lond Ser B 1981;292:155-166.

34. Jerison HJ. The evolutionary biology of intelli-gence; afterthoughts. In: Intelligence and evolutionarybiology (Jerison H, and Jerison I, eds). New York:Springer-Vertag; 1988:447-466.

35. Konlgsberg L, Falk D, Hildebolt C, Cheverud J,Hetmkamp RC, and Vannier M. External brain mor-

56 The Journal of Heredity 199081(1)

phology in rhesus macaques (Macaca mulatto). J Hum 41. Radinsky L. The fossil record of primate brain evo- (Chepko-Sade BD, and Halpln ZT, eds). Chicago: Uni-Evol (in press). luUon. Forty-ninth James Arthur Lecture. New York: versity of Chicago Press; 1987:225-241.36. Lande R. Quantitative genetic analysis of multi- T h e A r a e r i C a n M u S e U m of N a t u r a l H i s t O r * 1 9 7 9 ' 46. Sade D, Chepko-Sade BD, Schneider j , Roberts S,variate evolution applied to brain: body size allometry. 42. Riska B, and Atchley W. Genetics of growth predict and Richtsmeier J. Basic demographic observationsEvolution 1979:33:402-416. patterns of brain-size evolution. Science 1985; 229:668- on free-ranging rhesus monkeys. Human Relations Area

37. Leamy U Morphometric studies In inbred and hy 6 7 1 ' F i l e s ' N e w H a v e n ' Connecticut; 1985.brid house mice. VI A genetical analysis of brain and 43. Roderick T, Wlmer R, and Wimer C. Genetic ma- 47. Sade D, Schneider J, Figueroa A, Kaplan J, Cushingbody size. Behav Genet 1985; 15:251-263. nipulation of neuroanatomical traits In: Knowing, K,CushingP, DunaifJ, Morse T.Rhodes D, and Stewart38 Leamy L Genetic and maternal influences on brain thinking and believing (Petrinovich L, and McGaugh M. Population dynamics in relation to social structureand body size in randombred house mice. Evolution J ' «*»»)• N e w Y o r f c P l e n u m : 1976.143-178. onCayoSantiago. YrbkPhys Anthropol 1976.1977;20:1988; 42:42-53. 44. Roderick T, Wimer R, Wimer C, and Schwartzkroin39. Ober C Olivier T, Sade D, Schneider J, Cheverud P- Genetic and phenotypic variation in weight of brain **• Smouse P, Long J, and Sokal R. Multiple regressionJ and Buettner-Janusch J Demographic components and spinal cord between inbred strains of mice. Brain and correlation extensions of the Mantel test of matrixof gene frequency change in f r e e z i n g macaques on Res 1973; 64:345-353 correspondence. Syst Zool 1986; 35:627-^32.Cayo Santiago. Am j Phys Anthropol 1984; 64:223-231. 4 5 ^ „_ c h e p k l > S a d e B D ^ M , ^ c h e v e m d j .

40 Radinsky L A new approach to mammalian cranial Definition and measurement of migration in age-struc-analysis, illustrated by examples of prosimian pri- tured populations. In: Mammalian dispersal patterns:mates. J Morphol 1968; 124:167-180 the eBects of social structure on population genetics

Oieverud et al • Brain Size and Surface Features in Rhesus Macaques 57