Professional Documents
Culture Documents
Research Institute, and c Neurosciences Graduate Program and Departments of d Neurosciences and e Radiology,
12 55
10
6 50
Latitude
4
2
0 45
0 1,000 2,000 3,000 4,000
a Geographic distance (km)
40
1.0 1.0
Proportion classified
Error (km)
0.8
0.8 0.6 > 1,200
4001,200
0.6 0.4 <400
0.2 0 (correct)
0.4 0
0.2
0 e
0 500 1,000 1,500 2,000 2,500
d Predicted location error (km)
Fig. 1. Cranial morphology reflects geography across Europe. population locations. Average 8 SD plotted for 100 bootstrap
a Pairwise distances between 27 European populations. Cranio- replications (black) and random permutations (gray). *p ! 0.001.
metric distance is significantly correlated (rM = 0.51, p ! 1 ! d Individual female skulls were identified with correct or nearby
10 5) with geographic distance. b Non-metric multi-dimensional populations based on cranial morphometry (solid) significantly
scaling ordination of craniometric distances aligned to geograph- better than chance (dotted). e Proportion of female skulls that
ic coordinates of populations. Population symbols identify 4 clus- were correctly classified (black) and misclassified with popula-
ters, and lines form a minimum spanning tree. c Distances be- tions at different distances (gray shades). Sample sizes are listed
tween predicted locations based on craniometric ordination and after population name.
and similar to the cross-validated classification perfor- maximizes the variation of the 37 cranial measures under
mance of the male skull training set. Misclassified crania the condition that this projection is a linear combination
are more likely than is expected by chance (p = 6.4 ! of longitude and latitude. The first principal component
105) to be from geographically proximal populations is statistically significant (p = 0.005) and explains 12.8%
(fig.1d, e; online suppl. fig. S2), which is consistent with of total craniometric variation, more than a third of the
the spatial structure found at the population level in this variation (30.5%) explained by the first component of an
study. unconstrained PCA. Population eigenvalues for the first
We then tested the hypothesis that skulls exhibit clin- principal component were interpolated by ordinary krig-
al variation along geographic axes within Europe. A di- ing and plotted to create an isocline map of Europe
rectional Mantel correlogram shows a monotonic de- (fig.2a). Cross-validation indicated that predicted popu-
crease in craniometric similarity with distance in two or- lation eigenvalues were highly correlated with observed
thogonal directions, NW-SE and NE-SW (online suppl. values (r = 0.88). Significantly, this map shows a clear gra-
fig. S3B), and this result motivated us to search for a geo- dient along a NW-SE axis that was not specified a priori
graphic axis that can explain a significant fraction of the and emerged from the redundancy analysis as the direc-
craniometric variation between populations. Redundan- tion of maximum cranial variation. Therefore, a subset of
cy analysis, a constrained version of principal compo- cranial measures exhibits clinal variation along this geo-
nents analysis (PCA), was used to find a projection that graphic axis.
174.6.226.46 - 8/29/2017 8:47:05 AM
185
24.5
112
175
2 1 0 1 2 2 1 0 1 2 2 1 0 1 2
ancestry (online suppl. fig. S5). For each ADNI subject,
NW-SE loc. (au) NW-SE genetic ancestry was inferred by projecting the
rotated principal components onto a 45 line oriented
NOL ASB BBH
R2 = 0.35; p = 1.1e03 R2 = 0.27; p = 6e03 R2 = 0.25; p = 7.6e03 NW and SE in figure 3a. This PCA-based assignment of
185
130
109
b 2 1 0 1 2 2 1 0 1 2 2 1 0 1 2
[26] found that a PCA plot of genetic distances placed in-
dividuals with grandparents of both NW and SE Euro-
pean and Ashkenazi Jewish origin in between these refer-
Fig. 2. Cranial measures show significant variation along a NW- ence populations.
SE axis within Europe. a Isoclines of interpolated eigenvalues for
first spatially constrained component of a redundancy analysis
We found that ADNI individuals with a NW ancestry
and geographic locations of populations. b Cranial measures plot- are on average 4 cm taller than ADNI individuals with a
ted in order of their contribution to this map. Negative abscissas SE or Ashkenazi Jewish ancestry (p = 7.3 ! 106), consis-
correspond to a more NW location. Proportion of variance ex- tent with previously observed differences in height across
plained (R 2) and nominal p values are indicated. NLB = Nasal Europe [30]. Our sample is evenly distributed along the
breadth; M28 = sagittal occipital arc; GOL = glabello-occipital
length; NOL = nasio-occipital length; ASB = biasterionic breadth; NW-SE axis based on MCI (p = 0.84) and AD (p = 0.13)
BBH = basion-bregma height. diagnosis. Our sample is slightly unevenly distributed
based on age (p = 0.014) and sex (p = 0.002) as ADNI in-
dividuals with a NW ancestry are on average 2.5 years
younger and include a greater number of female subjects.
Six measures are significantly associated with NW-SE We tested the correlation between degree of estimated
location, and this NW-SE axis explains 2544% of the NW-SE European ancestry and 12 summary measures of
variance of these individual measures (fig. 2b; online MRI-derived brain morphology, while controlling for
suppl. fig. S4). Of these measures, glabello-occipital height, weight, BMI, sex, age, and neurological diagnosis.
length (maximum length) and nasio-occipital length, bi- Intracranial and brain volumes and total cortical surface
asterionic breadth, and sagittal occipital arc are approxi- area are significantly (p ! 0.0006) correlated with degree
mately 5% longer in NW populations, while basion-breg- of ancestry along the NW-SE axis (fig.3b). These three
174.6.226.46 - 8/29/2017 8:47:05 AM
0.04
5.0
0.02 2.0
PC1 (rotated)
0 2.0
5.0
0.02
0
0 0
likely to be an artifact caused by differences in sex or neu-
rological diagnosis.
2 105 Intracranial and brain volumes and cortical surface
4 104 3 105
0.02 0.04 0.08 0.02 0.04 0.08 0.02 0.04 0.08
area progressively increase with the amount of inferred
NW-SE loc. (au) NW European ancestry (fig.3b), and these measures are
b
approximately 5% larger in the 10% of individuals with
the most NW European ancestry compared to the 10%
Fig. 3. Structural brain measures follow a predicted trend in a with the most SE European ancestry. This percentage in-
group of individuals with European ancestry. a The first two prin- crease matches the percentage increase in cranial length
cipal components of genotypes of ADNI subjects (yellow/small and breadth observed along the same NW-SE geographic
points; color refers to online version only) and individuals from
European reference populations (gray crosses) rotated 18 to align axis in the skull data set (fig.2b) and cannot be attributed
with a map of Europe. For each reference population (see online to a correlation with body size since we controlled for
suppl. table S3 for labels), the average (SD) of principal compo- height and weight. This correlation involves specific not
nents for all individuals in that population are indicated by disc global brain morphology because hippocampal, basal
position (diameter). Geographic origin of each population is in- ganglia, ventricular, and cerebellar volumes and average
dicated by disc shade of gray from NW (black) to SE (light gray)
Europe. ADNI subjects are spread out primarily along a NW-SE cortical thickness are not associated with NW-SE ances-
axis and form two distinct clusters corresponding to NW Euro- try.
pean and Ashkenazi Jewish ancestry (see also online suppl. fig. Next, we performed both a region of interest analysis
S5). b Brain structural measures tested for association with esti- and vertex-based tests across the cortex to test whether
mated NW-SE ancestry, while controlling for height, weight, BMI, the surface area of specific cortical regions showed more
age, sex, and diagnosis. Negative abscissas correspond to a larger
proportion of NW ancestry. significant association with the degree of NW-SE ances-
try. We found that cortical surface area predominantly
in the frontal and temporal lobes from both hemispheres
is significantly associated (online suppl. table S4) and is
brain measures remain significantly associated (p ! 0.05) 49% larger among 10% of individuals with the most
with NW-SE ancestry among males (n = 368), females NW European ancestry compared to 10% with the most
(n = 258), and subjects without a diagnosis of AD (n = SE European ancestry. We found a similar frontotempo-
486), and cortical surface area shows a trend level asso- ral pattern of association with the degree of NW-SE an-
ciation (p = 0.09) in the small sample of healthy controls cestry with a vertex-based analysis (fig.4; online suppl.
(n = 181). Therefore, the observed association between fig. S6).
174.6.226.46 - 8/29/2017 8:47:05 AM
References
1 Panizzon MS, Fennema-Notestine C, Eyler 9 Manica A, Amos W, Balloux F, Hanihara T: 18 Sokal RR, Uytterschaut H, Rosing FW,
LT, Jernigan TL, Prom-Wormley E, Neale M, The effect of ancient population bottlenecks Schwidetzky I: A classification of European
Jacobson K, Lyons MJ, Grant MD, Franz CE, on human phenotypic variation. Nature skulls from three time periods. Am J Phys
Xian H, Tsuang M, Fischl B, Seidman L, Dale 2007;448:346348. Anthropol 1987;74:120.
A, Kremen WS: Distinct genetic influences 10 Martnez-Abadas N, Gonzlez-Jos R, 19 Hanihara T, Ishida H: Os incae: variation in
on cortical surface area and cortical thick- Gonzlez-Martn A, Van der Molen S, Tala- frequency in major human population
ness. Cereb Cortex 2009;19:27282735. vera A, Hernndez P, Hernndez M: Pheno- groups. J Anat 2001;198:137152.
2 Joyner AH, Roddey JC, Bloss CS, Bakken TE, typic evolution of human craniofacial mor- 20 Hanihara T, Ishida H: Metric dental varia-
Rimol LM, Melle I, Agartz I, Djurovic S, phology after admixture: a geometric mor- tion of major human populations. Am J Phys
Topol EJ, Schork NJ, Andreassen OA, Dale phometrics approach. Am J Phys Anthropol Anthropol 2005; 128:287298.
AM: A common MECP2 haplotype associ- 2006;129:387398. 21 Hubbe M, Hanihara T, Harvati K: Climate
ates with reduced cortical surface area in hu- 11 Relethford J: Population-specific deviations signatures in the morphological differentia-
mans in two independent populations. Proc of global human craniometric variation tion of worldwide modern human popula-
Natl Acad Sci USA 2009;106:1548315488. from a neutral model. Am J Phys Anthropol tions. Anat Rec 2009;292:17201733.
3 Rimol LM, Agartz I, Djurovic S, Brown AA, 2010;142:105111. 22 R: A Language and Environment for Statisti-
Roddey JC, Kahler AK, Mattingsdal M, 12 Martnez-Abadas N, Paschetta C, de Aze- cal Computing, version 2.11.1. Vienna, R
Athanasiu L, Joyner AH, Schork NJ, Halgren vedo S, Esparza M, Gonzlez-Jos R: Devel- Foundation for Statistical Computing, 2010.
E, Sundet K, Melle I, Dale AM, Andreassen opmental and genetic constraints on neuro- 23 Hewitt JE, Legendre P, McArdle BH, Thrush
OA: Sex-dependent association of common cranial globularity: insights from analyses of SF, Bellehumeur C, Lawrie SM: Identifying
variants of microcephaly genes with brain deformed skulls and quantitative genetics. relationships between adult and juvenile bi-
structure. Proc Natl Acad Sci USA 2010;107: Evol Biol 2009; 36:3756. valves at different spatial scales. J Exp Mar
384388. 13 Hallgrmsson B, Lieberman DE, Liu W, Biol Ecol 1997; 216:7798.
4 Schumann CM, Bloss CS, Barnes CC, Wide- Ford-Hutchinson AF, Jirik FR: Epigenetic 24 Purcell S, Neale B, Todd-Brown K, Thomas
man GM, Carper RA, Akshoomoff N, Pierce interactions and the structure of phenotypic L, Ferreira MA, Bender D, Maller J, Sklar P,
K, Hagler D, Schork N, Lord C, Courchesne variation in the cranium. Evol Dev 2007; 9: de Bakker PI, Daly MJ, Sham PC: Plink: a
E: Longitudinal magnetic resonance imag- 7691. tool set for whole-genome association and
ing study of cortical development through 14 Novembre J, Johnson T, Bryc K, Kutalik Z, population-based linkage analyses. Am J
early childhood in autism. J Neurosci 2010; Boyko AR, Auton A, Indap A, King KS, Berg- Hum Genet 2007;81:559575.
30:44194427. mann S, Nelson MR, Stephens M, Bustaman- 25 Price AL, Patterson NJ, Plenge RM, Wein-
5 Brans R, van Haren N, van Baal G, Schnack te CD: Genes mirror geography within Eu- blatt ME, Shadick NA, Reich D: Principal
H, Kahn R, Pol H: Heritability of changes in rope. Nature 2008; 456:98101. components analysis corrects for stratifica-
brain volume over time in twin pairs discor- 15 Lao O, Lu TT, Nothnagel M, Junge O, Frei- tion in genome-wide association studies. Na-
dant for schizophrenia. Arch Gen Psychiatry tag-Wolf S, Caliebe A, Balascakova M, Ber- ture Genet 2006;38:904909.
2008;65:1259. tranpetit J, Bindoff LA, Comas D, Holmlund 26 Need AC, Kasperaviciute D, Cirulli ET,
6 Van Essen DC, Dierker D, Snyder AZ, G, Kouvatsi A, Macek M, Mollet I, Parson W, Goldstein DB: A genome-wide genetic signa-
Raichle ME, Reiss AL, Korenberg J: Symme- Palo J, Ploski R, Sajantila A, Tagliabracci A, ture of Jewish ancestry perfectly separates
try of cortical folding abnormalities in Wil- Gether U, Werge T, Rivadeneira F, Hofman individuals with and without full Jewish an-
liams syndrome revealed by surface-based A, Uitterlinden AG, Gieger C, Wichmann cestry in a large random sample of European
analyses. J Neurosci 2006;26:54705483. HE, Ruther A, Schreiber S, Becker C, Nurn- Americans. Genome Biol 2009; 10:R7.
7 White NS, Alkire MT, Haier RJ: A voxel- berg P, Nelson MR, Krawczak M, Kayser M: 27 Price AL, Butler J, Patterson N, Capelli C,
based morphometric study of nondemented Correlation between genetic and geographic Pascali VL, Scarnicci F, Ruiz-Linares A,
adults with Down syndrome. Neuroimage structure in Europe. Curr Biol 2008;18:1241 Groop L, Saetta AA, Korkolopoulou P,
2003;20:393403. 1248. Seligsohn U, Waliszewska A, Schirmer C,
8 Howells WW: Cranial Variation in Man: A 16 Nothnagel M, Lu TT, Kayser M, Krawczak Ardlie K, Ramos A, Nemesh J, Arbeitman L,
Study by Multivariate Analysis of Patterns of M: Genomic and geographic distribution of Goldstein DB, Reich D, Hirschhorn JN: Dis-
Difference among Recent Human Popula- SNP-defined runs of homozygosity in Euro- cerning the ancestry of European Americans
tions. Papers of the Peabody Museum of Ar- peans. Hum Mol Genet 2010;19:29272935. in genetic association studies. PLoS Genet
chaeology and Ethnology No. 67. Cam- 17 Sokal RR, Uytterschaut H: Cranial variation 2008;4:e236.
bridge, MA, Harvard University Press, 1973. in European populations: a spatial autocor-
relation study at three time periods. Am J
Phys Anthropol 1987; 74:2138.
174.6.226.46 - 8/29/2017 8:47:05 AM