Content uploaded by Abby Drake
Author content
All content in this area was uploaded by Abby Drake on Feb 02, 2015
Content may be subject to copyright.
vol. 175, no. 3 the american naturalist march 2010
Large-Scale Diversification of Skull Shape in Domestic
Dogs: Disparity and Modularity
Abby Grace Drake
*
and Christian Peter Klingenberg
Faculty of Life Sciences, University of Manchester, Michael Smith Building, Oxford Road, Manchester M13 9PT, United Kingdom
Submitted June 1, 2009; Accepted October 27, 2009; Electronically published January 22, 2010
abstract: The variation among domestic dog breeds offers a unique
opportunity to study large-scale diversification by microevolutionary
mechanisms. We use geometric morphometrics to quantify the di-
versity of skull shape in 106 breeds of domestic dog, in three wild
canid species, and across the order Carnivora. The amount of shape
variation among domestic dogs far exceeds that in wild species, and
it is comparable to the disparity throughout the Carnivora. The
greatest shape distances between dog breeds clearly surpass the max-
imum divergence between species in the Carnivora. Moreover, do-
mestic dogs occupy a range of novel shapes outside the domain of
wild carnivorans. The disparity among companion dogs substantially
exceeds that of other classes of breeds, suggesting that relaxed func-
tional demands facilitated diversification. Much of the diversity of
dog skull shapes stems from variation between short and elongate
skulls and from modularity of the face versus that of the neuro-
cranium. These patterns of integration and modularity apply to var-
iation among individuals and breeds, but they also apply to fluc-
tuating asymmetry, indicating they have a shared developmental
basis. These patterns of variation are also found for the wolf and
across the Carnivora, suggesting that they existed before the do-
mestication of dogs and are not a result of selective breeding.
Keywords: Canis familiaris, Carnivora, geometric morphometrics,
morphological integration, novelty, selection.
Introduction
Evolutionary change in morphological traits can be very
rapid, as has been documented by a range of studies (Rez-
nick et al. 1997; Hendry and Kinnison 1999; Huey et al.
2000; Grant and Grant 2006). Most of the changes found
in natural populations are at a relatively small scale, how-
ever, and therefore the question is raised as to whether
there are inherent limits to the amounts of change that
can occur by microevolutionary mechanisms or whether
special macroevolutionary processes are required to
* Corresponding author. Present address: Department of Biology, College of
the Holy Cross, P.O. Box B, Worcester, Massachusetts 01610; e-mail:
abbygracedrake@googlemail.com.
Am. Nat. 2010. Vol. 175, pp. 289–301. 䉷2010 by The University ofChicago.
0003-0147/2010/17503-51313$15.00. All rights reserved.
DOI: 10.1086/650372
achieve large-scale change (Stanley 1998; Gould 2002).
Similarly, studies of adaptive radiation face difficulties in
extrapolating from microevolutionary data recorded in ex-
tant species to the larger scale of the entire radiation (e.g.,
Gavrilets and Losos 2009). These problems can be ad-
dressed directly in domesticated organisms, where sus-
tained selection by breeders has long been known to cause
large-scale phenotypic change (e.g., Darwin 1868).
Although large-scale variation has been indicated for
other domesticated species such as pigeons (Darwin 1868;
Helms and Brugmann 2007), and the morphological var-
iation of these species has been studied quantitatively (e.g.,
Johnston 1992), the variation among the breeds of the
domestic dog (Canis familiaris) is particularly suitable for
examining large morphological changes that originated ex-
clusively from microevolutionary processes (Darwin 1868;
Stockard 1941; Wayne 1986a, 1986b; Clutton-Brock 1995;
Coppinger and Coppinger 2001; Chase et al. 2002; Fondon
and Garner 2004; Parker et al. 2004; Lindblad-Toh et al.
2005; Wayne and Ostrander 2007). Many dog breeds were
established by selecting limited numbers of individuals for
breeding from larger populations (such as a regional pool
of farm dogs) or from one or more existing breeds; there-
fore, these breeds are equivalent to new lineages resulting
from founder events and hybridization (Parker et al. 2004;
Lindblad-Toh et al. 2005; Sampson and Binns 2006; Ra¨ber
2008). The reproductive separation and limited effective
population sizes of established breeds provide extensive
opportunity for divergence by genetic drift (Parker et al.
2004; Wayne and Ostrander 2007; Bjo¨rnerfeldt et al. 2008;
Calboli et al. 2008). Moreover, breeds are under selection
for a variety of morphological and behavioral traits (Cop-
pinger and Coppinger 2001; Parker et al. 2004; Ostrander
et al. 2006; Ra¨ber 2008). For instance, selection by breeders
for specific traits of the head has been shown to yield
sustained and substantial change in skull shape (Drake
and Klingenberg 2008). As a result, domestic dogs have a
vast spectrum of cranial variation (Darwin 1868; Stockard
1941; Nussbaumer 1982; Wayne 1986a; Young and Ban-
nasch 2006), which has clear functional consequences, for
290 The American Naturalist
instance, on bite forces (Ellis et al. 2009) or breathing
(Koch et al. 2003). In many ways, this situation corre-
sponds to an adaptive radiation (Schluter 2000; Gavrilets
and Losos 2009), but at the intraspecific level.
To what extent morphological variation is modular or
integrated throughout an entire structure is widely con-
sidered to be a key determinant of evolutionary flexibility
(Cheverud 1984; Wagner and Altenberg 1996; Klingenberg
2005). Because strong integration among parts may act as
a constraint if they are under opposing selection regimes,
modularity may enhance the capacity of morphological
traits to evolve. Moreover, the patterns of integration can
influence the direction of evolution and therefore may
have long-term consequences, for instance, if there are
multiple fitness peaks (Schluter 1996; Arthur 2001). In
turn, the patterns of morphological integration are also
expected to evolve under the influence of selection and,
as a consequence, to reflect functional associations among
traits (Cheverud 1984; Wagner and Altenberg 1996). Both
integration and modularity can vary in strength and are
not all-or-nothing phenomena (Klingenberg et al. 2003).
Integration and modularity can be studied by analyzing
the covariation among traits. Covariation of traits among
individuals reflects all differences, including evolutionary
divergence of taxa or breeds, whereas the covariation of
fluctuating asymmetry can reveal integration and modu-
larity of the developmental processes that are generating
the traits of interest (Klingenberg 2003, 2005).
Domestic dogs show abundant variation in the arrange-
ment of the nasomaxillary complex in relation to the neu-
rocranium (Stockard 1941; Nussbaumer 1982; Fondon
and Garner 2004; Drake and Klingenberg 2008), suggest-
ing that the two cranial complexes may be separate mod-
ules. Moreover, there is integrated shape variation along
a spectrum, from dolichocephalic breeds, which have slen-
der and elongate skulls, to brachycephalic breeds, which
have short and broad skull shapes (Wayne 1986a; Haworth
et al. 2001; Young and Bannasch 2006). Although parts of
the skull can be involved to different degrees, this type of
variation tends to have effects on most of the parts and
suggests integration throughout the whole skull. Similar
patterns of variation were also reported in domestic cats
(Ku¨nzel et al. 2003), in interspecific comparisons in Car-
nivora, and even in marsupials (Sears et al. 2007; Wroe
and Milne 2007), which raises the question as to whether
the process of domestication and divergence of breeds dis-
rupted ancestral constraints or exploited pre-existing pat-
terns of modular and integrated variation.
Here we use geometric morphometrics to characterize
and quantify shape variation in the skulls of domestic dogs,
and we compare this variation to variation in the three
most closely related wild species (Lindblad-Toh et al.
2005), the gray wolf (Canis lupus), the coyote (Canis la-
trans), and the golden jackal (Canis aureus). Because cra-
nial variation in domestic dogs is known to exceed that
in the family Canidae (Wayne 1986a), we use represen-
tatives from the entire order Carnivora as the basis for
comparison. In addition to quantifying the shape disparity
in these taxa, we also examine the patterns of integration
in the skull. Integration in the average shapes of the left
and right sides of the skull provides information about
variation among individuals in the samples, including the
differentiation of breeds or species. Cranial asymmetries,
however, arise from perturbations in the development of
each individual and can thus provide information about
the inherent tendency of skull development to produce
new morphological variation (Klingenberg 2005, 2008b).
By comparing integration in domestic dogs, in the wolf,
and across the Carnivora, we investigate whether domes-
tication changed the patterns of integration and modu-
larity, thereby disrupting ancestral evolutionary con-
straints, or whether the diversification of domestic dogs
used conserved patterns of cranial covariation.
Material and Methods
Data Set
Our study includes 677 adult dogs from 106 breeds (num-
bers of individuals per breed range from one to 62). The
sample of Carnivora includes one species each from 122
genera representing all major groups within the order
(Flynn et al. 2005), and care was taken to include the full
range of disparity. In addition, we examined samples of
adult gray wolves ( ), coyotes ( ), andnp288 np57
golden jackals ( ). Skulls were obtained from thenp49
following collections: the Smithsonian Institution’s Na-
tional Museum of Natural History (Washington, DC); the
private collection of Bonnie Dazell; the University of
Alaska Museum in Fairbanks, Alaska; the Museum of Ver-
tebrate Zoology at the University of California in Berkeley,
California; the Natural History Museum (London); the
Natural History Museum in Bern, Switzerland; the Powell-
Cotton Museum (Birchington, United Kingdom); and the
Oxford University Museum of Natural History (Oxford,
United Kingdom).
A set of 50 landmarks (12 median landmarks and 19
pairs on the left and right sides; fig. 1; table 1) was digitized
in three dimensions on the dorsal and the ventral aspects
of each skull using a MicroScribe digitizer. The landmarks
of the dorsal and the ventral aspects were combined by a
least squares fit (translation and rotation only) using four
landmarks that were digitized from both the dorsal and
the ventral views. Every skull was digitized twice.
Diversification of Skull Shape in Dogs 291
Figure 1: Landmarks on the dorsal and ventral sides of the skull that
were digitized (for anatomical definitions, see table 1). The dashed circles
indicate the hypothesis of modularity; the landmarks inside these circles
belong to the neurocranial module, whereas the landmarks outside the
circles are included in the facial module.
Shape Analysis
We quantify shape variation in the skulls of dogs by using
the methods of geometric morphometrics as implemented
in the MorphoJ software (Klingenberg 2008a). Shape var-
iation was extracted from the coordinate data by a full
Procrustes fit and projection to the shape-tangent space
(Dryden and Mardia 1998), using a procedure that takes
into account the object symmetry of the skull (Klingenberg
et al. 2002). An initial Procrustes ANOVA (Klingenberg et
al. 2002) confirmed that the Procrustes mean squares for
individual variation and fluctuating asymmetry substan-
tially exceeded measurement error, such that measurement
error is negligible (particularly for the analyses of variation
among individuals). For the remaining analyses, we av-
eraged the shape data from the replicate measurements of
each specimen. To explore the shape variation in the total
data set, we used a multivariate ordination of skull shapes
by a principal component (PC) analysis based on the co-
variance matrix of the symmetric component of shape
variation (Klingenberg et al. 2002). Visualizations of
shapes at the extremes of the PC axes were performed by
warping the scanned surface of a wolf skull, using the
Landmark software (Wiley et al. 2005).
Comparisons of Disparity
As a direct measure of the differences in shape, we computed
Procrustes distances between all possible pairs of specimens
in the dog sample and in the carnivoran sample. Procrustes
distances between specimens were computed as Euclidean
distances in tangent space (Dryden and Mardia 1998). Even
the greatest pairwise distances between specimens were
within the range of distances where the tangent-space ap-
proximation performs well (Marcus et al. 2000).
Two measures of shape disparity were computed. The
first is the Procrustes variance of observations in each
group, which is the mean squared Procrustes distance of
each specimen or breed average from the mean shape of
the respective group or, equivalently, the sumof the sample
variances of all Procrustes coordinates (Klingenberg and
McIntyre 1998; Zelditch et al. 2003). Procrustes variance
quantifies the average dispersion of data points around the
mean shape. The second measure is the volume of the
convex hull (de Berg et al. 2000) enclosing the data points
of each group, which quantifies the portion of shape space
occupied by the group. This is a measure of the degree of
difference among opposite extremes in each group, and
therefore it does not consider observations located near
the center of the scatter of data points. Convex hulls were
computed from the first three PCs because they contained
most of the variation in the sample and because com-
putation of higher-dimensional volumes presented com-
putational difficulties (dimensions with small amounts of
variation produce volumes near 0 for all samples, which
led to problems with numerical precision). The statistical
significance for pairwise comparisons of samples was es-
tablished with permutation tests (Good 2000) that sim-
ulated the null hypothesis of equal dispersion within
groups by randomly exchanging the deviations of data
points from the respective sample mean.
Allometry is a major factor in the diversification of dogs
(e.g., Wayne 1986a). To correct for the effects of allometry
on shape disparity, we computed the same disparity mea-
sures for the residuals from pooled within-group regres-
sion of shape on size. Because of the large amount of
variation in and the sample size of the dogs, this common
estimate of allometry is likely to be a better fit for dogs
than for the other groups in this study. It is therefore
expected to eliminate the allometric component of shape
variation more effectively from dogs than from the other
groups and, thus, to reduce the total shape disparity for
the dogs to a greater degree than for the other samples.
Analyses of disparity were conducted to compare dogs,
Table 1 : Osteometric landmarks collected on each skull, and their definitions
Landmark Definition
1 Midline point on the premaxilla at the inferior tip of the bony septum between the upper central incisors (F)
2 Premaxillary-maxillary suture, anterior, left side (F)
3 Premaxillary-maxillary suture, anterior, right side (F)
4 Nasal, anterior tip, left side (F)
5 Nasal, anterior tip, right side (F)
6 Nasale, nasal, anterior, midline (F)
7 Premaxillary-maxillary suture, posterior end in dorsal view, left side (F)
8 Premaxillary-maxillary suture, posterior end in dorsal view, right side (F)
9 Frontal-maxillary-nasal suture, left side (F)
10 Frontal-maxillary-nasal suture, right side (F)
11 Nasion, nasal-frontal suture, midline (F)
12 Frontal-maxillary suture, posterior, left side (F)
13 Frontal-maxillary suture, posterior, right side (F)
14 Frontal-parietal-sphenoid suture, left side (N)
15 Bregma, frontal-parietal suture, midline (N)
16 Frontal-parietal-sphenoid suture, right side (N)
17 Lambda, parietal-occipital suture, midline (N)
18 Asterion, posterior at occipital-parietal-temporal suture, left side (N)
19 Asterion, posterior at occipital-parietal-temporal suture, right side (N)
20 Opsithion, dorsal lip of foramen magnum, midline (N)
21 Occipital condyle (widest point of foramen magnum), left side (N)
22 Occipital condyle (widest point of foramen magnum), right side (N)
23 Basion, ventral lip of foramen magnum, midline (N)
24 Zygo-maxillare inferior, left side (F)
25 Squamosal-jugal suture, anterior projection of zygomatic process of temporal bone, left side (F)
26 Optic canal (ventral lip), left side (N)
27 Squamosal-jugal suture, posterior projection of jugal, ventral, left side (F)
28 External auditory meatus, posterior, left side (N)
29 Zygo-maxillare inferior, right side (F)
30 Squamosal-jugal suture, anterior projection of zygomatic process of temporal bone, right side (F)
31 Optic canal (ventral lip), right side (N)
32 Squamosal-jugal suture, posterior projection of jugal, ventral, right side (F)
33 External auditory meatus, posterior, right side (N)
34 Premaxillary-maxillary suture, lateral end in ventral view, left side (F)
35 Premaxillary-maxillary suture, lateral end in ventral view, right side (F)
36 Premaxillary-maxillary suture, posterior at midline (F)
37 Maxillary-palatine suture, anterior at midline (F)
38 Palatine, posterior at midline (F)
39 Presphenoid, anterior tip at midline (F)
40 Palatine-pterygoid suture posterior, right side (N)
41 Palatine-pterygoid suture posterior, left side (N)
42 Presphenoid-basisphenoid suture, midline (N)
43 Tympanooccipital fissure, anterior lip, right side (N)
44 Tympanooccipital fissure, anterior lip, left side (N)
45 Canine (posterior buccal corner), right side (F)
46 Premolar 3 (posterior buccal corner), right side (F)
47 Premolar 4 (posterior buccal corner), right side (F)
48 Canine (posterior buccal corner), left side (F)
49 Premolar 3 (posterior buccal corner), left side (F)
50 Premolar 4 (posterior buccal corner), left side (F)
Note: The letters F and N indicate whether a landmark belongs to the face or to the neurocranium, respectively.
Diversification of Skull Shape in Dogs 293
Figure 2: Hypothesis of modularity and adjacency graph for the land-
marks. The filled circles denote the landmarks of the facial module; the
open circles denote the landmarks of the neurocranial module (see also
fig. 1 and table 1). The lines connect landmarks that are deemed to be
anatomically adjacent to one other, and they are used to define spatial
contiguity of partitions of landmarks (Klingenberg 2009). A subset of
landmarks is spatially contiguous if all of its landmarks are connected
by the edges of the adjacency graph. A partition of the configuration is
contiguous if all subsets of landmarks are contiguous themselves (Klin-
genberg 2009).
the three wild canid species, and the Carnivora (see “Data
Set” for sample sizes). Moreover, we also compared the
disparity of dogs according to the following functional
groups (as defined by the United Kennel Club): Com-
panion Dog ( , 20 breeds), Guardian Dog (np138 np
, 23 breeds), Gun Dog ( , nine breeds), Herding246 np42
Dog ( , 10 breeds), Northern Breed ( , 12np44 np51
breeds comprising various sledge, hunting, and guard
dogs), Scenthound ( , six breeds), Sighthound andnp19
Pariah ( , 13 breeds, including the dingo and thenp68
New Guinea singing dog), and Terrier ( , 13 breeds).np69
Integration and Modularity
Because differences in the arrangement of the snout rel-
ative to the braincase—for instance, the variation between
klinorhynchy and airorhynchy (Nussbaumer 1982; Fon-
don and Garner 2004)—are an important component of
cranial variation in domestic dogs, we examine the hy-
pothesis that the face and the neurocranium are separate
modules (fig. 1; table 1). Modules are regions that are
integrated internally but that are relatively independent of
each other. The hypothesis of modularity therefore implies
that the covariation between the landmarks of the face and
those of the neurocranium should be weaker than the
covariation for other partitions of the landmarks into sub-
sets of the corresponding sizes (Klingenberg et al. 2003;
Klingenberg 2008b, 2009).
To assess the strength of modularity, we compared the
strength of covariation between subsets of landmarks for
the hypothesized modules with that of alternative partitions
of the total set of landmarks into subsets (Klingenberg
2009). To quantify the strength of covariation between sub-
sets of landmarks, we used the RV coefficient (Escoufier
1973), which can be interpreted as a multivariate general-
ization of the bivariate R
2
value (for detailed explanations,
see Klingenberg 2009; Laffont et al. 2009). The RV coeffi-
cient for the subdivision of landmarks into facial and neu-
rocranial regions was compared with the distribution of RV
coefficients for randomly generated subdivisions. Random
subdivisions were formed from the paired landmarks of one
side and the median landmarks such that the numbers of
landmarks matched those in the facial and neurocranial
regions (Klingenberg 2009). We limited this comparison to
those subdivisions where both subsets of landmarks were
spatially contiguous (i.e., connected by the edges of a graph
indicating landmarks that are anatomically adjacent; fig. 2;
for details, see Klingenberg 2009).
For the analysis of modularity in the symmetric com-
ponent of shape variation in the Carnivora, we used in-
dependent contrasts (Felsenstein 1985) to take into ac-
count the phylogenetic structure of the data, using the
supertree of Bininda-Emonds et al. (2007). This adjust-
ment was not required for the asymmetric component,
because the left-right differences are computed within in-
dividuals and are therefore independent with respect to
phylogeny.
For quantifying overall similarity of covariance matrices,
we computed matrix correlations and the associated per-
mutation tests, using procedures adapted for geometric
morphometrics and object symmetry (Klingenberg and
McIntyre 1998; Klingenberg et al. 2002; Klingenberg
2008a).
Results
Patterns of Morphological Diversification and Novelty
The first three PCs account for 71.8% of the total shape
variation, and therefore, they provide a reasonable ap-
proximation of the total variation (no other PC accounts
for more than 5%). PC1 primarily contrasts brachyce-
phalic and dolichocephalic skulls; PC2 opposes elongate
skulls, where the braincase is aligned posterior to the muz-
zle, to broader and higher skulls, where the braincase is
raised above the rostrum; and, finally, PC3 sets skulls with
enlarged faces and broad and high rostra against others
with relatively larger braincases and muzzles that taper
markedly toward the front (fig. 3D). The scatter of PC
scores (fig. 3A–3C) shows that the amount of variation is
much greater in dogs than it is in wolves, coyotes, and
golden jackals and that the dispersion of data points for
dogs is comparable to that across the Carnivora.
294 The American Naturalist
Figure 3: Principal component (PC) analysis for skull shape in the complete data set. A–C, Plots of the PC scores. D, Shape changes associated
with the PC axes. For each PC, the shapes corresponding to the observed extremes in the positive and negative directions are shown as a warped
surface of a wolf skull (Wiley et al. 2005).
Closer inspection of the PC plots for shape variation
(fig. 3) clearly shows that domestic dogs have only limited
overlap with the other carnivores (see particularly fig. 3C).
The samples of wolves, coyotes, and golden jackals are
located in the region of the intersection between dogs and
the other carnivores. It is clear from this analysis not only
that domestic dogs occupy a large region of shape space
that is outside of the range of the ancestral species, the
wolf, and the other members of the family Canidae (Wayne
1986a) but also that much of the shape variation in do-
mestic dogs is novel relative to the range of skull shapes
in the order Carnivora as a whole.
Quantifying Disparity
The greatest Procrustes distance between specimens in the
dog sample is 0.477 (between a Collie and a Pekingese),
which exceeds the distance of 0.424 between the most
divergent specimens in the Carnivora sample (walrus
Odobenus rosmarus and falanouc Eupleres goudotii).
Among domestic dogs, pairwise distances greater than 0.45
involve other pairings of breeds (Borzoi-Pekingese, Collie–
Japanese Chin, Borzoi–Japanese Chin, Borzoi-Pug, Aire-
dale Terrier–Pekingese, Scottish Deerhound–Pekingese,
Collie-Pug). All of these contrasts are between dolicho-
cephalic and brachycephalic breeds. In the sample of the
Carnivora, all other Procrustes distances greater than 0.4
also involve the walrus, in combination with the mountain
coati (Nasuella olivacea, 0.405), the Malagasy civet (Fossa
fossana, 0.404), Owston’s palm civet (Chrotogale owstoni,
0.404), and the servaline genet (Genetta servalina, 0.401).
To compare the amounts of shape variation in the entire
sample of domestic dogs, the three wild species, and the
Carnivora, we used two measures of disparity that capture
Diversification of Skull Shape in Dogs 295
Figure 4: Skull shape disparity in the different samples. A, Total shape
disparity, quantified by Procrustes variance. B, Total shape disparity,
quantified by the volumes of the convex hulls for the first three dimen-
sions. C, Shape disparity corrected for the effects of size, quantified by
Procrustes variance. D, Shape disparity corrected for the effects of size,
quantified by the volumes of convex hulls for the first three dimensions.
Figure 5: Skull shape disparity in the different groups of dogs, according
to the breed classification of the United Kennel Club. A, Total shape
disparity, quantified by Procrustes variance. B, Total shape disparity,
quantified by the volumes of the convex hulls for the first three dimen-
sions. C, Shape disparity corrected for the effects of size, quantified by
Procrustes variance. D, Shape disparity corrected for the effects of size,
quantified by the volumes of convex hulls for the first three dimensions.
The last column in each panel (“All non-Companion”) includes all groups
other than Companion Dog. The size correction for Cand Dwas com-
puted for dogs separately, and therefore it differs from the one used in
figure 4; Aand Bare directly comparable to figure 4Aand 4B, respectively.
different aspects of shape diversification within samples:
Procrustes variance and the volume of the convex hull
enclosing the data points. Both measures of shape disparity
give similar results (fig. 4A,4B). The variation among
average shapes of dog breeds is less than that among in-
dividual dogs (significantly so for the volume of convex
hulls, ; nonsignificantly so for Procrustes vari-Pp.0004
ance, ). The disparity within dogs, both amongPp.12
individuals and among breed means, is consistently much
greater than the disparity within wolves, coyotes, and
golden jackals (for both measures of disparity; all P≤
). Finally, the variation among individual dogs is of a.002
magnitude similar to that in the carnivore sample (sig-
nificant difference for Procrustes variance, ; noPp.0028
significant difference for the volume of convex hulls,
). The fact that this match between dogs and car-Pp.99
nivores is closer for the volumes of the convex hulls (fig.
4B) than for the Procrustes variances (fig. 4A) reflects the
fact that the disparity among dogs is dominated particu-
larly by some breeds that are highly divergent, whereas
many breeds retain a skull shape that is closer to the an-
cestral shape (fig. 3).
Because allometry is a major factor in the diversification
of dog breeds, we repeated these comparisons with data
that were corrected for the effects of allometry. The esti-
mates of disparity based on the residuals from allometric
regression (fig. 4C,4D) are similar to the uncorrected
values, with substantially greater disparity for domestic
dogs than for the wild canid species. Because the estimate
of allometry is dominated by domestic dogs, the amounts
of variation are reduced more for the dogs than for the
other groups (fig. 4C,4D; the slight increase for the car-
nivore sample is due to differences in allometries). Nev-
ertheless, this analysis shows conclusively that the dispro-
portionate amount of shape variation in domestic dogs is
not simply due to allometric scaling.
Among the groups of dog breeds as defined by the
United Kennel Club, the Companion Dog group has a far
greater disparity than do any of the other groups, regard-
less of which of the two disparity measures is used or
whether a correction for allometric effects is made (fig. 5;
all ). If the Companion Dog group is contrastedP≤.0001
to all other groups jointly (rightmost columns in the panels
of fig. 5), it still has a much greater Procrustes variance
( ), but it has a comparable volume of the convexP!.0001
hull ( and .82 for analyses without and with thePp.99
correction for allometric effects, respectively). The differ-
ence in the results reflects the manner in which the two
measures of disparity consider the average versus the ex-
treme deviations from the shape averages in the groups
under comparison.
296 The American Naturalist
Figure 6: Analysis of modularity in the skull. Graphs show the RV
coefficients (interpreted as multivariate generalizations of the bivariate
R
2
values; Escoufier 1973) for the subdivision of landmarks into facial
and neurocranial regions (arrows) and the distribution of RV coefficients
for 10,000 alternative partitions of landmarks into anatomically contig-
uous subsets (histograms). A, Domestic dogs, symmetric component of
variation. This includes the shape variation among breeds. B, Domestic
dogs, asymmetric component. This is the component of fluctuating asym-
metry, and it provides a measure of the tendency of cranial development
to produce new variation. C, Wolves, symmetric component of variation.
D, Wolves, asymmetric component. E, Carnivora, independent contrasts
(Felsenstein 1985) for the symmetric component of variation. F, Car-
nivora, asymmetric component.
Integration and Modularity of the Skull
In the sample of domestic dogs, the RV coefficient between
facial and neurocranial landmarks is 0.87 for the sym-
metric component of variation, which indicates a very
strong association. This value shows the tight integration
of the entire skull in dogs (PC1, which represents a contrast
between dolichocephalic and brachycephalic breeds, alone
accounts for more than 63% of the symmetric variance in
the sample). Nevertheless, the RV coefficient between the
facial and neurocranial subsets is lower than those for most
other subdivisions of the landmarks ( ; fig. 6A).Pp.0475
Analyses with a correction for allometric variation yield
very similar results. Overall, therefore, the face and the
neurocranium possess a degree of modular separation de-
spite the strong integration, particularly in the differences
between breeds (e.g., breeds with short skulls vs. those
with long skulls).
For asymmetry in dogs, integration is generally lower
than it is for the symmetric component (fig. 6B), and the
RV coefficient between facial and neurocranial regions is
0.32. This RV coefficient is lower than those for most
alternative partitions ( ). The matrix correlationPp.0052
between the covariance matrices for the symmetric and
the asymmetric components of variation in dogs is 0.20
( ), indicating a weak but statistically discerniblePp.044
relationship between the diversification of skull shapes
across breeds and the intrinsic pattern of cranial variation.
Integration in the symmetric component of shape var-
iation in the wolf is substantially weaker than it is for the
domestic dogs (fig. 6C). The RV coefficient between facial
and neurocranial landmarks is 0.32, which is lower than
the RV coefficients for most alternative partitions (Pp
). For asymmetry in wolves, the RV coefficient for.0538
the subdivision into facial and neurocranial sets is 0.21,
and only 26 of 10,000 random subdivisions yield a lower
covariation. The range of RV coefficients is fairly limited
for both symmetric and asymmetric components (fig. 6C,
6D), because all partitions have a similar, low-to-moderate
covariation. Whereas this result is consistent with the hy-
pothesis of facial and neurocranial modules, the limited
range of RV coefficients indicates a fairly weak modularity.
The matrix correlation between the covariance matrices
for the symmetric and asymmetric components of varia-
tion in the wolf sample is 0.63 ( ). The asymmetricP!.0001
components of the dog and the wolf samples have a matrix
correlation of 0.77 ( ) and are thus quite similar,P!.0001
suggesting shared patterns of developmental variation in
wolves and dogs. The patterns of variation in the sym-
metric component are less similar (matrix correlation:
0.47; ).P!.0001
In the analysis of the symmetric component of shape
variation in the Carnivora, the RV coefficient for the sub-
division into facial and neurocranial sets is 0.44, and it is
less than the values for most random partitions (Pp
; fig. 6E). For fluctuating asymmetry in the Carnivora,.0293
the RV coefficient between the hypothesized modules is
0.29, which is lower than the RV coefficients for most
random partitions ( ; fig. 6F). These results arePp.0231
consistent with the hypothesis of modularity. The patterns
of variation of the symmetric and asymmetric components
are moderately similar (matrix correlation: 0.45; Pp
). The correspondence between the asymmetric com-.0016
ponent in the Carnivora and those in dogs and wolves is
considerably better (matrix correlation: 0.73 and 0.72, re-
spectively; both ). Moreover, there is a significantP!.0001
similarity between the covariance matrix of independent
contrasts in the Carnivora and that for the symmetric
Diversification of Skull Shape in Dogs 297
component of variation in domestic dogs (matrix corre-
lation: 0.54; ), indicating that the patterns of di-P!.0001
versification in domestic dogs are moderately similar to
those in the Carnivora.
Discussion
The breeds of domestic dogs have long been known for
their great variety of skull shapes (Darwin 1868; Stockard
1941; Nussbaumer 1982; Wayne 1986a; Clutton-Brock
1995; Young and Bannasch 2006). Our analyses show that
the variation of cranial shape in dogs is comparable to
that across the entire order Carnivora (figs. 3, 4) and that
differences in skull shape between extreme dog breeds even
exceed the maximal distances we found among the species
of Carnivora.
This massive disparity among dogs has evolved in a few
hundred to several thousand years (Clutton-Brock 1995;
Lindblad-Toh et al. 2005), a very brief time span by com-
parison with the evolution of natural clades with com-
parable disparity, which raises the question as to whether
the diversification in dogs can be representative of large-
scale evolution in nature. Studies in natural populations
have documented rapid change due to selection (Reznick
et al. 1997; Hendry and Kinnison 1999; Huey et al. 2000)
and, in some cases, also hybridization (Grant and Grant
2002), but the changes observed in these populations were
much smaller than the differences among dog breeds. At
a larger scale, bursts of relatively rapid increase in disparity
have been inferred for clades of species (e.g., Harmon et
al. 2003; Ricklefs 2004), and speciation itself may be a
factor in this process (Ricklefs 2004; Mattila and Bokma
2008). Because dogs are a single species, however, with
separation of breeds maintained only by human interven-
tion, it is not speciation per se that is the driving force
producing morphological disparity, although population
bottlenecks and founder effects during establishment of
breeds might cause genetic conditions similar to those in
new species or island populations (Meffert 2006; Millien
2006; Losos and Ricklefs 2009). As a consequence of these
effects, nonsynonymous mutations have accumulated in
the dog genome (Cruz et al. 2008), which may facilitate
morphological divergence. Some of the divergence be-
tween dog breeds has been linked to single mutational
steps with large effects (Sutter et al. 2007; Parker et al.
2009), but there are also reports of mechanisms producing
incremental change in dog breeds as well as among car-
nivoran species (Fondon and Garner 2004; Sears et al.
2007). There is little information on the dynamics of
change in dog breeds, but a detailed study of the historical
change of skull shape in St. Bernard dogs found that the
features that were described as desirable in the breed stan-
dard gradually became more accentuated (Drake and Klin-
genberg 2008). Overall, it appears that the processes of
morphological diversification in dog breeds are broadly
comparable to those in natural lineages. Dogs are therefore
an excellent model system for the study of evolutionary
processes, including adaptive radiation and the origin of
large-scale morphological disparity.
By comparison, there is only little quantitative infor-
mation about the degree of variation in other domesticated
animals. Although cranial variation in domestic cats is
considerable (Ku¨nzel et al. 2003), it is not clear whether
it is comparable to the variation across the family Felidae
(Christiansen 2008). This apparently smaller scale of var-
iation may reflect the fact that genetic distance among cat
breeds tends to be clearly less than that among dog breeds
(Menotti-Raymond et al. 2007). For domestic pigeons,
widely different cranial shapes have been reported (Helms
and Brugmann 2007), but no quantitative studies of cra-
nial variation have been published. Divergence of skeletal
traits among pigeon breeds is much greater than the dif-
ferences between rock pigeons (Columba livia) and feral
pigeon populations (Johnston 1992). Unfortunately, the
differences in measurements and methods make it difficult
to compare across these studies, and none of these studies
includes more than a small number of breeds. Moreover,
except for those of domestic dogs (this article; Wayne
1986a, 1986b), no study directly compares the morpho-
logical variation in a domesticated species with that in the
larger clade to which it belongs.
In addition to the great amount of variation in domestic
dogs, a striking result is that most of the range of skull
shape variation is outside the range of that for the Car-
nivora (fig. 3C). Even if allowance is made for intraspecific
variation, for which the range of variation in the three
wild canid species can serve as a basis of comparison, a
substantial part of the domain of variation for the dogs is
still outside the zone of overlap (fig. 3). These shapes
represent novel arrangements of the skull in domestic dogs
that are not found in other carnivorans.
These novel skull shapes, as well as the disparity in dog
breeds and in wild species (figs. 4, 5), suggest that diver-
sification of dog skull shapes is due, at least in part, to the
radical change of the selective regime, as dog breeds were
derived from wolves through domestication and the later
establishment of modern breeds (Lindblad-Toh et al. 2005;
Pollinger et al. 2005; Bjo¨rnerfeldt et al. 2006; Wayne and
Ostrander 2007). Domestication relaxed selective pressures
in contexts such as foraging, and the requirement to process
hard or tough foods was reduced. This new selective regime
may tolerate changes in cranial morphology even if they
affect aspects of performance such as bite force (Ellis et al.
2009) or breathing, which is impeded in brachycephalic dogs
(Koch et al. 2003; similar problems for cats are discussed
in Ku¨nzel et al. 2003). One would expect the functional
298 The American Naturalist
demands on skull shape to be most permissive for the Com-
panion Dog group, intermediate for breeds selected to per-
form work (such as hunting, herding, or guarding), and
most stringent for the wild species. In agreement with this
expectation, the greatest amounts of disparity are found in
the Companion Dog group, fewer are found in the other
groups of breeds, and the smallest amounts are found in
the three wild species (figs. 4, 5). Furthermore, this argu-
ment predicts that feral dogs should be less variable than
pet dogs and other domestic breeds. This is difficult to
assess, because only a few specimens were available: three
dingos and two New Guinea singing dogs, which are ancient
populations derived from feral dogs (Savolainen et al. 2004).
The Sighthound and Pariah group, which contains both
breeds, is not unusually variable by comparison with the
other functional groups. Stronger support for such an ar-
gument was found for skeletal measurements in pigeons,
where feral pigeons resemble wild rock pigeons, whereas
domestic pigeon breeds are much more different from each
other, suggesting that feral pigeons were under selection for
trait values similar to those of their wild ancestor but that
domestic pigeons were not (Johnston 1992; Sol 2008). No
comparable evidence appears to be available for cranial
shape in pigeons or for any skeletal structure in other feral
animals.
Since the establishment of modern dog breeds, artificial
selection for behaviors or appearance desired by breeders
specifically favored divergence (Kemp et al. 2005; Drake
and Klingenberg 2008). Direct selection for head shape is
expected to be strongest for pet dogs, which is consistent
with the large disparity of the Companion Dog group (fig.
5). In addition, the process of domestication itself may
bring about a variety of changes due to selection for tame-
ness (Trut et al. 2004) and due to changes in the selective
regime (Bjo¨rnerfeldt et al. 2006) and direct mechanical
influence on skull growth (O’Regan and Kitchener 2005).
In addition to selection, bottleneck events in the estab-
lishment of new breeds and subsequent inbreeding (Lind-
blad-Toh et al. 2005; Bjo¨rnerfeldt et al. 2008; Calboli et
al. 2008) may have further enhanced diversification
through the fixation of mutations that affect cranial shape
(Haworth et al. 2001; Neff et al. 2004; Pollinger et al. 2005;
Bjo¨rnerfeldt et al. 2006; Cruz et al. 2008).
Morphological Integration and Modularity
The large-scale divergence of dog breeds in response to
selection by breeders raises the question about the devel-
opmental basis of variation. Was there any particular pre-
disposition in the developmental system of the skull that
facilitated this diversification? Or, did the processes of do-
mestication and artificial selection themselves produce a
reorganization of the patterns of integration and modu-
larity? If genetic and developmental modularity are molded
by functional associations of traits and selection (Cheverud
1984; Wagner and Altenberg 1996), a drastic change of the
selective regime such as domestication may be expected
to affect the integration and modularity of the skull.
The results of the comparisons of RV coefficients for
dogs (fig. 4) are generally consistent with the hypothesis
that the face and the neurocranium are morphological
modules, as the subdivision into face and neurocranium
produces RV coefficients that are lower than those for most
alternative partitions. Moreover, the relatively limited
range of RV coefficients in each of the analyses indicates
that modularity is relatively weak, that is, that the co-
variation within modules is not much stronger than the
covariation between modules (Klingenberg et al. 2003;
Klingenberg 2009). The high RV coefficients, particularly
for the symmetric component in dogs, also reflect the large
proportion of variation that is integrated throughout the
skull, such as the spectrum between brachycephalic and
dolichocephalic skull shapes (e.g., PC1 in fig. 3; Stockard
1941; Wayne 1986a; Haworth et al. 2001; Young and Ban-
nasch 2006). For fluctuating asymmetry, which originates
from random variation produced by the developmental
system, integration is weaker and the range of RV coef-
ficients tends to be greater than that for the symmetric
component of variation. Particularly for the dogs, the
modular structure of the skull is therefore more apparent
for this spontaneous developmental variation than it is for
the symmetric component of shape variation, which in-
cludes the evolved differences among breeds. A lower de-
gree of integration of asymmetry than for the among-
individual variation has also been found in similar analyses
for rodents and fly wings (e.g., Klingenberg 2009; Laffont
et al. 2009) and may therefore be a widespread phenom-
enon.
The patterns of developmental integration, as charac-
terized by the covariance matrices for fluctuating asym-
metry (Klingenberg 2003, 2008b), are similar in domestic
dogs, in the wolf, and across the order Carnivora (all ma-
trix correlations 10.7 and ). Accordingly, there isP!.0001
no indication that domestication changed the way in which
new morphological variation is generated. On the contrary,
the results are consistent with the possibility that the an-
cestral developmental system was mostly conserved
throughout the evolution of the carnivores and the do-
mestication of dogs, despite the long evolutionary time
spans and the large scale of morphological diversification.
We are not aware of similar comparisons of the patterns
of integration for fluctuating asymmetry, except at the
intraspecific level (Debat et al. 2006, 2008, 2009). At the
level of intraspecific variation, where factors other than
development may also influence the observed patterns, a
comparison across the order Carnivora found that patterns
Diversification of Skull Shape in Dogs 299
of cranial integration within species are mostly similar
(Goswami 2006b) and that there are some shared patterns
of integration even across therian mammals (Goswami
2006a; Porto et al. 2009). Despite this large-scale conser-
vation, patterns of integration can vary at small taxonomic
scales (Steppan 1997; Mitteroecker and Bookstein 2008;
Jamniczky and Hallgrı´msson 2009; Kulemeyer et al. 2009).
The matrix correlation between the symmetric and the
asymmetric components of shape variation for the wolf
(0.63) is stronger than the low-to-moderate matrix cor-
relations for domestic dogs (0.20) or Carnivora (0.45). For
both dogs and Carnivora, the symmetric component of
variation is dominated by the evolved differences among
breeds or species, whereas in the wolf it is mainly within-
population variation. The relatively high matrix correla-
tion between symmetric and asymmetric components of
the wolf suggests that new variation generated by the de-
velopmental system is incorporated into the pool of
within-population variation in a fairly equitable manner.
In contrast, the lower matrix correlations for the Carnivora
and for dogs suggest that patterns of diversification among
taxa or breeds differ substantially from the patterns of
variation spontaneously produced by the developmental
system because phylogenetic divergence of Carnivora and
artificial selection by dog breeders disproportionately fa-
vored some directions in shape space over others. These
matrix correlations correspond to the mixed results found
in similar comparisons between patterns of symmetric var-
iation and fluctuating asymmetry for other animals (e.g.,
Klingenberg and McIntyre 1998; Debat et al. 2000; Klin-
genberg et al. 2002; Willmore et al. 2005; Breuker et al.
2006), with the difference being that all of those studies
were based on intrapopulation variation.
The diversification of both the Carnivora and the do-
mestic dogs relies on a combination of the input of new
variation, which retains a mostly conserved pattern, and
selection that can favor particular aspects of shape. As a
result, diversification in skull shape is responsive to specific
selective inputs, but it follows general patterns concerning
the relative length of the skull (brachycephalic vs. doli-
chocephalic skulls) and the relative arrangement of the
facial and neurocranial parts of the skull (airorhynchy vs.
klinorhynchy, but possibly other rearrangements as well).
Investigating the contributions of the developmental sys-
tem and specific selective events, in dogs and other do-
mesticated species as well as in naturally evolved clades,
will be a promising approach for understanding the pro-
cesses responsible for morphological diversification.
Acknowledgments
We are grateful to the curators of the various collections
(see “Material and Methods”) for access to skulls in their
care. We thank E. Sherratt for crucial technical help with
visualizing three-dimensional surfaces and M. Nussbau-
mer and two anonymous reviewers for helpful comments
on drafts of this article. This research was supported by
grants from the Leverhulme Trust and the Royal Society.
Literature Cited
Arthur, W. 2001. Developmental drive: an important determinant of
the direction of phenotypic evolution. Evolution andDevelopment
3:271–278.
Bininda-Emonds, O. R. P., M. Cardillo, K. E. Jones, R. D. E.MacPhee,
R. M. D. Beck, R. Grenyer, S. A. Price, R. A. Vos, J. L. Gittleman,
and A. Purvis. 2007. The delayed rise of present-day mammals.
Nature 446:507–512.
Bjo¨ rnerfeldt, S., M. T. Webster, and C. Vila` . 2006. Relaxation of
selective constraint on dog mitochondrial DNA following domes-
tication. Genome Research 16:990–994.
Bjo¨ rnerfeldt, S., F. Hailer, M. Nord, and C. Vila` . 2008. Assortative
mating and fragmentation within dog breeds. BMC Evolutionary
Biology 8:28.
Breuker, C. J., J. S. Patterson, and C. P. Klingenberg. 2006. A single
basis for developmental buffering of Drosophila wing shape. PLoS
ONE 1:e7.
Calboli, F. C. F., J. Sampson, N. Fretwell, and D. J. Balding. 2008.
Population structure and inbreeding from pedigree analysis of
purebred dogs. Genetics 179:593–601.
Chase, K., D. R. Carrier, F. R. Adler, T. Jarvik, E. A. Ostrander, T.
D. Lorentzen, and K. G. Lark. 2002. Genetic basis for systems of
skeletal quantitative traits: principal component analysis of the
canid skeleton. Proceedings of the National Academy of Sciences
of the USA 99:9930–9935.
Cheverud, J. M. 1984. Quantitative genetics and developmental con-
straints on evolution by selection. Journal of Theoretical Biology
110:155–171.
Christiansen, P. 2008. Evolution of skull and mandible shape in cats
(Carnivora: Felidae). PLoS ONE 3:e2807.
Clutton-Brock, J. 1995. Origins of the dog: domestication and early
history. Pages 7–20 in J. Serpell, ed. The domestic dog: its evo-
lution, behaviour and interactions with people. Cambridge Uni-
versity Press, Cambridge.
Coppinger, R., and L. Coppinger. 2001. Dogs: a startling new un-
derstanding of canine origin, behavior, and evolution. Scribner,
New York.
Cruz, F., C. Vila` , and M. T. Webster. 2008. The legacy of domesti-
cation: accumulation of deleterious mutations in the dog genome.
Molecular Biology and Evolution 25:2331–2336.
Darwin, C. 1868. Animals and plants under domestication. Vol. 1.
J. Murray, London.
Debat, V., P. Alibert, P. David, E. Paradis, and J.-C. Auffray. 2000.
Independence between developmental stability and canalization in
the skull of the house mouse. Proceedings of the Royal Society B:
Biological Sciences 267:423–430.
Debat, V., C. C. Milton, S. Rutherford, C. P. Klingenberg, and A. A.
Hoffmann. 2006. Hsp90 and the quantitative variation of wing
shape in Drosophila melanogaster. Evolution 60:2529–2538.
Debat, V., R. Cornette, A. B. Korol, E. Nevo, D. Soulet, and J. R.
David. 2008. Multidimensional analysis of Drosophila wing vari-
ation in Evolution Canyon. Journal of Genetics 87:407–419.
Debat, V., A. Debelle, and I. Dworkin. 2009. Plasticity, canalization,
300 The American Naturalist
and developmental stability of the Drosophila wing: joint effects
of mutations and developmental temperature. Evolution 63:2864–
2876.
de Berg, M., M. van Krefeld, M. Overmars, and O. Schwarzkopf.
2000. Computational geometry: algorithms and applications.
Springer, Berlin.
Drake, A. G., and C. P. Klingenberg. 2008. The pace of morphological
change: historical transformation of skull shape in St. Bernard
dogs. Proceedings of the Royal Society B: Biological Sciences 275:
71–76.
Dryden, I. L., and K. V. Mardia. 1998. Statistical shape analysis.Wiley,
Chichester.
Ellis, J. L., J. Thomason, E. Kebreab, K. Zubair, and J. France. 2009.
Cranial dimensions and forces of biting in the domestic dog. Jour-
nal of Anatomy 214:362–373.
Escoufier, Y. 1973. Le traitement des variables vectorielles. Biometrics
29:751–760.
Felsenstein, J. 1985. Phylogenies and the comparative method.Amer-
ican Naturalist 125:1–15.
Flynn, J. J., J. A. Finarelli, S. Zehr, J. Hsu, and M. A. Nedbal. 2005.
Molecular phylogeny of the Carnivora (Mammalia): assessing the
impact of increased sampling on resolving enigmatic relationships.
Systematic Biology 54:317–337.
Fondon, J. W., III, and H. R. Garner. 2004. Molecular origins of
rapid and continuous morphological evolution. Proceedings of the
National Academy of Sciences of the USA 101:18058–18063.
Gavrilets, S., and J. B. Losos. 2009. Adaptive radiation: contrasting
theory with data. Science 323:732–737.
Good, P. 2000. Permutation tests: a practical guide to resampling
methods for testing hypotheses. Springer, New York.
Goswami, A. 2006a. Cranial modularity shifts during mammalian
evolution. American Naturalist 168:270–280.
———. 2006b. Morphological integration in the carnivoran skull.
Evolution 60:169–183.
Gould, S. J. 2002. The structure of evolutionary theory. Harvard
University Press, Cambridge, MA.
Grant, P. R., and B. R. Grant. 2002. Unpredictable evolution in a 30-
year study of Darwin’s finches. Science 296:707–711.
———. 2006. Evolution of character displacement in Darwin’s
finches. Science 313:224–226.
Harmon, L. J., J. A. Schulte II, A. Larson, and J. B. Losos. 2003.
Tempo and mode of evolutionary radiation in iguanian lizards.
Science 301:961–964.
Haworth, K. E., I. Islam, M. Breen, W. Putt, E. Makrinou, M. Binns,
D. Hopkinson, and Y. Edwards. 2001. Canine TCOF1: cloning,
chromosome alignment and genetic analysis in dogs with different
head types. Mammalian Genome 12:622–629.
Helms, J. A., and S. A. Brugmann. 2007. The origins of species-
specific facial morphology: the proof is in the pigeon. Integrative
and Comparative Biology 47:338–342.
Hendry, A. P., and M. T. Kinnison. 1999. The pace of modern life:
measuring rates of contemporary microevolution. Evolution 53:
1637–1653.
Huey, R. B., G. W. Gilchrist, M. L. Carlson, D. Berrigan, and L. Serra.
2000. Rapid evolution of a geographic cline in size in an introduced
fly. Science 287:308–309.
Jamniczky, H. A., and B. Hallgrı´msson. 2009. A comparison of co-
variance structure in wild and laboratory muroid crania. Evolution
63:1540–1556.
Johnston, R. F. 1992. Evolution in the rock dove: skeletal morphology.
Auk 109:530–542.
Kemp, T. J., K. N. Bachus, J. A. Nairn, and D. R. Carrier. 2005.
Functional trade-offs in the limb bones of dogs selected for run-
ning versus fighting. Journal of Experimental Biology 208:3475–
3482.
Klingenberg, C. P. 2003. Developmental instability as a research tool:
using patterns of fluctuating asymmetry to infer the developmental
origins of morphological integration. Pages 427–442 in M. Polak,
ed. Developmental instability: causes and consequences. Oxford
University Press, New York.
———. 2005. Developmental constraints, modules and evolvability.
Pages 219–247 in B. Hallgrı´msson and B. K. Hall, eds. Variation:
a central concept in biology. Elsevier, Burlington, MA.
———. 2008a. MorphoJ. Faculty of Life Sciences, University of Man-
chester, Manchester. http://www.flywings.org.uk/MorphoJ_page
.htm.
———. 2008b. Morphological integration and developmental mod-
ularity. Annual Review of Ecology, Evolution, and Systematics 39:
115–132.
———. 2009. Morphometric integration and modularity in config-
urations of landmarks: tools for evaluating a-priori hypotheses.
Evolution and Development 11:405–421.
Klingenberg, C. P., and G. S. McIntyre. 1998. Geometric morpho-
metrics of developmental instability: analyzing patterns of fluc-
tuating asymmetry with Procrustes methods. Evolution 52:1363–
1375.
Klingenberg, C. P., M. Barluenga, and A. Meyer. 2002. Shape analysis
of symmetric structures: quantifying variation among individuals
and asymmetry. Evolution 56:1909–1920.
Klingenberg, C. P., K. Mebus, and J.-C. Auffray. 2003. Developmental
integration in a complex morphological structure: how distinct
are the modules in the mouse mandible? Evolution and Devel-
opment 5:522–531.
Koch, D. A., S. Arnold, M. Hubler, and P. M. Montavon. 2003.
Brachycephalic syndrome in dogs. Compendium Continuing Ed-
ucation for Veterinarians 25:48–55.
Kulemeyer, C., K. Asbahr, P. Gunz, S. Frahnert, and F. Bairlein. 2009.
Functional morphology and integration of corvid skulls: a 3D
geometric morphometric approach. Frontiers in Zoology 6:2.
Ku¨ nzel, W., S. Breit, and M. Oppel. 2003. Morphometric investi-
gations of breed-specific features in feline skulls and considerations
on their functional implications. Anatomia Histologia Embryol-
ogia 32:218–223.
Laffont, R., E. Renvoise´ , N. Navarro, P. Alibert, and S. Montuire.
2009. Morphological modularity and assessment of developmental
processes within the vole dental row (Microtus arvalis, Arvicolinae,
Rodentia). Evolution and Development 11:302–311.
Lindblad-Toh, K., C. M. Wade, T. S. Mikkelsen, E. K. Karlsson, D.
B. Jaffe, M. Kamal, M. Clamp, et al. 2005. Genome sequence,
comparative analysis and haplotype structure of the domestic dog.
Nature 438:803–819.
Losos, J. B., and R. E. Ricklefs. 2009. Adaptation and diversification
on islands. Nature 457:830–836.
Marcus, L. F., E. Hingst-Zaher, and H. Zaher. 2000. Application of
landmark morphometrics to skulls representing the orders of living
mammals. Hystrix 11:27–47.
Mattila, T., and F. Bokma. 2008. Extant mammal body masses suggest
punctuated equilibrium. Proceedings of the Royal Society B: Bi-
ological Sciences 275:2195–2199.
Diversification of Skull Shape in Dogs 301
Meffert, L. M. 2006. Population bottlenecks and founder effects.
Pages 414–425 in C. W. Fox and J. B. Wolf, eds. Evolutionary
genetics: concepts and case studies. Oxford University Press, New
Yor k .
Menotti-Raymond, M., V. A. David, S. M. Pflueger, K. Lindblad-Toh,
C. M. Wade, S. J. O’Brien, and W. E. Johnson. 2007. Patterns of
molecular genetic variation among cat breeds. Genomics 91:1–11.
Millien, V. 2006. Morphological evolution is accelerated among island
mammals. PLoS Biology 4:e321.
Mitteroecker, P., and F. L. Bookstein. 2008. The evolutionary role of
modularity and integration in the hominoid cranium. Evolution
62:943–958.
Neff, M. W., K. R. Robertson, A. K. Wong, N. Safra, K. W. Broman,
M. Slatkin, K. L. Mealey, and N. C. Perdersen. 2004. Breed dis-
tribution and history of canine mdr1-1D, a pharmacogenetic mu-
tation that marks the emergence of breeds from the collie lineage.
Proceedings of the National Academy of Sciences of the USA 101:
11725–11730.
Nussbaumer, M. 1982. U
¨ber die Variabilita¨t der dorso-basalen Scha¨d-
elknickungen bei Haushunden. Zoologischer Anzeiger 209:1–32.
O’Regan, H. J., and A. C. Kitchener. 2005. The effects of captivity
on the morphology of captive, domesticated feral mammals. Mam-
mal Review 35:215–230.
Ostrander, E. A., U. Giger, and K. Lindblad-Toh, eds. 2006. The dog
and its genome. Cold Spring Harbor Laboratory Press, Cold Spring
Harbor, NY.
Parker, H. G., L. V. Kim, N. B. Sutter, S. Carlson, T. D. Lorentzen,
T. B. Malek, G. S. Johnson, H. B. De France, E. A. Ostrander, and
L. Kruglyak. 2004. Genetic structure of the purebred domestic dog.
Science 304:1160–1164.
Parker, H. G., B. M. VonHoldt, P. Quignon, E. H. Margulies, S. Shao,
D. S. Mosher, T. C. Spady, et al. 2009. An expressed Fgf4 retrogene
is associated with breed-defining chondrodyspasia in domestic
dogs. Science 325:995–998.
Pollinger, J. P., C. D. Bustamante, A. Fledel-Alon, S. Schmutz, M.
M. Gray, and R. K. Wayne. 2005. Selective sweep mapping of genes
with large phenotypic effects. Genome Research 15:1809–1819.
Porto, A., F. B. De Oliveira,L. T. Shirai, V. De Conto, and G.Marroig.
2009. The evolution of modularity in the mammalian skull. I.
Morphological integration patterns and magnitudes. Evolutionary
Biology 36:118–135.
Ra¨ber, H. 2008. Die Schweizer Hunderassen. Schweizerische Kyn-
ologische Gesellschaft, Bern.
Reznick, D. N., F. H. Shaw, F. H. Rodd, and R. G. Shaw. 1997.
Evaluation of the rate of evolution in natural populations of gup-
pies (Poecilia reticulata). Science 275:1934–1937.
Ricklefs, R. E. 2004. Cladogenesis and morphological diversification
in passerine birds. Nature 430:338–341.
Sampson, J., and M. M. Binns. 2006. The Kennel Club and the early
history of dog shows and breed clubs. Pages 19–30 in E. A. Os-
trander, U. Giger, and K. Lindblad-Toh, eds. The dog and its
genome. Cold Spring Harbor Laboratory Press, Cold Spring
Harbor, NY.
Savolainen, P., T. Leitner, A. N. Wilton, E. Matisoo-Smith, and J.
Lundeberg. 2004. A detailed picture of the origin of the Australian
dingo, obtained from the study of mitochondrial DNA. Proceed-
ings of the National Academy of Sciences of the USA 101:12387–
12390.
Schluter, D. 1996. Adaptive radiation along genetic lines of least
resistance. Evolution 50:1766–1774.
———. 2000. The ecology of adaptive radiation. Oxford University
Press, Oxford.
Sears, K. E., A. Goswami, J. J. Flynn, and L. A. Niswander. 2007.
The correlated evolution of Runx2 tandem repeats, transcriptional
activity, and facial length in Carnivora. Evolution and Develop-
ment 9:550–565.
Sol, D. 2008. Artificial selection, naturalization, and fitness:Darwin’s
pigeons revisited. Biological Journal of the Linnean Society 93:
657–665.
Stanley, S. M. 1998. Macroevolution: patternand process. JohnsHop-
kins University Press, Baltimore.
Steppan, S. J. 1997. Phylogenetic analysis of phenotypic covariance
structure. I. Contrasting results from matrix correlation and com-
mon principal component analyses. Evolution 51:571–586.
Stockard, C. R. 1941. The genetic and endocrinic basis fordifferences
in form and behavior as elucidated by studies of contrasted pure-
line dog breeds and their hybrids. Wistar Institute, Philadelphia.
Sutter, N. B., C. D. Bustamante, K. Chase, M. M. Gray, K. Zhao, L.
Zhu, B. Padhukasahasram, et al. 2007. A single IGF1 allele is a
major determinant of small size in dogs. Science 316:112–115.
Trut, L. N., I. Z. Plyusnina, and I. N. Oskina. 2004. An experiment
on fox domestication and debatable issues of evolution of the dog.
Russian Journal of Genetics 40:644–655.
Wagner, G. P., and L. Altenberg. 1996. Complex adaptations and the
evolution of evolvability. Evolution 50:967–976.
Wayne, R. K. 1986a. Cranial morphology of domestic and wild ca-
nids: the influence of development on morphological change. Evo-
lution 40:243–261.
———. 1986b. Limb morphology of domestic and wild canids: the
influence of development on morphologic change. Journal of Mor-
phology 187:301–319.
Wayne, R. K., and E. A. Ostrander. 2007. Lessons learned from the
dog genome. Trends in Genetics 23:557–567.
Wiley, D. F., N. Amenta, D. A. Alcantara, D. Ghosh, Y. J. Kil, E.
Delson, W. Harcourt-Smith, F. J. Rohlf, K. St. John, and B. Ha-
mann. 2005. Evolutionary morphing. Pages 431–438 in Proceed-
ings of the IEEE Visualization 2005 (VIS ’05), Minneapolis.
Willmore, K. E., C. P. Klingenberg, and B. Hallgrı´msson. 2005. The
relationship between fluctuating asymmetry and environmental
variance in rhesus macaque skulls. Evolution 59:898–909.
Wroe, S., and N. Milne. 2007. Convergence and remarkably consis-
tent constraint in the evolution of carnivore skull shape. Evolution
61:1251–1260.
Young, A., and D. Bannasch. 2006. Morphological variation in the
dog. Pages 47–65 in E. A. Ostrander, U. Giger, and K. Lindblad-
Toh, eds. The dog and its genome. Cold Spring Harbor Laboratory
Press, Cold Spring Harbor, NY.
Zelditch, M. L., H. D. Sheets, and W. L. Fink. 2003. The ontogenetic
dynamics of shape disparity. Paleobiology 29:139–156.
Associate Editor: Stephen B. Heard
Editor: Mark A. McPeek